Steps to an Ecology of Mind




Bateson, Gregory.
Steps to an ecology of mind.




Part I: Metalogues………………………………………….12

Metaloque: Why Do Things Get in a Muddle? 13
Metalogue: Why Do Frenchmen? 19
Metalogue: About Games and Being Serious 24
Metalogue: How Much Do You Know? 31
Metalogue: Why Do Things Have Outlines? 37
Metalogue: Why a Swan? 43
Metaloque: What Is an Instinct? 48

Part II: Form and Pattern in Anthropology……70
Culture Contact and Schismogenesis 71
Experiments in Thinking About Observed
Ethnological Material 83
Morale and National Character 98

Bali: The Value System of a Steady State 116
Style, Grace, and Information in Primitive Art 137
Comment on Part II 162

Part III: Form and Pathology in Relationship
Social Planning and the Concept of

A Theory of Play and Fantasy 183

Epidemiology of a Schizophrenia 199

Toward a Theory of Schizophrenia 205

The Group Dynamics of Schizophrenia 233

Minimal Requirements for a Theory of

Schizophrenia 249

Double Bind, 1969 276

The Logical Categories of Learning and Communication 284

The Cybernetics of “Self”: A Theory of Alcoholism 315

Comment on Part III 345

Part IV: Biology and Evolution……………………347

On Empty-Headedness Among Biologists and
State Boards of Education

The Role of Somatic Change in Evolution
Problems in Cetacean and Other

Mammalian Communication*

A Re-examination of “Bateson’s Rule”

Comment on Part IV 404

Part V: Epistemology and Ecology………………406
Cybernetic Explanation 407
Redundancy and Coding 419
Conscious Purpose versus Nature 434
Effects of Conscious Purpose on Human Adaptation 447
Form, Substance, and Difference 455
Comment on Part V 472

Part VI: Crisis in the Ecology of Mind………..474
From Versailles to Cybernetics 475
Pathologies of Epistemology 484
The Roots of Ecological Crisis 494
Ecology and Flexibility in Urban Civilization 499



Some men seem able to go on working steadily with little
success and no reassurance from outside. I am not one of these. I have
needed to know that somebody else believed that my work had
promise and direction, and I have often been surprised that others
had faith in me when I had very little in myself. I have, at times, even
tried to shrug off the responsibility which their continued faith
imposed on me by thinking, “But they don’t really know what I am
doing. How can they know when I myself do not?”

My first anthropological field work among the Baining of New
Britain was a failure, and I had a period of partial failu60re in research
with dolphins. Neither of these failures has ever been held against

I therefore have to thank many people and institutions for backing
me, at times when I did not consider myself a good bet.

First, I have to thank the Council of Fellows of St. John’s
College, Cambridge, who elected me to a Fellowship immediately
after my failure among the Baining.

Next, in chronological order, I owe a deep debt to Margaret Mead,
who was my wife and very close co-worker in Bali and New Guinea,
and who since then has continued as a friend and professional

In 1942, at a Macy Foundation conference, I met Warren
McCulloch and Julian Bigelow, who were then talking excitedly
about “feedback.” The writing of Naven had brought me to the very
edge of what later became cybernetics, but I lacked the concept of
negative feedback. When I returned from overseas after the war, I
went to Frank Fremont-Smith of the Macy Foundation to ask for a
conference on this then-mysterious matter. Frank said that he had just
arranged such a conference with McCulloch as chair-man. It thus
happened that I was privileged to be a member of the famous Macy
Conferences on Cybernetics. My debt to Warren McCulloch, Norbert
Wiener, John von Neumann, Evelyn Hutchinson, and other members
of these conferences is evident in everything that I have written since
World War II.

In my first attempts to synthesize cybernetic ideas with anthropological
data, I had the benefit of a Guggenheim Fellowship.

In the period of my entry into the psychiatric field, it was Jurgen
Ruesch, with whom I worked in the Langley Porter Clinic, who initiated
me into many of the curious features of the psychiatric world.

From 1949 to 1962, I had the title of “Ethnologist” in the Veterans
Administration Hospital at Palo Alto, where I was given singular
freedom to study whatever I thought interesting. I was protected from
outside demands and given this freedom by the director of the
hospital, Dr. John J. Prusmack.

In this period, Bernard Siegel suggested that the Stanford University
Press republish my book, Naven, which had fallen flat on its
face when first published in 1936; and I was lucky enough to get film
footage of a sequence of play between otters in the Fleishhacker Zoo
which seemed to me of such theoretical interest as to justify a small
research program.

I owe my first research grant in the psychiatric field to the late
Chester Barnard of the Rockefeller Foundation, who had kept a copy
of Naven for some years by his bedside. This was a grant to study
“the role of the Paradoxes of Abstraction in Communication.”

Under this grant, Jay Haley, John Weakland, and Bill Fry joined
me to form a small research team within the V.A. Hospital.

But again there was failure. Our grant was for only two years,
Chester Barnard had retired, and in the opinion of the Foundation
staff we did not have enough results to justify renewal. The grant ran
out, but my team loyally stayed with me without pay. The work
went on, and, a few days after the end of the grant, while I was writing
a desperate letter to Norbert Wiener for his advice about where
to get the next grant, the double bind hypothesis fell into place.

Finally Frank Fremont-Smith and the Macy Foundation saved us.

After that there were grants from the Foundations Fund for Psychiatry
and from the National Institute of Mental Health.

Gradually it appeared that for the next advances in the study of
logical typing in communication I should work with animal material,
and I started to work with octopus. My wife, Lois, worked with me,
and for over a year we kept a dozen octopuses in our living room.
This preliminary work was promising but needed to be repeated and
extended under better conditions. For this no grants were available.


At this point, John Lilly came forward and invited me to be the
director of his dolphin laboratory in the Virgin Islands. I worked there
for about a year and became interested in the problems of cetacean
communication, but I think I am not cut out to administer a
laboratory dubiously funded in a place where the logistics are intolerably

It was while I was struggling with these problems that I received a
Career Development Award under the National Institute of Mental
Health. These awards were administered by Bert Boothe, and I owe
much to his continued faith and interest.

In 1963, Taylor Pryor of the Oceanic Foundation in Hawaii invited
me to work in his Oceanic Institute on cetacean and other problems of
animal and human communication. It is here that I have written more
than half of the present book, including the whole of Part V.

While in Hawaii, I have also been working recently with the Culture
Learning Institute of the East-West Center in the University of
Hawaii, and owe some theoretical insights regarding Learning III to
discussions held in that Institute.

My debt to the Wenner-Gren Foundation is evident from the fact
that the book contains no less than four position papers written for
Wenner-Gren conferences. I wish also to thank personally Mrs. Lita
Osmundsen, the Director of Research of that Foundation.

Many also have labored along the road to help me. Most of these
cannot be mentioned here, but I must particularly thank Dr. Vern
Carroll, who prepared the bibliography, and my secretary, Judith Van
Slooten, who labored with accuracy through long hours in preparing
this book for press.

Finally there is the debt that every man of science owes to the giants
of the past. It is no mean comfort, at times when the next idea
cannot be found and the whole enterprise seems futile, to remember that
greater men have wrestled with the same problems. My personal
inspiration has owed much to the men who over the last 200 years
have kept alive the idea of unity between mind and body: Lamarck,
the founder of evolutionary theory, miserable, old, and blind, and
damned by Cuvier, who believed in Special Creation; William Blake,
the poet and painter, who saw “through his eyes, not with them,” and
knew more about what it is to be human than any other man;
Samuel Butler, the ablest contemporary critic of Darwinian

evolution and the first analyst of a schizophrenogenic family; R. G.
Collingwood, the first man to recognize—and to analyze in crystalline
prose—the nature of context; and William Bateson, my father,
who was certainly ready in 1894 to receive the cybernetic ideas.

Selection and Arrangement of Items

The book contains almost everything that I have written, with the
exception of items too long to be included, such as books and extensive
analyses of data; and items too trivial or ephemeral, such as
book reviews and controversial notes. A complete personal bibliography
is appended.

Broadly, I have been concerned with four sorts of subject matter:
anthropology, psychiatry, biological evolution and genetics, and the
new epistemology which comes out of systems theory and ecology.
Essays on these subjects make up Parts II, III, IV, and V of the book,
and the order of these parts corresponds to the chronological order
of four overlapping periods in my life in which these subjects have
been central to my thinking. Within each part, the essays are in
chronological order.

I recognize that readers are likely to attend most carefully to those
parts of the book dealing with their particular subjects. I have
therefore not edited out some repetition. The psychiatrist interested
in alcoholism will encounter in “The Cybernetics of `Self’ ” ideas
which appear again in more philosophic dress in “Form, Substance,
and Difference.”

Oceanic Institute, Hawaii Apra 16, 1971



The Science of Mind and Order*

The title of this book of collected essays and lectures is intended
precisely to define the contents. The essays, spread over thirty-five
years, combine to propose a new way of thinking about ideas and
about those aggregates of ideas which I call “minds.” This way of
thinking I call the “ecology of mind,” or the ecology of ideas. It is a
science which does not yet exist as an organized body of theory or

But the definition of an “idea” which the essays combine to propose
is much wider and more formal than is conventional. The essays
must speak for themselves, but here at the beginning let me
state my belief that such matters as the bilateral symmetry of an animal,
the patterned arrangement of leaves in a plant, the escalation of an
armaments race, the processes of courtship, the nature of play, the
grammar of a sentence, the mystery of biological evolution, and the
contemporary crises in man’s relationship to him environment,
can only be understood in terms of such an ecology of
ideas as I propose.

The questions which the book raises are ecological: How do
ideas interact? Is there some sort of natural selection which
determines the survival of some ideas and the extinction or death of
others? What sort of economics limits the multiplicity of ideas in a
given region of mind? What are the necessary conditions for
stability (or survival) of such a system or subsystem?

Some of these questions are touched upon in the essays, but the
main thrust of the book is to clear the way so that such questions can be
meaningfully asked.

It was only in late 1969 that I became fully conscious of what I had
been doing. With the writing of the Korzybski Lecture, “Form,

* This essay, written in 1971, has not been published else-where.

Substance, and Difference,” I found that in my work with primitive
peoples, schizophrenia, biological symmetry, and in my discontent with
the conventional theories of evolution and learning, I had identified a
widely scattered set of bench marks or points of reference from which a
new scientific territory could be defined. These bench marks I have
called “steps” in the title of the book.

In the nature of the case, an explorer can never know what he is
exploring until it has been explored. He carries no Baedeker in his
pocket, no guidebook which will tell him which churches he should
visit or at which hotels he should stay. He has only the ambiguous
folklore of others who have passed that way. No doubt deeper levels of
the mind guide the scientist or the artist toward experiences and
thoughts which are relevant to those problems which are somehow his,
and this guidance seems to operate long before the scientist has any
conscious knowledge of his goals. But how this happens we do not

I have often been impatient with colleagues who seemed unable to
discern the difference between the trivial and the profound. But when
students have asked me to define that difference, I have been struck
dumb. I have said vaguely that any study which throws light upon the
nature of “order” or “pattern” in the universe is surely nontrivial.

But this answer only begs the question.

I used to teach an informal course for psychiatric residents in the
Veterans Administration Hospital at Palo Alto, trying to get them to
think some of the thoughts that are in these essays. They would attend
dutifully and even with intense interest to what I was saying, but every
year the question would arise after three or four sessions of the class:
“What is this course all about?”

I tried various answers to this question. Once I drew up a sort of
catechism and offered it to the class as a sampling of the questions
which I hoped they would be able to discuss after completing the
course. The questions ranged from “What is a sacrament?’ to “What is
entropy?” and “What is play?”

As a didactic maneuver, my cathechism was a failure: it silenced the
class. But one question in it was useful:

A certain mother habitually rewards her small son
with ice cream after he eats his spinach. What

additional information would you need to be able to
predict whether the child will: a. Come to love or hate
spinach, b. Love or hate ice cream, or c. Love or hate

We devoted one or two sessions of the class to exploring the many
ramifications of this question, and it became clear to me that all the
needed additional information concerned the context of the mother’s
and son’s behavior. In fact, the phenomenon of context and the closely
related phenomenon of “meaning” defined a division between the
“hard” sciences and the sort of science which I was trying to build.

Gradually I discovered that what made it difficult to tell the class
what the course was about was the fact that my way of thinking was
different from theirs. A clue to this difference came from one of the
students. It was the first session of the class and I had talked about the
cultural differences between England and America—a matter which
should always be touched on when an Englishman must teach
Americans about cultural anthropology. At the end of the session, one
resident came up. He glanced over his shoulder to be sure that the
others were all leaving, and then said rather hesitantly, “I want to ask a
question.” “Yes.” “It’s—do you want us to learn what you are telling
us?” I hesitated a moment, but he rushed on with, “Or is it all a sort of
example, an illustration of something else?” “Yes, indeed!”

But an example of what?

And then there was, almost every year, a vague complaint which
usually came to me as a rumor. It was alleged that “Bateson knows
something which he does not tell you,” or “There’s something be-hind
what Bateson says, but he never says what it is.”

Evidently I was not answering the question, “An example of what?”

In desperation, I constructed a diagram to describe what I conceive
to be the task of the scientist. By use of this diagram, it became clear that
a difference between my habits of thought and those of my students
sprang from the fact that they were trained to think and argue
inductively from data to hypotheses but never to test hypotheses
against knowledge derived by deduction from the fundamentals of
science or philosophy.

The diagram had three columns. On the left, I listed various sorts
of uninterpreted data, such as a film record of human or animal

behavior, a description of an experiment, a description or photograph
of a beetle’s leg, or a recorded human utterance. I stressed
the fact that “data” are not events or objects but always records or
descriptions or memories of events or objects. Always there is a
transformation or recoding of the raw event which intervenes
between the scientist and his object. The weight of an object is measured
against the weight of some other object or registered on a me-ter.
The human voice is transformed into variable magnetizations of tape.
Moreover, always and inevitably, there is a selection of data because
the total universe, past and present, is not subject to observation from
any given observer’s position.

In a strict sense, therefore, no data are truly “raw,” and every
record has been somehow subjected to editing and transformation
either by man or by his instruments.

But still the data are the most reliable source of information, and
from them the scientist must start. They provide his first inspiration and
to them he must later return.

In the middle column, I listed a number of imperfectly defined
explanatory notions which are commonly used in the behavioral
sciences—”ego,” “anxiety,” “instinct,” “purpose,” “mind,” “self,”
“fixed action pattern,” “intelligence,” “stupidity,” “maturity,” and the
like. For the sake of politeness, I call these “heuristic” concepts; but, in
truth, most of them are so loosely derived and so mutually irrelevant
that they mix together to make a sort of conceptual fog which does
much to delay the progress of science.

In the right-hand column, I listed what I call “fundamentals.”
These are of two kinds: propositions and systems of propositions
which are truistical, and propositions or “laws” which are generally
true. Among the truistical propositions I included the “Eternal Verities”
of mathematics where truth is tautologically limited to the do-mains
within which man-made sets of axioms and definitions obtain: “If
numbers are appropriately defined and if the operation of addition is
appropriately defined; then 5 + 7 = 12.” Among propositions which I
would describe as scientifically or generally and empirically true, I
would list the conservation “laws” for mass and energy, the Second
Law of Thermodynamics, and so on. But the line between tautological
truths and empirical generalizations is not sharply definable, and,
among my “fundamentals,” there are many propositions whose truth no

sensible man can doubt but which can-not easily be classified as either
empirical or tautological. The “laws” of probability cannot be stated so
as to be understood and not be believed, but it is not easy to decide
whether they are empirical or tautological; and this is also true of
Shannon’s theorems in Information Theory.

With the aid of such a diagram, much can be said about the
whole scientific endeavor and about the position and direction of
any particular piece of inquiry within it. “Explanation” is the
mapping of data onto fundamentals, but the ultimate goal of science
is the increase of fundamental knowledge.

Many investigators, especially in the behavioral sciences, seem to
believe that scientific advance is predominantly inductive and should
be inductive. In terms of the diagram, they believe that progress is
made by study of the “raw” data, leading to new heuristic concepts.
The heuristic concepts are then to be regarded as “working
hypotheses” and tested against more data. Gradually, it is hoped, the
heuristic concepts will be corrected and improved until at last they are
worthy of a place in the list of fundamentals. About fifty years of work
in which thousands of clever men have had their share have, in fact,
produced a rich crop of several hundred heuristic concepts, but, alas,
scarcely a single principle worthy of a place in the list of

It is all too clear that the vast majority of the concepts of contemporary
psychology, psychiatry, anthropology, sociology, and economics
are totally detached from the network of scientific fundamentals.

Moliere, long ago, depicted an oral doctoral examination in which
the learned doctors ask the candidate to state the “cause and reason”
why opium puts people to sleep. The candidate triumphantly answers
in dog Latin, “Because there is in it a dormitive principle (virtus

Characteristically, the scientist confronts a complex interactive
system—in this case, an interaction between man and opium. He
observes a change in the system — the man falls asleep. The scientist
then explains the change by giving a name to a fictitious “cause,” located
in one or other component of the interacting system. Either the
opium contains a reified dormitive principle, or the man contains a

reified need for sleep, an adormitosis, which is “expressed” in his
response to opium.

And, characteristically, all such hypotheses are “dormitive” in the
sense that they put to sleep the “critical faculty” (another reified
fictitious cause) within the scientist himself.

The state of mind or habit of thought which goes from data to
dormitive hypothesis and back to data is self-reinforcing. There is,
among all scientists, a high value set upon prediction, and, indeed, to
be able to predict phenomena is a fine thing. But prediction is a
rather poor test of an hypothesis, and this is especially true of “dormitive
hypotheses.” If we assert that opium contains a dormitive
principle, we can then devote a lifetime of research to studying the
characteristics of this principle. Is it heat-stable? In which fraction of
a distillate is it located? What is its molecular formula? And so on.
Many of these questions will be answerable in the laboratory and will
lead on to derivative hypotheses no less “dormitive” than that from
which we started.

In fact, the multiplication of dormitive hypotheses is a symptom of
excessive preference for induction, and this preference must al-ways
lead to something like the present state of the behavioral sciences— a
mass of quasi-theoretical speculation unconnected with any core of
fundamental knowledge.

In contrast, I try to teach students— and this collection of essays is
very much concerned with trying to communicate this thesis—that in
scientific research you start from two beginnings, each of which has its
own kind of authority: the observations cannot be denied, and the
fundamentals must be fitted. You must achieve a sort of pincers

If you are surveying a piece of land, or mapping the stars, you have
two bodies of knowledge, neither of which can be ignored. There are
your own empirical measurements on the one hand and there is
Euclidean geometry on the other. If these two cannot be made to fit
together, then either the data are wrong or you have argued wrongly
from them or you have made a major discovery leading to a revision of
the whole of geometry.

The would-be behavioral scientist who knows nothing of the basic
structure of science and nothing of the 3000 years of careful philosophic
and humanistic thought about man — who cannot define either entropy

or a sacrament —had better hold his peace rather than add to the existing
jungle of half-baked hypotheses.

But the gulf between the heuristic and the fundamental is not solely
due to empiricism and the inductive habit, nor even to the seductions of
quick application and the faulty educational system which makes
professional scientists out of men who care little for the fundamental
structure of science. It is due also to the circumstance that a very large
part of the fundamental structure of nineteenth-century science was
inappropriate or irrelevant to the problems and phenomena which
confronted the biologist and behavioral scientist.

For at least 200 years, say from the time of Newton to the late
nineteenth century, the dominant preoccupation of science was with
those chains of cause and effect which could be referred to forces and
impacts. The mathematics available to Newton was preponderantly
quantitative, and this fact, combined with the central focus upon forces
and impacts, led men to measure with remarkable accuracy quantities of
distance, time, matter, and energy.

As the measurements of the surveyor must jibe with Euclidean
geometry, so scientific thought had to jibe with the great conservative
laws. The description of any event examined by a physicist or chemist
was to be founded upon budgets of mass and energy, and this rule gave
a particular kind of rigor to the whole of thought in the hard sciences.

The early pioneers of behavioral science not unnaturally began their
survey of behavior by desiring a similar rigorous base to guide their
speculations. Length and mass were concepts which they could hardly
use in describing behavior (whatever that might be), but energy seemed
more handy. It was tempting to relate “energy” to already existing
metaphors such as “strength” of emotions or character or “vigor.” Or to
think of “energy” as somehow the opposite of “fatigue” or “apathy.”
Metabolism obeys an energy budget (within the strict meaning of
“energy”), and energy expended in behavior must surely be included in
this budget; therefore it seemed sensible to think of energy as a
determinant of behavior.

It would have been more fruitful to think of lack of energy as preventive
of behavior, since in the end a starving man will cease to behave.
But even this will not do: an amoeba, deprived of food, be-comes
for a time more active. Its energy expenditure is an inverse function of
energy input.

The nineteeth-century scientists (notably Freud) who tried to establish
a bridge between behavioral data and the fundamentals of
physical and chemical science were, surely, correct in insisting upon the
need for such a bridge but, I believe, wrong in choosing “energy” as the
foundation for that bridge.

If mass and length are inappropriate for the describing of behavior,
then energy is unlikely to be more appropriate. After all, energy is Mass
x Velocity2, and no behavioral scientist really insists that “psychic
energy” is of these dimensions.

It is necessary, therefore, to look again among the fundamentals for
an appropriate set of ideas against which we can test our heuristic

But some will argue that the time is not yet ripe; that surely the
fundamentals of science were all arrived at by inductive reasoning from
experience, so we should continue with induction until we get a
fundamental answer.

I believe that it is simply not true that the fundamentals of science
began in induction from experience, and I suggest that in the
search for a bridgehead among the fundamentals we should go back to
the very beginnings of scientific and philosophic thought; certainly to
a period before science, philosophy, and religion had be-come separate
activities separately pursued by professionals in separate disciplines.

Consider, for example, the central origin myth of Judaeo-Christian
peoples. What are the fundamental philosophic and scientific problems
with which this myth is concerned?

In the beginning God created the heaven and the earth.
And the earth was without form, and void; and darkness was
upon the face of the deep. And the Spirit of God moved upon
the face of the waters.

And God said, Let there be light: and there was light.And God
saw the light, that it was good: and God divided the light from
the darkness.And God called the light Day, and the darkness he
called Night. And the evening and the morning were the first

And God said, Let there be a firmament in the midst of the
waters, and let it divide the waters from the waters. And God
made the firmament, and divided the waters which were under
the firmament from the waters which were above the
firmament: and it was so. And God called the firmament
Heaven. And the evening and the morning were the second

And God said, Let the waters under the heaven be
gathered together unto one place, and let the dry land
appear: and it was so. And God called the dry land Earth;
and the gathering together of the waters called he Seas: and
God saw that it was good.


Authorized version

Out of these first ten verses of thunderous prose, we can draw
some of the premises or fundamentals of ancient Chaldean thought
and it is strange, almost eerie, to note how many of the fundamentals
and problems of modern science are foreshadowed in the ancient

(1) The problem of the origin and nature of matter is summarily
(2) The passage deals at length with the problem of the origin of
(3) A separation is thus generated between the two sorts of problem.
It is possible that this separation of problems was an error, but
—error or not—the separation is maintained in the fundamentals of
modern science. The conservative laws for matter and energy are
still separate from the laws of order, negative entropy, and
(4) Orde is seen as a matter of sorting and dividing. But the essential
notion in all sorting is that some difference shall cause some other
difference at a later time. If we are sorting black balls from white balls,
or large balls from small balls, a difference among the balls is to be
followed by a difference in their location—balls of one class to one sack
and balls of another class to another. For such an operation, we need

something like a sieve, a threshold, or, par excellence, a sense organ.
It is understandable, therefore, that a perceiving Entity should have
been invoked to perform this function of creating an otherwise
improbable order.

(5) Closely linked with the sorting and dividing is the mystery of
classification, to be followed later by the extraordinary human
achievement of naming.
It is not at all clear that the various components of this myth are all
products of inductive reasoning from experience. And the mat-ter
becomes still more puzzling when this origin myth is compared with
others which embody different fundamental premises.

Among the Iatmul of New Guinea, the central origin myth, like
the Genesis story, deals with the question of how dry land was separated
from water. They say that in the beginning the crocodile
Kavwokmali paddled with his front legs and with his hind legs; and his
paddling kept the mud suspended in the water. The great culture
hero, Kevembuangga, came with his spear and killed Kavwokmali.
After that the mud settled and dry land was formed. Kevembuangga
then stamped with his foot on the dry land, i.e., he proudly
demonstrated “that it was good.”

Here there is a stronger case for deriving the myth from experience
combined with inductive reasoning. After all, mud does remain
in suspension if randomly stirred and does settle when the stir-ring
ceases. Moreover, the Iatmul people live in the vast swamps of the
Sepik River valley where the separation of land from water is
imperfect. It is understandable that they might be interested in the
differentiation of land from water.

In any case, the Iatmul have arrived at a theory of order which is
almost a precise converse of that of the book of Genesis. In Iatmul
thought, sorting will occur if randomization is prevented. In Genesis,
an agent is invoked to do the sorting and dividing.

But both cultures alike assume a fundamental division between the
problems of material creation and the problems of order and

Returning now to the question of whether the fundamentals of
science and/or philosophy were, at the primitive level, arrived at by
inductive reasoning from empirical data, we find that the answer is
not simple. It is difficult to see how the dichotomy between


substance and form could be arrived at by inductive argument. No
man, after all, has ever seen or experienced formless and unsorted
matter; just as no man has ever seen or experienced a “random” event.
If, therefore, the notion of a universe “without form and void” was
arrived at by induction, it was by a monstrous—and perhaps
erroneous — jump of extrapolation.

And even so, it is not clear that the starting point from which the
primitive philosophers took off was observation. It is at least equally
likely that dichotomy between form and substance was an unconscious
deduction from the subject-predicate relation in the structure of
primitive language. This, however, is a matter beyond the reach of
useful speculation.

Be that as it may, the central—but usually not explicit — subject
matter of the lectures which I used to give to psychiatric residents
and of these essays is the bridge between behavioral data and the
“fundamentals” of science and philosophy; and my critical comments
above about the metaphoric use of “energy” in the behavioral sciences
add up to a rather simple accusation of many of my colleagues, that
they have tried to build the bridge to the wrong half of the ancient
dichotomy between form and substance. The conservative laws for
energy and matter concern substance rather than form. But mental
process, ideas, communication, organization, differentiation, pattern,
and so on, are matters of form rather than substance.

Within the body of fundamentals, that half which deals with form
has been dramatically enriched in the last thirty years by the
discoveries of cybernetics and systems theory. This book is concerned
with building a bridge between the facts of life and behavior and
what we know today of the nature of pattern and order.

Part I: Metalogues

DEFINITION: A metalogue is a conversation about some
problematic subject. This conversation should be such that not
only do the participants discuss the problem but the structure of the
conversation as a whole is also relevant to the same subject. Only
some of the conversations here presented achieve this double

Notably, the history of evolutionary theory is inevitably a
metalogue between man and nature, in which the creation and
interaction of ideas must necessarily exemplify evolutionary

Metaloque: Why Do Things Get in a

Daughter: Daddy, why do things get in a muddle?
Father: What do you mean? Things? Muddle?

D: Well, people spend a lot of time tidying things, but they never
seem to spend time muddling them. Things just seem to get in a
muddle by themselves. And then people have to tidy them up
F: But do your things get in a muddle if you don’t touch them?
D: No—not if nobody touches them. But if you touch them—or if
anybody touches them—they get in a muddle and it’s a worse
muddle if it isn’t me.
F: Yes—that’s why I try to keep you from touching the things on my
desk. Because my things get in a worse muddle if they are
touched by somebody who isn’t me.
D: But do people always muddle other people’s things? Why do
they, Daddy?
F: Now, wait a minute. It’s not so simple. First of all, what do you
mean by a muddle?
D: I mean—so I can’t find things, and so it looks all muddled up.
The way it is when nothing is straight
F: Well, but are you sure you mean the same thing by muddle that
anybody else would mean?
D: But, Daddy, I’m sure I do—because I’m not a very tidy person
and if I say things are in a muddle, then I’m sure everybody else
would agree with me.
F: All right—but do you think you mean the same thing by “tidy”
that. other people would? If your mummy makes your things
tidy, do you know where to find them?
D: Hmm . . . sometimes—because, you see, I know where she puts
things when she tidies up
F: Yes, I try to keep her away from tidying my desk, too. I’m sure
that she and I don’t mean the same thing by “tidy.”
* Written in 1948; not previously published.

D: Daddy, do you and I mean the same thing by “tidy?” F: I doubt it,
my dear—I doubt it.
D: But, Daddy, isn’t that a funny thing—that everybody means the
same when they say “muddled” but every-body means something
different by “tidy.” But “tidy” is the opposite of “muddled,” isn’t
F: Now we begin to get into more difficult questions. Let’s start
again from the beginning. You said “Why do things always get in
a muddle?” Now we have made a step or two—and let’s change
the question to “Why do things get in a state which Cathy calls
‘not tidy?’ ” Do you see why I want to make that change?
D: … Yes, I think so—because if I have a special meaning for “tidy”
then some of other people’s “tidies” will look like muddles to me
—even if we do agree about most of what we call muddles
F: That’s right. Now—let’s look at what you call tidy. When your
paint box is put in a tidy place, where is it? D: Here on the end of
this shelf.
F: Okay—now if it were anywhere else?
D: No, that would not be tidy.
F: What about the other end of the shelf, here? Like this?
D: No, that’s not where it belongs, and anyhow it would have to be
straight, not all crooked the way you put it.
F: Oh—in the right place and straight.
D: Yes.
F: Well, that means that there are only very few places which are
“tidy” for your paint box
D: Only one place—
F: No—very few places, because if I move it a little bit, like this, it
is still tidy.
D: All right—but very, very few places.
F: All right, very, very few places. Now what about the teddy bear
and your doll, and the Wizard of Oz and your sweater, and your
shoes? It’s the same for all the things, isn’t it, that each thing has
only a very, very few places which are “tidy” for that thing?
D: Yes, Daddy—but the Wizard of Oz could be any-where on that
shelf. And Daddy—do you know what? I hate, hate it when my
books get all mixed up with your books and Mummy’s books.
F : Yes, I know. (Pause)

D: Daddy, you didn’t finish. Why do my things get the way I say
isn’t tidy?
F: But I have finished—it’s just because there are more ways which
you call “untidy” than there are ways which you call “tidy.”
D: But that isn’t a reason why
F: But, yes, it is. And it is the real and only and very important
D: Oh, Daddy! Stop it.
F: No, I’m not fooling.
That is the reason, and all of science is
hooked up with that reason. Let’s take an-other example. If I put
some sand in the bottom of this cup and put some sugar on the
top of it, and now stir it with a teaspoon, the sand and the sugar
will get mixed up, won’t they?
D: Yes, but, Daddy, is it fair to shift over to talking about “mixed
up” when we started with “muddled up?”
F: Hmm . . . I wonder . .. but I think so—Yes—because let’s say we
can find somebody who thinks it is more tidy to have all the sand
underneath all the sugar. And if you like I’ll say I want it that way
D: Hmm…
F: All right—take another example. Sometimes in the movies you
will see a lot of letters of the alphabet all scattered over the
screen, all higgledy-piggledy and some even upside down. And
then something shakes the table so that the letters start to move,
and then as the shaking goes on, the letters all come together to
spell the title of the film.
D: Yes, I’ve seen that—they spelled DONALD.
F: It doesn’t matter what they spelled. The point is that you saw
something being shaken and stirred up and in-stead of getting
more mixed up than before, the letters came together into an
order, all right way up, and spelled a word—they made up
something which a lot of people would agree is sense.
D: Yes, Daddy, but you know .. .
F: No, I don’t know; what I am trying to say is that in the real world
things never happen that way. It’s only in the movies.
D: But, Daddy .. .
F: I tell you it’s only in the movies that you can shake things and
they seem to take on more order and sense than they had before ..

D: But, Daddy .. .
F: Wait till I’ve finished this time . . . And they make it look like that
in the movies by doing the whole thing backwards. They put the
letters all in order to spell DONALD and then they start the
camera and then they start shaking the table.
D: Oh, Daddy—I knew that and I did so want to tell you that—and
then when they run the film, they run it backwards so that it
looks as though things had happened forwards. But really the
shaking happened back-wards. And they have to photograph it
upside down … Why do they, Daddy?
F: Oh God.
D: Why do they have to fix the camera upside down, Daddy?
F: No, I won’t answer that question now because we’re in the middle
of the question about muddles.
D: Oh—all right, but don’t forget, Daddy, you’ve got to answer that
question about the camera another day. Don’t forget! You won’t
forget, will you, Daddy? Be-cause I may not remember. Please,
F: Okay—but another day. Now, where were we? Yes, about things
never happening backwards. And I was trying to tell you why it
is a reason for things to hap-pen in a certain way if we can show
that that way has more ways of happening than some other way.
D: Daddy—don’t begin talking nonsense.
F: I’m not talking nonsense. Let’s start again. There’s only one way
of spelling DONALD. Agreed?
D: Yes.
F: All right. And there are millions and millions and mil-lions of
ways of scattering six letters on the table. Agreed?
D: Yes. I suppose so. Can some of these be upside down?
F: Yes—just in the sort of higgledy-piggledy muddle they were in in
the film. But there could be millions and millions and millions of
muddles like that, couldn’t there? And only one DONALD?
D: All right—yes. But, Daddy, the same letters might spell OLD
F: Never mind. The movie people don’t want them to spell OLD
DAN. They only want DONALD.
D: Why do they?
F: Damn the movie people.

D: But you mentioned them first, Daddy.
F: Yes—but that was to try to tell you why things happen that way in
which there are most ways of their happening. And now it’s your
D: But, Daddy, you never did finish telling me why things happen
that way—the way that has most ways.
F: All right. But don’t start any more hares running—one is quite
enough. Anyhow, I am tired of DONALD, let’s take another
example. Let’s take tossing pennies.
D: Daddy? Are you still talking about the same question we started
with? “Why do things get in a muddle?”
F: Yes.
D: Then, Daddy, is what you are trying to say true about pennies,
and about DONALD, and about sugar and sand, and about my
paint box, and about pennies?
F: Yes—that’s right.
D: Oh—I was just wondering, that’s all.
F: Now, let’s see if I can get it said this time. Let’s go back to the
sand and the sugar, and let’s suppose that somebody says that
having the sand at the bottom is “tidy” or “orderly.”
D: Daddy, does somebody have to say something like that before
you can go on to talk about how things are going to get mixed up
when you stir them?
F: Yes—that’s just the point. They say what they hope will happen
and then I tell them it won’t happen because there are so many
other things that might happen. And I know that it is more likely
that one of the many things will happen and not one of the few.
D: Daddy, you’re just an old bookmaker, backing all the other
horses against the one horse that I want to bet on.
F: That’s right, my dear. I get them to bet on what they call the
“tidy” way—I know that there are infinitely many muddled
ways—so things will always go toward muddle and mixedness.
D: But why didn’t you say that at the beginning, Daddy? I could
have understood that all right.
F: Yes, I suppose so. Anyhow, it’s now bedtime.
D: Daddy, why do grownups have wars, instead of just fighting the
way children do?

F: No—bedtime. Be off with you. We’ll talk about wars another

Metalogue: Why Do Frenchmen?*

Daughter: Daddy, why do Frenchmen wave their arms about?
Father: What do you mean?

D: I mean when they talk. Why do they wave their arms and all that?
F: Well—why do you
smile? Or why do you stamp your foot
D: But that’s not the same thing, Daddy. I don’t wave my arms about
like a Frenchman does. I don’t believe they can stop doing it,
Daddy. Can they?
F: I don’t know—they might find it hard to stop. . . . Can you stop
D: But Daddy, I don’t smile all the time. It’s hard to stop when I feel
like smiling. But I don’t feel like it all the time. And then I stop.
F: That’s true—but then a Frenchman doesn’t wave his arms in the
same way all the time. Sometimes he waves them in one way and
sometimes in another—and sometimes, I think, he stops waving

F: What do you think? I mean, what does it make you think when a
Frenchman waves his arms?
D: I think it looks silly, Daddy. But I don’t suppose it looks like that
to another Frenchman. They cannot all look silly to each other.
Because if they did, they would stop it. Wouldn’t they?
F: Perhaps—but that is not a very simple question. What else do
they make you think?
D: Well—they look all excited .. .
F: All right—”silly” and “excited.”
D: But are they really as excited as they look? If I were as excited as
that, I would want to dance or sing or hit somebody on the nose
* This metalogue is reprinted from Impulse 1951, an annual of contemporary
dance, by permission of Impulse Publications, Inc. It has also appeared in ETC.: A
Re-view of General Semantics, Vol. X, 1953.

… but they just go on waving their arms. They can’t be really

F: Well—are they really as silly as they look to you? And anyhow,
why do you sometimes want to dance and sing and punch
somebody on the nose?
D: Oh. Sometimes I just feel like that.
F: Perhaps a Frenchman just feels
“like that” when he waves his
arms about.
D: But he couldn’t feel like that all the time, Daddy, he just couldn’t.
F: You mean—the Frenchman surely does not feel when he waves
his arms exactly as you would feel if you waved yours. And
surely you are right.
D: But, then, how does he feel?
F: Well—let us suppose you are talking to a Frenchman and he is
waving his arms about, and then in the middle of the
conversation, after something that you have said, he suddenly
stops waving his arms, and just talks. What would you think
then? That he had just stopped being silly and excited?
D: No . . . I’d be frightened. I’d think I had said something that hurt
his feelings and perhaps he might be really angry.
F: Yes—and you might be right.
* * *

D: All right—so they stop waving their arms when they start being
Wait a minute. The question, after all, is what does one
Frenchman tell another Frenchman by waving his arms? And we
have part of an answer—he tells him something about how he
feels about the other guy. He tells him he is not seriously angry—
that he is willing and able to be what you call “silly.”
D: But—no—that’s not sensible. He cannot do all that work so that
later he will be able to tell the other guy that he is angry by just
keeping his own arms still. How does he know that he is going to
be angry later on?
F: He doesn’t know. But, just in case .. .
D: No, Daddy, it doesn’t make sense. I don’t smile so as to be able to
tell you I am angry by not smiling later on.

F: Yes—I think that that is part of the reason for smiling. And there
are lots of people who smile in order to tell you that they are not
angry—when they really are.
D: But that’s different, Daddy. That’s a sort of telling lies with one’s
face. Like playing poker.
F: Yes.
F: Now where are we? You don’t think it sensible for Frenchmen to
work so hard to tell each other that they are not angry or hurt.
But after all what is most conversation about? I mean, among
D: But, Daddy, it’s about all sorts of things—baseball and ice cream
and gardens and games. And people talk about other people and
about themselves and about what they got for Christmas.
F: Yes, yes—but who listens? I mean—all right, so they talk about
baseball and gardens. But are they exchanging information? And,
if so, what information?
D: Sure—when you come in from fishing, and I ask you “did you
catch anything?” and you say “nothing,” I didn’t know that you
wouldn’t catch anything till you told me.
F: Hmm.

F: All right-so you mention my fishing—a matter about which I am
sensitive—and then there is a gap, a silence in the conversation
—and that silence tells you that I don’t like cracks about how
many fish I didn’t catch. It’s just like the Frenchman who stops
waving his arms about when he is hurt.
D: I’m sorry, Daddy, but you did say .. .
F: No—wait a minute—let’s not get confused by being sorry—I
shall go out fishing again tomorrow and I shall still know that I
am unlikely to catch a fish .. .
But, Daddy, you said all conversation is only telling
other people that you are not angry with them .. .
F: Did I? No—not all conversation, but much of it. Some-times if
both people are willing to listen carefully, it is possible to do
more than exchange greetings and good wishes. Even to do more

than exchange information. The two people may even find out
something which neither of them knew before.

* * *

F: Anyhow, most conversations are only about whether people are
angry or something. They are busy telling each other that they
are friendly—which is sometimes a lie. After all, what happens
when they cannot think of anything to say? They all feel
D: But wouldn’t that be information, Daddy? I mean—information
that they are not cross?
F: Surely, yes. But it’s a different sort of information from “the cat is
on the mat.”
* * *

D: Daddy, why cannot people just say “I am not cross at you” and
let it go at that?
F: Ah, now we are getting to the real problem. The point is that the
messages which we exchange in gestures are really not the same
as any translation of those gestures into words.
D: I don’t understand.
F: I mean—that no amount of telling somebody in mere words that
one is or is not angry is the same as what one might tell them by
gesture or tone of voice.
D: But, Daddy, you cannot have words without some tone of voice,
can you? Even if somebody uses as little tone as he can, the other
people will hear that he is holding himself back—and that will be
a sort of tone, won’t it?
F: Yes—I suppose so. After all that’s what I said just now about
gestures—that the Frenchman can say something special by
stopping his gestures.
* * *

F: But then, what do I mean by saying that “mere words” can never
convey the same message as gestures—if there are no “mere
D: Well, the words might be written.

F: No—that won’t let
me out of the difficulty. Because written
words still have some sort of rhythm and they still have
overtones. The point is that no mere words exist. There are only
words with either gesture or tone of voice or something of the
sort. But, of course, gestures without words are common enough.

D: Daddy, when they teach us French at school, why don’t they
teach us to wave our hands?
F: I don’t know. I’m sure I don’t know. That is probably one of the
reasons why people find learning languages so difficult.

F: Anyhow, it is all nonsense. I mean, the notion that language is
made of words is all nonsense—and when I said that gestures
could not be translated into “mere words,” I was talking
nonsense, because there is no such thing as “mere words.” And
all the syntax and grammar and all that stuff is nonsense. It’s all
based on the idea that “mere” words exist—and there are none.
D: But, Daddy .. .
F: I tell you—we have to start all over again from the beginning and
assume that language is first and fore-most a system of gestures.
Animals after all have only gestures and tones of voice—and
words were invented later. Much later. And after that they
invented school-masters.
D: Daddy?
F: Yes.
D: Would it be a good thing if people gave up words and went back
to only using gestures?
F: Hmm. I don’t know. Of course we would not be able to have any
conversations like this. We could only bark, or mew, and wave
our arms about, and laugh and grunt and weep. But it might be
fun—it would make life a sort of ballet—with dancers making
their own music.

Metalogue: About Games and Being

Daughter: Daddy, are these conversations serious?
Father: Certainly they are.

D: They’re not a sort of game that you play with me?
F: God forbid . . . but they are a sort of game that we play together.
D: Then they’re not serious!
* * *

F: Suppose you tell me what you would understand by the words
“serious” and a “game.”
D: Well . . . if you’re . . . I don’t know.
F: If I am what?
D: I mean . . . the conversations are serious for me, but if you are
only playing a game .. .
F: Steady now. Let’s look at what is good and what is bad about
“playing” and “games.” First of all, I don’t mind —not much—
about winning or losing. When your questions put me in a tight
spot, sure, I try a little harder to think straight and to say clearly
what I mean. But I don’t bluff and I don’t set traps. There is no
temptation to cheat.
D: That’s just it. It’s not serious to you. It’s a game. People who
cheat just don’t know how to play. They treat a game as though it
were serious.
F: But it is serious.
D: No, it isn’t—not for you it isn’t.
F: Because I don’t even want to cheat?
D: Yes—partly that.
* This metalogue is reprinted by permission from ETC.: A Review of General
Semantics, Vol. X, 1953.

F: But do you want to cheat and bluff all the time? D: No—of
course not.
F: Well then?
D: Oh—Daddy—you’ll never understand.
F: I guess I never will.
F: Look, I scored a sort of debating point just now by forcing you to
admit that you don’t want to cheat—and then I tied onto that
admission the conclusion that therefore the conversations are not
“serious” for you either. Was that a sort of cheating?
D: Yes—sort of.
F: I agree—I think it was. I’m sorry.
D: You see, Daddy—if I cheated or wanted to cheat, that would
mean that I was not serious about the things we talk about. It
would mean that I was only playing a game with you.
F: Yes, that makes sense.

D: But it doesn’t make sense, Daddy. It’s an awful muddle.
F: Yes—a muddle—but still a sort of sense.
D: How, Daddy?

F: Wait a minute. This is difficult to say. First of all—I think that we
get somewhere with these conversations. I enjoy them very much
and I think you do. But also, apart from that, I think that we get
some ideas straight and I think that the muddles help. I mean—
that if we both spoke logically all the time, we would never get
anywhere. We would only parrot all the old cliches that
everybody has repeated for hundreds of years.
D: What is a cliche, Daddy?
F: A cliche? It’s a French word, and I think it was originally a
printer’s word. When they print a sentence they have to take the
separate letters and put them one by one into a sort of grooved
stick to spell out the sentence. But for words and sentences
which people use often, the printer keeps little sticks of letters
ready made up. And these ready-made sentences are called

D: But I’ve forgotten now what you were saying about cliches,
F: Yes—it was about the muddles that we get into in these talks
and how getting into muddles makes a sort of sense. If we
didn’t get into muddles, our talks would be like playing rummy
without first shuffling the cards.
D: Yes,
Daddy—but what about those things—the ready-made
sticks of letters?
F: The cliches? Yes—it’s the same thing. We all have lots of ready-
made phrases and ideas, and the printer has ready-made sticks
of letters, all sorted out into phrases. But if the printer wants to
print something new—say, something in a new language, he
will have to break up all that old sorting of the letters. In the
same way, in order to think new thoughts or to say new things,
we have to break up all our ready-made ideas and shuffle the
D: But, Daddy, the printer would not shuffle all the letters? Would
he? He wouldn’t shake them all up in a bag. He would put them
one by one in their places—all the a’s in one box and all the b’s
in another, and all the commas in another, and so on.
F: Yes—that’s right. Otherwise he would go mad trying to find an a
when he wanted it.
* * *

F: What are you thinking?
D: No—it’s only that there are so many questions. F: For example?
D: Well, I see what you mean about our getting into muddles. That
that makes us say new sorts of things. But I am thinking about
the printer. He has to keep all his little letters sorted out even
though he breaks up all the ready-made phrases. And I am
wondering’ about our muddles. Do we have to keep the little
pieces of our thought in some sort of order—to keep from going
F: I think
so—yes—but I don’t know what sort of order. That
would be a terribly hard question to answer. I don’t think we
could get an answer to that question today.


F: You said there were “so many questions.” Do you have another?
D: Yes—about games and being
serious. That’s what we started
from, and I don’t know how or why that led us to talk about our
muddles. The way you confuse everything—it’s a sort of
F: No, absolutely not.

F: You brought up two questions. And really there are a lot more . . .
We started from the question about these conversations—are they
serious? Or are they a sort of game? And you felt hurt that I
might be playing a game, while you were serious. It looks as
though a conversation is a game if a person takes part in it with
one set of emotions or ideas—but not a “game” if his ideas or
emotions are different.
D: Yes, it’s if your ideas about the conversation are different from
mine .. .
F: If we both had the game idea, it would be all right? D: Yes—of
F: Then it seems to be up to me to make clear what I mean by the
game idea. I know that I am serious—whatever that means—
about the things that we talk about. We talk about ideas. And I
know that I play with the ideas in order to understand them and
fit them together. It’s “play” in the same sense that a small child
“plays” with blocks . . . And a child with building blocks is
mostly very serious about his “play.”
D: But is it a game, Daddy? Do you play against me?
F: No. I think
of it as you and I playing together against the
building blocks—the ideas. Sometimes competing a bit—but
competing as to who can get the next idea into place. And
sometimes we attack each other’s bit of building, or I will try to
defend my built-up ideas from your criticism. But always in the
end we are working together to build the ideas up so that they
will stand.

* * *

D: Daddy,
do our talks have rules? The difference between
a game and just playing is that a game has rules.
F: Yes. Let me think about that. I think we do have a sort of rules …
and I think a child playing with blocks has rules. The blocks
themselves make a sort of rules. They will balance in certain
positions and they will not balance in other positions. And it
would be a sort of cheating if the child used glue to make the
blocks stand up in a position from which they would otherwise
D: But what rules do we have?
F: Well, the ideas that we play with bring in a sort of rules. There
are rules about how ideas will stand up and sup-port each
other. And if they are wrongly put together the whole building
falls down.
D: No glue, Daddy?
F: No—no glue. Only logic.
* * *

D: But you said that if we always talked logically and did not get
into muddles, we could never say anything new. We could only
say ready-made things. What did you call those things?
F: Cliches. Yes. Glue is what cliches are stuck together with.
D: But you said “logic,” Daddy.
F: Yes, I know. We’re in a muddle again. Only I don’t see a way out
of this particular muddle.
* * *

D: How did we get into it, Daddy?
F: All right, let’s see if we can retrace our steps. We were talking
about the “rules” of these conversations. And I said that the ideas
that we play with have rules of logic .. .
D: Daddy! Wouldn’t it be a good thing if we had a few more rules
and obeyed them more carefully? Then we might not get into
these dreadful muddles.

F: Yes. But wait. You mean that I get us into these muddles because
I cheat against rules which we don’t have. Or put it this way. That
we might have rules which would stop us from getting into
muddles—as long as we obeyed them.
D: Yes, Daddy, that’s what the rules of a game are for.
F: Yes, but do you want to turn these conversations into that sort of
a game? I’d rather play canasta—which is fun too.
D: Yes, that’s right. We can play canasta whenever we want to. But
at the moment I would rather play this game. Only I don’t know
what sort of a game this is. Nor what sort of rules it has.
F: And yet we have been playing for some time.
D: Yes. And it’s been fun.
F: Yes.

F: Let’s go back to the question which you asked and which I said
was too difficult to answer today. We were talking about the
printer breaking up his cliches, and you said that he would still
keep some sort of order among his letters—to keep from going
mad. And then you asked “What sort of order should we cling to
so that when we get into a muddle we do not go mad?” It seems
to me that the “rules” of the game is only an-other name for that
sort of order.
D: Yes—and cheating is what gets us into muddles.
F: In a sense, yes. That’s right. Except that the whole point of the
game is that we do get into muddles, and do come out on the
other side, and if there were no muddles our “game” would be
like canasta or chess—and that is not how we want it to be.
D: Is it you that make the rules, Daddy? Is that fair?
F: That, daughter, is a dirty crack. And probably an unfair one. But
let me accept it at face value. Yes, it is I who make the rules—
after all, I do not want us to go mad.
D: All right. But, Daddy, do you also change the rules? Sometimes?
F: Hmm,
another dirty crack. Yes, daughter, I change them
constantly. Not all of them, but some of them.
D: I wish you’d tell me when you’re going to change them!

F: Hmm—yes—again. I wish I could. But it isn’t like that. If it were
like chess or canasta, I could tell you the rules, and we could, if
we wanted to, stop playing and discuss the rules. And then we
could start a new game with the new rules. But what rules would
hold us between the two games? While we were discussing the
D: I don’t understand.
F: Yes. The point is that the purpose of these conversations is to
discover the “rules.” It’s like life—a game whose purpose is to
discover the rules, which rules are always changing and always
D: But I don’t call that a game, Daddy.
F: Perhaps not. I would call it a game, or at any rate “play.” But it
certainly is not like chess or canasta. It’s more like what kittens
and puppies do. Perhaps. I don’t know.
* * *

D: Daddy, why do kittens and puppies play?
F: I don’t know—I don’t know.

Metalogue: How Much Do You Know?*

Daughter: Daddy, how much do you know?
Father: Me? Hmm—I have about a pound of knowledge.

D: Don’t be silly. Is it a pound sterling or a pound weight? I mean
really how much do you know?
F: Well, my brain weighs about two pounds and I suppose I use
about a quarter of it—or use it at about a quarter efficiency. So
let’s say half a pound.
D: But do you know more than Johnny’s daddy? Do you know more
than I do?
F: Hmm—I once knew a little boy in England who asked his father,
“Do fathers always know more than sons?” and the father said,
“Yes.” The next question was, “Daddy, who invented the steam
engine?” and the father said, “James Watt.” And then the son
came back with “—but why didn’t James Watt’s father invent it?”

D: I know. I know more than that boy because I know why James
Watt’s father didn’t. It was because some-body else had to think
of something else before anybody could make a steam engine. I
mean something like—I don’t know—but there was somebody
else who had to discover oil before anybody could make an
F: Yes—that makes a difference. I mean, it means that knowledge is
all sort of knitted together, or woven, like cloth, and each piece
of knowledge is only meaningful or useful because of the other
pieces—and . . .
D: Do you think we ought to measure it by the yard?
F: No. I don’t.
D: But that’s how we buy cloth.
* This metalogue is reprinted by permission from ETC.:A Review of General
Semantics, Vol. X, 1953. 91

F: Yes. But I didn’t mean that it is cloth.
Only it’s like it—and
certainly would not be flat like cloth—but in three dimensions—
perhaps four dimensions.
D: What do you mean, Daddy?
F: I really don’t know, my dear. I was just trying to think.
F: I don’t think we are doing very well this morning. Sup-pose we
start out on another tack. What we have to think about is how the
pieces of knowledge are woven together. How they help each
D: How do they?
F: Well—it’s as if sometimes two facts get added together and all
you have is just two facts. But sometimes instead of just adding
they multiply—and you get four facts.
D: You cannot multiply one by one and get four. You know you
F: Oh.
* * *

F: But yes I can, too. If the things to be multiplied are pieces of
knowledge or facts or something like that. Because every one of
them is a double something.
D: I don’t understand.
F: Well—at least a double something.
D: Daddy!
F: Yes—take
the game of Twenty Questions. You think of
something. Say you think of “tomorrow.” All right. Now I ask “Is
it abstract?” and you say “Yes.” Now from your “yes” I have got a
double bit of information. I know that it is abstract and I know
that it isn’t concrete. Or say it this way—from your “yes” I can
halve the number of possibilities of what the thing can be. And
that’s a multiplying by one over two.
D: Isn’t it a division?
F: Yes—it’s the same thing. I mean—all right—it’s a multiplication
by .5. The important thing is that it’s not just a subtraction or an
D: How do you know it isn’t?

F: How do I know it?—Well, suppose I ask another question which
will halve the possibilities among the abstractions. And then
another. That will have brought down the total possibilities to an
eighth of what they were at the beginning. And two times two
times two is eight.
D: And two and two and two is only six.
F: That’s right.
D: But, Daddy, I don’t see—what happens with Twenty Questions?
F: The point is that if I pick my questions properly I can decide
between two times two times two times two twenty times over
things—220 things. That’s over a mil-lion things that you might
have thought of. One question is enough to decide between two
things; and two questions will decide between four things—and
so on.
D: I don’t like arithmetic, Daddy.
F: Yes, I know. The working it out is dull, but some of the ideas in it
are amusing. Anyhow, you wanted to know how to measure
knowledge, and if you start measuring things that always leads to
D: We haven’t measured any knowledge yet.
F: No. I know. But we have made a step or two toward knowing
how we would measure it if we wanted to. And that means we
are a little nearer to knowing what knowledge is.
D: That would be
a funny sort of knowledge, Daddy. I mean
knowing about knowledge—would we measure that sort of
knowing the same way?
F: Wait a minute—I don’t know—that’s really the $64 Question on
this subject. Because—well, let’s go back to the game of Twenty
Questions. The point that we never mentioned is that those
questions have to be in a certain order. First the wide general
question and then the detailed question. And it’s only from
answers to the wide questions that I know which detailed
questions to ask. But we counted them all alike. I don’t know. But
now you ask me if knowing about knowledge would be measured
the same way as other knowledge. And the answer must surely be
no. You see, if the early questions in the game tell me what
questions to ask later, then they must be partly questions about
knowing. They’re exploring the business of knowing.

D: Daddy—has anybody ever measured how much any-body knew.
F: Oh yes. Often. But I don’t quite know what the answers meant.
They do it with examinations and tests and quizzes, but it’s like
trying to find out how big a piece of paper is by throwing stones
at it.
D: How do you mean?
F: I mean—if you throw stones at two pieces of paper from the
same distance and you find that you hit one piece more often
than the other, then probably the one that you hit most will be
bigger than the other. In the same way, in an examination you
throw a lot of questions at the students, and if you find that you
hit more pieces of knowledge in one student than in the others,
then you think that student must know more. That’s the idea.
D: But could one measure a piece of paper that way?
F: Surely one could. It might even be quite a good way of doing it.
We do measure a lot of things that way. For example, we judge
how strong a cup of coffee is by looking to see how black it is—
that is, we look to see how much light is stopped. We throw light
waves at it instead of stones, it’s the same idea.
D: Oh.
* * *

D: But then—why shouldn’t we measure knowledge that way?
F: How? By quizzes? No—God forbid. The trouble is that that sort
of measuring leaves out your point—that there are different sorts
of knowledge—and that there’s knowing about knowledge. And
ought one to give higher marks to the student who can answer the
widest question? Or perhaps there should be a different sort of
marks for each different sort of question.
D: Well, all right. Let’s do that and then add the marks together and
then .. .
F: No—we couldn’t add them together. We might multiply or divide
one sort of marks by another sort but we couldn’t add them.
D: Why not, Daddy?
F: Because—because
we couldn’t. No wonder you don’t like
arithmetic if they don’t tell you that sort of thing at school—What

do they tell you? Golly—I wonder what the teachers think
arithmetic is about.

D: What is it about, Daddy?
F: No. Let’s stick to the question of how to measure knowledge—
Arithmetic is a set of tricks for thinking clearly and the only fun
in it is just its clarity. And the first thing about being clear is not
to mix up ideas which are really different from each other. The
idea of two oranges is really different from the idea of two miles.
Because if you add them together you only get fog in your head.
D: But, Daddy, I can’t keep ideas separate. Ought I to do that?
F: No— No— Of course not. Combine them. But don’t add them.
That’s all. I mean—if the ideas are numbers and you want to
combine two different sorts, the thing to do is to multiply them
by each other. Or divide them by each other. And then you’ll get
some new sort of idea, a new sort of quantity. If you have miles
in your head, and you have hours in your head, and you divide
the miles by the hours, you get “miles per hour”—that’s a speed.
D: Yes, Daddy. What would I get if I multiplied them?
F: Oh—er—I suppose you’d get mile-hours. Yes. I know what they
are. I mean, what a mile-hour is. It’s what you pay a taxi driver.
His meter measures miles and he has a clock which measures
hours, and the meter and the clock work together and multiply
the hours by the miles and then it multiplies the mile-hours by
something else which makes mile-hours into dollars.
D: I did an experiment once.
F: Yes?
D: I wanted to find out if I could think two thoughts at the same
time. So I thought “It’s summer” and I thought “It’s winter.” And
then I tried to think the two thoughts together.
F: Yes?
D: But I found I wasn’t having two thoughts. I was only having one
thought about having two thoughts.
F: Sure, that’s just it. You can’t mix thoughts, you can only combine
them. And in the end, that means you can’t count them. Because
counting is really only adding things together. And you mostly
can’t do that.
D: Then really do we only have one big thought which has lots
of branches and lots and lots of branches?

F: Yes. I think so. I don’t know. Anyhow I think that is a clearer
way of saying it. I mean it’s clearer
than talking about bits of knowledge and trying to count them.

* * *

D: Daddy, why don’t you use the other three-quarters of your brain?
F: Oh, yes—that—you see the trouble is that I had school-teachers
too. And they filled up about a quarter of my brain with fog. And
then I read newspapers and listened to what other people said,
and that filled up another quarter with fog.
D: And the other quarter, Daddy?
F: Oh—that’s fog that I made for myself when I was trying to think.

Metalogue: Why Do Things Have

Daughter: Daddy, why do things have outlines?
Father: Do they? I don’t know. What sort of things do you mean?

D: I mean when I draw things, why do they have outlines? F: Well,
what about other sorts of things—a flock of sheep? or a
conversation? Do they have outlines?
D: Don’t be silly. I can’t draw a conversation. I mean things.
F: Yes—I was trying to find out just what you meant. Do you mean
“Why do we give things outlines when we draw them?” or do
you mean that the things have out-lines whether we draw them or
D: I don’t know, Daddy. You tell me. Which do I mean?
F: I don’t know, my dear. There was a very angry artist once who
scribbled all sorts of things down, and after he was dead they
looked in his books and in one place they found he’d written
“Wise men see outlines and therefore they draw them” but in
another place he’d written “Mad men see outlines and therefore
they draw them.”
D: But which does he mean? I don’t understand.
F: Well, William Blake—that was his name—was a great artist and a
very angry man. And sometimes he rolled up his ideas into little
spitballs so that he could throw them at people.
D: But what was he mad about, Daddy?
F: But what was he mad about? Oh, I see—you mean “angry.” We
have to keep those two meanings of “mad” clear if we are going
to talk about Blake. Because a lot of people thought he was mad
—really mad—crazy. And that was one of the things he was
mad-angry about. And then he was mad-angry, too, about some
artists who painted pictures as though things didn’t have outlines.
He called them “the slobbering school.”
D: He wasn’t very tolerant, was he, Daddy?
* Reprinted by permission from ETC.: A Review of General Semantics, Vol. XI,

F: Tolerant? Oh, God. Yes, I know—that’s what they drum into
you at school. No, Blake was not very tolerant. He didn’t even
think tolerance was a good thing. It was just more slobbering.
He thought it blurred all the outlines and muddled everything—
that it made all cats gray. So that nobody would be able to see
anything clearly and sharply.
D: Yes, Daddy.
F: No, that’s not the answer. I mean “Yes, Daddy” is not the answer.
All that says is that you don’t know what your opinion is—and
you don’t give a damn what I say or what Blake says and that
the school has so befuddled you with talk about tolerance that
you can-not tell the difference between anything and anything
D: (Weeps.)
F: Oh, God. I’m sorry, but I was angry. But not really angry with
you. Just angry at the general mushiness of how people act and
think—and how they preach muddle and call it tolerance.
D: But, Daddy
F: Yes?
D: I don’t know. I don’t seem able to think very well. It’s all in a
F: I’m sorry. I suppose I muddled you by starting to let off steam.
* * *

D: Daddy? F: Yes?
D: Why is that something to get angry about?
F: Is what something to get angry about?
D: I mean—about whether things have outlines. You said William
Blake got angry about it. And then you get angry about it. Why
is that, Daddy?
F: Yes, in a way I think it is. I think it matters. Perhaps in a way, is
the thing that matters. And other things only matter because
they are part of this.
D: What do you mean, Daddy?
F: I mean, well, let’s talk about tolerance. When Gentiles want to
bully Jews because they killed Christ, I get intolerant. I think

the Gentiles are being muddle-headed and are blurring all the
outlines. Because the Jews didn’t kill Christ, the Italians did it.

D: Did they, Daddy?
F: Yes, only the ones who did are called Romans today, and we
have another word for their descendants. We call them Italians.
You see there are two muddles and I was making the second
muddle on purpose so we could catch it. First there’s the
muddle of getting the history wrong and saying the Jews did it,
and then there’s the muddle of saying that the descendants
should be responsible for what their ancestors didn’t do. It’s all
D: Yes, Daddy.
F: All right, I’ll try not to get angry again. All I’m trying to say is
that muddle is something to get angry about. D: Daddy?
F: Yes?
D: We were talking about muddle the other day. Are we really
talking about the same thing now?
F: Yes. Of course we are. That’s why it’s important—what we said
the other day.
D: And you said that getting things clear was what Science was
F: Yes, that’s the same thing again.

D: I don’t seem to understand it all very well. Everything seems to
be everything else, and I get lost in it.
F: Yes, I know it’s difficult. The point is that our conversations do
have an outline, somehow—if only one could see it clearly.

F: Let’s think about
a real concrete out-and-out muddle, for a
change, and see if that will help. Do you remember the game of
croquet in Alice in Wonderland?
D: Yes—with flamingos?
F: That’s right.
D: And porcupines for balls?

F: No,
hedgehogs. They were hedgehogs. They don’t have
porcupines in England.
D: Oh. Was it in England, Daddy? I didn’t know.
F: Of
course it was in England. You don’t have duchesses in
America either.
D: But there’s the Duchess of Windsor, Daddy.
F: Yes, but she doesn’t have quills, not like a real porcupine.
D: Go on about Alice and don’t be silly, Daddy.
F: Yes, we were talking about flamingos. The point is that the man
who wrote Alice was thinking about the same things that we are.
And he amused himself with little Alice by imagining a game of
croquet that would be all muddle, just absolute muddle. So he
said they should use flamingos as mallets because the flamingos
would bend their necks so the player wouldn’t know even
whether his mallet would hit the ball or how it would hit the ball.
D: Anyhow the ball might walk away of its own accord because it
was a hedgehog.
F: That’s right. So that it’s all so muddled that nobody can tell at all
what’s going to happen.
D: And the hoops walked around, too, because they were soldiers.
F: That’s right—everything could move and nobody could tell how it
would move.
D: Did
everything have to be alive so as to make a complete
F: No—he could have made it a muddle by . . . no, I suppose you’re
right. That’s interesting. Yes, it had to be that way. Wait a minute.
It’s curious but you’re right. Because if he’d muddled things any
other way, the players could have learned how to deal with the
muddling details. I mean, suppose the croquet lawn was bumpy,
or the balls were a funny shape, or the heads of the mallets just
wobbly instead of being alive, then the people could still learn
and the game would only be more difficult—it wouldn’t be
impossible. But once you bring live things into it, it becomes
impossible. I wouldn’t have expected that.
D: Wouldn’t you, Daddy? I would have. That seems natural to me.
F: Natural? Sure—natural enough. But I would not have expected
it to work that way.
D: Why not? That’s what I would have expected.

F: Yes. But this is the thing that I would not have expected. That
animals, which are themselves able to see things ahead and act
on what they think is going to happen—a cat can catch a mouse
by jumping to land where the mouse will probably be when she
has completed her jump—but it’s just the fact that animals are
capable of seeing ahead and learning that makes them the only
really unpredictable things in the world. To think that we try to
make laws as though people were quite regular and predictable.
D: Or
do they make the laws just because people are not
predictable, and the people who make the laws wish the other
people were predictable?
F: Yes, I suppose so.

D: What were we talking about?
F: I don’t quite know—not yet. But you started a new line by asking
if the game of croquet could be made into a real muddle only by
having all the things in it alive. And I went chasing after that
question, and I don’t think I’ve caught up with it yet. There is
some-thing funny about that point.
D: What?
F: I don’t quite know—not yet. Something about living things and
the difference between them and the things that are not alive—
machines, stones, so on. Horses don’t fit in a world of
automobiles. And that’s part of the same point. They’re
unpredictable, like flamingos in the game of croquet.
D: What about people, Daddy?
F: What about them?
D: Well, they’re alive. Do they fit? I mean on the streets?
F: No, I suppose they don’t really fit—or only by working pretty
hard to protect themselves and make themselves fit. Yes, they
have to make themselves predictable, be-cause otherwise the
machines get angry and kill them.
D: Don’t be silly. If the machines can get angry, then they would ,not
be predictable. They’d be like you, Daddy. You can’t predict
when you’re angry, can you?
F: No, I suppose not.

D: But, Daddy, I’d rather have you unpredictable—sometimes.
* * *

D: What did you mean by a conversation having an out-line? Has
this conversation had an outline?
F: Oh, surely, yes. But we cannot see it yet because the conversation
isn’t finished. You cannot ever see it while you’re in the middle of
it. Because if you could see it, you would be predictable—like
the machine. And I would be predictable—and the two of us
together would be predictable
D: But I don’t understand. You say it is important to be clear about
things. And you get angry about people who blur the outlines.
And yet we think it’s better to be unpredictable and not to be like
a machine. And you say that we cannot see the outlines of our
conversation till it’s over. Then it doesn’t matter whether we’re
clear or not. Because we cannot do anything about it then.
F: Yes, I know—and I don’t understand it myself. . . . But anyway,
who wants to do anything about it?

Metalogue: Why a Swan?*

Daughter: Why a swan?
Father: Yes—and why a puppet in Petroushka?

D: No—that’s different. After all a puppet is sort of human—and
that particular puppet is very human. F: More human than the
D: Yes.
F: But still only sort of human? And after all the swan is also sort
of human.
D: Yes.

D: But what about the dancer? Is she human? Of course she really
is, but, on the stage, she seems inhuman or impersonal—perhaps
superhuman. I don’t know.
F: You mean—that while the swan is only a sort of swan and has
no webbing between her toes, the dancer seems only sort of
D: I don’t know—perhaps it’s something like that.

F: No—I get confused when I speak of the “swan” and the dancer as
two different things. I would rather say that the thing I see on the
stage—the swan figure—is both “sort of” human and “sort of”
D: But then you would be using the word “sort of” in two senses.
F: Yes, that’s so. But anyhow, when I say that the swan figure is
“sort of” human, I don’t mean that it (or she) is a member of that
species or sort which we call human. D: No, of course not.
F: Rather that she (or it) is a member of another subdivision of a
larger group which would include Petroushka puppets and ballet
swans and people.
* This metalogue appeared in Impulse 1954 and is re-printed by permission of
Impulse Publications, Inc.

D: No, it’s not like genera and species. Does your larger group
include geese?
F: All right. Then I evidently do not know what the word “sort of”
means. But I do know that the whole of fantasy, poetry, ballet,
and art in general owes its meaning and importance to the
relationship which I refer to when I say that the swan figure is a
“sort of” swan—or a “pretend” swan.
D: Then we shall never know why the dancer is a swan or a puppet
or whatever, and shall never be able to say what art or poetry is
until someone says what is really meant by “sort of.”
F: Yes.
F: But we don’t have to avoid puns. In French the phrase espece de
(literally “sort of”) carries a special sort of punch. If one man
calls another “a camel” the insult may be a friendly one. But if he
calls him an espece de chameau—a sort of camel—that’s bad. It’s
still worse to call a man an espece d’espece—a sort of a sort. D:
A sort of a sort of what?
F: No—just a sort of a sort. On the other hand, if you say of a man
that he is a true camel, the insult carries a flavor of grudging
D: But when a Frenchman calls a man a sort of camel, is he using
the phrase sort of in anything like the same way as I, when I say
the swan is sort of human?
* * *

F: It’s like—there’s a passage in Macbeth. Macbeth is talking to the
murderers whom he is sending out to kill Banquo. They claim to
be men, and he tells them they are sort of men.
Ay—in the catalogue ye go for men.
as hounds and greyhounds, mongrels, spaniels, curs,
shoughs, water-rugs and demi-wolves are clept
all by the name of dogs.
(Macbeth, Act III, Scene 1)

D: No—that’s
what you said just now. What was it? “Another
subdivision of a larger group?” I don’t think that’s it at all.

F: No, it’s not only that. Macbeth, after all, uses dogs in his simile.
And “dogs” means either noble hounds or scavengers. It would
not be the same if he had used the domestic varieties of cats—or
the subspecies of wild roses.
D: All right, all right. But what is the answer to my question? When
a Frenchman calls a man a “sort of” camel, and I say that the
swan is “sort of” human, do we both mean the same thing by
“sort of”?

F: All right, let’s try to analyze what “sort of” means. Let’s take a
single sentence and examine it. If I say “the puppet Petroushka is
sort of human,” I state a relation-ship.
D: Between what and what?
F: Between ideas, I think.
D: Not between a puppet and people?
F: No. Between some ideas that I have about a puppet and some
ideas that I have about people.
D: Oh.

D: Well then, what sort of a relationship?
F: I don’t know. A metaphoric relationship?

F: And then there is that other relationship which is emphatically
not “sort of.” Many men have gone to the stake for the
proposition that the bread and wine are not “sort of” the body
and blood.
D: But is that
the same thing? I mean—is the swan ballet a
F: Yes—I think so—at least for some people. In Protestant language
we might say that the swanlike costume and movements of the
dancer are “outward and visible signs of some inward and
spiritual grace” of woman. But in Catholic language that would
make the ballet into a mere metaphor and not a sacrament.

D: But you said that for some people it is a sacrament. You mean
for Protestants?
F: No, no. I mean that if for some people the bread and wine are
only a metaphor, while for others—Catholics —the bread and
wine are a sacrament; then, if there be some for whom the
ballet is a metaphor, there may be others for whom it is
emphatically more than a metaphor—but rather a sacrament.
D: In the Catholic sense?
F: Yes.
* * *

F: I
mean that if we could say clearly what is meant by the
proposition “the bread and wine is not `sort of’ the body and
blood”; then we should know more about what we mean when
we say either that the swan is “sort of” human or that the ballet is
a sacrament.
D: Well—how do you tell the difference?
F: Which difference?
D: Between a sacrament and a metaphor.
* * *

F: Wait a minute. We are, after all, talking about the per-former or
the artist or the poet, or a given member of the audience. You ask
me how I tell the difference between a sacrament and a metaphor.
But my answer must deal with the person and not the message.
You ask me how I would decide whether a certain dance on a
certain day is or is not sacramental for the particular dancer.
D: All right—but get on with it.
F: Well—I think it’s a sort of a secret.
D: You mean you won’t tell me?
F: No—it’s not that sort of secret. It’s not something that one must
not tell. It’s something that one cannot tell.
D: What do you mean? Why not?
F: Let us suppose I asked the dancer, “Miss X, tell me, that dance
which you perform—is it for you a sacrament or a mere
metaphor?” And let us imagine that I can make this question
intelligible. She will perhaps put me off by saying, “You saw it

—it is for you to decide, if you want to, whether or not it is
sacramental for you.” Or she might say, “Sometimes it is and
sometimes it isn’t. ” Or “How was I, last night?” But in any case
she can have no direct control over the matter.


D: Do you mean that anybody who knew this secret would have it in
their power to be a great dancer or a great poet?
F: No, no, no. It isn’t like that at all. I mean first that great art and
religion and all the rest of it is about this secret; but knowing the
secret in an ordinary conscious way would not give the knower

D: Daddy, what has happened? We were trying to find out what “sort
of” means when we say that the swan is “sort of” human. I said
that there must be two senses of “sort of.” One in the phrase “the
swan figure is a `sort of’ swan, and another in the phrase “the
swan figure is `sort of’ human.” And now you are talking about
mysterious secrets and control.
F: All right. I’ll start again. The swan figure is not a real swan but a
pretend swan. It is also a pretend-not human being. It is also
“really” a young lady wearing a white dress. And a real swan
would resemble a young lady in certain ways.
D: But which of these is sacramental?
F: Oh Lord, here we go again. I can only say this: that it is not one
of these statements but their combination which constitutes a
sacrament. The “pretend” and the “pretend-not” and the “really”
somehow get fused together into a single meaning.
D: But we ought to keep them separate.
F: Yes. That is what the logicians and the scientists try to do. But
they do not create ballets that way—nor sacraments.

Metaloque: What Is an Instinct?*

Daughter: Daddy, what is an instinct?

Father: An instinct, my dear, is a explanatory principle. D: But what
does it explain?

F: Anything—almost
anything at all. Anything you want it to
D: Don’t be silly. It doesn’t explain gravity.
F: No. But that
is because nobody wants “instinct” to explain
gravity. f they did, it would explain it. We could simply say that
the moon has an instinct whose strength varies inversely as the
square of the distance .. .
D: But that’s nonsense, Daddy.
F: Yes, surely. But it was you who mentioned “instinct,” not I.
D: All right—but then what does explain gravity?
F: Nothing, my dear, because gravity is an explanatory principle.
D: Oh.
D: Do you mean that you cannot use one explanatory principle to
explain another? Never?
F: Hmm . . . hardly ever. That is what Newton meant when he said,
“hypotheses non fingo.”
D: And what does that mean? Please.
F: Well, you know what “hypotheses” are. Any statement linking
together two descriptive statements is an hypothesis. If you say
that there was a full moon on February 1st and another on March
1st; and then you link these two observations together in any
way, the statement which links them is an hypothesis.
D: Yes—and I know what non means. But what’s fingo?
F: Well—fingo is a late Latin word for “make.” It forms a verbal
noun fictio from which we get the word “fiction.”
D: Daddy, do you mean that Sir Isaac Newton thought that all
hypotheses were just made up like stories?
F: Yes—precisely that.
* This metalogue is reprinted by permission of Mouton & Co. from Approaches
to Animal Communication, edited by Thomas A. Sebeok, 1969

D: But didn’t he discover gravity? With the apple? F: No, dear. He
invented it.
D: Oh…. Daddy, who invented instinct?
F: I don’t know. Probably biblical.
D: But
if the idea of gravity links together two descriptive
statements, it must be an hypothesis.
F: That’s right.
D: Then Newton did fingo an hypothesis after all.
F: Yes—indeed he did. He was a very great scientist. D : Oh.
D: Daddy,
is an explanatory principle the same thing as an
F: Nearly, but not quite. You see, an hypothesis tries to explain
some particular something but an explanatory principle—like
“gravity” or “instinct”—really explains nothing. It’s a sort of
conventional agreement between scientists to stop trying to
explain things at a certain point.
D: Then is that what Newton meant? If “gravity” explains nothing
but is only a sort of full stop at the end of a line of explanation,
then inventing gravity was not the same as inventing an
hypothesis, and he could say he did not fingo any hypotheses.
F: That’s right. There’s no explanation of an explanatory principle.
It’s like a black box.
D: Oh.
D: Daddy, what’s a black box?
F: A “black box” is a conventional agreement between scientists to
stop trying to explain things at a certain point. I guess it’s usually
a temporary agreement.
D: But that doesn’t sound like a black box.
F: No—but that’s what it’s called. Things often don’t sound like their
D: No.
F: It’s a word that comes from the engineers. When they draw a
diagram of a complicated machine, they use a sort of shorthand.
Instead of drawing all the details, they put a box to stand for a

whole bunch of parts and label the box with what that bunch of
parts is supposed to do.

D: So
a “black box” is a label for what a bunch of things are
supposed to do… .
F: That’s right. But it’s not an explanation of how the bunch works.
D: And gravity?
F: Is
a label for what gravity is supposed to do. It’s not an
explanation of how it does it.
D: Oh.
D: Daddy, what is an instinct?
F: It’s a label for what a certain black box is supposed to do.
D: But what’s it supposed to do?
F: Hm. That is a very difficult question .. .
D: Go on.
F: Well. It’s supposed to control—partly control—what an organism
D: Do plants have instincts?
F: No. If a botanist used the word “instinct,” when talking about
plants, he would be accused of zoomorphism. D: Is that bad?
F: Yes.
Very bad for botanists. For a botanist to be guilty of
zoomorphism is as bad as for a zoologist to be guilty of
anthropomorphism. Very bad, indeed.
D: Oh. I see.
D: What did you mean by “partly control”?
F: Well. If an animal falls down a cliff, its falling is con-trolled by
gravity. But if it wiggles while falling, that might be due to
D: Self-preservative instinct?
F: I suppose so.
D: What is a self, Daddy? Does a dog know it has a self?
F: I don’t know.
But if the dog does know it has a self, and it
wiggles in order to preserve that self, then its wiggling is
rational, not instinctive.
D: Oh. Then a “self-preservative instinct” is a contradiction. F: Well,
it’s a sort of halfway house on the road to anthropomorphism.
D: Oh. That’s bad.

F: But the dog might know it had a self and not know that that self
should be preserved. It would then be rational to not wiggle. So
if the dog still wiggles, this would be instinctive. But if it learned
to wiggle, then it would not be instinctive.
D: Oh.
D: What would not be
instinctive, Daddy? The learning or the
F: No—just the wiggling.
D: And the learning would be instinctive?
F: Well . . . yes. Unless the dog had to learn to learn. D : Oh.
D: But, Daddy, what is instinct supposed to explain?
F: I keep trying to avoid that question. You see, instincts were
invented before anybody knew anything about genetics, and most
of modern genetics was discovered before anybody knew
anything about communication theory. So it is doubly difficult to
translate “instinct” into modern terms and ideas.
D: Yes, go on.
F: Well, you know that in the chromosomes, there are genes; and
that the genes are some sort of messages which have to do with
how the organism develops and with how it behaves.
D: Is
developing different from behaving, Daddy? What’s the
difference? And which is learning? Is it “developing” or
F: No! No! Not
so fast. Let’s avoid those questions by putting
developing-learning-behavior all together in one basket. A single
spectrum of phenomena. Now let’s try to say how instinct
contributes to explaining this spectrum.
D: But is it a spectrum?
F: No—that’s only a loose way of talking.
D: Oh.
D: But isn’t instinct all on the behavior end of that “spectrum”?
And isn’t learning all determined by environment and not
F: Let’s get this clear—that there is no behavior and no anatomy
and no learning in the chromosomes them-selves.

D: Don’t they have their own anatomy?
F: Yes, of course. And their own physiology. But the anatomy and
physiology of the genes and chromosomes is not the anatomy
and physiology of the whole animal.
D: Of course not.
F: But
it is about the anatomy and physiology of the whole
D: Anatomy about anatomy?
F: Yes, just as letters and words have their own forms and shapes
and those shapes are parts of words or sentences and so on—
which may be about anything.
D: Oh.
D: Daddy, is the anatomy of the genes and chromosomes about the
anatomy of the whole animal? And the physiology of the genes
and chromosomes about the physiology of the whole animal?
F: No, no. There is no reason to expect that. It’s not like that.
Anatomy and physiology are not separate in that way.
D: Daddy, are you going to put anatomy and physiology together
in one basket, like you did developing-learning-behavior?
F: Yes. Certainly.
D : Oh.
D: The same basket?
F: Why not? I think developing is right in the middle of that
basket. Right smack in the middle.
D: Oh.
D: If chromosomes and genes have anatomy and physiology, they
must have development.
F: Yes. That follows.
D: Do you think their development could be about the development
of the whole organism?
F: I don’t even know what that question would mean.
D: I do.
It means that the chromosomes and genes would be
changing or developing somehow while the baby is developing,
and the changes in the chromosomes would be about the changes
in the baby. Controlling them or partly controlling them.
F: No. I don’t think so.

D: Oh.
D: Do chromosomes learn?
F: I don’t know.
D: They do sound rather like black boxes.
F: Yes, but if chromosomes or genes can learn, then they are much
more complicated black boxes than anybody at present believes.
Scientists are always assuming or hoping that things are simple,
and then discovering that they are not.
D: Yes, Daddy.
D: Daddy, is that an instinct?
F: Is what an instinct?
D: Assuming that things are simple.
F: No. Of course not. Scientists have to be taught to do that.
D: But I thought no organism could be taught to be wrong every
F: Young lady, you are being disrespectful and wrong. In the first
place, scientists are not wrong every time they assume that things
are simple. Quite often they are right or partly right and still
more often, they think they are right and tell each other so. And
that is enough reinforcement. And, anyhow you are wrong in
saying that no organism can be taught to be wrong every time.
D: When people say that something is “instinctive,” are they trying
to make things simple? F: Yes, indeed.
D: And are they wrong?
F: I don’t know. It depends on what they mean.
D: Oh.
D: When do they do it?
F: Yes, that’s a better way of asking the question. They do it when
they see a creature do something, and they are sure: first, that the
creature did not learn how to do that something and, second, that
the creature is too stupid to understand why it should do that.
D: Any other time?
F: Yes. When they see that all members of the species do the same
things under the same circumstances; and when they see the

animal repeating the same action even when the circumstances
are changed so that the action fails.

D: So there are four ways of knowing that it’s instinctive. F: No.
Four conditions under which scientists talk about instinct.
D: But what if one condition isn’t there? An instinct sounds rather
like a habit or a custom.
F: But habits are learned.
D: Yes.
D: Are habits always twice learned?
F: What do you mean?
D: I mean—when I learn a set of chords on the guitar, first I learn
them or find them; and then later when I practice, I get the habit
of playing them that way. And sometimes I get bad habits.
F: Learning to be wrong every time?
D: Oh—all right. But what about that twice-over business? Would
both parts of learning be not there if guitar playing were
F: Yes. If both parts of learning were clearly not there, scientists
might say that guitar playing is instinctive.
D: But what if only one part of learning was missing?
F: Then, logically, the missing part could be explained by “instinct.”
D: Could either part be missing?
F: I don’t know. I don’t think anybody knows.
D: Oh.
D: Do birds practice their songs?
F: Yes. Some birds are said to practice.
D: I guess instinct gives them the first part of singing, but they have
to work on the second part.
F: Perhaps.
D: Could practicing be instinctive?
F: I suppose it could be—but I am not sure what the word “instinct”
is coming to mean in this conversation.
D: It’s an explanatory principle, Daddy, just like you said… There’s
one thing I don’t understand.
F: Yes?
D: Is there a whole lot of instinct? Or are there lots of instincts?

F: Yes. That’s a good question, and scientists have talked a great deal
about it, making lists of separate instincts and then lumping them
together again.
D: But what’s the answer?
F: Well.
It’s not quite clear. But one thing is certain: That
explanatory principles must be not multiplied beyond necessity.
D: And that means? Please?
F: It’s the idea behind monotheism—that the idea of one big God is
to be preferred to the idea of two little gods.
D: Is God an explanatory principle?
F: Oh, yes—a very big one. You shouldn’t use two black boxes—or
two instincts—to explain what one black box would explain .. .
D: If it were big enough.
F: No. It means .. .
D: Are there big instincts and little instincts?
F: Well—as a matter of fact, scientists do talk as if there were. But
they call the little instincts by other names —”reflexes,” “innate
releasing mechanisms,” “fixed action patterns,” and so on.
D: I see—like having one big God to explain the universe and lots
of little “imps” or “goblins” to explain the small things that
F: Well, yes. Rather like that.
D: But, Daddy, how do they lump things together to make the big
F: Well, for example, they don’t say that the dog has one instinct
which makes it wiggle when it falls down the cliff and another
which makes it run away from fire.
D: You mean those would both be explained by a self-preservative
F: Something like that. Yes.
D: But if you put those different acts together under one instinct,
then you cannot get away from saying that the dog has the use of
the notion of “self.”
F: No, perhaps not.
D: What would you do about the instinct for the song and the
instinct for practicing the song?
F: Well—depending on what the song is used for. Both song and
practice might be under a territorial instinct or a sexual instinct.

D: I wouldn’t put them together.
F: No?
D: Because what if the
bird also practiced picking up seed or
something? You’d have to multiply the instincts —what is it?—
beyond necessity.
F: What do you mean?
D: I mean a food-getting instinct to explain the practicing picking up
seed, and a territory instinct for practicing song. Why not have a
practicing instinct for both? That saves one black box.
F: But then you would throw away the idea of lumping together
under the same instinct actions which have the same purpose.
D: Yes—because if the practicing is for a purpose—I mean, if the
bird has a purpose—then the practicing is rational and not
instinctive. Didn’t you say something like that?
F: Yes, I did say something like that.
D: Could we do without the idea of “instinct”?
F: How would you explain things then?
D: Well. I’d just look at the little things: When some-thing goes
“pop,” the dog jumps. When the ground is not under his feet, he
wiggles. And so on.
F: You mean—all the imps but no gods?
D: Yes, something like that.
F: Well.
There are scientists who try to talk that way, and it’s
becoming quite fashionable. They say it is more objective.
D: And is it?
F: Oh, yes.
D: What does “objective” mean?
F: Well. It means that you look very hard at those things which you
choose to look at.
D: That sounds right. But how do the objective people choose which
things they will be objective about?
F: Well.
They choose those things about which it is easy to be
D: You mean easy for them?
F: Yes.

D: But how do they know that those are the easy things?
F: I suppose they try different things and find out by experience.
D: So it’s a subjective choice?
F: Oh, yes. All experience is subjective.
D: But it’s human and subjective. They decide which bits of animal
behavior to be objective about by consulting human subjective
experience. Didn’t you say that anthropomorphism is a bad
F: Yes—but they do try to be not human.
D: Which things do they leave out?
F: What do you mean?
D: I mean—subjective experience shows them which things it is
easy to be objective about. So, they go and study those things.
But which things does their experience show are difficult? So
that they avoid those things. Which are the things they avoid?
F: Well, you mentioned earlier something called “practice.” That’s a
difficult thing to be objective about. And there are other things
that are difficult in the same sort of way. Play, for example. And
exploration. It’s difficult to be objective about whether a rat is
really exploring or really playing. So they don’t investigate those
things. And then there’s love. And, of course, hate.
D: I see.
Those are the sorts of things that I wanted to invent
separate instincts for. F: Yes—those things. And don’t forget
D: Daddy—are animals objective?
F: I don’t know—probably not.
I don’t think they are subjective
either. I don’t think they are split that way.
D: Isn’t it true that people have a special difficulty about being
objective about the more animal parts of their nature?
F: I guess so. Anyhow Freud said so, and I think he was right. Why
do you ask?
D: Because, oh dear, those poor people. They try to study animals.
And they specialize in those things that they can study
objectively. And they can only be objective about those things in

which they themselves are least like animals. It must be difficult
for them.

F: No—that does not
necessarily follow. It is still possible for
people to be objective about some things in their animal nature.
You haven’t shown that the whole of animal behavior is within
the set of things that people cannot be objective about.
D: No?
D: What are the really big differences between people and animals?
F: Well—intellect, language, tools. Things like that.
D: And it
is easy for people to be intellectually objective in
language and about tools?
F: That’s right.
D: But that must mean that in people there is a whole set of ideas or
whatnot which are all tied together. A sort of second creature
within the whole person, and that second creature must have a
quite different way of thinking about everything. An objective
F: Yes. The royal road to consciousness and objectivity is through
language and tools.
D: But what happens when this creature looks at all those parts of
the person about which it is difficult for people to be objective?
Does it just look? Or does it meddle?
F: It meddles.
D: And what happens?
F: That’s a very terrible question.
D: Go on. If we are going to study animals, we must face that
F: Well . . . The poets and artists know the answer better than the
scientists. Let me read you a piece:
Thought chang’d the infinite to a serpent, that which pitieth
To a devouring flame; and man fled from its face and hid
In forests of night: then all the eternal forests were’ divided
Into earths rolling in circles of space, that like an ocean rush’d
And overwhelmed all except this finite wall of flesh.
Then was the serpent temple form’d, image of infinite
Shut up in finite revolutions; and man became an


Angel, Heaven a mighty circle turning, God a tyrant crown’d.*

D: I don’t understand it. It sounds terrible, but what does it mean?
F: Well. It’s not an objective statement, because it is talking about
the effect of objectivity—what the poet calls here “thought”
upon the whole person or the whole of life. “Thought” should
remain a part of the whole but instead spreads itself and
meddles with the rest.
D: Go on.
F: Well. It slices everything to bits.
D: I don’t understand.
F: Well, the first slice is between the objective thing and the rest.
And then inside the creature that’s made in the model of
intellect, language, and tools, it is natural that purpose will
evolve. Tools are for purposes and anything which blocks
purpose is a hindrance. The world of the objective creature gets
split into “helpful” things and “hindering” things.
D: Yes. I see that.
F: All right. Then the creature applies that split to the world of the
whole person, and “helpful” and “hindering” become Good and
Evil, and the world is then split between God and the Serpent.
And after that, more and more splits follow because the intellect
is always classifying and dividing things up.
D: Multiplying explanatory principles beyond necessity? F: That’s
D: So, inevitably, when the objective creature looks at animals, it
splits things up and makes the animals look like human beings
after their intellects have invaded their souls.
F: Exactly. It’s a sort of inhuman anthropomorphism.
D: And that is why the objective people study all the little imps
instead of the larger things?
F: Yes. It’s called S-R psychology. It’s easy to be objective about
sex but not about love.
* Blake, W., 1794, Europe a Prophecy, printed and published by the author.
(Italics added.)

D: Daddy, we’ve talked about two ways of studying animals—the
big instinct way and the S-R way, and neither way seemed very
sound. What do we do now?
F: I don’t know.
D: Didn’t you say that the royal road to objectivity and
consciousness is language and tools? What’s the royal road to
the other half?
F: Freud said dreams.
D: Oh.
D: What are dreams? How are they put together?
F: Well—dreams are bits and pieces of the stuff of which we are
made. The non-objective stuff.
D: But how are they put together?
F: Look.
Aren’t we getting rather far from the question of
explaining animal behavior?
D: I don’t know, but I don’t think so. It looks as if we are going to
be anthropomorphic in one way or another, whatever we do.
And it is obviously wrong to build our anthropomorphism on
that side of man’s nature in which he is most unlike the animals.
So let’s try the other side. You say dreams are the royal road to
the other side. So . . .
F: I didn’t. Freud said it. Or something like it.
D: All right. But how are dreams put together?
F: Do you mean how are two dreams related to each other?
D: No. Because, as you said, they are only bits and pieces. What I
mean is: How is a dream put together inside itself? Could
animal behavior be put together in the same sort of way?
F: I don’t know where to begin.
D: Well. Do dreams go by opposites?
F: Oh Lord! The old folk idea. No. They don’t predict the future.
Dreams are sort of suspended in time. They don’t have any
D: But if a person is afraid of something which he knows will
happen tomorrow, he might dream about it to-night?

F: Certainly. Or about something in his past. Or about both past
and present. But the dream contains no label to tell him what it
is “about” in this sense. It just is.
D: Do you mean it’s as if the dream had no title page?
F: Yes.
It’s like an old manuscript or a letter that has lost its
beginning and end, and the historian has to guess what it’s all
about and who wrote it and when—from what’s inside it.
D: Then we’re going to have to be objective, too?
F: Yes indeed. But we know that we have to be careful about it.
We have to watch that we don’t force the concepts of the
creature that deals in language and tools upon the dream
D: How do you mean?
F: Well. For example: if dreams somehow have not tenses and are
somehow suspended in time, then it would be forcing the wrong
sort of objectivity to say that a dream “predicts” something.
And equally wrong to say it is a statement about the past. It’s
not history.
D: Only propaganda?
F: What do you mean?
D: I mean—is it like the sort of stories that propagandists write
which they say are history but which are really only fables?
F: All right. Yes. Dreams are in many ways like myths and fables.
But not consciously made up by a propagandist. Not planned.
D: Does a dream always have a moral?
F: I don’t know about always. But often, yes. But the moral is not
stated in the dream. The psychoanalyst tries to get the patient to
find the moral. Really the whole dream is the moral.
D: What does that mean?
F: I don’t quite know.
D: Well. Do dreams go by opposites? Is the moral the opposite of
what the dream seems to say?
F: Oh yes. Often. Dreams often have an ironic or sarcastic twist. A
sort of reductio ad absurdum.
D: For example?
F: All right. A friend of mine was a fighter pilot in World War II.
After the war he became a psychologist and had to sit for his

Ph. D. oral exam. He began to be terrified of the oral, but, the
night before the exam, he had a nightmare in which he
experienced again being in a plane which had been shot down.
Next day he went into the examination without fear.

D: Why?
F: Because it was silly for a fighter pilot to be afraid of a bunch of
university professors who couldn’t really shoot him down.
D: But how did he know that? The dream could have been telling
him that the professors would shoot him down. How did he
know it was ironic?
F: Hmm. The answer is he didn’t know. The dream doesn’t have a
label on it to say it is ironic. And when people are being ironic
in waking conversation, they often don’t tell you they are being
D: No. That’s true. I always think it’s sort of cruel. F: Yes. It often
D: Daddy, are animals ever ironic or sarcastic?
F: No. I guess not. But I am not sure that those are quite the words
we should use. “Ironic” and “sarcastic” are words for the
analysis of message material in language. And animals don’t
have language. It’s perhaps part of the wrong sort of objectivity.
D: All right. Then do animals deal in opposites?
F: Well, yes. As a matter of fact, they do. But I’m not sure it’s the
same thing .. .
D: Go on. How do they? And when?
F: Well. You know how a puppy lies on his back and presents his
belly to a bigger dog. That’s sort of inviting the bigger dog to
attack. But it works in the opposite way. It stops the bigger dog
from attacking.
D: Yes. I see. It is a sort of use of opposites. But do they know
F: You mean does the big dog know that the little dog is saying the
opposite of what he means? And does the little dog know that
that is the way to stop the big dog?
D: Yes.
F: I don’t know.
I sometimes think the little dog knows a little
more about it than the big dog. Anyhow, the little dog does not

give any signals to show that he knows. He obviously couldn’t
do that.

D: Then it’s like the dreams. There’s no label to say that the dream
is dealing in opposites.
F: That’s right.
D: I think we’re getting somewhere. Dreams deal in opposites, and
animals deal in opposites, and neither carries labels to say when
they are dealing in opposites.
F: Hmm.
D: Why do animals fight?
F: Oh, for many reasons. Territory, sex, food .. .
D: Daddy, you’re talking like instinct theory. I thought we agreed
not to do that.
F: All right. But
what sort of an answer do you want to the
question, why animals fight?
D: Well. Do they deal in opposites?
F: Oh.
Yes. A lot of fighting ends up in some sort of peacemaking.
And certainly playful fighting is partly a way of affirming
friendship. Or discovering or rediscovering friendship.
D: I thought so. . . .
D: But why are the labels missing? Is it for the same reason in both
animals and dreams?
F: I don’t know. But, you know, dreams do not always deal in
D: Does a dream always have a moral?
F: I don’t know about always. But often, yes. But the moral is not
stated in the dream. The psychoanalyst tries to get the patient to
find the moral. Really the whole dream is the moral.
D: What does that mean?
F: I don’t quite know.
D: Well. Do dreams go by opposites? Is the moral the opposite of
what the dream seems to say?
F: Oh yes. Often. Dreams often have an ironic or sarcastic twist. A
sort of reductio ad absurdum.
D: For example?

F: All right. A friend of mine was a fighter pilot in World War II.
After the war he became a psychologist and had to sit for his Ph.
D. oral exam. He began to be terrified of the oral, but, the night
before the exam, he had a nightmare in which he experienced
again being in a plane which had been shot down. Next day he
went into the examination without fear.
D: Why?
F: Because it was silly for a fighter pilot to be afraid of a bunch of
university professors who couldn’t really shoot him down.
D: But how did he know that? The dream could have been telling
him that the professors would shoot him down. How did he
know it was ironic?
F: Hmm. The answer is he didn’t know. The dream doesn’t have a
label on it to say it is ironic. And when people are being ironic in
waking conversation, they often don’t tell you they are being
D: No. That’s true. I always think it’s sort of cruel.
F: Yes. It often is.
D: Daddy, are animals ever ironic or sarcastic?
F: No. I guess not. But I am not sure that those are quite the words
we should use. “Ironic” and “sarcastic” are words for the analysis
of message material in language. And animals don’t have
language. It’s perhaps part of the wrong sort of objectivity.
D: All right. Then do animals deal in opposites?
F: Well, yes. As a matter of fact, they do. But I’m not sure it’s the
same thing .. .
D: Go on. How do they? And when?
F: Well. You know how a puppy lies on his back and presents his
belly to a bigger dog. That’s sort of inviting the bigger dog to
attack. But it works in the opposite way. It stops the bigger dog
from attacking.
D: Yes. I see. It is a sort of use of opposites. But do they know that?
F: You mean does the big dog know that the little dog is saying the
opposite of what he means? And does the little dog know that
that is the way to stop the big dog?
D: Yes.

F: I don’t know. I sometimes think the little dog knows a little more
about it than the big dog. Anyhow, the little dog does not give
any signals to show that he knows. He obviously couldn’t do that.
D: Then it’s like the dreams. There’s no label to say that the dream is
dealing in opposites.
F: That’s right.
D: I think we’re getting somewhere. Dreams deal in opposites, and
animals deal in opposites, and neither carries labels to say when
they are dealing in opposites.
F: Hmm.
D: Why do animals fight?
F: Oh, for many reasons. Territory, sex, food . . .
D: Daddy, you’re talking like instinct theory. I thought we agreed not
to do that.
F: All right. But what sort of an answer do you want to the question,
why animals fight?
D: Well. Do they deal in opposites?
F: Oh. Yes. A lot of fighting ends up in some sort of peace-making.
And certainly playful fighting is partly a way of affirming
friendship. Or discovering or rediscovering friendship.
D: I thought so. . . .
D: But why are the labels missing? Is it for the same reason in both
animals and dreams?
F: I don’t know.
But, you know, dreams do not always deal in
D: No—of course not—nor do animals.
F: All right then.
D: Let’s go back to that dream. Its total effect on the man was the
same as if somebody had said to him, ” `you in a fighter plane’ is
not equal to `you in an oral exam.’ ”
F: Yes. But the dream didn’t spell that out. It only says, “you in a
fighter plane. It leaves out the “not,” and it leaves out the
instruction to compare the dream with something else and it
doesn’t say what he should compare it with.
D: All right. Let’s take the “not” first. Is there any “not” in animal

F: How could there be?
D: I mean can an animal say by its actions, “I will not bite you”?
F: Well, to begin with. Communication by actions cannot possibly
have tenses. They are only possible in language.
D: Didn’t you say that dreams have no tenses?
F: Hmm. Yes, I did.
D: Okay. But what about “not”. Can the animal say, “I am not biting
F: That still has a tense in it. But never mind. If the animal is not
biting the other, he’s not biting it, and that’s it.
D: But he might be not doing all sorts of other things, sleeping,
eating, running, and so on. How can he say, “It’s biting that I’m
not doing”?
F: He can only do that if biting has somehow been mentioned.
D: Do you mean that he could say, “I am not biting you” by first
showing his fangs and then not biting?
F: Yes. Something like that.
D: But what about two animals? They’d both have to show their
F: Yes.
D: And, it seems to me, they might misunderstand each other, and
get into a fight.
F: Yes. There is always that danger when you deal in opposites and
do not or cannot say what you are doing, especially when you do
not know what you are doing. D: But the animals would know
that they bared their fangs in order to say, “I won’t bite you.”
F: I doubt whether they would know.
Certainly neither animal
knows it about the other. The dreamer doesn’t know at the
beginning of the dream how the dream is going to end.
D: Then it’s a sort of experiment… .
F: Yes.
D: So they might get into a fight in order to find out whether
fighting was what they had to do.
F: Yes—but I’d rather put it less purposively—that the fight shows
them what sort of relationship they have, after it. It’s not planned.
D: Then “not” is really not there when the animals show their fangs?
F: I guess not. Or often not. Perhaps old friends might engage in
playful fighting and know at the beginning what they are doing.

D: All right. Then the “not” is absent in animal behavior because
“not” is part of verbal language, and there can-not be any action
signal for, “not.” And because there is no “not,” the only way to
agree on a negative is to act out the whole reductio ad
absurdum. You have to act out the battle to prove it isn’t one, and
then you have to act out the submission to prove that the other
won’t eat you.
F: Yes.
D: Did the animals have to think that out?
F: No.
Because it’s all necessarily true. And that which is
necessarily true will govern what you do regardless of whether
you know that it is necessarily true. If you put two apples with
three apples you will get five apples—even though you cannot
count. It’s another way of “explaining” things.
D : Oh.
D: But, then, why does the dream leave out the “not”?
F: I think really for a rather similar reason. Dreams are mostly made
of images and feelings, and if you are going to communicate in
images and feelings and such, you again are governed by the fact
that there is no image for “not.”
D: But you could dream of a “Stop” sign with a line through it,
which would mean “No Stopping.”
F: Yes. But that’s halfway toward language. And the deleting line
isn’t the word “not.” It’s the word “don’t.” “Don’t” can be
conveyed in action language—if the other person makes a move
to mention what you want to forbid. You can even dream in
words, and the word “not” might be among them. But I doubt if
you can dream a “not” which is about the dream. I mean a “not”
which means “This dream is not to be taken literally.”
Sometimes, in very light sleep, one knows that one is dreaming.
D: But, Daddy, you still haven’t answered the question about how
dreams are put together.
F: I think really I have answered it. But let me try again. A dream is
a metaphor or a tangle of metaphors. Do you know what a
metaphor is?

D: Yes. If I say you are like a pig that is a simile. But if I say you are
a pig, that is a metaphor.
F: Approximately, yes. When a metaphor is labeled as a metaphor it
becomes a simile.
D: And it’s that labeling that a dream leaves out.
F: That’s right. A metaphor compares things without spelling out the
comparison. It takes what is true of one group of things and
applies it to another. When we say a nation “decays,” we are
using a metaphor, suggesting that some changes in a nation are
like changes which bacteria produce in fruit. But we don’t stop to
mention the fruit or the bacteria.
D: And a dream is like that?
F: No. It’s the other way around. The dream would mention the fruit
and possibly the bacteria but would not mention the nation. The
dream elaborates on the relationship but does not identify the
things that are related.
D: Daddy, could you make a dream for me?
F: You mean, on this recipe? No. Let’s take the piece of verse which
I read you just now and turn it into a dream. It’s almost dream
material the way it stands. For most of it, you have only to
substitute images for the words. And the words are vivid enough.
But the whole string of metaphors or images is pegged down,
which would not be so in a dream.
D: What do you mean by “pegged down”?
F: I mean by the first word: “Thought.” That word the writer is using
literally, and that one word tells you what all the rest is about.
D: And in a dream?
F: That word, too, would have been metaphoric. Then the whole
poem would have been much more difficult.
D: All right—change it then.
F: What about “Barbara changed the infinite . . .” and so on.
D: But why? Who is she?
F: Well, she’s barbarous, and she’s female, and she is the mnemonic
name of a syllogistic mood. I thought she would do rather well as
a monstrous symbol for “Thought.” I can see her now with a pair
of calipers, pinching her own brain to change her universe.
D: Stop it.

F: All right. But you see what I mean by saying that in dreams the
metaphors are not pegged down.
D: Do animals peg down their metaphors?
F: No. They don’t have to. You see, when a grown-up bird makes
like a baby bird in approaching a member of the opposite sex,
he’s using a metaphor taken from the relationship between child
and parent. But he doesn’t have to peg down whose relationship
he is talking about. It’s obviously the relationship between himself
and the other bird. They’re both of them present.
D: But don’t they ever use metaphors—act out metaphors —about
something other than their own relationships?
F: I don’t think so. No—not mammals. And I don’t think birds do
either. Bees—perhaps. And, of course, people.
D: There’s one thing I don’t understand.
F: Yes?
D: We’ve found a whole lot of things in common between dreams
and animal behavior. They both deal in opposites, and they both
have no tenses, and they both have no “not,” and they both work
by metaphor, and neither of them pegs the metaphors down. But
what I don’t understand is—why, when the animals do these
things, it makes sense. I mean for them to work in opposites. And
they don’t have to peg down their metaphors—but I don’t see
why dreams should be like that, too.
F: Nor do I.
D: And there’s another thing.
F: Yes?
D: You
talked about genes and chromosomes carrying messages
about development. Do they talk like animals and dreams? I
mean in metaphors and with no “nots”? Or do they talk like us?
F: I don’t know. But I am sure their message system contains no
simple transform of Instinct Theory.

Part II: Form and
Pattern in

Culture Contact and Schismogenesis*

The Memorandum written by a Committee of the Social Sciences
Research Council (Man, 1935, 162) has stimulated me to put
forward a point of view which differs considerably from theirs; and,
though the beginning of this article may appear to be critical of their
Memorandum, I wish to make it clear from the outset that I regard
as a real contribution any serious attempt to devise categories for the
study of culture contact. Moreover, since there are several passages
in the Memorandum (among them the Definition) which I do not
perfectly understand, my criticisms are offered with some hesitation,
and are directed not so much against the Committee as against
certain errors prevalent among anthropologists.

(1) The uses of such systems of categories. In general it
is unwise to construct systems of this sort until the problems which
they are designed to elucidate have been clearly formulated; and so
far as I can see, the categories drawn up by the Committee have
been constructed not in reference to any specifically defined
problems, but to throw a general light on “the problem” of
acculturation, while the problem itself remains vague.
(2) From this it follows that our immediate need is not so much
the construction of a set of categories which will throw a light on all
the problems, but rather the schematic formulation of the problems
in such a way that they may be separately investigable.
(3) Although the Committee leave their problems undefined, we
may from a careful reading of the categories gather roughly what
questions they are asking of the material. It seems that the
Committee have, as a matter of fact, been influenced by the sort of
questions which administrators ask of anthropologists—”Is it a good
thing to use force in culture contacts?” “How can we make a given
* The whole controversy of which this article was a part has been reprinted in
Beyond the Frontier, edited by Paul Bohannon and Fred Plog. But the
ripples of this controversy have long since died down, and the article is included
here only for its positive contributions. It is reprinted, unchanged, from Man,
Article 199, Vol. XXXV, 1935, by permission of the Royal Anthropological
Institute of Great Britain and Ireland.

people accept a certain sort of trait?” and so on. In response to this
type of question we find in the definition of acculturation an emphasis
upon difference in culture between the groups in contact and
upon the resulting changes; and such dichotomies as that between
“elements forced upon a people or received voluntarily by them”1
may likewise be regarded as symptomatic of this thinking in terms
of administrative problems. The same may be said of the categories
V, A, B, and C, “acceptance,” “adaptation” and “reaction.”

(4) We may agree that answers are badly needed to these
questions of administration and, further, that a study of culture
contacts is likely to give these answers, But it is almost certain that
the scientific formulation of the problems of con-tact will not follow
these lines. It is as if in the construction of categories for the study
of criminology we started with a dichotomy of individuals into
criminal and noncriminal —and, indeed, that curious science was
hampered for a long while by this very attempt to define a “criminal
(5) The Memorandum is based upon a fallacy: that we can
classify the traits of a culture under such headings as economic,
religious, etc. We are asked, for example, to classify traits into three
classes, presented respectively because of: (a) economic profit or
political dominance; (b) desirability of bringing about conformity to
values of donor group; and (c) ethical and religious considerations.
This idea, that each trait has either a single function or at least some
one function which overtops the rest, leads by extension to the
idea that a culture can be subdivided into “institutions” where the
bundle of traits which make up one institution are alike in their
major functions. The weakness of this method of subdividing a
culture has been conclusively demonstrated by Malinowski and
his pupils, who have shown that almost the whole of a culture
may be seen variously as a mechanism for modifying and
satisfying the sexual needs of the individuals, or for the
enforcement of the norms of behavior, or for supplying the
individuals with food.2 From this exhaustive demonstration we
must expect that any single trait of a culture will prove on
1 In any case it is clear that in a scientific study of processes and natural laws this
invocation of free will can have no place.


examination to be not simply economic or religious or structural,
but to partake of all these qualities according to the point of view
from which we look at it. If this be true of a culture seen in
synchronic section, then it must also apply to the diachronic
processes of culture contact and change; and we must expect that for
the offering, acceptance or refusal of every trait that are
simultaneous causes of an economic, structural, sexual, and
religious nature.

(6) From this it follows that our categories “religious,”
“economic,” etc., are not real subdivisions which are present in the
cultures which we study, but are merely abstractions which we
make for our own convenience when we set out to describe cultures
in words. They are not phenomena present in culture, but are labels
for various points of view which we adopt in our studies. In
handling such abstractions we must be careful to avoid Whitehead’s
“fallacy of misplaced concreteness,” a fallacy into which, for
2 Cf. Malinowski, Sexual Life and Crime and Custom; A. I. Richards, Hunger
and Work. This question of the subdivision of a culture into “institutions” is not
quite as simple as I have indicated; and, in spite of their own works, I believe that
the London School still adheres to a theory that some such division is practicable. It
is likely that confusion arises from the fact that certain native peoples—perhaps all,
but in any case those of Western Europe—actually think that their culture is so
subdivided. Various cultural phenomena also contribute something toward such a
subdivision, e.g., (a) the division of labor and differentiation of norms of behavior
between different groups of individuals in the same community, and (b) an
emphasis, present in certain cultures, upon the subdivisions of place and time upon
which behavior is ordered. These phenomena lead to the possibility, in such
cultures, of dubbing all behavior which, for example, takes place in church between

11.30 and 12.30 on Sundays as “religious.” But even in the study of such cultures
the anthropologist must look with some suspicion upon his classification of traits
into institutions and must expect to find a great deal of over-lapping between
various institutions.
An analogous fallacy occurs in psychology, and consists in regarding
behavior as classifiable according to the impulses which inspire it, e.g., into
such categories as self-protective, assertive, sexual, acquisitive, etc. Here, too,
confusion results from the fact that not only the psychologist, but also the
individual studied, is prone to think in terms of these categories. The
psychologists would do well to accept the probability that every bit of behavior
is—at least in a well-integrated individual —simultaneously relevant to all
these abstractions

example, the Marxian historians fall when they maintain that
economic “phenomena” are “primary.”

With this preamble, we may now consider an alternative scheme
for the study of contact phenomena.

(7) Scope of the inquiry I suggest that we should consider under
the head of “culture contact” not only those cases in which the
contact occurs between two communities with different cultures and
results in profound disturbance of the culture of one or both groups;
but also cases of contact within a single community. In these cases
the contact is between differentiated groups of individuals, e.g.,
between the sexes, between old and young, between aristocracy and
plebs, between clans, etc., groups which live together in
approximate equilibrium. I would even extend the idea of “contact”
so widely as to include those processes whereby a child is molded
and trained to fit the culture into which he was born,3 but for the
present we may confine ourselves to contacts between groups of
individuals, with different cultural norms of behavior in each group.
(8) If we consider the possible end of the drastic disturbances
which follow contacts between profoundly different communities,
we see that the changes must theoretically result in one or other of
the following patterns:
(a) the complete fusion of the originally different groups
(b) the elimination of one or both groups
(c) the
persistence of both groups in dynamic equilibrium
within one major community
(9) My purpose in extending the idea of contact to cover the
conditions of differentiation inside a single culture is to use our
knowledge of these quiescent states to throw light upon the factors
which are at work in states of disequilibrium. It may be easy to
obtain a knowledge of the factors from their quiet working, but
impossible to isolate them when they are violent. The laws of
3 The present scheme is oriented toward the study of social rather than
psychological processes, but a closely analogous scheme might be constructed for
the study of psychopathology. Here the idea of “contact” would be studied,
especially in the contexts of the molding of the individual, and the processes of
schismogenesis would be seen to play an important part not only in accentuating the
maladjustments of the deviant, but also in assimilating the normal individual to his


gravity cannot conveniently be studied by observation of houses
collapsing in an earth-quake.

(10) Complete fusion Since this is one of the possible ends of the
process we must know what factors are present in a group of
individuals with consistent homogeneous pat-terns of behavior in all
members of the group. An approach to such conditions may be
found in any community which is in a state of approximate
equilibrium but, unfortunately, our own communities in Europe are
in a state of such flux that these conditions scarcely occur.
Moreover, even in primitive communities. the conditions are usually
complicated by differentiation, so that we must be content with
studies of such homogeneous groups as can be observed within the
major differentiated communities.
Our first task will be to ascertain what sorts of unity obtain
within such groups, or rather—bearing in mind that we are
concerned with aspects and not classes of phenomena—what
aspects of the unity of the body of traits we must describe in order to
get a whole view of the situation. I submit that the material, to be
fully understood, must be examined in, at least, the following five
separable aspects:

(a) A structural aspect of unity The behavior of any one
individual in any one context is, in some sense, cognitively
consistent with the behavior of all the other individuals in all other
contexts. Here we must be prepared to find that the inherent logic
of one culture differs profoundly from that of others. From this
point of view we shall see, for example, that when individual A
gives a drink to individual B, that behavior is consistent with other
norms of behavior obtaining within the group which contains A
and B.
This aspect of the unity of the body of behavior patterns may be
restated in terms of a standardization of the cognitive aspects of the
personalities of the individuals. We may say that the patterns of
thought of the individuals are so standardized that their behavior
appears to them logical.

(b) Affective aspects of unity In studying the culture from this
point of view, we are concerned to show the emotional setting of all
the details of behavior. We shall see the whole body of behavior as a

concerted mechanism oriented toward affective satisfaction and
dissatisfaction of the individuals.

This aspect of a culture may also be described in terms of a
standardization of affective aspects of the personalities of the
individuals, which are so modified by their culture that their
behavior is to them emotionally consistent.

(c) Economic unity Here we shall see the whole body of behavior
as a mechanism oriented toward the production and distribution of
material objects.
(d) Chronological and spatial unity Here we shall see the
behavior patterns as schematically ordered according to time and
place. We shall see A as giving the drink to B “because it is Saturday
evening in the Blue Boar.”
(e) Sociological unity Here we shall see the behavior of the
individuals as oriented toward the integration and disintegration of
the major unit, the Group as a whole. We shall see the giving of
drinks as a factor which promotes the solidarity of the group.
(11) In addition to studying the behavior of members of the
homogeneous group from all these points of view, we must examine
a number of such groups to discover the effects of standardization of
these various points of view in the people we are studying. We have
stated above that every bit of behavior must be regarded as probably
relevant to all these viewpoints, but the fact remains that some
peoples are more inclined than others to see and phrase their own
behavior as “logical” or “for the good of the State.”
(12) With this knowledge of the conditions which obtain in
homogeneous groups, we shall be in a position to examine the
processes of fusion of two diverse groups into one. We may even be
able to prescribe measures which will either promote or retard such
fusion, and predict that a trait which fits the five aspects of unity can
be added to a culture with-out other changes. If it does not fit, then
we can search for appropriate modifications either of the culture or
of the trait.
(13) The elimination of one or both groups This end result is
perhaps scarcely worth studying, but we should at least examine any
material that is available, to determine what sort of effects such
hostile activity has upon the culture of the survivors. It is possible,
for example, that the patterns of behavior associated with

elimination of other groups may be assimilated into their culture so
that they are impelled to eliminate more and more.

(14) Persistence of both groups in dynamic equilibrium This is
probably the most instructive of the possible end results of contact,
since the factors active in the dynamic equilibrium are likely to be
identical or analogous with those which, in disequilibrium, are
active in cultural change. Our first task is to study the relationships
obtaining between groups of individuals with differentiated behavior
patterns, and later to consider what light these relationships throw
upon what are more usually called “contacts.” Every anthropologist
who has been in the field has had opportunity of studying such
differentiated groups.
(15) The possibilities of differentiation of groups are by no
means infinite, but fall clearly into two categories (a) cases in which
the relationship is chiefly symmetrical, e.g., in the differentiation of
moieties, clans, villages and the nations of Europe; and (b) cases in
which the relationship is complementary, e.g., in the differentiation
of social strata, classes, castes, age grades, and, in some cases, the
cultural differentiation between the sexes.4 Both these types of
differentiation contain dynamic elements, such that when certain
restraining factors are removed the differentiation or split between
the groups increases progressively toward either breakdown or a
new equilibrium.
(16) Symmetrical differentiation To this category may be referred
all those cases in which the individuals in two groups A and B have
the same aspirations and the same behavior patterns, but are
differentiated in the orientation of these patterns. Thus members of
group A exhibit behavior patterns A,B,C in their dealings with each
other, but adopt the patterns X,Y,Z in their dealings with members of
group B. Similarly, group B adopt the patterns A,B,C among them4
Cf. Margaret Mead, Sex and Temperament, 1935. Of the communities
described in this book, the Arapesh and the Mundugumor have a preponderantly
symmetrical relationship between the sexes, while the Chambuli have a
complementary relationship. Among the Iatmul, a tribe in the same area, which I
have studied, the relationship between the sexes is complementary, but on rather
different lines from that of the Chambuli. I hope shortly to publish a book on the
Iatmul with sketches of their culture from the points of view a, b, and e out-lined in
paragraph 10. (See Bibliography, items 1936 and 1958 B.)


selves, but exhibit X,Y,Z in dealing with group A. Thus a position is
set up in which the behavior X,Y,Z is the standard reply to X,Y,Z.
This position contains elements which may lead to progressive
differentiation or schismogenesis along the same lines. If, for
example, the patterns X,Y,Z include boasting, we shall see that there
is a Iikelihood, if boasting is the reply to boasting, that each group
will drive the other into excessive emphasis of the pattern, a process
which if not re-strained can only lead to more and more extreme
rivalry and ultimately to hostility and the breakdown of the whole

(17) Complementary differentiation To this category we may
refer all those cases in which the behavior and aspirations of the
members of the two groups are fundamentally different. Thus
members of group A treat each other with patterns L,M,N, and
exhibit the patterns O,P,Q in dealings with group B. In reply to
O,P,Q, the members of group B exhibit the patterns U,V,W, but
among themselves they adopt patterns R,S,T. Thus it comes about
that O,P,Q is the reply to U,V,W, and vice versa. This differentiation
may be-come progressive. If, for example, the series, O,P,Q includes
patterns culturally regarded as assertive, while U,V,W includes
cultural submissiveness, it is likely that submissiveness will promote
further assertiveness which in turn will promote further
submissiveness. This schismogenesis, unless it is re-strained, leads
to a progressive unilateral distortion of the personalities of the
members of both groups, which results in mutual hostility between
them and must end in the break-down of the system.
(18) Reciprocity Though relationships between groups can
broadly be classified into two categories, symmetrical and
complementary, this subdivision is to some extent blurred by
another type of differentiation which we may describe as reciprocal.
In this type the behavior patterns X and Y are adopted by members
of each group in their dealings with the other group, but instead of
the symmetrical system whereby X is the reply to X and Y is the
reply to Y, we find here that X is the reply to Y. Thus in every single
in-stance the behavior is asymmetrical, but symmetry is regained
over a large number of instances since sometimes group A exhibit X
to which group B reply with Y, and sometimes group A exhibit Y
and group B reply with X. Cases in which group A sometimes sell

sago to group B and the latter some-times sell the same commodity
to A, may be regarded as reciprocal; but if group A habitually sell
sago to B while the latter habitually sell fish to A, we must, I think,
regard the pattern as complementary. The reciprocal pattern, it may
be noted, is compensated and balanced within itself and therefore
does not tend toward schismogenesis.

(19) Points for investigation:
(a) We need a proper survey of the types of behavior which can
lead to schismogeneses of the symmetrical type. At present it is only
possible to point to boasting and commercial rivalry, but no doubt
there are many other patterns which will be found to be
accompanied by the same type of effect.
(b) We need a survey of the types of behavior which are mutually
complementary and lead to schismogeneses of the second type. Here
we can at present only cite assertiveness versus submissiveness,
exhibitionism versus admiration, fostering versus expressions of
feebleness and, in addition, the various possible combinations of
these pairs.
(c) We need verification of the general law assumed above, that
when two groups exhibit complementary behavior to each other, the
internal behavior between members of group A must necessarily
differ from the internal behavior between members of group B.
(d) We need a systematic examination of schismogeneses of both
types from the various points of view outlined in paragraph 10. At
present I have only looked at the matter from the ethological and
structural points of view (paragraph 10, aspects a and b) . In
addition to this, the Marxian historians have given us a picture of the
economic aspect of complementary schismogenesis in Western
Europe. It is likely, however, that they themselves have been
influenced unduly by the schismogenesis which they studied and
have been thereby prompted into exaggeration.
(e) We need to know something about the occurrence of
reciprocal behavior in relationships which are preponderantly either
symmetrical or complementary.
(20) Restraining factors But, more important than any of the
problems in the previous paragraph, we need a study of the factors
which restrain both types of schismogenesis. At the present moment,
the nations of Europe are far advanced in symmetrical schismo

genesis and are ready to fly at each other’s throats; while within each
nation are to be observed growing hostilities between the various
social strata, symptoms of complementary schismogenesis. Equally,
in the countries ruled by new dictatorships we may observe early
stages of complementary schismogenesis, the behavior of his
associates pushing the dictator into ever greater pride and

The purpose of the present article is to suggest problems and
lines of investigation rather than to state the answers, but,
tentatively, suggestions may be offered as to the factors controlling

(a) It is possible that, actually, no healthy equilibrated
relationship between groups is either purely symmetrical or purely
complementary, but that every such relationship contains elements
of the other type. It is true that it is easy to classify relationships into
one or the other category according to their predominant emphases,
but it is possible that a very small admixture of complementary
behavior in a symmetrical relationship, or a very small admixture of
symmetrical behavior in a complementary relationship, may go a
long way toward stabilizing the position. Examples of this type of
stabilization are perhaps common. The squire is in a predominantly
complementary and not always comfortable relationship with his
villagers, but if he participate in village cricket (a symmetrical
rivalry) but once a year, this may have a curiously disproportionate
effect upon his relationship with them.
(b) It is certain that, as. in the case quoted above in which group
A sell sago to B while the latter sell fish to A, complementary
patterns may sometimes have a real stabilizing effect by promoting a
mutual dependence between the groups.
(c) It is possible that the presence of a number of truly reciprocal
elements in a relationship may tend to stabilize it, preventing the
schismogenesis which otherwise might result either from
symmetrical or complementary elements. But this would seem to be
at best a very weak defense: on the one hand, if we consider the
effects of symmetrical schismogenesis upon the reciprocal behavior
patterns, we see that the latter tend to be less and less exhibited.
Thus, as the individuals composing the nations of Europe become
more and more involved in their symmetrical international rivalries,

they gradually leave off behaving in a reciprocal manner, deliberately
reducing to a minimum their former reciprocal commercial
behavior.5 On the other hand, if we consider the effects of
complementary schismogenesis upon the reciprocal behavior
patterns, we see that one-half of the reciprocal pat-tern is liable to
lapse. Where formerly both groups exhibited both X and Y, a system
gradually evolves in which one of the groups exhibits only X, while
the other exhibits only Y. In fact, behavior which was formerly
reciprocal is reduced to a typical complementary pattern and is
likely after that to contribute to the complementary schismogenesis.

(d) It is certain that either type of schismogenesis between two
groups can be checked by factors which unite the two groups either
in loyalty or opposition to some outside element. Such an outside
element may be either a symbolic individual, an enemy people or
some quite impersonal circumstance—the lion will lie down with
the lamb if only it rain hard enough. But it must be noted that where
the outside element is a person or group of persons, the relationship
of the combined groups A and B to the outside group will always be
itself a potentially schismogenic relationship of one or the other
type. Examination of multiple systems of this kind is badly needed
and especially we need to know more about the systems (e.g.,
military hierarchies) in which the distortion of personality is
modified in the middle groups of the hierarchy by permitting the
individuals to exhibit respect and submission in dealings with
higher groups while they exhibit assertiveness and pride in dealing
with the lower.
(e) In the case of the European situation, there is one other
possibility—a special case of control by diversion of attention to
outside circumstances. It is possible that those responsible for the
policy of classes and nations might become conscious of the
processes with which they are playing and cooperate in an attempt
to solve the difficulties. This, how-ever, is not very likely to occur
5 In this, as in the other examples given, no attempt is made to consider the
schismogenesis from all the points of view outlined in paragraph 10. Thus, inasmuch
as the economic aspect of the matter is not here being considered, the effects of the
slump upon the schismogenesis are ignored. A complete study would be subdivided
into separate sections, each treating one of the aspects of the phenomena.


since anthropology and social psychology lack the prestige
necessary to advise; and, with-out such advice, governments will
continue to react to each other’s reactions rather than pay attention to

(21) In conclusion, we may turn to the problems of the
administrator faced with a black-white culture contact. His first task
is to decide which of the end results outlined in paragraph 8 is
desirable and possible of attainment. This decision he must make
without hypocrisy. If he chooses fusion, then he must endeavor to
contrive every step so as to promote the conditions of consistency
which are outlined (as problems for investigation) in paragraph 10.
If he chooses that both groups shall persist in some form of dynamic
equilibrium, then he must contrive to establish a system in which the
possibilities of schismogenesis are properly compensated or
balanced against each other. But at every step in the scheme which I
have outlined there are problems which must be studied by trained
students and which when solved will contribute, not only to applied
sociology, but to the very basis of our understanding of human
beings in society.

Experiments in Thinking About
Ethnological Material*

As I understand it, you have asked me for an honest, introspective—
personal—account of how I think about anthropological
material, and if I am to be honest and personal about my thinking,
then I must be impersonal about the results of that thinking. Even if
I can banish both pride and shame for half an hour, honesty will still
be difficult.

Let me try to build up. a picture of how I think by giving you an
autobiographical account of how I have acquired my kit of
conceptual tools and intellectual habits. I do not mean an academic
biography or a list of what subjects I have studied, but something
more significant than that—a list rather of the motifs of thought in
various scientific subjects which left so deep an impression on my
mind that when I came to work on anthropological material, I
naturally used those borrowed motifs to guide my approach to this
new material.

I owe the greatest part of this kit of tools to my father, William
Bateson, who was a geneticist. In schools and universities they do
very little to give one an idea of the basic principles of scientific
thinking, and what I learned of this came in very large measure from
my father’s conversation and perhaps especially from the overtones
of his talk. He himself was inarticulate about philosophy and
mathematics and logic, and he was articulately distrustful of such
subjects, but still, in spite of himself, I think, he passed on to me
something of these matters.

The attitudes which I got from him were especially those which
he had denied in himself. In his early—and as I think he knew—his
best work he posed the problems of animal symmetry, segmentation,

* This paper was given at the Seventh Conference on Methods in Philosophy
and the Sciences, held at the New School for Social Research, April 28, 1940. It is
here -reprinted from Philosophy of Science, Vol. 8, No. 1, copyright 1941, The
Williams & Wilkins Co. Reproduced by permission.

serial repetition of parts, patterns, etc. Later he turned away from
this field into Mendelism, to which he devoted the remainder of his
life. But he had always a hankering after the problems of pattern and
symmetry, and it was this hankering and the mysticism that in-spired
it that I picked up and which, for better or worse, I called “science.”

I picked up a vague mystical feeling that we must look for the
same sort of processes in all fields of natural phenomena—that we
might expect to find the same sort of laws at work in the structure of
a crystal as in the structure of society, or that the segmentation of an
earthworm might really be comparable to the process by which
basalt pillars are formed.

I should not preach this mystical faith in quite those terms today
but would say rather that I believe that the types of mental operation
which are useful in analyzing one field may be equally useful in
another—that the framework (the eidos) of science, rather than the
framework of Nature, is the same in all fields. But the more mystical
phrasing of the matter was what I vaguely learnt, and it was of paramount
importance. It lent a certain dignity to any scientific
investigation, implying that when I was analyzing the pat-terns of
partridges’ feathers, I might really get an answer or a bit of an
answer to the whole puzzling business of pattern and regularity in
nature. And further, this bit of mysticism was important because it
gave me freedom to use my scientific background, the ways of
thought that I had picked up in biology and elementary physics and
chemistry; it encouraged me to expect these ways of thought to fit in
with very different fields of observation. It enabled me to regard all
my training as potentially useful rather than utterly irrelevant to

When I came into anthropology there was a considerable reaction
taking place against the use of loose analogies, especially against the
Spencerian analogy between the Organism and Society. Thanks to
this mystical belief in the pervading unity of the phenomena of the
world, I avoided a great deal of intellectual waste. I never had any
doubt that this analogy was fundamentally sound; since to doubt
would have been emotionally expensive. Nowadays, of course, the
emphasis has shifted. Few would seriously doubt that the ways of
analysis which have been found useful in analyzing one complex
functioning system are likely to be of use in analyzing any other


similar system. But the mystical prop was useful then, though its
phrasing was bad.

There is another way, too, in which that mysticism has helped—a
way which is especially relevant to my thesis. I want to emphasize
that whenever we pride ourselves upon finding a newer, stricter way
of thought or exposition; when-ever we start insisting too hard upon
“operationalism” or symbolic logic or any other of these very
essential systems of tramlines, we lose something of the ability to
think new thoughts. And equally, of course, whenever we rebel
against the sterile rigidity of formal thought and exposition and let
our ideas run wild, we likewise lose. As I see it, the advances in
scientific thought come from a combination of loose and strict
thinking, and this combination is the most precious tool of science.

My mystical view of phenomena contributed specifically to build
up this double habit of mind—it led me into wild “hunches” and, at
the same time, compelled more formal thinking about those
hunches. It encouraged looseness of thought and then immediately
insisted that that looseness be measured up against a rigid
concreteness. The point is that the first hunch from analogy is wild,
and then, the moment I begin to work out the analogy, I am brought
up against the rigid formulations which have been devised in the
field from which I borrow the analogy.

Perhaps it is worth giving an example of this; it was a matter of
formulating the social organization of a New Guinea tribe,—the
Iatmul. The Iatmul social system differs from ours in one very
essential point. Their society completely lacks any sort of
chieftainship, and I phrased this matter loosely by saying that the
control of the individual was achieved by what I called “lateral”
sanctions rather than by “sanctions from above.” Going over my
material, I found further that in general the subdivisions of the
society—the clans, moieties, etc.—had virtually no means of
punishing their own members. I had a case in which a ceremonial
house owned by a particular junior age grade had been defiled, and
though the other members of the grade were very angry with the
defiler, they could do nothing about it. I asked whether they would
kill one of his pigs or take any of his property, and they replied “No,
of course not. He is a member of their own initiatory grade.” If the
same thing had happened in the big senior ceremonial house which

belongs to several grades, then the defiler would be punished. His
own grade would defend him but the others would start a brawl.6

I then began looking for more concrete cases which could be
compared with the contrast between this system and our own. I said,
“It’s like the difference between the radially symmetrical animals
(jellyfish, sea anemones, etc.) and the animals which have transverse
segmentation (earthworms, lobsters, man, etc.).”

Now in the field of animal segmentation we know very little
about the mechanisms concerned, but at least the problems are more
concrete than in the social field. When we compare a social problem
with a problem of animal differentiation, we are at once provided
with a visual diagram, in terms of which we may be able to talk a
little more precisely. And for the transversely segmented animals, at
least, we have something more than a merely anatomical diagram.
Thanks to the work that has been done on experimental embryology
and axial gradients, we have some idea of the dynamics of the
system. We know that some sort of asymmetrical relation obtains
between the successive segments, that each segment would, if it
could (I speak loosely) form a head, but that the next anterior
segment prevents this. Further, this dynamic asymmetry in the
relations between successive segments is reflected morphologically;
we find in most such animals a serial difference—what is called
metameric differentiation—between the. successive segments.

Their appendages, though they can be shown to conform to a
single basic structure, differ one from another as we go down the
series. (The legs of the lobster provide a familiar example of the
sort of thing I mean.)

In contrast with this, in the radially symmetrical animals, the
segments, arranged around the center like sectors of a circle, are
usually all alike.

As I say, we do not know much about the segmentation of
animals, but at least here was enough for me to take back to the
problem of Iatmul social organization. My “hunch” had provided
me with a set of stricter words and diagrams, in terms of which I
could try to be more precise in my thinking about the Iatmul

6 For details of this and other similar incidents cf. Naven, pp. 98-107,
Cambridge, Cambridge University Press, 1936.


problem. I could now look again at the Iatmul material to
determine whether the relationship between the clans was really in
some sense symmetrical and to determine whether there was
anything that could be compared with the lack of metameric
differentiation. I found that the “hunch” worked. I found that so far
as opposition, control, etc. between the clans was concerned, the
relations between them were reasonably symmetrical, and further,
as to the question of differentiation between them, it could be
shown that, though there were considerable differences, these
followed no serial pattern. Additionally, I found that there was a
strong tendency for clans to imitate each other, to steal bits of each
other’s mythological history and to incorporate these into their own
past—a sort of fraudulent heraldry, each clan copying the others so
that the whole system tended to diminish the differentiation
between them. (The system perhaps also contained tendencies in
an opposite direction, but this question I need not discuss now.)

I followed up the analogy in another direction. Impressed by the
phenomena of metameric differentiation, I made the point that in our
society with its hierarchical systems (comparable to the earthworm
or the lobster), when a group secedes from the parent society, it is
usual to find that the line of fission, the division between the new
group and the old, marks a differentiation of mores. The Pilgrim
Fathers wander off in order to be different. But among the Iatmul,
when two groups in a village quarrel, and one half goes off and
founds a new community, the mores of the two groups remain
identical. In our society, fission tends to be heretical (a following
after other doctrines or mores), but in Iatmul, fission is rather
schismatic (a following after other leaders without change of

You will note that . here I overrode my analogy at one point and
that this matter is still not perfectly clear. When a transverse fission
or a lateral budding occurs in a transversely segmented animal, the
products of that bud or fission are identical, the posterior half which
was held in check by the anterior is relieved of this control and
develops into a normal, complete animal. I am therefore not in step
with my analogy when I regard the differentiation which accompanies
fission in a hierarchical society as comparable with that
which exists before fission in a transversely segmented animal. This

divergence from the analogy will surely be worth investigation; it
will take us into a more precise study of the asymmetrical relations
which obtain between the units in the two cases and raise questions
about the reactions of the subordinate member to its position in the
asymmetry. This aspect of the matter I have not yet examined.

Having got some sort of conceptual frame within which to
describe the interrelations between clans, I went on from this to
consider the interrelations between the various age grades in terms
of this same frame. Here, if anywhere, where age might be expected
to provide a basis for serial differentiation, we ought to expect to
find some analogue of the transverse segmentation with
asymmetrical relations between the successive grades—and to a
certain extent the age-grade system fitted this picture. Each grade
has its ceremonies and its secrets of initiation into that grade; and in
these ceremonies and secrets it was perfectly easy to trace a metameric
differentiation. Ceremonies which are fully developed at the
top of the system are still recognizable in their basic form in the
lower levels—but more rudimentary at each level as we go down the

But the initiatory system contains one very interesting element
which was brought into sharp relief when my point of view was
defined in terms of animal segmentation. The grades alternate, so
that the whole system consists of two opposed groups, one group
made up of grades 3, 5, 7, etc. (the odd numbers), and the other
made up of 2, 4, 6, etc.; and these two groups maintain the type of
relationship which I had already described as “symmetrical”—each
providing sanctions by quarreling with the other when their rights
are infringed.

Thus even where we might expect the most definite hierarchy,
the Iatmul have substituted for it a headless system in which one
side is symmetrically opposed to the other.

From this conclusion my enquiry, influenced by many other
types of material, will go on to look at the matter from other points
of view—especially the psychological problems of whether a
preference for symmetrical rather than asymmetrical relationships
can be implanted in the individual, and what the mechanisms of
such character formation may be. But we need not go into that now.


Enough has been said to bring out the methodological theme—
that a vague “hunch” derived from some other science leads into the
precise formulations of that other science in terms of which it is
possible to think more fruit-fully about our own material.

You will have noticed that the form in which I used the
biological findings was really rather different from that in which a
zoologist would talk about his material. Where the zoologist might
talk of axial gradients, I talked about “asymmetrical relationships
between successive segments,” and in my phrasing I was prepared to
attach to the word “successive” two simultaneous meanings—in
referring to the animal material it meant a morphological series in a
three-dimensional concrete organism, while in referring to the
anthropological material the word “successive” meant some
abstracted property of a hierarchy.

I think it would be fair to say that I use the analogies in some
curiously abstract form—that, as for “axial gradients” I substitute
“asymmetrical relationships,” so also I endow the word “successive”
with some abstract meaning which makes it applicable to both sorts
of cases.

This brings us to another very important motif in my thinking—a
habit of constructing abstractions which refer to terms of
comparison between entities; and to illustrate this I can clearly
remember the first occasion on which I was guilty of such an
abstraction. It was in my Zoological Tripos examination at
Cambridge, and the examiner had tried to compel me to answer at
least one question on each branch of the subject. Comparative
anatomy I had always regarded as a waste of time, but I found
myself face to face with it in the examnation and had not the
necessary detailed knowledge. I was asked to compare the
urinogenital system of the amphibia with that of the mammalia, and
I did not know much about it.

Necessity was the mother of invention. I decided that I ought to
be able to defend the position that comparative anatomy was a
muddled waste of time, and so I set to work to attack the whole
emphasis on homology in zoological theory. As you probably will
know, zoologists conventionally deal in two sorts of comparability
between organs—homology and analogy. Organs are said to be
“homologous” when it can be shown that they have similar structure

or bear similar structural relations to other organs, e.g., the trunk of
the elephant is homologous with the nose and lip of a man be-cause
it has the same formal relation to other parts—eyes, etc.; but the
trunk of an elephant is analogous to the hand of a man because both
have the same uses. Fifteen years ago comparative anatomy
revolved endlessly around these two sorts of comparability, which
incidentally are good examples of what I mean by “abstractions
which define the terms of a comparison between entities.”

My attack on the system was to suggest that there might be other
sorts of comparability and that these would con-fuse the issue to
such a degree that mere morphological analysis would not suffice. I
argued that the bilateral fins of a fish would conventionally be
regarded as homologous with the bilateral limbs of a mammal, but
that the tail of a fish, a median organ, would conventionally be
regarded a “different from” or at most only “analogous to” the fins.
But what about the double-tailed Japanese goldfish? In this animal
the factors causing an anomaly of the tail also cause the same
anomaly in the bilateral fins; therefore there was here another sort of
comparability, an equivalence in terms of processes and laws of
growth. Well, I don’t know what mark I got for my answer. I found
out much later that, as a matter of fact, the lateral fins of the goldfish
are scarcely, if at all, affected by the factors which cause the
anomaly in the tail, but I doubt if the examiner caught me in my
bluff; and I found also that, curiously, Haekel in 1854 had actually
coined the word “homonomy” for the very type of equivalence that I
was inventing. The word is, so far as I know, obsolete, and was
obsolete when I wrote my answer.

So far as I was concerned, however, the idea was new and I had
thought of it myself. I felt that I had discovered how to think. That
was in 1926, and this same old clue—recipe, if you like—has
remained with me ever since. I did not realize that I had a recipe;
and it was not until ten years later that I fully grasped the
significance of this analogyhomology-homonomy business.

Perhaps it will be of interest to recount in some detail my various
brushes with these concepts and the recipe which they contained.
Soon after the examination to which I have referred, I went into
anthropology and for some time stopped thinking—wondering
rather what could be made of this subject, but not getting anything


clear except a repudiation of most of the conventional approaches
which, to me, seemed meaningless. I wrote a little skit on the
concept of totemism in 1930, first proving that the totemism of the
Iatmul is true totemism because it contains a “high percentage” of
characteristics of totemism listed in “Notes and Queries on Anthropology”
issued more or less ex cathedra by the Royal
Anthropological Institute, and then going on to the question, what
sort of equivalence we thought we were referring to when we equate
some bits of Iatmul culture with the totem-ism of North America,
and dragging in homology-homonomy, etc.

In this discussion of “true” totemism I still had the homonomyhomology
abstractions perfectly clear and was using the concepts
with a clean (though inarticulate) understanding of what sort of
abstractions they were—but it is interesting that I afterwards made
some other comparable abstractions for the analysis of latmul
material and muddled the issues through forgetting this very thing.

I was especially interested in studying what I called the “feel” of
culture, and I was bored with the conventional study of the more
formal details. I went out to New Guinea with that much vaguely
clear—and in one of my first letters home I complained of the
hopelessness of putting any sort of salt on the tail of such an
imponderable concept as the “feel” of culture. I had been watching a
casual group of natives chewing betel, spitting, laughing, joking,
etc., and I felt acutely the tantalizing impossibility of what.I wanted
to do.

A year later, still in New Guinea, I read Arabia Deserta and
recognized with a thrill that Doughty had in a sense done what I
wanted to do. He had put salt on the tail of the very bird that I was
hunting. But I realized also—sadlythat he had used the wrong kind
of salt. I was not interested in achieving a literary or artistic
representation of the “feel” of the culture; I was interested in a
scientific analysis of it.

On the whole I think that Doughty was an encouragement to me,
and the greatest encouragement I got from him was due to a
fallacious bit of thinking which he prompted. It appeared to me that
it was impossible to understand the behavior of his Arabs apart from
the “feel” of their culture, and from this it seemed to follow that the
“feel” of the culture was in some way causative in shaping native

behavior. This encouraged me to go on thinking that I was trying
after something that was important—so far so good. But it also
guided me into regarding the “feel” of the culture as much more
concrete and causally active than I had any right to do.

This false concreteness was reinforced later by an accident of
language. Radcliffe-Brown called to my attention the old word
“ethos” and told me that that was what I was trying to study. Words
are dangerous things, and it so hap-pens that “ethos” is in some
ways a very bad word. If I had been compelled to make up my own
word for what I wanted to say, I might have done better and saved
myself a great deal of confusion. I would, I hope, have put forward
something like “ethonomy,” which would have reminded me that I
was referring to an abstraction of the same order as homology or
homonomy. The trouble with the word “ethos” is just this—that it is
too short. It is a unit word, a single Greek substantive, and as such
helped me to go on thinking that it referred to a unit something
which I could still regard as causative. I handled the word as if it
were a category of behavior or a sort of factor which shaped

We are all familiar with this loose use of words in such phrases
as: “the causes of war are economic,” “economic behavior,” “he was
influenced by his emotions,” “his symptoms are the result of conflict
between his superego and his id.” (I am not sure how many of these
fallacies are contained in that last example; at a rough count, there
seem to be five with a possible sixth, but there may be more.
Psychoanalysis has erred sadly in using words that are too short and
there-fore appear more concrete than they are.) I was guilty of just
this sort of shoddy thinking in my handling of the word “ethos,” and
you must excuse me if I have gathered moral support for this
confession by a digression to show that at any rate others have
committed the same crime.

Let us examine the stages by which I got into the fallacy and the
way in which I got out of it. I think the first step toward an escape
from sin was to multiply offenses—and there is a good deal to be
said for this method. Vice is after all a dull business whether it be
physical or intellectual, and an effective cure can sometimes be
achieved by indulgence to the point at which the patient realizes
this. It is a way of proving that a given line of thought or conduct


will not do, by experimentally extrapolating it to infinity, when its
absurdities become evident.

I multiplied my offenses by creating several more concepts of
about the same degree of abstraction as “ethos”—I had “eidos,”
“cultural structure,” “sociology”—and all these I handled as though
they were concrete entities. I pictured the relations between ethos
and cultural structure as being like the relation between a river and
its banks—”The river molds the banks and the banks guide the river.
Similarly, the ethos molds the cultural structure and is guided by it.”
I was still looking for physical analogies, but now the position was
not quite the same as when I was looking for analogies in order to
get concepts which I could use in analyzing observed material. I was
looking now for physical analogies which I could use in analyzing
my own concepts, and that is a very much less satisfactory business.
I do not mean, of course, that the other sciences can give one no
help in the attempt to straighten out one’s thoughts; they surely can.
For example, the theory of Dimensions in physics may be of
enormous help in this field. What I mean is that when one is seeking
an analogy for the elucidation of material of one sort, it is good to
look at the way analogous material has been analyzed. But when one
is seeking an elucidation of one’s own concepts, then one must look
for analogies on an equally abstract level. However, these similes
about rivers and their banks seemed pretty to me and I treated them
quite seriously.

Here I must digress for a moment to describe a trick of thought
and speech, which I have found useful. When I am faced with a
vague concept and feel that the time is not yet ripe to bring that
concept into strict expression, I coin some loose expression for
referring to this concept and do not want to prejudge the issue by
giving the concept too meaningful a term. I therefore dub it hastily
with some brief concrete colloquial term—generally Anglo-Saxon
rather than Latin—I will speak of the “stuff” of culture, or “bits” of
culture, or the “feel” of culture. These brief Anglo-Saxon terms
have for me a definite feeling tone which reminds me all the time
that the concepts behind them are vague and await analysis. It is a
trick like tying a knot in a handkerchief—but has the advantage
that it still permits me, if I may so express it, to go on using the
handkerchief for other purposes. I can go on using the vague

concept in the valuable process of loose thinking—still continually
reminded that my thoughts are loose.

But these similes about ethos being the river and the
formulations of culture or “cultural structure” being its banks were
not Anglo-Saxon reminders that I was leaving some-thing for
analysis at a later date. They were, as I thought, the real thing—a
real contribution to our understanding of how culture works. I
thought that there was one sort of phenomenon which I could call
“ethos” and another sort which I could call “cultural structure” and
that these two worked together—had mutual effect one on the
other. All that remained for me to do was to discriminate clearly
between these various sorts of phenomena so that other people
could perform the same sort of analysis that I was doing.

This effort of discrimination I postponed, feeling perhaps that
the problem was not quite ripe—and I went on with the cultural
analysis. And did what I still think was good work. I want to
emphasize this last point—that, as a matter of fact, considerable
contributions to science can be made with very blunt and crooked
concepts. We may joke about the way misplaced concreteness
abounds in every word of psycho-analytic writing—but in spite of
all the muddled thinking that Freud started, psychoanalysis
remains as the outstanding contribution, almost the only
contribution to our understanding of the family—a monument to
the importance and value of loose thinking.

Finally I had completed my book on Iatmul culture, with the
exception of the last chapter, the writing of which was to be the final
testing and review of my various theoretical concepts and
contributions. I planned that this chapter should contain some
attempt to discriminate between the sort of thing that I called “ethos”
and the sort of thing that I called “eidos,” etc.

I was in a state approximating that panic in the examination room
which formerly produced the concept of homonomy. I was due to
sail for my next field trip—my book had to be finished before I
sailed—the book could not stand without some clear statement
about the interrelations of these concepts of mine.

Here I will quote what finally appeared in the book in this last


“I began to doubt the validity of my own categories, and
performed an experiment. I chose three bits of culture: (a) a wau
(mother’s brother) giving food to a laua (sister’s son); a pragmatic
bit, (b) a man scolding his wife; an ethological bit, and (c) a man
marrying his father’s sister’s daughter; a structural bit. Then I drew a
lattice of nine squares on a large piece of paper, three rows of
squares with three squares in each row. I labeled the horizontal rows
with my bits of culture and the vertical columns with my categories.
Then I forced myself to see each bit as conceivably belonging to
each category. I found that it could be done.

“I found that I could think of each bit of culture structurally; I
could see it as in accordance with a consistent set of rules or
formulations. Equally, I could see each bit as `pragmatic,’ either as
satisfying the needs of individuals or as contributing to the
integration of society. Again, I could see each bit ethologically, as an
expression of emotion.

“This experiment may seem puerile, but to me it was very
important, and I have recounted it at length because there may be
some among my readers who tend to regard such concepts as
`structure’ as concrete parts which `interact’ in culture, and who find,
as I did, a difficulty in thinking of these concepts as labels merely
for points of view adopted either by the scientist or by the natives. It
is instructive to perform the same experiment with such concepts as
economics, etc.”7

In fact, “ethos” and the rest were finally reduced to abstractions
of the same general order as “homology,” “homonomy,” etc.; they
were labels for points of view voluntarily adopted by the
investigator. I was, as you may imagine, enormously excited at
getting this tangle straightened out—but I was also worried because
I thought I should be compelled to rewrite the whole book. But I
found that this was not so. I had to tune up the definitions, check
through to see that each time the technical term appeared I could
substitute the new definition for it, mark the more egregious pieces
of nonsense with footnotes warning the reader that these passages
might be taken as a warning of how not to say things—and so on.

7 Loc. cit., p. 261.

But the body of the book was sound enough—all that it needed was
new castors on its legs.

So far I have spoken of my own personal experiences with strict
and loose thinking, but I think actually the story which I have
narrated is typical of the whole fluctuating business of the advance
of science. In my case, which is a small one and comparatively
insignificant in the whole advance of science, you can see both
elements of the alternating process—first the loose thinking and
the building up of a structure on unsound foundations and then the
correction to stricter thinking and the substitution of a new
underpinning beneath the already constructed mass. And that, I
believe, is a pretty fair picture of how science advances, with this
exception, that usually the edifice is larger and the individuals who
finally contribute the new underpinning are different people from
those who did the initial loose thinking. Sometimes, as in physics,
we find centuries between the first building of the edifice and the
later correction of the foundations—but the process is basically the

And if you ask me for a recipe for speeding up this process, I
would say first that we ought to accept and enjoy this dual nature
of scientific thought and be willing to value the way in which the
two processes work together to give us advances in understanding
of the world. We ought not to frown too much on either process, or
at least to frown equally on either process when it is
unsupplemented by the other. There is, I think, a delay in science
when we start to specialize for too long either in strict or in loose
thinking. I suspect, for example, that the Freudian edifice has been
al-lowed to grow too big before the corrective of strict thought is
applied to it—and now when investigators start rephrasing the
Freudian dogmas in new stricter terms there may be a lot of ill
feeling, which is wasteful. (At this point I might perhaps throw out
a word of comfort to the orthodox in psychoanalysis. When the
formulators begin rooting about among the most basic of analytic
premises and questioning the concrete reality of such concepts as
the “ego” or “wishes” or the “id” or the “libido”—as indeed they
are already be-ginning to root—there is no need to get alarmed and
to start having terror dreams of chaos and storms at sea. It is
certain that most of the old fabric of analysis will still be left


standing after the new underpinning has been inserted. And when
the concepts, postulates, and premises have been straightened out,
analysts will be able to embark upon a new and still more fruitful
orgy of loose thinking, until they reach a stage at which again the
results of their thinking must be strictly conceptualized. I think that
they ought to enjoy this alternating quality in the progress of
science and not delay the progress of science by a refusal to accept
this dualism.)

Further than this, besides simply not hindering progress, I think
we might do something to hasten matters, and I have suggested
two ways in which this might be done. One is to train scientists to
look among the older sciences for wild analogies to their own
material, so that their wild hunches about their own problems will
land them among the strict formulations. The second method is to
train them to tie knots in their handkerchiefs whenever they leave
some mat-ter unformulated—to be willing to leave the matter so
for years, but still leave a warning sign in the very terminology
they use, such that these terms will forever stand, not as fences
hiding the unknown from future investigators, but rather as

signposts which read: ” UNEXPLORED BEYOND THIS POINT.

Morale and National Character*

We shall proceed as follows: (1) We shall examine some of the
criticisms which can be urged against our entertaining any concept
of “national character.” (2) This examination will enable us to state
certain conceptual limits within which the phrase “national
character” is likely to be valid. (3) We shall then go on, within these
limits, to outline what orders of difference we may expect to find
among Western nations, trying, by way of illustration, to guess more
concretely at some of these differences. (4) Lastly, we shall consider
how the problems of morale and international relations are affected
by differences of this order.

Barriers to Any Concept of “National Character”

Scientific enquiry has been diverted from questions of this type
by a number of trains of thought which lead scientists to regard all
such questions as unprofitable or unsound. Be-fore we hazard any
constructive opinion as to the order of differences to be expected
among European populations, therefore, these diverting trains of
thought must be examined.

It is, in the first place, argued that not the people but rather the
circumstances under which they live differ from one community to
another; that we have to deal with differences either in historical
background or in current conditions, and that these factors are
sufficient to account for all differences in behavior without our
invoking any differences of character in the individuals concerned.
Essentially this argument is an appeal to Occam’s Razor—an
assertion that we ought not to multiply entities beyond necessity.
The argument is that, where observable differences in circumstance
exist, we ought to invoke those rather than mere inferred differences
in character, which we cannot observe.

* This essay appeared in Civilian Morale, edited by Goodwin Watson, copyright
1942 by the Society for the Psychological Study of Social Issues. It is here reprinted
by permission of the publisher. Some introductory material has been edited out.

The argument may be met in part by quoting experimental data,
such as Lewin’s experiments (unpublished material), which showed
that there are great differences in the way in which Germans and
Americans respond to failure in an experimental setting. The
Americans treated failure as a challenge to increase effort; the
Germans responded to the same failure with discouragement. But
those who argue for the effectiveness of conditions rather than
character can still reply that the experimental conditions are not, in
fact, the same for both groups; that the stimulus value of any
circumstance depends upon how that circumstance stands out
against the background of other circumstances in the life of the
subject, and that this contrast cannot be the same for both groups.

It is possible, in fact, to argue that since the same circumstances
never occur for individuals of different cultural back-ground, it is
therefore unnecessary to invoke such abstractions as national
character. This argument breaks down, I believe, when it is pointed
out that, in stressing circumstance rather than character, we would
be ignoring the known facts about learning. Perhaps the best
documented generalization in the field of psychology is that, at any
given moment, the behavioral characteristics of any mammal, and
especially of man, depend upon the previous experience and
behavior of that individual. Thus in presuming that character, as well
as circumstance, must be taken into account, we are not multiplying
entities beyond necessity; we know of the significance of learned
character from other types of data, and it is this knowledge which
compels us to consider the additional “entity.”

A second barrier to any acceptance of the notion of “national
character” arises after the first has been negotiated. Those who grant
that character must be considered can still doubt whether any
uniformity or regularity is likely to obtain within such. a sample of
human beings as constitutes a nation. Let us grant at once that
uniformity obviously does not occur, and let us proceed to consider
what sorts of regularity may be expected.

The criticism which we are trying to meet is likely to take five
forms. (1) The critic may point to the occurrence of subcultural
differentiation, to differences between the sexes, or between classes,
or between occupational groups within the community. (2) He may
point to the extreme heterogeneity and confusion of cultural norms

which can be observed in “melting-pot” communities. (3) He may
point to the accidental deviant, the individual who has undergone
some “accidental” traumatic experience, not usual among those in
his social environment. (4) He may point to the phenomena of
cultural change, and especially to the sort of differentiation which
results when one part of the community lags behind some other in
rate of change. (5) Lastly, he may point to the arbitrary nature of
national boundaries.

These objections are closely interrelated, and the replies to them
all derive ultimately from two postulates: first, that the individual,
whether from a physiological or a psycho-logical point of view, is a
single organized entity, such that all its “parts” or “aspects” are
mutually modifiable and mutually interacting; and second, that a
community is like-wise organized in this sense.

If we look at social differentiation in a stable community—say, at
sex differentiation in a New Guinea tribe8—we find that it is not
enough to say that the habit system or the character structure of one
sex is different from that of another. The significant point is that the
habit system of each sex cogs into the habit system of the other; that
the behavior of each promotes the habits of the other.9 We find, for
example, between the sexes, such complementary patterns as
spectatorship-exhibitionism, dominance-submission, and succoring-
dependence, or mixtures of these. Never do we find mutual
irrelevance between such groups.

Although it is unfortunately true that we know very little about
the terms of habit differentiation between classes, sexes,
occupational groups, etc., in Western nations, there is, I think, no
danger in applying this general conclusion to all cases of stable
differentiation between groups which are living in mutual contact. It

8 Cf. M. Mead (Sex and Temperament in Three Primitive Societies, New York,
Morrow, 1935), especially Part III, for an analysis of sex differentiation among the
Chambuli; also G. Bateson (Naven, Cambridge, Cambridge University Press, 1936)
for an analysis of sex differentiation among adults in Iatmul, New Guinea.

9 We are considering here only those cases in which ethological differentiation
follows the sex dichotomy. It is also probable that, where the ethos of the two sexes is
not sharply differentiated, it would still be correct to say that the ethos of each
promotes that of the other, e.g., through such mechanisms as competition and mutual
imitation. Cf. M. Mead (op. cit.).

is, to me, inconceivable that two differing groups could exist side by
side in a community with-out some sort of mutual relevance
between the special characteristics of one group and those of the
other. Such an occurrence would be contrary to the postulate that a
community is an organized unit. We shall, therefore, presume that
this generalization applies to all stable social differentiation.

Now, all that we know of the mechanics of character formation—
especially the processes of projection, reaction formation,
compensation, and the like—forces us to regard these bipolar
patterns as unitary within the individual. If we know that an
individual is trained in overt expression of one-half of one of these
patterns, e.g., in dominance behavior, we can predict with certainty
(though not in precise language) that the seeds of the other half—
submission—are simultaneously sown. in his personality. We have
to think of the individual, in fact, as trained in dominance-
submission, not in either dominance or submission. From this it
follows that where we are dealing with stable differentiation within
a community, we are justified in ascribing common character to the
members of that community, provided we take the precaution of
describing that common character in terms of the motifs of
relationship between the differentiated sections of the community.

The same sort of considerations will guide us in dealing with our
second criticism—the extremes of heterogeneity, such as occur in
modern “melting-pot” communities. Suppose we attempted to
analyze out all the motifs of relationship between individuals and
groups in such a community as New York City; if we did not end in
the madhouse long before we had completed our study, we should
arrive at a picture of common character that would be almost
infinitely complex—certainly that would contain more fine
differentiations than the human psyche is capable of resolving
within itself. At this point, then, both we and the individuals whom
we are studying are forced to take a short cut: to treat heterogeneity
as a positive characteristic of the common environment, sui generis.
When, with such an hypothesis, we begin to look for common
motifs of behavior, we note the very clear tendencies toward
glorying in heterogeneity for its own sake (as in the Robinson
Latouche “Ballad for Americans”) and toward regarding the world

as made up of an infinity of disconnected quiz-bits (like Ripley’s
“Believe It or Not”).

The third objection, the case of the individual deviant, falls in
the same frame of reference as that of the differentiation of stable
groups. The boy on whom an English public-school education does
not take, even though the original roots of his deviance were laid
in some “accidental” traumatic incident, is reacting to the public-
school system. The behavioral habits which he acquires may not
follow the norms which the school intends to implant, but they are
acquired in reaction to those very norms. He may (and often does)
acquire patterns the exact opposite of the normal; but he cannot
conceivably acquire irrelevant patterns. He may become a “bad”
public-school Englishman, he may become insane, but still his
deviant characteristics will be systematically related to the norms
which he is resisting. We may describe his character, indeed, by
saying that it is as systematically related to the standard public-
school character as the character of Iatmul natives of one sex is
systematically related to the character of the other sex. His
character is oriented to the motifs and patterns of relationship in
the society in which he lives.

The same frame of reference applies to the fourth consideration,
that of changing communities and the sort of differentiation which
occurs when one section of a community lags behind another in
change. Since the direction in which a change occurs will
necessarily be conditioned by the, status quo ante, the new
patterns, being reactions to the old, will be systematically related
to the old. As long as we confine ourselves to the terms and
themes of this systematic relationship, therefore, we are
entitled to expect regularity of character in the individuals.
Furthermore, the expectation and experience of change may, in
some cases, be so important as to become a common character


determining factor10 sui generis, in the same sort of way that
“heterogeneity” may have positive effects.

Lastly, we may consider cases of shifting national boundaries,
our fifth criticism. Here, of course, we cannot expect that a
diplomat’s signature on a treaty will immediately modify the
characters of the individuals whose national allegiance is thereby
changed. It may even happen—for example, in cases where a
preliterate native population is brought for the first time in contact
with Europeans—that, for some time after the shift, the two parties
to such a situation will behave in an exploratory or almost random
manner, each retaining its own norms and not yet developing any
special adjustments to the situation of contact. During this period,
we should still not expect any generalizations to apply to both
groups. Very soon, however, we know that each side does develop
special patterns of behavior to use in its contacts with the other.11 At
this point, it becomes meaningful to ask what systematic terms of
relationship will describe the common character of the two groups;
and from this point on, the degree of common character structure
will increase until the two groups become related to each other just
as two classes or two sexes in a stable, differentiated society.12

In sum, to those who argue that human communities show too
great internal differentiation or contain too great a random element
for any notion of common character to apply, our reply would be
that we expect such an approach to be useful (a) provided we

10 For a discussion of the role played by “change” and “heterogeneity” in
melting-pot communities, cf. M. Mead (“Educative effects of social
environment as disclosed by studies of primitive societies.” Paper read at the
Symposium on Environment and Education, University of Chicago, September 22,
1941). Also F. Alexander (“Educative influence of personality factors in the
environment.” Paper read at the Symposium on Environment and Education,
University of Chicago, September 22,1941).

11 In the South Seas, those special modes of behavior which Europeans
adopt toward native peoples, and those other modes of behavior which the native
adopts toward Europeans, are very obvious. Apart from analyses of “pidgin”
languages, we have, however, no psychological data on these patterns. For a
description of the analogous patterns in Negro-white relationships, cf. J.
Dollard (Caste and Class in a Southern Toivn, New Haven, Yale University
Press, 1937), especially Chapter XII, Accommodation Attitudes of Negroes.

12 Cf. G. Bateson, “Culture Contact and Schismogenesis,” Man, 1935, 8: 199.
(Reprinted in this volume.)


describe common character in terms of the themes of relationship
between groups and individuals within the community, and (b)
provided that we allow sufficient time to elapse for the community
to reach some degree of equilibrium or to accept either change or
heterogeneity as a characteristic of their human environment.

Differences Which We May Expect Between National Groups

The above examination of “straw men” in the case against
“national character” has very stringently limited the scope of this
concept. But the conclusions from this examination are by no means
simply negative. To limit the scope of a concept is almost
synonymous with defining it.

We have added one very important tool to our equipment —the
technique of describing the common character (or the “highest
common factor” of character) of individuals in a human community
in terms of bipolar adjectives. Instead of despairing in face of the
fact that nations are highly differentiated, we shall take the
dimensions of that differentiation as our clues to the national
character. No longer content to say, “Germans are submissive,” or
“Englishmen are aloof,” we shall use such phrases as “dominant-
submissive” when relationships of this sort can be shown to occur.
Similarly, we shall not refer to “the paranoidal element in German
character,” unless we can show that by “paranoidal” we mean some
bipolar characteristic of German-German or German-foreign
relationships. We shall not describe varieties of character by
defining a given character in terms of its position on a continuum
between extreme dominance and extreme submissiveness, but we
shall, instead, try to use for our descriptions some such continua as
“degree of interest in, or orientation toward, dominance-

So far, we have mentioned only a very short list of bipolar
characteristics: dominance-submission, succoring-dependence, and
exhibitionism-spectatorship. One criticism will certainly be
uppermost in the reader’s mind, that, in short, all three of these
characteristics are clearly present in all West-ern cultures. Before


our method becomes useful, therefore, we must try to expand it to
give us sufficient scope and discriminatory power to differentiate
one Western culture from another.

As this conceptual frame develops, no doubt, many further
expansions and discriminations will be introduced. The present
paper will deal with only three such types of expansion.

Alternatives to Bipolarity

When we invoked bipolarity as a means of handling differentiation
within society without foregoing some notion of common
character structure, we considered only the possibility of simple
bipolar differentiation. Certainly this pattern is very common in
Western cultures; take, for instance, Republican-Democrat, political
Right-Left, sex differentiation, God and the devil, and so on. These
peoples even try to impose a binary pattern upon phenomena which
are not dual in nature—youth versus age, labor versus capital, mind
versus matter—and, in general, lack the organizational devices for
handling triangular systems; the inception of any “third” party is
always regarded, for example, as a threat to our political
organization. This clear tendency toward dual systems ought not,
however, to blind us to the occurrence of other patterns.13

There is, for example, a very interesting tendency in English
communities toward the formation of ternary systems, such as
parents-nurse-child, king-ministers-people, officers-N.C.O.’sprivates.
14 While the precise motifs of relationship in these ternary

13 The Balinese social system in the mountain communities is almost entirely
devoid of such dualisms. The ethological differentiation of the sexes is rather slight;
political factions are completely absent. In the plains, there is a dualism which has
resulted from the intrusive Hindoo caste system, those with caste being discriminated
from those without caste. At the symbolic level (partly as a result of Hindoo
influence) dualisms are much more frequent, however, than they are in the social
structure (e.g., Northeast vs. Southwest, Gods vs. demons, symbolic Left vs. Right,
symbolic Male vs. Female, etc.).

14 A fourth instance of this threefold pattern occurs in some great public schools
(as in Charterhouse), where the authority is divided between the quieter, more


systems remain to be investigated, it is important to note that these
systems, to which I refer as “ternary,” are neither “simple
hierarchies” nor “triangles.” By a pure hierarchy, I should mean a
serial system in which face-to-face relations do not occur between
members when they are separated by some intervening member; in
other words, systems in which the only communication between A
and C passes through B. By a triangle I should mean a threefold
system with no serial properties. The ternary system, parent-nursechild,
on the other hand, is very different from either of these other
forms. It contains serial elements, but face-to-face contact does
occur between the first and the third members. Essentially, the
function of the middle member is to instruct and discipline the third
member in the forms of behavior which he should adopt in his
contacts with the first. The nurse teaches the child how to behave
toward its parents, just as the N.C.O. teaches and disciplines the
private in how he should behave toward officers. In psychoanalytic
terminology, the process of introjection is done indirectly, not by
direct impact of the parental personality upon the child.15 The faceto-
face contacts between the first and third members are, however,
very important. We may refer, in this connection, to the vital daily
ritual in the British Army, in which the officer of the day asks the
assembled privates and N.C.O.’s whether there are any complaints.

Certainly, any full discussion of English character ought to allow
for ternary, as well as bipolar patterns.

Symmetrical Motifs

So far, we have considered only what we have called “complementary”
patterns of relationship, in which the behavior patterns
at one end of the relationship are different from, but fit in with, the
behavior patterns at the other end (dominance-submission, etc.).

polished, intellectual leaders (“monitors”) and the rougher, louder, athletic leaders
(captain of football, head of long room, etc.), who have the duty of seeing to it that
the “fags” run when the monitor calls.

15 For a general discussion of cultural variants of the Oedipus situation and the
related systems of cultural sanctions, cf. M. Mead (“Social change and cultural


There exists, however, a whole category of human interpersonal
behavior which does not conform to this description. In addition to
the contrasting complementary patterns, we have to recognize the
existence of a series of symmetrical patterns, in which people
respond to what others are doing by themselves doing something
similar. In particular, we have to consider those competitive16 patterns
in which individual or group A is stimulated to more of any
type of behavior by perceiving more of that same type of behavior
(or greater success in that type of behavior) in individual or group B.

There is a very profound contrast between such competitive
systems of behavior and complementary dominance-submission
systems—a highly significant contrast for any discussion of national
character. In complementary striving, the stimulus which prompts A
to greater efforts is the relative weakness in B; if we want to makeA
subside or submit, we ought to show him that B is stronger than he
is. In fact, the complementary character structure may be
summarized by the phrase “bully-coward,” implying the
combination of these characteristics in the personality. The
symmetrical competitive systems, on the other hand, are an almost
precise functional opposite of the complementary. Here the stimulus
which evokes greater striving in A is the vision of greater strength
or greater striving in B; and, inversely, if we demonstrate to A that B
is really weak, A will relax his efforts.

It is probable that these two contrasting patterns are alike
available as potentialities in all human beings; but clearly, any
individual who behaves in both ways at once will risk internal
confusion and conflict. In the various national groups, consequently,
different methods of resolving this discrepancy have developed. In
England and in America, where children and adults are subjected to
an almost continuous barrage of disapproval whenever they exhibit
the complementary patterns, they inevitably come to accept the

16 The term “cooperation,” which is sometimes used as the opposite of
“competition,” covers a very wide variety of patterns, some of them symmetrical and
others complementary, some bipolar and others in which the cooperating
individuals are chiefly oriented to some personal or impersonal goal. We may
expect that some careful analysis of these patterns will give us vocabulary for
describing other sorts of national characteristics. Such an analysis cannot be attempted
in this paper.


ethics of “fair play.” Responding to the challenge of difficulties, they
cannot, without guilt, kick the underdog.17 For British morale
Dunkirk was a stimulus, not a depressant.

In Germany, on the other hand, the same cliches are apparently
lacking, and the community is chiefly organized on the basis of a
complementary hierarchy in terms of dominance-submission. The
dominance behavior is sharply and clearly developed; yet the picture
is not perfectly clear and needs further investigation. Whether a pure
dominance-submission hierarchy could ever exist as a stable system
is doubtful. It seems that in the case of Germany, the submission end
of the pattern is masked, so that overt submissive behavior is almost
as strongly tabooed as it is in America or England. In place of
submission, we find a sort of parade-ground impassivity.

A hint as to the process by which the submissive role is modified
and rendered tolerable comes to us out of the inter-views in a
recently begun study of German life histories.18 One German subject
described how different was the treatment which he, as a boy,
received in his South German home, from that which his sister
received. He said that much more was demanded of him; that his
sister was allowed to evade discipline; that whereas he was always
expected to click his heels and obey with precision, his sister was
allowed much more freedom. The interviewer at once began to look
for intersex sibling jealousy, but the subject declared that it was a
greater honor for the boy to obey. “One doesn’t expect too much of
girls,” he said. “What one felt they (boys) should accomplish and do
was very serious, because they had to be prepared for life.” An
interesting inversion of noblesse oblige.

17 io It is, however, possible that in certain sections of these nations,
complementary patterns occur with some frequency—particularly among groups who
have suffered from prolonged insecurity and uncertainty, e.g., racial minorities,
depressed areas, the stock exchange, political circles, etc.

18 G. Bateson, unpublished research for the Council on Human Relations.

Combinations of Motifs

Among the complementary motifs, we have mentioned only
three—dominance-submission, exhibitionism-spectatorship, and
succorance-dependence—but these three will suffice to illustrate the
sort of verifiable hypotheses at which we can arrive by describing
national character in this hyphenated terminology.19

Since, clearly, all three of these motifs occur in all Western
cultures, the possibilities for international difference are limited to
the proportions and ways in which the motifs are combined. The
proportions are likely to be very difficult to detect, except where the
differences are very large. We may be sure ourselves that Germans
are more oriented toward dominance-submission than are
Americans, but to demonstrate this certainty is likely to be difficult.
To estimate differences in the degree of development of
exhibitionismspectatorship or succorance-dependence in the various
nations will, indeed, probably be quite impossible.

If, however, we consider the possible ways in which these motifs
may be combined together, we find sharp qualitative differences
which are susceptible of easy verification. Let us assume that all
three of these motifs are developed in all relationships in all Western
cultures, and from this assumption go on to consider which
individual plays which role.

It is logically possible that in one cultural environment A will be
dominant and exhibitionist, while B is submissive and spectator;
while in another culture X may be dominant and spectator, while Y
is submissive and exhibitionist.

Examples of this sort of contrast rather easily come to mind.
Thus we may note that whereas the dominant Nazis preen
themselves before the people, the czar of Russia kept his private
ballet, and Stalin emerges from seclusion only to review his troops.
We might perhaps present the relationship between the Nazi Party
and the people thus:

19 “For a fuller study, we ought to consider such other motifs as aggression-
passivity, possessive-possessed, agent-tool, etc. And all of these motifs will require
somewhat more critical definition than can be attempted in this paper.


Party People

Dominance Submission

Exhibitionism Spectatorship

While the czar and his ballet would be represented:

Czar Ballet

Dominance Submission

Spectatorship Exhibitionism

Since these European examples are comparatively unproved, it is
worthwhile at this point to demonstrate the occurrence of such
differences by describing a rather striking ethnographic difference
which has been documented more fully. In Europe, where we tend to
associate succoring behavior with social superiority, we construct
our parent symbols accordingly. Our God, or our king, is the
“father” of his people. In Bali, on the other hand, the gods are the
“children” of the people, and when a god speaks through the mouth
of a person in trance, he addresses anyone who will listen as
“father.” Similarly, the rajah is sajanganga (“spoilt” like a child) by
his people. The Balinese, further, are very fond of putting children in
the combined roles of god and dancer; in mythology, the perfect
prince is polished and narcissistic. Thus the Balinese pattern might
be summarized thus:

High Status Low Status

Dependence Succoring

Exhibitionism Spectatorship

And this diagram would imply, not only that the Balinese feel
dependence and exhibitionism and superior status to go naturally
together, but also that a Balinese will not readily combine succoring
with exhibitionism (that is, Bali completely lacks the ostentatious
gift-giving characteristic of many primitive peoples) or will be
embarrassed if forced by the context to attempt such a combination.


Although the analogous diagrams for our Western cultures
cannot be drawn with the same certainty, it is worthwhile to attempt
them for the parent-child relationships in English, American, and
German cultures. One extra complication must, however, be faced;
when we look at parent-child relationships instead of at relationships
between princes and people, we have to make specific allowance for
the changes in the pattern which occur as the child grows older. Succorance-
dependence is undoubtedly a dominant motif in early
childhood, but various mechanisms later modify this extreme
dependence, to bring about some degree of psychological independence.

The English upper-and middle-class system would be represented
diagrammatically thus:

Parents Children
Dominance Submission
(modified by “ternary” nurse system)
Succoring Dependence(dependence habits broken by separation—children sent to school)
Exhibitionism Spectatorship
(children listen silently at meals)

In contrast with this, the analogous American pattern seems tobe:

Parents Children

Dominance (slight) Submission (slight)

Succoring Dependence

Spectatorship Exhibitionism

And this pattern differs from the English not only in the reversal
of the spectatorship-exhibitionism roles, but also in the content of
what is exhibited. The American child is encouraged by his parents
to show off his independence. Usually the process of psychological
weaning is not accomplished by sending the child away to a

boarding school; instead, the child’s exhibitionism is played off
against his independence, until the latter is neutralized. Later, from
this beginning in the exhibition of independence, the individual may
sometimes go on in adult life to show off succorance, his wife and
family becoming in some degree his “exhibits.”

Though the analogous German pattern probably resembles the
American in the arrangement of the paired complementary roles,
certainly it differs from the American in that the father’s dominance
is much stronger and much more consistent, and especially in that
the content of the boy’s exhibitionism is quite different. He is, in
fact, dominated into a sort of heel-clicking exhibitionism which
takes the place of overt submissive behavior. Thus, while in the
American character exhibitionism is encouraged by the parent as a
method of psychological weaning, both its function and its content
are for the German entirely different.

Differences of this order, which may be expected in all European
nations, are probably the basis of many of our naive and often
unkind international comments. They may, indeed, be of
considerable importance in the mechanics of international relations,
in as much as an understanding of them might dispel some of our
misunderstandings. To an American eye, the English too often
appear “arrogant,” whereas to an English eye the American appears
to be “boastful.” If we could show precisely how much of truth and
how much of distortion is present in these impressions, it might be a
real contribution to interallied cooperation.

In terms of the diagrams above, the “arrogance” of the
Englishman would be due to the combination of dominance and
exhibitionism. The Englishman in a performing role (the parent at
breakfast, the newspaper editor, the political spokesman, the
lecturer, or what not) assumes that he is also in a dominant role—
that he can decide in accordance with vague, abstract standards what
sort of performance to give —and the audience can “take it or leave
it.” His own arrogance he sees either as “natural” or as mitigated by
his humility in face of the abstract standards. Quite unaware that
his behavior could conceivably be regarded as a comment upon his
audience, he is, on the contrary, aware only of be-having in the
performer’s role, as he understands that role. But the American
does not see it thus. To him, the “arrogant” behavior of the


Englishman appears to be directed against the audience, in which
case the implicit invocation of some abstract standard appears only
to add insult to injury.

Similarly, the behavior which an Englishman interprets as
“boastful” in an American is not aggressive, although the
Englishman may feel that he is being subjected to some sort of
invidious comparison. He does not know that, as a matter of fact,
Americans will only behave like this to people whom they rather
like and respect. According to the hypothesis above, the “boasting”
pattern results from the curious linkage whereby exhibition of self-
sufficiency and independence is played off against
overdependence. The American, when he boasts, is looking for
approval of his upstanding independence; but the naive
Englishman interprets this behavior as a bid for some sort of
dominance or superiority.

In this sort of way, we may suppose that the whole flavor of one
national culture may differ from that of another, and that such
differences may be considerable enough to lead to serious
misunderstandings. It is probable, however, that these differences
are not so complex in their nature as to be beyond the reach of
investigation. Hypotheses of the type which we have advanced
could be easily tested, and research on these lines is urgently

National Character and American Morale

Using the motifs of interpersonal and intergroup relation-ship as
our clues to national character, we have been able to indicate certain
orders of regular difference which we may expect to find among the
peoples who share our Western civilization. Of necessity, our
statements have been theoretical rather than empirical; still, from the
theoretical structure which we have built up, it is possible to extract
certain formulas which may be useful to the builder of morale.

All of these formulas are based upon the general assumption
that people will respond most energetically when the context is
structured to appeal to their habitual patterns of reaction. It is not

sensible to encourage a donkey to go up hill by offering him raw
meat, nor will a lion respond to grass.

(1) Sinnce all Western nations tend to think and behave in
bipolar terms, we shall do well, in building American morale, to
think of our various enemies as a single hostile entity. The
distinctions and gradations which intellectuals might prefer are
likely to be disturbing.
(2) Since both Americans and English respond most energetically
to symmetrical stimuli, we shall be very unwise if we
soft-pedal the disasters of war. If our enemies defeat us at any
point, that fact ought to be used to the maximum as a challenge
and a spur to further effort. When our forces have suffered some
reverse, our newspapers ought to be in no hurry to tell us that
“enemy advances have been checked.” Military progress is always
intermittent, and the moment to strike, the moment when
maximum morale is needed, occurs when the enemy is solidifying
his position and preparing the next blow. At such a moment, it is
not sensible to reduce the aggressive energy of our leaders and
people by smug re-assurance.
(3) There is, however, a superficial discrepancy between the
habit of symmetrical motivation and the need for showing self-
sufficiency. We have suggested that the American boy learns to
stand upon his own feet through those occasions in childhood
when his parents are approving spectators of his self-sufficiency. If
this diagnosis is correct, it would follow that a certain bubbling up
of self-appreciation is normal and healthy in Americans and is
perhaps an essential ingredient of American independence and
A too literal following of the formula above, therefore, a too
great insistence upon disasters and difficulties, might lead to some
loss of energy through the damming up of this spontaneous
exuberance. A rather concentrated diet of “blood, sweat, and tears”
may be good for the English; but Americans, while no less
dependent upon symmetrical motivation, cannot feel their oats when
fed on nothing but disaster. Our public spokesmen and newspaper
editors should never softpedal the fact that we have a man-sized job
on our hands, but they will do well to insist also that America is a
man-sized nation. Any sort of attempt to reassure Americans by


minimizing the strength of the enemy must be avoided, but frank
boasts of real success are good.

(4) Because our vision of the peace is a factor in our war-making
morale, it is worthwhile to ask at once what light the study of
national differences may throw upon the problems of the peace
We have to devise a peace treaty (a) such that Americans and
British will fight to achieve it, and (b) such that it will bring out the
best rather than the worst characteristics of our enemies. If we
approach it scientifically, such a problem is by no means beyond our

The most conspicuous psychological hurdle to be negotiated, in
imagining such a peace treaty, is the contrast between British and
American symmetrical patterns and the German complementary
pattern, with its taboo on overt sub-missive behavior. The allied
nations are not psychologically equipped to enforce a harsh treaty;
they might draw up such a treaty, but in six months they would tire
of keeping the underdog down. The Germans, on the other hand, if
they see their role as “submissive,” will not stay down without harsh
treatment. We have seen that these considerations applied even to
such a mildly punitive treaty as was devised at Versailles; the allies
omitted to enforce it, and the Germans refused to accept it. It is,
therefore, useless to dream of such a treaty, and worse than useless
to repeat such dreams as a way of raising our morale now, when we
are angry with Germany. To do that would only obscure the issues in
the final settlement.

This incompatibility between complementary and symmetrical
motivation means, in fact, that the treaty cannot be organized around
simple dominance-submissive motifs; hence we are forced to look
for alternative solutions. We must ex-amine, for example, the motif
of exhibitionism-spectatorship —what dignified role is each of the
various nations best fitted to play?—and that of succoringdependence—
in the starving postwar world, what motivational
patterns shall we evoke between those who give and those who
receive food? And, alternative to these solutions, we have the
possibility of some threefold structure, within which both the allies
and Germany would submit, not to each other, but to some abstract

Bali: The Value System of a Steady

“Ethos” and “Schismogenesis”

It would be an oversimplification—it would even be false —to
say that science necessarily advances by the construction and
empirical testing of successive working hypotheses. Among thephysicists and chemists there may be some who really proceed inthis oithoclox manner, but among the social scientists there is
perhaps not one. Our concepts are loosely defined—a haze of
chiaroscuro prefiguring sharper lines still undrawn—and our
hypotheses are still so vague that rarely can we imagine any crucialinstance whose investigation will test them.

The present paper is an attempt to make more precise an idea
which I published in 193620 and which has lain fallow since that
time. The notion of ethos had proved a useful conceptual tool for me,
and with it I had been able to get a sharper understanding of Iatmul
culture. But this experience by no means proved that this tool would
necessarily be useful in other hands or for the analysis of other
cultures. The most general conclusion I could draw was of this
order: that my own mental processes had certain characteristics;
that the sayings, actions, and organization of the Iatmul had certain
characteristics; and that the abstraction, “ethos,” performed some
role—catalytic, perhaps—in easing the relation between these two
specificities, my mind and the data which I myself had collected.

Immediately after completing the manuscript of Naven, I went to
Bali with the intention of trying upon Balinese data this tool which
had been evolved for the analysis of Iatmul. For one reason or
another, however, I did not do this, partly because in Bali Margaret

* This essay appeared in Social Structure: Studies Presented to A. R. Radcliffe-
Brown, edited by Meyer Fortes, 1949. It is reprinted by permission of the Clarendon
Press. Preparation of the essay was aided by a Guggenheim Fellowship.
20 G. Bateson, Naven, Cambridge, Cambridge University Press, 1936.


Mead and I were engaged in devising other tools—photographic
methods of record and description—and partly because I was
learning the techniques of applying genetic psychology to cultural
data, but more especially because at some inarticulate level I felt
that the tool was unsuitable for this new task.

It was not that ethos was in any sense disproved—indeed, a tool
or a method can scarcely be proved false. It can only be shown to be
not useful, and in this case there was not even a clear demonstration
of uselessness. The method remained almost untried, and the most I
could say was that, after that surrender to the data which is the first
step in all anthropological study, ethological analysis did not seem
to be the next thing to do.

It is now possible to show with Balinese data what peculiarities
of that culture may have influenced me away from ethological
analysis, and this demonstration will lead to a greater generalization
of the abstraction; ethos. We shall in the process make certain
heuristic advances which may guide us to more rigorous descriptive
procedures in dealing with other cultures.

(1) The analysis of Iatmul data led to the definition of ethos as
“The expression of a culturally standardized system of organization
of the instincts and emotions of the individuals.”21
(2) Analysis of Iatmul ethos—consisting in the ordering of data
so as to make evident certain recurrent “emphases” or “themes”—led
to recognition of schismogenesis. It appeared that the working of
latmul society involved inter alia two classes of regenerative22 or
“vicious” circles. Both of these were sequences of social interaction
21 Naven, p. 118.

22 The terms “regenerative” and “degenerative” are borrowed from communications
engineering. A regenerative or “vicious” circle is a chain of variables of the
general type: increase in A causes increase in B; increase in B causes increase in C;
.. increase in N causes increase in A. Such a system, if provided with the necessary
energy sources and if external factors permit, will clearly operate at a greater and
greater rate or intensity. A “degenerative” or “self-corrective” circle differs from a regenerative
circle in containing at least one link of the type: “increase in N causes
decrease in M.” The house thermostat or the steam engine with a governor are examples
of such self-correcting systems. It will be noted that in many instances the
same material circuit may be either regenerative or degenerative according to the
amount of loading, frequency of impulses transmitted around the path, and time
characteristics of the total path.


such that A’s acts were stimuli for B’s acts, which in turn became
stimuli for more intense action on the part of A, and so on, A and B
being persons acting either as individuals or as group members.

(3) These schismogenic sequences could be classified into two
classes: (a) symmetrical schismogenesis, where the mutually
promoting actions of A and B were essentially similar, e.g., in cases
of competition, rivalry, and the like; and ( b) complementary
schismogenesis, where the mutually promoting actions are
essentially dissimilar but mutually appropriate, e.g., in cases of
dominance-submission, succoring-dependence, exhibitionismspectatorship,
and the like.
(4) In 1939 a considerable advance was made in defining the
formal relations between the concepts of symmetrical and
complementary schismogenesis. This came from an attempt to state
schismogenic theory in terms of Richardson’s equations for
international armaments races23. The equations for rivalry evidently
gave a first approximation to what I had called “symmetrical
schismogenesis.” These equations assume that the intensity of A’s
actions (the rate of his arming, in Richardson’s case) is simply
proportional to the amount by which B is ahead of A. The stimulus
term in fact is (B —A), and when this term is positive it is expected
that A will en-gage in efforts to arm. Richardson’s second equation
makes the same assumption mutatis mutandis about B’s actions.
These equations suggested that other simply rivalrous or competitive
phenomena—e.g., boasting—though not subject to such simple
measurement as expenditure on armament, might yet when
ultimately measured be reducible to a simply analogous set of
The matter was, however, not so clear in the case of complementary
schismogenesis. Richardson’s equations for “submission”
evidently define a phenomenon somewhat different from a
progressive complementary relationship, and the form of his
equations describes the action of a factor “submissiveness” which
slows down and ultimately reverses the sign of warlike effort. What
was, however, required to describe complementary schismogenesis

23 L. F. Richardson, “Generalized Foreign Politics,” British Journal of
Psychology, Monograph Supplement xxiii, 1939.

was an equational form giving. a sharp and discontinuous reversal of
sign. Such an equational form is achieved by supposing A’s actions
in a complementary relationship to be proportional to a stimulus
term of the type (A —B) . Such a form has also the advantage of
automatically defining the actions of one of the participants as
negative, and thus gives some mathematical analogue for the
apparent psychological relatedness of domination to submission,
exhibitionism to spectatorship, succoring to dependence, etc.

Notably this formulation is itself a negative of the formulation
for rivalry, the stimulus term being the opposite. It had been
observed that symmetrical sequences of actions tend sharply to
reduce the strain of excessively complementary relationships
between persons or groups.24 It is tempting to ascribe this effect to
some hypothesis which would make the two types of
schismogenesis in some degree psychologically incompatible, as is
done by the above formulation.

(5) It is of interest to note that all the modes associated with the
erogenous zones,25 though not clearly quantifiable, define themes for
complementary relationship.
(6) The link with erogenous zones suggested in 5, above,
indicates that we ought, perhaps, not to think of simple rising
exponential curves of intensity limited only by factors analogous to
fatigue, such as Richardson’s equations would imply; but rather that
we should expect our curves to be bounded by phenomena
comparable to orgasm—that the achievement of a certain degree of
bodily or neural involvement or intensity may be followed by a
release of schismogenic tension. Indeed, all that we know about
human beings in various sorts of simple contests would seem to
24 Naven, p. 173.

25 E. H. Homburger, “Configurations in Play: Psycho-logical Notes,”
Psychoanalytical Quarterly, 1937, vi: 138-214. This paper, one of the most
important in the literature seeking to state psychoanalytic hypotheses in more
rigorous terms, deals with the “modes” appropriate to the various erogenous zones
—intrusion, incorporation, retention, and the like—and shows how these modes may
be transferred from one zone to another. This leads the writer to a chart of the
possible permutations and combinations of such transferred modalities. This chart
provides precise means of describing the course of the development of a large
variety of different types of character structure (e.g., as met with in different


indicate that this is the case, and that the conscious or unconscious
wish for release of this kind is an important factor which draws the
participants on and prevents them from simply withdrawing from
contests which would otherwise not commend themselves to
“common sense.” If there be any basic human characteristic which
makes man prone to struggle, it would seem to be this hope of
release from tension through total involvement. In the case of war
this factor is undoubtedly often potent. (The real truth—that in
modern warfare only a very few of the participants achieve this
climactic release—seems hardly to stand against the insidious myth
of “total” war.)

(7) In 1936 it was suggested that the phenomenon of “falling in
love” might be comparable to a schismogenesis with the signs
reversed, and even that “if the course of true love ever ran smooth it
would follow an exponential curve.”26 Richardson27 has since,
independently, made the same point in more formal terms.
Paragraph 6, above, clearly indicates that the “exponential curves”
must give place to some type of curve which will not rise
indefinitely but will reach a climax and then fall. For the rest,
however, the obvious relation-ship of these interactive phenomena
to climax and orgasm very much strengthens the case for regarding
schismogenesis and those cumulative sequences of interaction
which lead to love as often psychologically equivalent. (Witness the
curious confusions between fighting and lovemaking, the symbolic
identifications of orgasm with death, the recurrent use by mammals
of organs of offense as ornaments of sexual attracttion, etc.)
(8) Schismogenic sequences were not found in Bali. This
negative statement is of such importance and conflicts with so many
theories of social opposition and Marxian determinism that, in order
to achieve credibility, I must here de-scribe schematically the
process of character formation, the resulting Balinese character
structure, the exceptional in-stances in which some sort of
cumulative interaction can be recognized, and the methods by which
quarrels and status differentiation are handled. (Detailed analysis of
the various points and the supporting data cannot here be
26 Naven, p. 197.
27 Op. cit., 1939.


reproduced, but references will be given to published sources where
the data can be examined.)28

Balinese Character

(a) The most important exception to the above generalization
occurs in the relationship between adults (especially parents) and
children. Typically, the mother will start a small flirtation with the
child, pulling its penis or otherwise stimulating it to interpersonal
activity. This will excite the child, and for a few moments
cumulative interaction will occur. Then just as the child,
approaching some small climax, flings its arms around the mother’s
neck, her attention wanders. At this point the child will typically
start an alternative cumulative interaction, building up toward
temper tantrum. The mother will either play a spectator’s role,
enjoying the child’s tantrum, or, if the child actually attacks her, will
brush off his attack with no show of anger on her part. These sequences
can be seen either as an expression of the mother’s distaste
for this type of personal involvement or as context in which the
child acquires a deep distrust of such involvement. The perhaps
basically human tendency towards cumulative personal interaction
is thus muted.29 It is possible that some sort of continuing plateau
of intensity is substituted for climax as the child becomes more
fully adjusted to Balinese life. This cannot at present be clearly
documented for sexual relations, but there are indications that a
plateau type of sequence is characteristic for trance and for
quarrels (see d, below).
(b)Similar sequences have the effect of diminishing the child’s
tendencies toward competitive and rivalrous behavior. The mother
will, for example, tease the child by suckling the baby of some

28 See especially G. Bateson and M. Mead, Balinese Character: A Photographic
Analysis. Since this photo-graphic record is available, no photographs are included
in the present paper.

29 Balinese Character: A Photographic Analysis, pl. 47, and pp. 32-6.

other woman and will enjoy her own child’s efforts to push the
intruder from the breast.30

(c)In general the lack of climax is characteristic for Balinese
music, drama, and other art forms. The music typically has a
progression, derived from the logic of its formal structure, and
modifications of intensity determined by the duration and
progress of the working out of these formal relations. It does not
have the sort of rising intensity and climax structure characteristic
of modern Occidental music, but rather a formal progression.31

(d) Balinese culture includes definite techniques for dealing with
quarrels. Two men who have quarrelled will go formally to the
office of the local representative of the Rajah and will there register
their quarrel, agreeing that whichever speaks to the other shall pay a
fine or make an offering to the gods. Later, if the quarrel terminates,
this contract may be formally nullified. Smaller—but similar—
avoidances (pwik) are practiced, even by small children in their
quarrels. It is significant, perhaps, that this procedure is not an
attempt to influence the protagonists away from hostility and toward
friendship. Rather, it is a formal recognition of the state of their
mutual relationship, and possibly, in some sort, a pegging of the
relationship at that state. If this interpretation is correct, this method
of dealing with quarrels would correspond to the substitution of a
plateau for a climax.
(e) In regard to warfare, contemporary comment on the old wars
between the Rajahs indicates that in the period when the comments
were collected (1936–39) war was thought of as containing large
elements of mutual avoidance. The village of Bajoeng Gede was
surrounded by an old vallum and foss, and the people explained the
functions of these fortifications in the following terms: “If you and I
had a quarrel, then you would go and dig a ditch around your house.
Later I would come to fight with you, but I would find the ditch and
then there would be no fight”—a sort of mutual Maginot Line
psychology. Similarly the boundaries between neighboring
kingdoms were, in general, a deserted no-man’s land inhabited only
by vagrants and exiles. (A very different psychology of warfare was
30 lbid., pls. 49, 52, 53, and 69-72.

31 See Colin McPhee, “The Absolute Music of Bali,” Modern Music, 1935; and
A House in Bali, London, Gollancz, 1947.


no doubt developed when the kingdom of Karangasem embarked on
the conquest of the neighboring island of Lombok in the beginning
of the eighteenth century. The psychology of this militarism has not
been investigated, but there is reason to believe that the time
perspective of the Balinese colonists in Lombok is today significantly
different from that of Balinese in Bali.)32

(f) The formal techniques of social influence—oratory and the
like—are almost totally lacking in Balinese culture. To demand the
continued attention of an individual or to exert emotional influence
upon a group are alike distasteful and virtually impossible; because
in such circumstances the attention of the victim rapidly wanders.
Even such continued speech as would, in most cultures, be used for
the telling of stories does not occur in Bali. The narrator will,
typically, pause after a sentence or two, and wait for some member
of the audience to ask him a concrete question about some detail of
the plot. He will then answer the question and so resume his
narration. This procedure apparently breaks the cumulative tension
by irrelevant interaction.
(g) The principal hierarchical structures in the society—the caste
system and the hierarchy of full citizens who are the village council
—are rigid. There are no contexts in which one individual could
conceivably compete with another for position in either of these
systems. An individual may lose his membership in the hierarchy for
various acts, but his place in it cannot be altered. Should he later
return to orthodoxy and be accepted back, he will return to his
original position in relation to the other members.33
The foregoing descriptive generalizations are all partial answers
to a negative question—”Why is Balinese society nonschismogenic?”—
and from the combination of these generalizationswe arrive at a picture of a society differing very markedly from our
own, from that of the Iatmul, from those systems of social
opposition which Radcliffe-Brown has analyzed, and from any
social structure postulated by Marxian analysis.

32 See G. Bateson, “An Old Temple and a New Myth,” Djawa, xvii, Batavia, 1937.

33 See M. Mead, “Public Opinion Mechanisms among Primitive Peoples,”
Public Opinion Quarterly, 1937, is 5-16.


We started with the hypothesis that human beings have a
tendency to involve themselves in sequences of cumulative
interaction, and this hypothesis is still left virtually intact. Among
the Balinese the babies, at least, evidently have such tendencies. But
for sociological validity this hypothesis must now be guarded with a
parenthetical clause stipulating that these tendencies are operative in
the dynamics of society only if the childhood training is not such as
to prevent their expression in adult life.

We have made an advance in our knowledge of the scope of
human character formation in demonstrating that these tendencies
toward cumulative interaction are subject to some sort of
modification, deconditioning, or inhibition.34 And this is an
important advance. We know how it is that the Balinese are
nonschismogenic and we know how their distaste for schismogenic
patterns is expressed in various details of the social organization—
the rigid hierarchies, the institutions for the handling of quarrels, etc.
—but we still know nothing of the positive dynamics of the society.
We have answered only the negative question.

Balinese Ethos

The next step, therefore, is to ask about Balinese ethos. What
actually are the motives and the values which accompany the
complex and rich cultural activities of the Balinese? What, if not
competitive and other types of cumulative interrelationship, causes
the Balinese to carry out the elaborate patterns of their lives?

(1) It is immediately clear to any visitor to Bali that the driving
force for cultural activity is not either acquisitiveness or crude
physiological need. The Balinese, especially in the plains, are not
hungry or poverty-stricken. They are wasteful of food, and a very
considerable part of their activity goes into entirely nonproductive
activities of an artistic or ritual nature in which food and wealth are
34 As is usual in anthropology, the data are not sufficiently precise to give us any clue
as to the nature of the learning processes involved. Anthropology, at best, is only
able to raise problems of this order. The next step must be left for laboratory


lavishly expended. Essentially, we are dealing with an economy of
plenty rather than an economy of scarcity. Some, indeed, are rated
“poor” by their fellows, but none of these poor are threatened by
starvation, and the suggestion that human beings may actually starve
in great Occidental cities was, to the Balinese, unutterably shocking.

(2) In their economic transactions the Balinese show a great
deal of carefulness in their small dealings. They are “penny wise.”
On the other hand, this carefulness is counter-acted by occasional
“pound foolishness” when they will expend large sums of money
upon ceremonials and other forms of lavish consumption. There are
very few Balinese who have the idea of steadily maximizing their
wealth or property; these few are partly disliked and partly regarded
as oddities. For the vast majority the “saving of pennies” is done
with a limited time perspective and a limited level of aspiration.
They are saving until they have enough to spend largely on some
ceremonial. We should not describe Balinese economics in terms of
the individual’s attempt to maximize value, but rather compare it
with the relaxation oscillations of physiology and engineering,
realizing that not only is this analogy descriptive of their sequences
of transactions, but that they themselves see these sequences as
naturally having some such form.
(3) The Balinese are markedly dependent upon spatial
orientation. In order to be able to behave they must know their
cardinal points, and if a Balinese is taken by motor car over twisting
roads so that he loses his sense of direction, he may become
severely disorientated and unable to act (e.g., a dancer may become
unable to dance) until he has got back his orientation by seeing some
important landmark, such as the central mountain of the island
around which the cardinal points are structured. There is a
comparable dependence upon social orientation, but with this
difference: that where the spatial orientation is in a horizontal plane,
social orientation is felt to be, in the main, vertical. When two
strangers are brought together, it is necessary, before they can
converse with any freedom, that their relative caste positions be
stated. One will ask the other, “Where do you sit?” and this is a
metaphor for caste. It is asking, essentially, “Do you sit high or
low?” When each knows the caste of the other, each will then know

what etiquette and what linguistic forms he should adopt, and
conversation can then proceed. Lacking such orientation, a Balinese
is tongue-tied.

(4) It is common to find that activity (other than the “penny
wisdom” mentioned above) rather than being purposive, i.e., aimed
at some deferred goal, is valued for itself. The artist, the dancer, the
musician, and the priest may receive a pecuniary reward for their
professional activity, but only in rare cases is this reward adequate to
recompense the artist even for his time and materials. The reward is
a token of appreciation, it is a definition of the context in which the
theatrical company performs, but it is not the economic main-stay of
the troupe. The earnings of the troupe may be saved up to enable
them to buy new costumes, but when finally the costumes are
bought it is usually necessary for every member to make a
considerable contribution to the common fund in order to pay for
them. Similarly, in regard to the offerings which are taken to every
temple feast, there is no purpose in this enormous expenditure of
artistic work and real wealth. The god will not bring any benefit
because you made a beautiful structure of flowers and fruit for the
calendric feast in his temple, nor will he avenge your abstention.
Instead of deferred purpose there is an immediate and immanent
satisfaction in performing beautifully, with everybody else, that
which it is correct to perform in each particular context.
(5) In general there is evident enjoyment to be had from doing
things busily with large crowds of other people.35 Conversely there
is such misfortune inherent in the loss of group membership that the
threat of this loss is one of the most serious sanctions in the culture.
(6) It is of great interest to note that many Balinese actions are
articulately accounted for in sociological terms rather than in terms
of individual goals or values.36
This is most conspicuous in regard to all actions related to the
village council, the hierarchy which includes all full citizens. This
body, in its secular aspects, is referred to as I Desa (literally, “Mr.
Village”), and numerous rules and procedures are rationalized by
reference to this abstract personage. Similarly, in its sacred aspects,

35 Bateson and Mead, op. cit., p1. 5.

36 Cf. Naven, pp. 250 if., where it was suggested that we must expect to find that some
peoples of the world would relate their actions to the sociological frame.


the village is deified as Betara Desa (God Village), to whom shrines
are erected and offerings brought. (We may guess that a
Durkheimian analysis would seem to the Balinese to be an obvious
and appropriate approach to the understanding of much of their
public culture.)

In particular all money transactions which involve the village
treasury are governed by the generalization, “The village does not
lose” (Desanne sing dadi potjol). This generalization applies, for
example, in all cases in which a beast is sold from the village herd.
Under no circumstances can the village accept a price less than that
which it actually or nominally paid. (It is important to note that the
rule takes the form of fixing a lower limit and is not an injunction to
maximize the village treasury.)

A peculiar awareness of the nature of social processes is evident
in such incidents as the following: A poor man was about to undergo
one of the important and expensive rites de passage which are
necessary for persons as they approach the top of the council
hierarchy. We asked what would hap-pen if he refused to undertake
this expenditure. The first answer was that, if he were too poor, I
Desa would lend him the money. In response to further pressing as
to what would happen if he really refused, we were told that nobody
ever had refused, but that if somebody did, nobody would go
through the ceremony again. Implicit in this answer and in the fact
that nobody ever does refuse is the assumption that the ongoing
cultural process is itself to be valued.

(7) Actions which are culturally correct (patoet) are acceptable
and aesthetically valued. Actions which are permissible (dadi) are
of more or less neutral value; while actions which are not
permissible (sing dadi) are to be deprecated and avoided. These
generalizations, in their translated form, are no doubt true in many
cultures, but it is important to get a clear understanding of what the
Balinese mean by dadi. The notion is not to be equated with our
“etiquette” or “law,” since each of these invokes the value judgment
of some other person or sociological entity. In Bali there is no
feeling that actions have been or are categorized as dadi or sing dadi
by some human or supernatural authority. Rather, the statement that
such-and-such an action is dadi is an absolute generalization to the

effect that under the given circumstances this action is regular.37 It is
wrong for a casteless person to address a prince in other than the
“polished language,” and it is wrong for a menstruating woman to
enter a temple. The prince or the deity may express annoyance, but
there is no feeling that either the prince, the deity, or the casteless
per-son made the rules. The offense is felt to be against the order
and natural structure of the universe rather than against the actual
person offended. The offender, even in such serious matters as incest
(for which he may be extruded from the society)38 is not blamed for
anything worse than stupidity and clumsiness. Rather, he is “an
unfortunate person” (anak latfoer), and misfortune may come to
any of us “when it is our turn.” Further, it must be stressed that these
patterns which define correct and permissible behavior are
exceedingly complex (especially the rules of language) and that the
individual Balinese (even to some degree inside his own family) has
continual anxiety lest he make an error. Moreover, the rules are not
of such a kind that they can be summarized either in a simple recipe
or an emotional attitude. Etiquette cannot be deduced from some
comprehensive statement about the other person’s feelings or from
respect for superiors. The details are too complex and too various
for this, and so the individual Balinese is forever picking his way,
like a tightrope walker, afraid at any moment lest he make some

(8) The metaphor from postural balance used in the last
paragraph is demonstrably applicable in many contexts of Balinese
(a)The fear of loss of support is an important theme in
Balinese childhood.39
(b)Elevation (with its attendant problems of physical and
metaphorical balance) is the passive complement of respect.40
(c)The Balinese child is elevated like a superior per-son or agod.41

37 The word dadi is also used as a copula referring to changes in social status. I
Anoe dadi Koebajan means “So-and-so has become a village official.”

38 Mead, “Public Opinion Mechanisms among Primitive Peoples,” loc. cit., 1937.

39 Bateson and Mead, op. cit., pls. 17, 67, and 79.

40 Ibid., pls. 10-14.

41 Ibid., p1. 45.


(d)In cases of actual physical elevation42 the duty of
balancing the system falls on the supporting lower person, but
control of the direction in which the system will move is in the
hands of the elevated. The little girl in the figure standing in
trance on a man’s shoulders can cause her bearer to go wherever
she desires by merely leaning in that direction. He must then
move in that direction in order to maintain the balance of the

(e)A large proportion of our collection of 1200 Balinese
carvings shows preoccupation on the part of the artist with
problems of balance.43

(f)The Witch, the personification of fear, frequently uses agesture called kapar, which is described as that of a man falling
from a coconut palm on suddenly seeing a snake. In this
gesture the arms are raised sideways to a position some-whatabove the head.

(g)The ordinary Balinese term for the period before the
coming of the white man is “when the world was steady”

(doegas goemine enteg).

Applications of the Von Neumannian Game

Even this very brief listing of some of the elements in Balinese
ethos suffices to indicate theoretical problems of prime importance.
Let us consider the matter in abstract terms. One of the hypotheses
underlying most sociology is that the dynamics of the social
mechanism can be described by assuming that the individuals
constituting that mechanism are motivated to maximize certain
variables. In conventional economic theory it is assumed that the
individuals will maximize value, while in schismogenic theory it
was tacitly assumed that the individuals would maximize intangible
but still simple variables such as prestige, self-esteem, or even

42 Ibid., p1. 10, fig. 3.
43 At present it is not possible to make such a statement in sharply defined
quantitative terms, the available judgments being subjective and Occidental.

submissiveness. The Balinese, however, do not maximize any such
simple variables.

In order to define the sort of contrast which exists between the
Balinese system and any competitive system, let us start by
considering the premisses of a strictly competitive Von
Neumannian game and proceed by considering what changes we
must make in these premisses in order to approximate more closely
to the Balinese system.

(1) The players in a Von Neumannian game are, by hypothesis,
motivated only in terms of a single linear (sc. monetary) scale of
value. Their strategies are determined: (a) by the rules of the
hypothetical game; and (b) by their intelligence, which is, by
hypothesis, sufficient to solve all problems presented by the game.
Von Neumann shows that, under certain definable circumstances
depending upon the number of players and upon the rules,
coalitions of various sorts will be formed by the players, and in
fact Von Neumann’s analysis concentrates mainly upon the
structure of these coalitions and the distribution of value among
the members. In comparing these games with human societies we
shall regard social organizations as analogous to coalition systems.44
(2) Von Neumannian systems differ from human societies in the
following respects:
(a)His “players” are from the start completely intelligent, whereas
human beings learn. For human beings we must expect that the rules
of the game and the conventions associated with any particular set of
coalitions will become incorporated into the character structures of
the individual players.

44 Alternatively, we might handle the analogy in another way. A social system is, as
Von Neumann and Morgenstern point out, comparable to a non-zero sum game in
which one or more coalitions of people play against each other and against nature. The
non-zero sum characteristic is based on the fact that value is continually extracted from
the natural environment. Inasmuch as Balinese society exploits nature, the total entity,
including both environment and people, is clearly comparable to a game requiring
coalition between people. It is possible, however, that that subdivision of the total
game comprising the people only might be such that the formation of coalitions
within it would not be essential—that is, Balinese society may differ from most other
societies in that the “rules” of the relationship between people de-fine a “game” of the
type Von Neumann would call “non-essential.” This possibility is not here examined.
(See Von Neumann and Morgenstern, op. cit.)


(b)The mammalian value scale is not simple and mono-tone, but
may be exceedingly complex. We know, even at a physiological
level, that calcium will not replace vitamins, nor will an amino acid
replace oxygen. Further, we know that the animal does not strive to
maximize its supply of any of these discrepant commodities, but
rather is required to maintain the supply of each within tolerable
limits. Too much may be as harmful as too little. It is also doubtful
whether mammalian preference is always transitive.

(c) In the Von Neumannian system the number of moves in a
given “play” of a game is assumed to be finite. The strategic
problems of the individuals are soluble because the individual can
operate within a limited time perspective. He need only look
forward a finite distance to the end of the play when the gains and
losses will be paid up and every-thing will start again from a tabula
rasa. In human society life is not punctuated in this way, and each
individual faces a vista of unknowable factors whose number
increases (probably exponentially) into the future.
(d) The Von Neumannian players are, by hypothesis, not
susceptible either to economic death or to boredom. The losers can
go on losing forever, and no player can withdraw from the game,
even though the outcome of every play is definitely predictable in
probability terms.
(3) Of these differences between Von Neumannian and human
systems, only the differences in value scales and the possibility of
“death” concern us here. For the sake of simplicity we shall assume
that the other differences, though very profound, can for the moment
be ignored.
(4) Curiously, we may note that, although men are mammals and
therefore have a primary value system which is multidimensional
and nonmaximizing, it is yet possible for these creatures to be put
into contexts in which they will strive to maximize one or a few
simple variables (money, prestige, power, etc.) .
(5) Since the multidimensional value system is apparently
primary, the problem presented by, for example, Iatmul social
organization is not so much to account for the behavior of Iatmul
individuals by invoking (or abstracting) their value system; we
should also ask how that value system is imposed on the mammalian
individuals by the social organization in which they find themselves.

Conventionally in anthropology this question is attacked through
genetic psychology. We endeavor to collect data to show how the
value system implicit in the social prganization is built into the
character structure of the individuals in their childhood. There is,
however, an alternative approach which would momentarily ignore,
as Von Neumann does, the phenomena of learning and consider
merely the strategic implications of those contexts which must occur
in accordance with the given “rules” and the coalition system. In this
connection it is important to note that competitive contexts—
provided the individuals can be made to recognize the contexts as
competitive—inevitably reduce the complex gamut of values to very
simple and even linear and monotone terms.45 Considerations of this
sort, plus descriptions of the regularities in the process of character
formation, probably suffice to de-scribe how simple value scales
are imposed upon mammalian individuals in competitive societies
such as that of the Iatmul or twentieth-century America.

(6) In Balinese society, on the other hand, we find an. entirely
different state of affairs. Neither the individual nor the village is
concerned to maximize any simple variable. Rather, they would
seem to be concerned to maximize some-thing which we may call
stability, using this term perhaps in a highly metaphorical way.
(There is, in fact, one simple quantitative variable which does
appear to be maximized. This variable is the amount of any fine
imposed by the village. When first imposed the fines are mostly
very small, but if payment is delayed the amount of the fine is
increased very steeply, and if there be any sign that the offender is
refusing to pay—”opposing the village”—the fine is at once raised
to an enormous sum and the offender is deprived of membership in
the community until he is willing to give up his opposition. Then a
part of the fine may be excused.)
(7) Let us now consider an hypothetical system consisting of a
number of identical players, plus an umpire who is concerned with
the maintenance of stability among the players. Let us further
suppose that the players are liable to economic death, that our
umpire is concerned to see that this shall not occur, and that the
umpire has power to make certain alterations in the rules of the
45 L. K. Frank, “The Cost of Competition,” Plan Age, 1940, vi : 314-24.


game or in the probabilities associated with chance moves. Clearly
this umpire will be in more or less continual conflict with the
players. He is striving to maintain a dynamic equilibrium or steady
state, and this we may rephrase as the attempt to maximize the
chances against the maximization of any single simple variable.

(8.) Ashby has pointed out in rigorous terms that the steady state
and continued existence of complex interactive systems depend
upon preventing the maximization of any variable, and that any
continued increase in any variable will inevitably result in, and be
limited by, irreversible changes in the system. He has also pointed
out that in such systems it is very important to permit certain
variables to alter.46 The steady state of an engine with a governor is
unlikely to be maintained if the position of the balls of the
governor is clamped. Similarly a tightrope walker with a balancing
pole will not be able to maintain his balance except by varying the
forces which he exerts upon the pole.

(9) Returning now to the conceptual model suggested in
paragraph 7, let us take one further step toward making this model
comparable with Balinese society. Let us substitute for the umpire a
village council composed of all the players. We now have a system
which presents a number of analogies to our balancing acrobat.
When they speak as members of the village council, the players by
hypothesis are interested in maintaining the steady state of the
system—that is, in preventing the maximization of any simple
variable the excessive increase of which would produce irreversible
change. In their daily life, however, they are still engaged in simple
competitive strategies.
(10) The next step toward making our model resemble Balinese
society more closely is clearly to postulate in the character structure
of the individuals and/or in the contexts of their daily life those
factors which will motivate them toward maintenance of the steady
state not only when they speak in council, but’also in their other
interpersonal relations. These factors are in fact recognizable in Bali
and have been enumerated above. In our analysis of why Balinese
society is nonschismogenic, we noted that the Balinese child learns
46 W. R. Ashby, “Effect of Controls on Stability,” Nature, clv, no. 3930,
February 24, 1945, 242-43.


to avoid cumulative interaction, i.e., the maximization of certain
simple variables, and that the social organization and contexts of
daily life are so constructed as to preclude competitive interaction.
Further, in our analysis of the Balinese ethos, we noted recurrent
valuation: (a) of the clear and static definition of status and spatial
orientation, and (b) of balance and such movement as will conduce
to balance.

In sum it seems that the Balinese extend to human relationships
attitudes based upon bodily balance, and that they generalize the
idea that motion is essential to balance. This last point gives us, I
believe, a partial answer to the question of why the society not
only continues to function but functions rapidly and busily,
continually undertaking ceremonial and artistic tasks which are not
economically or competitively determined. This steady state is
maintained by continual nonprogressive change.

Schismogenic System versus the Steady State

I have discussed two types of social system in such schematic
outline that it is possible to state clearly a contrast between them.
Both types of system, so far as they are capable of maintaining
themselves without progressive or irreversible change, achieve the
steady state. There are, how-ever, profound differences between
them in the manner in which the steady state is regulated.

The Iatmul system, which is here used as a prototype of
schismogenic systems, includes a number of regenerative causal
circuits or vicious circles. Each such circuit consists of two or more
individuals (or groups of individuals) who participate in potentially
cumulative interaction. Each human individual is an energy source
or “relay,” such that the energy used in his responses is not derived
from the stimuli but from his own metabolic processes. It therefore
follows that such a schismogenic system is—unless controlled—
liable to excessive increase of those acts which characterize the
schismogeneses. The anthropologist who attempts even a qualitative
description of. such a system must therefore identify: (1) the
individuals and groups involved in schismogenesis and the routes of


communication between them; (2) the categories of acts and
contexts characteristic of the schismogeneses; (3) the processes
whereby the individuals become psychologically apt to perform
these acts and/or the nature of the contexts which force these acts
upon them; and lastly, (4) he must identify the mechanisms or
factors which control the schismogeneses. These controlling factors
may be of at least three distinct types: (a) degenerative causal loops
may be superposed upon the schismogeneses so that when the latter
reach a certain intensity some form of restraint is a p p l i e d as
occurs in Occidental systems when a government intervenes to limit
economic competition; (b) there may be, in addition to the
schismogeneses already considered, other cumulative interactions
acting in an opposite sense and so promoting social integration
rather than fission; (c) the increase in schismogenesis may be
limited by factors which are internally or externally environmental
to the parts of the schismogenic circuit. Such factors which have
only small restraining effect at low intensities of schismogenesis
may increase with increase of intensity. Friction, fatigue, and
limitation of energy supply would be examples of such factors.

In contrast with these schismogenic systems, Balinese society is
an entirely different type of mechanism, and in de-scribing it the
anthropologist must follow entirely different procedures, for which
rules cannot as yet be laid down. Since the class of
“nonschismogenic” social systems is defined only in negative terms,
we cannot assume that members of the class will have common
characteristics. In the analysts of the Balinese system, however, the
following steps occurred, and it is possible that some at least of
these may be applicable in the analysis of other cultures of this
class: (1) it was observed that schismogenic sequences are rare in
Bali; (2) the exceptional cases in which such sequences occur were
investigated; (3) from this investigation it appeared, (a) that in
general the contexts which recur in Balinese social life preclude
cumulative interaction and ( b) that childhood experience trains the
child away from seeking climax in personal interaction; (4) it was
shown that certain positive values—related to balance—recur in the
culture and are incorporated into the character structure during
child-hood, and, further, that these values may be specifically related
to the steady state; (5) a more detailed study is now required to

arrive at a systematic statement about the self-correcting
characteristics of the system. It is evident that the ethos alone is
insufficient to maintain the steady state. From time to time the
village or some other entity does step in to correct infractions. The
nature of these instances of the working of the corrective mechanism
must be studied; but it is clear that this intermittent mechanism is
very different from the continually acting restraints which must be
present in all schismogenic systems.


Style, Grace, and Information in
Primitive Art*


This paper consists of several still-separate attempts to map a
theory associated with culture and the nonverbal arts. Since no one
of these attempts is completely successful, and since the attempts do
not as yet meet in the middle of the territory to be mapped, it. may
be useful to state, in non-technical language, what it is I am after.

Aldous Huxley used to say that the central problem for humanity
is the quest for grace. This word he used in what he thought was the
sense in, which it is used in the New Testament. He explained the
word, however, in his own terms. He argued—like Walt Whitman—
that the communication and behavior of animals has a naivete, a
simplicity, which man has lost. Man’s behavior is corrupted by
deceit—even self-deceit—by purpose, and by self-consciousness. As
Aldous saw the matter, man has lost the “grace” which animals still

In terms of this contrast, Aldous argued that God resembles the
animals rather than man: He is ideally unable to deceive and
incapable of internal confusions.

In the total scale of beings, therefore, man is as if displaced
sideways and lacks that grace which the animals have and which
God has.

I argue that art is a part of man’s quest for grace; some-times his
ecstasy in partial success, sometimes his rage and agony at failure.

I argue also that there are many species of grace within the
major genus; and also that there are many kinds of failure and

* This essay was a position paper for the Wenner-Gren Conference on Primitive
Art, 1967. It is here reprinted from A Study of Primitive Art, edited by Dr.
Anthony Forge, to be published by Oxford University Press, by permission of the

frustration and departure from grace. No doubt each culture has its
characteristic species of grace toward which its artists strive, and
its own species of failure.

Some cultures may foster a negative approach to this difficult
integration, an avoidance of complexity by crass preference either
for total consciousness or total unconsciousness. Their art is
unlikely to be “great.”

I shall argue that the problem of grace is fundamentally a
problem of integration and that what is to be integrated is the
diverse parts of the mind—especially those multiple levels of
which one extreme is called “consciousness” and the other the
“unconscious.” For the attainment of grace, the reasons of the heart
must be integrated with the reasons of the reason.

Edmund Leach has confronted us, in this conference, with the
question: How is it that the art of one culture can have meaning or
validity for critics raised in a different culture? My answer would
be that, if art is somehow expressive of something like grace or
psychic integration, then the success of this expression might well
be recognizable across cultural barriers. The physical grace of cats
is profoundly different from the physical grace of horses, and yet a
man who has the physical grace of neither can evaluate that of

And even when the subject matter of art is the frustration of
integration, cross-cultural recognition of the products of this
frustration is not too surprising.

The central question is: In what form is information about
psychic integration contained or coded in the work of art?


Style and Meaning

They say that “every picture tells a story,” and this generalization
holds for most of art if we exclude “mere” geometric ornamentation.
But I want precisely to avoid analyzing the “story.” That aspect of
the work of art which can most easily be reduced to words—the
mythology connected with the subject matter—is not what I want to
discuss. I shall not even mention the unconscious mythology of
phallic symbol-ism, except at the end.

I am concerned with what important psychic information is in
the art object quite apart from what it may “represent.” “Le style
est l’homme meme” (“The style is the man him-self”) (Buffon).
What is implicit in style, materials, composition, rhythm, skill, and
so on?

Clearly this subject matter will include geometrical ornamentation
along with the composition and stylistic aspects of more
representational works.

The lions in Trafalgar Square could have been eagles or
bulldogs and still have carried the same (or similar) messages
about empire and about the cultural premises of nineteenth-century
England. And yet, how different might their message have been
had they been made of wood!

But representationalism as such is relevant. The extremely
realistic horses and stags of Altamira are surely not about the same
cultural premises as the highly conventionalized black outlines of a
later period. The code whereby perceived objects or persons (or
supernaturals) are transformed into wood or paint is a source of
information about the artist and his culture.

It is the very rules of transformation that are of interest to me—
not the message, but the code.

My goal is not instrumental. I do not want to use the
transformation rules when discovered to undo the transformation
or to “decode” the message. To translate the art object into
mythology and then examine the mythology would be only a neat
way of dodging or negating the problem of “what is art?”

I ask, then, not about the meaning of the encoded message but
rather about the meaning of the code chosen. But still that most
slippery word “meaning” must be defined.

It will be convenient to define meaning in the most general
possible way in the first instance.

“Meaning” may be regarded as an approximate synonym of
pattern, redundancy, information, and “restraint,” within a
paradigm of the following sort:

Any aggregate of events or objects (e.g., a sequence of
phonemes, a painting, or a frog, or a culture) shall be said to
contain “redundancy” or “pattern” if the aggregate can be divided
in any way by a “slash mark,” such that an observer perceiving
only what is on one side of the slash mark can guess, with better
than random success, what is on the other side of the slash mark.
We may say that what is on one side of the slash contains
information or has meaning about what is on the other side. Or, in
engineer’s language, the aggregate contains “redundancy.” Or,
again, from the point of view of a cybernetic observer, the
information available on one side of the slash will restrain (i.e.,
reduce the probability of) wrong guessing. Examples:

The letter T in a given location in a piece of written English
prose proposes that the next letter is likely to be an H or an R or a
vowel. It is possible to make a better than random guess across a
slash which immediately follows the T. English spelling contains

From a part of an English sentence, delimited by a slash, it is
possible to guess at the syntactic structure of the remainder of the

From a tree visible above ground, it is possible to guess at the
existence of roots below ground. The top provides information
about the bottom.

From an arc of a drawn circle, it is possible to guess at the
position of other parts of the circumference. (From the diameter of
an ideal circle, it is possible to assert the length of the
circumference. But this is a matter of truth within a tautological
system. )

From how the boss acted yesterday, it may be possible to guess
how he will act today.


From what I say, it may be possible to make predictions about
how you will answer. My words contain meaning or information
about your reply.

Telegraphist A has a written message on his pad and sends this
message over wire to B, so that B now gets the same sequence of
letters on his message pad. This transaction (or “language game” in
Wittgenstein’s phrase) has created a redundant universe for an
observer O. If 0 knows what was on A’s pad, he can make a better
than random guess at what is on B’s pad.

The essence and raison d’ etre of communication is the creation
of redundancy, meaning, pattern, predictability, information, and/or
the reduction of the random by “restraint.”

It is, I believe, of prime importance to have a conceptual system
which will force us to see the “message” (e.g., the art object) as
both itself internally patterned and itself a part of a larger patterned
universe—the culture or some part of it.

The characteristics of objects of art are believed to be about, or
to be partly derived from, or determined by, other characteristics of
cultural and psychological systems. Our problem might therefore
he oversimply represented by the diagram:

[Characteristic, of art object/Characteristics of rest of culture]

where square brackets enclose the universe of relevance, and
where the oblique stroke represents a slash across which some
guessing is possible, in one direction or in both. The problem, then,
is to spell out what sorts of relationships, correspondences, etc.,
cross or transcend this oblique stroke.

Consider the case in which I say to you, “It’s raining,” and you
guess that if you look out the window you will see raindrops. A
similar diagram will serve:

[Characteristics of “It’s raining”/Perception of raindrops]

Notice, however, that this case is by no means simple. Only if
you know the language and have some trust in my veracity will you
be able to make a guess about the rain-drops. In fact, few people in
this situation restrain them-selves from seemingly duplicating their
information by looking out of the window. We like to prove that our

guesses are right, and that our friends are honest. Still more
important, we like to test or verify the correctness of our view of our
relationship to others.

This last point is nontrivial. It illustrates the necessarily
hierarchic structure of all communicational systems: the fact of
conformity or nonconformity (or indeed any other relationship)
between parts of a patterned whole may itself be informative as
part of some still larger whole. The matter may he diagrammed

[(“It’s raining”/raindrop.)/Yom—me relationship)

where redundancy across the slash mark within the smaller
universe enclosed in round brackets proposes (is a message about) a
redundancy in the larger universe enclosed in square brackets.

But the message “It’s raining” is itself conventionally coded and
internally patterned, so that several slash marks could be drawn
across the message indicating patterning within the message itself.

And the same is true of the rain. It, too, is patterned and
structured. From the direction of one drop, I could predict the
direction of others. And so on.

But the slash marks across the verbal message “It’s raining” will
not correspond in any simple way to the slash marks across the

If, instead of a verbal message, I had given you a picture of the
rain, some of the slashes on the picture would have corresponded
with slashes on the perceived rain.

This difference provides a neat formal criterion to separate the
“arbitrary” and digital coding characteristic of the verbal part of
language from the iconic coding of depiction.

But verbal description is often iconic in its larger structure. A
scientist describing an earthworm might start at the head end and
work down its length—thus producing a description iconic in its
sequence and elongation. Here again we observe a hierarchic
structuring, digital or verbal at one level and iconic at another.

Levels and Logical Types


“Levels” have been mentioned: (a) It was noted that the
combination of the message “It’s raining” with the perception of
raindrops can itself constitute a message about a universe of
personal relations; and (b) that when we change our focus of
attention from smaller to larger units of message material, we may
discover that a larger unit contains iconic coding though the smaller
parts of which it was made are verbal: the verbal description of an
earthworm may, as a whole, be elongated.

The matter of levels now crops up in another form which is
crucial for any epistemology of art:

The word “know” is not merely ambiguous in covering both
connaitre (to know through the senses, to recognize or perceive) and
savoir (to know in the mind), but varies —actively shifts— in
meaning for basic systemic reasons. That which we know through
the senses can become knowledge in the mind.

“I know the way to Cambridge” might mean that I have studied
the map and can give you directions. It might mean that I can recall
details all along the route. It might mean that when driving that
route I recognize many details even though I could recall only a few.
It might mean that when driving to Cambridge I can trust to “habit”
to make me turn at the right points, without having to think where I
am going. And so on.

In all cases, we deal with a redundancy or patterning of a quite
complex sort:

[(“I know . . .”/my mind)//the road]

and the difficulty is to determine the nature of the patterning
within the round brackets, or, to put the matter another way: what
parts of the mind are redundant with the particular message about

Last, there is a special form of “knowing” which is usually
regarded as adaptation rather than information. A shark is beautifully
shaped for locomotion in water, but the genome of the shark surely
does not contain direct information about hydrodynamics. Rather,
the genome must be supposed to contain information or instructions
which are the complement of hydrodynamics. Not hydrodynamics,
but what hydrodynamics requires, has been built up in the shark’s
genome. Similarly, a migratory bird perhaps does not know the way

to its destination in any of the senses outlined above, but the bird
may contain the complementary instructions necessary td cause it to
fly right.

“Le coeur a ses raisons que la raison ne connait point” (“The
heart has its reasons which the reason does not at all perceive”). It is
this—the complex layering of consciousness and unconsciousness—
that creates difficulty when we try to discuss art or ritual or
mythology. The matter of levels of the mind has been discussed
from many points of view, at least four of which must be mentioned
and woven into any scientific approach to art:

(1) Samuel Butler’s insistence that the better an organism
“knows” something, the less conscious it becomes of its knowledge,
i.e., there is a process whereby knowledge (or “habit” —whether of
action, perception, or thought) sinks to deeper and deeper levels of
the mind. This phenomenon, which is central to Zen discipline (cf.
Herrigel, Zen in the Art of Archery), is also relevant to all art and all
(2) Adalbert Ames’ demonstrations that the conscious, three-
dimensional visual images, which we make of that which we see,
are made by processes involving mathematical premises of
perspective, etc., of the use of which we are totally unconscious.
Over these processes, we have no voluntary control. A drawing of
a chair with the perspective of van Gogh affronts the conscious
expectations and, dimly, reminds the consciousness of what had
been (unconsciously) taken for granted.
(3) The Freudian (especially Fenichel’s) theory of dreams as
metaphors coded according to primary process. I shall consider
style—neatness, boldness of contrast, etc.—as metaphoric and
therefore as linked to those levels of the mind where primary
process holds sway.
(4) The Freudian view of the unconscious as the cellar or
cupboard to which fearful and painful memories are con-signed by
a process of repression.
Classical Freudian theory assumed that dreams were a secondary
product, created by “dream work.” Material unacceptable to
conscious thought was supposedly translated into the metaphoric
idiom of primary process to avoid waking the dreamer. And this may
be true of those items of information which are held in the


unconscious by the process of repression. As we have seen,
however, many other sorts of information are inaccessible to
conscious inspection, including most of the premises of mammalian
interaction. It would seem to me sensible to think of these items as
existing primarily in the idiom of primary process, only with
difficulty to be translated into “rational” terms. In other words, I
believe that much of early Freudian theory was upside down. At that
time many thinkers regarded conscious reason as normal and self-
explanatory while the unconscious was regarded as mysterious,
needing proof, and needing explanation. Repression was the
explanation, and the unconscious was filled with thoughts which
could have been conscious but which repression and dream work
had distorted. Today we think of consciousness as the mysterious,
and of the computational methods of the unconscious, e.g., primary
process, as continually active, necessary, and all-embracing.

These considerations are especially relevant in any at-tempt to
derive a theory of art or poetry. Poetry is not a sort of distorted and
decorated prose, but rather prose is poetry which has been stripped
down and pinned to a Procrustean bed of logic. The computer men
who would program the translation of languages sometimes forget
this fact about the primary nature of language. To try to construct a
machine to translate the art of one culture into the art of another
would be equally silly.

Allegory, at best a distasteful sort of art, is an inversion of the
normal creative process. Typically an abstract relation, e.g., between
truth and justice, is first conceived in rational terms. The
relationship is then metaphorized and dolled up to look like a
product of primary process. The abstractions are personified and
made to participate in a pseudomyth, and so on. Much advertising
art is allegorical in this sense, that the creative process is inverted.

In the cliche system of Anglo-Saxons, it is commonly assumed
that it would be somehow better if what is unconscious were made
conscious. Freud, even, is said to have said, “Where id was, there
ego shall be,” as though such an increase in conscious knowledge
and control would be both possible and, of course, an improvement.
This view is the product of an almost totally distorted epistemology
and a totally distorted view of what sort of thing a man, or any other
organism, is.

Of the four sorts of unconsciousness listed above, it is very clear
that the first three are necessary. Consciousness, for obvious
mechanical reasons,47 must always be limited to a rather small
fraction of mental process. If useful at all, it must therefore be
husbanded. The unconsciousness associated with habit is an
economy both of thought and of consciousness; and the same is true
of the inaccessability of the processes of perception. The conscious
organism does not require (for pragmatic purposes) to know how it
perceives —only to know what it perceives. (To suggest that we
might operate without a foundation in primary process would be to
suggest that the human brain ought to be differently structured.) Of
the four types, only the Freudian cupboard for skeletons is perhaps
undesirable and could be obviated. But there may still be advantages
in keeping the skeleton off the dining room table.

In truth, our life is such that its unconscious components are
continuously present in all their multiple forms. It follows that in our
relationships we continuously exchange messages about these
unconscious materials, and it becomes important also to exchange
metamessages by which we tell each other what order and species of
unconsciousness (or consciousness) attaches to our messages.

In a merely pragmatic way, this is important because the orders
of truth are different for different sorts of messages. Insofar as a
message is conscious and voluntary, it could be deceitful. I can tell
you that the cat is on the mat when in fact she is not there. I can tell
you “I love you” when in fact I do not. But discourse about
relationship is commonly accompanied by a mass of semivoluntary
kinesic and autonomic signals which provide a more trustworthy
comment on the verbal message.

Similarly with skill, the fact of skill indicates the presence of
large unconscious components in the performance.

It thus becomes relevant to look at any work of art with the
question: What components of this message material had what
orders of unconsciousness (or consciousness) for the artist? And this

47 Consider the impossibility of constructing a television set which would report upon
its screen all the workings of its component parts, including especially those parts
concerned in this reporting.


question, I believe, the sensitive critic usually asks, though perhaps
not consciously.

Art becomes, in this sense, an exercise in communicating about
the species of unconsciousness. Or, if you prefer it, a sort of play
behavior whose function is, amongst other things, to practice and
make more perfect communication of this kind.

I am indebted to Dr. Anthony Forge for a quotation from Isadora
Duncan: “If I could tell you what it meant, there would be no point
in dancing it.”

Her statement is ambiguous. In terms of the rather vulgar
premises of our culture, we would translate the statement to mean:
“There would then be no point in dancing it, be-cause I could tell it
to you, quicker and with less ambiguity, in words.” This
interpretation goes along with the silly idea that it would be a good
thing to be conscious of everything of which we are unconscious.

But there is another possible meaning of Isadora Duncan’s
remark: If the message were the sort of message that could be
communicated in words, there would be no point in dancing it, but it
is not that sort of message. It is, in fact, precisely the sort of message
which would be falsified if communicated in words, because the use
of words (other than poetry) would imply that this is a fully
conscious and voluntary message, and this would be simply untrue.

I believe that what Isadora Duncan or any artist is trying to
communicate is more like: “This is a particular sort of partly
unconscious message. Let us engage in this particular sort of partly
unconscious communication.” Or perhaps: “This is a message about
the interface between conscious and unconscious.

The message of skill of any sort must always be of this kind.
The sensations and qualities of skill can never be put in words, and
yet the fact of skill is conscious.

The artist’s dilemma is of a peculiar sort. He must practice in
order to perform the craft components of his job. But to practice has
always a double effect. It makes him, on the one hand, more able to
do whatever it is he is attempting; and, on the other hand, by the
phenomenon of habit formation, it makes him less aware of how he
does it.

If his attempt is to communicate about the unconscious
components of his performance, then it follows that he is on a sort of

moving stairway (or escalator) about whose position he is trying to
communicate but whose movement is itself a function of his efforts
to communicate.

Clearly, his task is impossible, but, as has been remarked, some
people do it very prettily.

Primary Process

“The heart has its reasons which the reason does not at all
perceive.” Among Anglo-Saxons, it is rather usual to think of the
“reasons” of the heart or of the unconscious as inchoate forces or
pushes or heavings—what Freud called Trieben. To Pascal, a
Frenchman, the matter was rather different, and he no doubt thought
of the reasons of the heart as a body of logic or computation as
precise and complex as the reasons of consciousness.

(I have noticed that Anglo-Saxon anthropologists some-times
misunderstand the writings of Claude Levi-Strauss for precisely
this reason. They say he emphasizes too much the intellect and
ignores the “feelings.” The truth is that he assumes that the heart
has precise algorithms.)

These algorithms of the heart, or, as they say, of the unconscious,
are, however, coded and organized in a manner totally
different from the algorithms of language. And since a great deal
of conscious thought is structured in terms of the logics of
language, the algorithms of the unconscious are doubly
inaccessible. It is not only that the conscious mind has poor access
to this material, but also the fact that when such access is achieved,
e.g., in dreams, art, poetry, religion, intoxication, and the like,
there is still a formidable problem of translation.

This is usually expressed in Freudian language by saying that
the operations of the unconscious are structured in terms of
primary process, while the thoughts of consciousness (especially
verbalized thoughts) are expressed in secondary process.

Nobody, to my knowledge, knows anything about secondary
process. But it is ordinarily assumed that everybody knows all


about it, so I shall not attempt to describe secondary process in any
detail, assuming that you know as much about it as I.

Primary process is characterized (e.g., by Fenichel) as lacking
negatives, lacking tense, lacking in any identification of linguistic
mood (i.e., no identification of indicative, subjunctive, optative,
etc.) and metaphoric. These characterizations are based upon the
experience of psychoanalysts, who must interpret dreams and the
patterns of free association.

It is also true that the subject matter of primary-process discourse
is different from the subject matter of language and consciousness.
Consciousness talks about things or persons, and attaches predicates
to the specific things or persons which have been mentioned. In
primary process the things or persons are usually not identified, and
the focus of the discourse is upon the relationships which are
asserted to obtain between them. This is really only another way of
saying that the discourse of primary process is metaphoric. A
metaphor retains unchanged the relationship which it “illustrates”
while substituting other things or persons for the relata. In a simile,
the fact that a metaphor is being used is marked by the insertion of
the words “as if” or “like.” In primary process (as in art) there are
no markers to indicate to the conscious mind that the message
material is metaphoric.

(For a schizophrenic, it is a major step towards a more
conventional sanity when he can frame his schizophrenic utterances
or the comments of his voices in an “as if” terminology.)

The focus of “relationship” is, however, somewhat more narrow
than would be indicated merely by saying that primary-process
material is metaphoric and does not identify the specific relata.
The subject matter of dream and other primary-process material is,
in fact, relationship in the more narrow sense of relationship
between self and other persons or between self and the

Anglo-Saxons who are uncomfortable with the idea that
feelings and emotions are the outward signs of precise and
complex algorithms usually have to be told that these matters, the
relationship between self and others, and the relationship between
self and environment, are, in fact, the subject matter of what are
called “feelings”—love, hate, fear, confidence, anxiety, hostility,

etc. It is unfortunate that these abstractions referring to patterns of
relationship have received names, which are usually handled in
ways that assume that the “feelings” are mainly characterized by
quantity rather than by precise pattern. This is one of the
nonsensical contributions of psychology to a distorted

Be all that as it may, for our present purposes it is important to
note that the characteristics of primary process as described above
are the inevitable characteristics of any communicational system
between organisms who must use only iconic communication. This
same limitation is characteristic of the artist and of the dreamer
and of the prehuman mammal or bird. (The communication of
insects is, perhaps, an-other matter.)

In iconic communication, there is no tense, no simple negative,
no modal marker.

The absence of simple negatives is of especial interest be-cause
it often forces organisms into saying the opposite of what they
mean in order to get across the proposition. that they mean the
opposite of what they say.

Two dogs approach each other and need to exchange the
message: “We are not going to fight.” But the only way in which
fight can be mentioned in iconic communication is by the showing
of fangs. It is then necessary for the dogs to discover that this
mention of fight was, in fact, only exploratory. They must,
therefore, explore what the showing of fangs means. They
therefore engage in a brawl; discover that neither ultimately
intends to kill the other; and, after that, they can be friends.

(Consider the peace-making ceremonials of the Andaman
Islanders. Consider also the functions of inverted statement or
sarcasm, and other sorts of humor in dream, art, and mythology.)

In general, the discourse of animals is concerned with relationship
either between self and other or self and environment. In
neither case is it necessary to identify the relata. Animal A tells B
about his relationship with B and he tells C about his relationship
with C. Animal A does not have to tell animal C about his
relationship with B. Always the relata are perceptibly present to
illustrate the discourse, and always the discourse is iconic in the
sense of being composed of part actions (“intention movements”)


which mention the whole action which is being mentioned. Even
when the cat asks you for milk, she cannot mention the object
which she wants (unless it be perceptibly present). She says,
“Mama, mama,” and you are supposed from this invocation of dependency
to guess that it is milk that she requires.

All this indicates that primary-process thoughts and the
communication of such thoughts to others are, in an evolutionary
sense, more archaic than the more conscious operations of language,
etc. This has implications for the whole economics and dynamic
structure of the mind. Samuel Butler was perhaps first to point out
that that which we know best is that of which we are least conscious,
i.e., that the process of habit formation is a sinking of knowledge
down to less conscious and more archaic levels. The unconscious
contains not only the painful matters which consciousness prefers to
not inspect, but also many matters which are so familiar that we do
not need to inspect them. Habit, therefore, is a major economy of
conscious thought. We can do things without consciously thinking
about them. The skill of an artist, or rather his demonstration of a
skill, becomes a message about these parts of his unconsciousness.
(But not perhaps a message from the unconscious.)

But the matter is not quite so simple. Some types of knowledge
can conveniently be sunk to unconscious levels, but other types
must be kept on the surface. Broadly, we can afford to sink those
sorts of knowledge which continue to be true regardless of changes
in the environment, but we must maintain in an accessible place all
those controls of behavior which must be modified for every
instance. The lion can sink into his unconscious the proposition
that zebras are his natural prey, but in dealing with any particular
zebra he must be able to modify the movements of his attack to fit
with the particular terrain and the particular evasive tactics of the
particular zebra.

The economics of the system, in fact, pushes organisms toward
sinking into the unconscious those generalities of relationship
which remain permanently true and toward keeping within the
conscious the pragmatics of particular instances.

The premises may, economically, be sunk, but particular
conclusions must be conscious. But the “sinking,” though economical,
is still done at a price—the price of inaccessibility. Since

the level to which things are sunk is characterized by iconic
algorithms and metaphor, it becomes difficult for the organism to
examine the matrix out or which his conscious conclusions spring.
Conversely, we may note that what is common to a particular
statement and a corresponding metaphor is of a generality
appropriate for sinking.

Quantitative Limits of Consciousness

A very brief consideration of the problem shows that it is not
conceivably possible for any system to be totally conscious.
Suppose that on the screen of consciousness there are reports from
many parts of the total mind, and consider the addition to
consciousness of those reports necessary to cover what is, at a given
stage of evolution, not already covered. This addition will involve a
very great increase in the circuit structure of the brain but still will
not achieve total coverage. The next step will be to cover the
processes and events occurring in the circuit structure which we
have just added. And so on.

Clearly, the problem is insoluble, and every next step in the
approach to total consciousness will involve a great in-crease in
the circuitry required.

It follows that all organisms must be content with rather little
consciousness, and that if consciousness has any useful functions
whatever (which has never been demonstrated but is probably
true), then economy in consciousness will be of the first
importance. No organism can afford to be conscious of matters
with which it could deal at unconscious levels.

This is the economy achieved by habit formation.

Qualitative Limits of Consciousness

It is, of course, true for the TV set that a satisfactory picture on
the screen is an indication that many parts of the machine are


working as they should; and similar considerations apply to the
“screen” of consciousness. But what is provided is only a very
indirect report of the working of all those parts. If the TV suffers
from a blown tube, or the man from a stroke, effects of this
pathology may be evident enough on the screen or to consciousness,
but diagnosis must still be done by an expert.

This matter has bearings upon the nature of art. The TV which
gives a distorted or otherwise imperfect picture is, in a sense,
communicating about its unconscious pathologies—exhibiting its
symptoms; and one may ask whether some artists are not doing
something similar. But this still won’t do.

It is sometimes said that the distortions of art (say, van Gogh’s
“Chair”) are directly representative of what the artist “sees.” If such
statements refer to “seeing” in the simplest physical sense (e.g.,
remediable with spectacles), I presume that they are nonsense. If van
Gogh could only see the chair in that wild way, his eyes would not
serve properly to guide him in the very accurate placing of paint on
canvas. And, conversely, a photographically accurate representation
of the chair on the canvas would also be seen by van Gogh in the
wild way. Re would see no need to distort the painting.

But suppose we say that the artist is painting today what he saw
yesterday—or that he is painting what he somehow knows that he
might see. “I see as well as you do—but do you realize that this
other way of seeing a chair exists as a human potentiality? And that
that potentiality is always in you and in me?” Is he exhibiting
symptoms which he might have, because the whole spectrum of
psychopathology is possible for us all?

Intoxication by alcohol or drugs may help us to see a distorted
world, and these distortions may be fascinating in that we recognize
the distortions as ours. In vino pars veritatis. We can be humbled or
aggrandized by realizing that this, too, is a part of the human self, a
part of Truth. But intoxication does not increase skill—at best it
may release skill previously acquired.

Without skill is no art.

Consider the case of the man who goes to the blackboard —or to
the side of his cave—and draws, freehand, a perfect reindeer in its
posture of threat. He cannot tell you about the drawing of the
reindeer (“If he could, there would be no point in drawing it”). “Do

you know that his perfect way of seeing—and drawing—a reindeer
exists as a human potentiality?” The consummate skill of the
draftsman validates the artist’s message about his relationship to the
animal—his empathy.

(They say the Altamira things were made for sympathetic
hunting magic. But magic only needs the crudest sort of representations.
The scrawled arrows which deface the beautiful
reindeer may have been magical—perhaps a vulgar attempt to
murder the artist, like moustaches scrawled on the Mona Lisa.)

The Corrective Nature of Art

It was noted above that consciousness is necessarily selective and
partial, i.e., that the content of consciousness is, at best, a small part
of truth about the self. But if this part be selected in any systematic
manner, it is certain that the partial truths of consciousness will be,
in aggregate, a distortion of the truth of some larger whole.

In the case of an iceberg, we may guess, from what is above
surface, what sort of stuff is below; but we cannot make the same
sort of extrapolation from the content of consciousness. It is not
merely the selectivity of preference, whereby the skeletons
accumulate in the Freudian unconscious, that makes such
extrapolation unsound. Such a selection by preference would only
promote optimism.

What is serious is the crosscutting of the circuitry of the mind. If,
as we must believe, the total mind is an integrated network (of
propositions, images, processes, neural pathology, or what have you
—according to what scientific language you prefer to use), and if the
content of consciousness is only a sampling of different parts and
localities in this net-work; then, inevitably, the conscious view of the
network as a whole is a monstrous denial of the integration of that
whole. From the cutting of consciousness, what appears above the
surface is arcs of circuits instead of either the complete circuits or
the larger complete circuits of circuits.

What the unaided consciousness (unaided by art, dreams, and the
like) can never appreciate is the systemic nature of mind.


This notion can conveniently be illustrated by an analogy: the
living human body is a complex, cybernetically integrated system.
This system has been studied by scientists—mostly medical men—
for many years. What they now know about the body may aptly be
compared with what the unaided consciousness knows about the
mind. Being doctors, they had purposes: to cure this and that. Their
research efforts were therefore focused (as attention focuses the
consciousness) upon those short trains of causality which they could
manipulate, by means of drugs or other intervention, to correct more
or less specific and identifiable states or symptoms. Whenever they
discovered an effective “cure” for something, research in that area
ceased and attention was directed elsewhere. We can now prevent
polio, but nobody knows much more about the systemic aspects of
that fascinating disease. Research on it has ceased or is, at best,
confined to improving the vaccines.

But a bag of tricks for curing or preventing a list of specified
diseases provides no overall wisdom. The ecology and population
dynamics of the species has been disrupted; parasites have been
made immune to antibiotics; the relationship between mother and
neonate has been almost destroyed; and so on.

Characteristically, errors occur wherever the altered causal chain
is part of some large or small circuit structure of system. And the
remainder of our technology (of which medical science is only a
part) bids fair to disrupt the rest of our ecology.

The point, however, which I am trying to make in this paper is
not an attack on medical science but a demonstration of an
inevitable fact: that mere purposive rationality unaided by such
phenomena as art, religion, dream, and the like, is necessarily
pathogenic and destructive of life; and that its virulence springs
specifically from the circumstance that life depends upon
interlocking circuits of contingency, while consciousness can see
only such short arcs of such circuits as human purpose may direct.

In a word, the unaided consciousness must always involve man
in the sort of stupidity of which evolution was guilty when she urged
upon the dinosaurs the common-sense values of an armaments race.
She inevitably realized her mistake a million years later and wiped
them out.

Unaided consciousness must always tend toward hate; not only
because it is good common sense to exterminate the other fellow,
but for the more profound reason that, seeing only arcs of circuits,
the individual is continually surprised and necessarily angered when
his hardheaded policies re-turn to plague the inventor.

If you use DDT to kill insects, you may succeed in reducing the
insect population so far that the insectivores will starve. You will
then have to use more DDT than be-fore to kill the insects which the
birds no longer eat. More probably, you will kill off the birds in the
first round when they eat the poisoned insects. If the DDT kills off
the dogs, you will have to have more police to keep down the
burglars. The burglars will become better armed and more cunning
… and so on.

That is the sort of world we live in—a world of circuit structures
—and love can survive only if wisdom (i.e., a sense or recognition
of the fact of circuitry) has an effective voice.

What has been said so far proposes questions about any
particular work of art somewhat different from those which have
been conventionally asked by anthropologists. The “culture and
personality school,” for example, has traditionally used pieces of art
or ritual as samples or probes to reveal particular psychological
themes or states.

The question has been: Does the art tell us about what sort of
person made it? But if art, as suggested above, has a positive
function in maintaining what I called “wisdom,” i.e., in correcting a
too purposive view of life and making the view more systemic, then
the question to be asked of the given work of art becomes: What
sorts of correction in the direction of wisdom would be achieved by
creating or viewing this work of art?

The question becomes dynamic rather than static.

Analysis of Balinese Painting

Turning now from the consideration of epistemology to a specific
art style, we note first what is most general and most obvious.


With almost no exceptions, the behaviors called art or their
products (also called art) have two characteristics: they require or
exhibit skill, and they contain redundancy or pattern.

But those two characteristics are not separate: the skill is first in
maintaining and then in modulating the redundancies.

The matter is perhaps most clear where the skill is that of the
journeyman and the redundancy is of comparatively low order. For
example, in the Balinese painting by Ida Bagus Djati Sura of the
village of Batuan, 1937 and in almost all painting of the Batuan
school, skill of a certain elementary but highly disciplined sort was
exercised or practiced in the background of foliage. The
redundancies to be achieved involve rather uniform and rhythmical
repetition of leaf forms, but this redundancy is, so to speak, fragile.
It would be broken or interrupted by smudges or irregularities of
size or tone in the painting of the successive leaves.

When a Batuan artist looks at the work of another, one of the first
things he examines is the technique of the leafy background. The
leaves are first drawn, in free outline in pencil; then each outline is
tightly redefined with pen and black ink. When this has been done
for all the leaves, the artist begins to paint with brush and Chinese
ink. Each leaf is covered with a pale wash. When these washes are
dry, each leaf receives a smaller concentric wash and after this
another still smaller, and so on. The final result is a leaf with an almost
white rim inside the inked outline, and successive steps of
darker and darker color toward the center of the leaf.

A “good” picture has up to five or six such successive washes on
every leaf. (This particular painting is not very “good” in this sense.
The leaves are done in only three or four steps.)

The skill and the patterning so far discussed depend upon
muscular rote and muscular accuracy—achieving the perhaps not
negligible artistic level of a well-laid out field of turnips.

I was watching a very gifted American carpenter-architect at
work on the woodwork of a house he had designed. I commented on
the sureness and accuracy of each step. He said, “Oh, that. That’s
only like using a typewriter. You have to be able to do that without

But on top of this level of redundancy is another. The uniformity
of the lower-level redundancy must be modulated to give higher

orders of redundancy. The leaves in one area must be different from
the leaves in another area, and these differences must be in some
way mutually redundant: they must be part of a larger pattern.

Indeed, the function and necessity of the first-level control is
precisely to make the second level possible. The perceiver of the
work of art must receive information that the artist can paint a
uniform area of leaves because without this information he will not
be able to accept as significant the variations in that uniformity.

Only the violinist who can control the quality of his notes can
use variations of that quality for musical purposes.

This principle is basic and accounts, I suggest, for the almost
universal linkage in aesthetics between skill and pattern. The
exceptions—e.g., the cult of natural landscapes, “found objects,”
inkblots, scattergrams, and the works of Jackson Pollock—seem to
exemplify the same rule in reverse. In these cases, a larger
patterning seems to propose the illusion that the details must have
been controlled. Inter-mediate cases also occur: e.g., in Balinese
carving, the natural grain of the wood is rather frequently used to
suggest de-tails of the form or surface of the subject. In these cases,
the skill lies not in the draftsmanship of the details, but in the artist’s
placement of his design within the three-dimensional structure of the
wood. A special “effect” is achieved, not by the mere
representationalism, but by the perceiver’s partial awareness that a
physical system other than that of draftsman-ship has contributed to
determine his perception.

We now turn to more complex matters, still concentrating
attention upon the most obvious and elementary.


(1) The delineation of leaves and other forms does not reach to
the edge of the picture but shades off into darkness so that almost all
around the rectangle there is a band of undifferentiated dark
pigment. In other words, the picture is framed within its own fade-
out. We are allowed to feel that the matter is in some sense “out of

this world”; and this in spite of the fact that the scene depicted is
familiar—the starting out of a cremation procession.

(2) The picture is filled. The composition leaves no open spaces.
Not only is none of the paper left unpainted, but no considerable
area is left in uniform wash. The largest such areas are the very dark
patches at the bottom between the legs of the men.
To Occidental eyes, this gives an effect of “fussiness.” To
psychiatric eyes, the effect is of “anxiety” or “compulsivity.” We
are all familiar with the strange look of those letters from cranks,
who feel that they must fill the page.

(3) But before trying too fast to diagnose or evaluate, we have to
note that the composition of the lower half of the picture, apart from
this filling of background space, is turbulent. Not merely a depiction
of active figures, but a swirling composition mounting upwards and
closed off by the contrasting direction of the gestures of the men at
the top of the pyramid.
The upper half of the picture, in contrast, is serene. Indeed, the
effect of the perfectly balanced women with offerings on their heads
is so serene that, at first glance, it appears that the men with musical
instruments must surely be sitting. (They are supposed to be moving
in procession.)

But this compositional structure is the reverse of the usual
Occidental. We expect the lower part of a picture to be the more
stable and expect to see action and movement in the upper part—if

(4) At this point, it is appropriate to examine the picture as a
sexual pun and, in this connection, the internal evidence for sexual
reference is at least as strong as it is in the case of the Tangaroa
figure discussed by Leach. All you have to do is to set your mind
in the correct posture and you will see an enormous phallic object
(the cremation tower) with two elephants’ heads at the base. This
object must pass through a narrow entrance into a serene courtyard
and thence onward and upward through a still more narrow
passageway. Around the base of the phallic object you see a
turbulent mass of homunculi, a crowd in which
Was none who would be foremost To
lead such dire attack;

But those behind cried “Forward!” And
those before cried “Back!”

And if you are so minded, you will find that Macaulay’s poem
about how Horatius kept the bridge is no less sexual than the present
picture. The game of sexual interpretation is easy if you want to play
it. No doubt the snake in the tree _ to the left of the picture could
also be woven into the sexual story.

It is still possible, however, that something is added to our
understanding of a work of art by the hypothesis that the subject
matter is double: that the picture represents both the start of a
cremation procession and a phallus with vagina. With a little
imagination, we could also see the picture as a symbolic
representation of Balinese social organization in which the smooth
relations of etiquette and gaiety metaphorically cover the turbulence
of passion. And, of course, “Horatius” is very evidently an idealized
myth of nineteenth-century imperial England.

It is probably an error to think of dream, myth, and art as being
about any one matter other than relationship. As was mentioned
earlier, dream is metaphoric and is not particularly about the relata
mentioned in the dream. In the conventional interpretation of dream,
another set of relata, often sexual, is substituted for the set in the
dream. But perhaps by doing this we only create another dream.
There indeed is no a priori reason for supposing that the sexual
relata are any more primary or basic than any other set.

In general, artists are very unwilling to accept interpretations of
this sort, and it is not clear that their objection is to the sexual nature
of the interpretation. Rather, it seems that rigid focusing upon any
single set of relata destroys for the artist the more profound
significance of the work. If the picture were only about sex or only
about social organization, it would be trivial. It is nontrivial or
profound precisely because it is about sex and social organization
and cremation, and other things. In a word, it is only about relationship
and not about any identifiable relata.

(5) It is appropriate then to ask how the artist has handled the
identification of his subject matter within the picture. We note first
that the cremation tower which occupies almost one-third of the
picture is almost invisible. It does not stand out against its

background as it should if the artist wanted to assert unequivocally
“this is a cremation.” Notably also, the coffin, which might be
expected to be a focal point, is appropriately placed just below the
center but even so does not catch the eye. In fact, the artist has
inserted details which label the picture as a cremation scene but
these details become almost whimsical asides, like the snake and the
little birds in the trees. The women are carrying the ritually correct
offerings on their heads, and two men appropriately bring bamboo
containers of palm toddy, but these details, too, are only whimsically
added. The artist plays down the subject identification and thereby
gives major stress to the contrast between the turbulent and the
serene mentioned in 3, above.

(6) In sum, it is my opinion that the crux of the picture is the
interwoven contrast between the serene and the turbulent. And a
similar contrast or combination was also present, as we have seen, in
the painting of the leaves. There, too, an exuberant freedom was
overlaid by precision.
In terms of this conclusion, I can now attempt an answer to the
question posed above: What sorts of correction, in the direction of
systemic wisdom, could be achieved by creating or viewing this
work of art? In final analysis, the picture can be seen as an
affirmation that to choose either turbulence or serenity as a human
purpose would be a vulgar error. The conceiving and creating of the
picture must have provided an experience which exposed this error.
The unity and integration of the picture assert that neither of these
contrasting poles can be chosen to the exclusion of the other,
because the poles are mutually dependent. This profound and
general truth is simultaneously asserted for the fields of sex, social
organization, and death.

Comment on Part II

Since World War II, it has been fashionable to engage in
“interdisciplinary” research, and this usually means, for example,
that an ecologist will need a geologist to tell him about the rocks and
soils of the particular terrain which he is investigating. But there is
another sense in which scientific work may claim to be

The man who studies the arrangement of leaves and branches in
the growth of a flowering plant may note an analogy between the
formal relations between stems, leaves, and buds, and the formal
relations that obtain between different sorts of words in a sentence.
He will think of a “leaf” not as something flat and green but as
something related in a particular way to the stem from which it
grows and to the secondary stem (or bud) which is formed in the
angle between leaf and primary stem. Similarly the modern linguist
thinks of a “noun” not as the “name of a person, place, or thing,” but
as a member of a class of words de-fined by their relationship in
sentence structure to “verbs” and other parts.

Those who think first of the “things” which are related (the
“relata”) will dismiss any analogy between grammar and the
anatomy of plants as far-fetched. After all, a leaf and a noun do not
at all resemble each other in outward appearance. But if we think
first of the relationships and consider the relata as defined solely by
their relationships, then we begin to wonder. Is there a profound
analogy between grammar and anatomy? Is there an
interdisciplinary science which should concern itself with such
analogies? What would such a science claim as its subject matter?
And why should we expect such far-flung analogies to have

In dealing with any analogy, it is important to define exactly
what is claimed when we say that the analogy is meaningful. In the
present example, it is not claimed that a noun should look like a leaf.
It is not even claimed that the relation between leaf and stem is the
same as the relation between noun and verb. What is claimed is,
first, that in both anatomy and grammar the parts are to be classified


according to the relations between them. In both fields, the relations
are to be thought of as somehow primary, the relata as secondary.
Beyond this, it is claimed that the relations are of the sort generated
by processes of information ex-change.

In other words, the mysterious and polymorphic relation between
context and content obtains in both anatomy and linguistics; and
evolutionists of the nineteenth century, preoccupied with what were
called “homologies,” were, in fact, studying precisely the contextual
structures of biological development.

All of this speculation becomes almost platitude when we realize
that both grammar and biological structure are products of
communicational and organizational process. The anatomy of the
plant is a complex transform of genotypic instructions, and the
“language” of the genes, like any other language, must of necessity
have contextual structure. More-over, in all communication, there
must be a relevance between the contextual structure of the message
and some structuring of the recipient. The tissues of the plant could
not “read” the genotypic instructions carried in the chromosomes of
every cell unless cell and tissue exist, at that given moment, in a
contextual structure.

What has been said above will serve as sufficient definition of
what is here meant by “form and pattern.” The focus of discussion
was upon form rather than content, upon context rather than upon
what occurs “in” the given con-text, upon relationship rather than
upon the related per-sons or phenomena.

The essays included range from a discussion of “schismogenesis”
(1935) to two essays written after the birth of cybernetics.

In 1935, I certainly had not clearly grasped the central
importance of “context.” I thought that the processes of schismogenesis
were important and nontrivial because in them I seemed
to see evolution at work: if interaction between persons could
undergo progressive qualitative change as in-tensity increased, then
surely this could be the very stuff of cultural evolution. It followed
that all directional change, even in biological evolution and
phylogeny, might—or must —be due to progressive interaction
between organisms. Under natural selection, such change in
relationships would favor progressive change in anatomy and

The progressive increase in size and armament of the dinosaurs
was, as I saw it, simply an interactive armaments race—a
schismogenic process. But I could not then see that the evolution
of the horse from Eohip pus was not a one-sided adjustment to
life on grassy plains. Surely the grassy plains themselves were
evolved pari passe with the evolution of the teeth and hooves of
the horses and other ungulates. Turf was the evolving response of
the vegetation to the evolution of the horse. It is the context
which evolves.

The classification of schismogenic process into “symmetrical”
and “complementary” was already a classification of con-texts of
behavior; and, already in this essay, there is a proposal to exmine
the possible combinations of themes in complementary behavior.
By 1942, I had completely for-gotten this old proposal, but I
attempted to do precisely what I had proposed seven years
previously. In 1942 many of us were interested in “national
character” and the con-, trast between England and America
fortunately brought into focus the fact that “spectatorship” is in
England a filial characteristic, linked with dependency and
submission, while in America spectatorship is a parental
characteristic linked with dominance and succoring.

This hypothesis, which I called “end-linkage,” marked a turning
point in my thinking. From that time on, I have consciously
focused upon the qualitative structure of con-texts rather than
upon intensity of interaction. Above all, the phenomena of end-
linkage showed that contextual structures could themselves be
messages—an important point which is not made in the 1942
article. An Englishman when he is applauding another is
indicating or signaling potential submission and/or dependency;
when he shows off or demands spectatorship, he is signaling.
dominance or superiority; and so on. Every Englishman who
writes a book must be guilty of this. For the American, the
converse must hold. His boasting is but a bid for quasiparental approval.

The notion of context reappears in the essay “Style, Grace, and
Information in Primitive Art,” but here the idea of context has
evolved to meet the related ideas of “redundancy,” “pattern,” and


Part III: Form and Pathology in

Social Planning andthe Concept ofDeutero-Learning*

Let me take as focus for this comment the last item1 in Dr.
Mead’s summary of her paper. To the layman who has not occupied
himself with the comparative study of human cultures, this
recommendation may appear strange; it may appear to be an ethical
or philosophical paradox, a suggestion that we discard purpose in

* This-article was my comment on Margaret Mead’s article “The Comparative
Study of Culture and the Purposive Cultivation of Democratic Values,” published as
Chapter IV of Science, Philosophy and Religion, Second Symposium, copyright
1942 by the Conference on Science, Philosophy and Religion, New York. It is here
reprinted by permission of the Conference and of Harper & Row, Inc.
I have italicized a parenthesis in footnote 5 which pre-figures the concept of the
“double bind.”

1 Dr. Mead writes: “. . those students who have de-voted themselves to studying
cultures as wholes, as systems of dynamic equilibrium, can make the following
contributions : . . .

“4. Implement plans for altering our present culture by recognizing the
importance of including the social scientist within his experimental material, and by
recognizing that by working toward defined ends we commit ourselves to the
manipulation of persons, and therefore to the negation of democracy. Only by
working in terms of values which are limited to defining a direction is it possible
for us to use scientific methods in the control of the process without the negation of
the moral autonomy of the human spirit.” (Italics hers.)


order to achieve our purpose; it may even call to mind some of the
basic aphorisms of Christianity and Taoism. Such aphorisms are
familiar enough; but the layman will be a little surprised to find
them coming from a scientist and dressed in all the paraphernalia
of analytic thought. To other anthropologists and social scientists,
Dr. Mead’s recommendations will be even more surprising, and
perhaps more meaningless, be-cause instrumentality and
“blueprints” are an essential ingredient in the whole structure of
life as science sees it. Likewise, to those in political life, Dr.
Mead’s recommendation will be strange, since they see decisions
as classifiable into policy-making decisions versus executive
decisions. The governors and the scientists alike (not to mention
the commercial world) see human affairs as patterned upon purpose,
means and ends, connation and satisfaction.

If anybody doubts that we tend to regard purpose and
instrumentality as distinctively human, let him consider the old
quip about eating and living. The creature who “eats to live” is the
highest human; he who “lives to eat” is coarser-grained, but still
human; but if he just “eats and lives,” without attributing
instrumentality or a spurious priority in time sequence to either
process, he is rated only among the animals, and some, less kind,
will regard him as vegetable.

Dr. Mead’s contribution consists in this—that she, fortified by
comparative study of other cultures, has been able to
transcend the habits of thought current in her own culture
and has been able to say virtually this: “Before we apply
social science to our own national affairs, we must re-examine and
change our habits of thought on the subject of means and ends. We
have learnt, in our cultural setting, to classify behavior into
`means’ and `ends’ and if we go on . defining ends as separate from
means and apply the social sciences as crudely instrumental
means, using the recipes of science to manipulate people, we shall
arrive at a totalitarian rather than a democratic system of life.” The
solution which she offers is that we look for the “direction,” and
“values” implicit in the means, rather than looking ahead to a
blueprinted goal and thinking of this goal as justifying or not
justifying manipulative means. We have to find the value of a
planned act implicit in and simultaneous with the act itself, not

separate from it in the sense that the act would derive its value
from reference to a future end or goal. Dr. Mead’s paper is, in fact,
not a direct preachment about ends and means; she does not say
that ends either do or do not justify the means. She is talking not
directly about ends and means, but about the way we tend to think
about ways and means, and about the dangers inherent in our habit
of thought.

It is specifically at this level that the anthropologist has most to
contribute to our problems. It is his task to see the highest common
factor implicit in a vast variety of human phenomena, or inversely,
to decide whether phenomena which appear to be similar are not
intrinsically different. He may go to one South Sea community,
such as the Manus, and there find that though everything that the
natives do is concretely different from our own behavior, yet the
underlying system of motives is rather closely comparable with
our own love of caution and wealth accumulation; or again he may
go to another society such as Bali and there find that, while the
outward appearance of the native religion is closely comparable
with our own—kneeling to pray, incense, intoned utterances
punctuated by a bell, etc.—the basic emotional attitudes are
fundamentally different. In Balinese religion we find an approval
accorded to rote, nonemotional performance of certain acts instead
of the insistence upon correct emotional attitude, characteristic of
Christian churches.

In every case the anthropologist is concerned not with mere
description but with a slightly higher degree of abstraction, a wider
degree of generalization. His first task is the meticulous collection
of masses of concrete observations of native life—but the next step
is not a mere summarizing of these data; it is rather to interpret the
data in an abstract language which shall transcend and comprehend
the vocabulary and notions explicit and implicit in our own
culture. It is not possible to give a scientific description of a native
culture in English words; the anthropologist must devise a more
abstract vocabulary in terms of which both our own and the native
culture can be described.

This then is the type of discipline which has enabled Dr. Mead to
point out that a discrepancy—a basic and fundamental discrepancy
—exists between “social engineering,” manipulating people in order


to achieve a planned blue-print society, and the ideals of democracy,
the “supreme worth and moral responsibility of the individual
human per-son.” The two conflicting motifs have long been implicit
in our culture, science has had instrumental leanings since before the
Industrial Revolution, and emphasis upon individual worth and
responsibility is even older. The threat of conflict between the two
motifs has only come recently, with increasing consciousness of,
and emphasis upon, the democratic motif and simultaneous spread
of the instrumental motif. Finally, the conflict is now a life-or-death
struggle over the role which the social sciences shall play in the
ordering of human relationships. It is hardly an exaggeration to say
that this war is ideologically about just this—the role of the social
sciences. Are we to reserve the techniques and the right to
manipulate people as the privilege of a few planning, goal-oriented,
and power-hungry individuals, to whom the instrumentality of
science makes a natural appeal? Now that we have the techniques,
are we, in cold blood, going to treat people as things? Or what are
we going to do with these techniques?

The problem is one of very great difficulty as well as urgency,
and it is doubly difficult because we, as scientists, are deeply
soaked in habits of instrumental t h o u g h t those of us, at least, for
whom science is a part of life, as well as a beautiful and dignified
abstraction. Let us try to surmount this additional source of
difficulty by turning the tools of science upon this habit of
instrumental thought and upon the new habit which Dr. Mead
envisages—the habit which looks for “direction” and “value” in the
chosen act, rather than in defined goals. Clearly, both of these
habits are ways of looking at time sequences. In the old jargon of
psychology, they represent different ways of apperceiving
sequences of behavior, or in the newer jargon of gestalt
psychology, they might both be described as habits of looking for
one or another sort of contextual frame for behavior. The problem
which Dr. Mead—who advocates a change in such habits—raises
is the problem of how habits of this abstract order are learned.

This is not the simple type of question which is posed in most
psychological laboratories, “Under what circumstances will a dog
learn to salivate in response to a bell?” or, “What variables govern
success in rote learning?” Our question is one degree more abstract,

and, in a sense, bridges the gap between the experimental work on
simple learning and the approach of the gestalt psychologists. We
are asking, “How does the dog acquire a habit of punctuating or apperceiving
the infinitely complex stream of events (including his
own behavior) so that this stream appears to be made up of one type
of short sequences rather than an-other?” Or, substituting the
scientist for the dog, we might ask, “What circumstances determine
that a given scientist will punctuate the stream of events so as to
conclude that all is predetermined, while another will see the stream
of events as so regular as to be susceptible of control?” Or, again, on
the same level of abstraction let us ask—and this question is very
relevant to the promotion of democracy—”What circumstances
promote that specific habitual phrasing of the universe which we
call `free will’ and those others which we call `responsibility,’
`constructiveness,’ `energy,’ `passivity,’ `dominance,’ and the rest?”
For all these abstract qualities, the essential stock-in-trade of the
educators, can be seen as various habits of punctuating the stream of
experience so that it takes on one or another sort of coherence and
sense. They are abstractions which begin to assume some
operational meaning when we see them take their place on a
conceptual level between the statements of simple learning and
those of gestalt psychology.

We can, for example, put our finger very simply on the process
which leads to tragedy and disillusion whenever men decide that the
“end justifies the means” in their efforts to achieve either a Christian
or a blueprinted heaven-on-earth. They ignore the fact that in social
manipulation, the tools are not hammers and screwdrivers. A
screwdriver is not seriously affected when, in an emergency, we use
it as a wedge; and a hammer’s outlook on life is not affected because
we sometimes use its handle as a simple lever. But in social
manipulation our tools are people, and people learn, and they
acquire habits which are more subtle and pervasive than the tricks
which the blueprinter teaches them. With the best intentions in the
world, he may train children to spy upon their parents in order to
eradicate some tendency prejudicial to the success of his blueprint,
but because the children are people they will do more than learn
this simple trick—they will build this experience into their whole
philosophy of life; it will color all their future attitudes to-ward


authority. Whenever they meet certain sorts of con-text, they will
tend to see these contexts as structured on an earlier familiar
pattern. The blueprinter may derive an initial advantage from the
children’s tricks; but the ultimate success of his blueprint may be
destroyed by the habits of mind which were learned with the trick.
(Unfortunately, there is no reason to believe that the Nazi blueprint
will break down for these reasons. It is probable that the unpleasant
attitudes here referred to are envisaged as basic both to
the plan itself and to the means of achieving it. The road to hell
can also be paved with bad intentions, though well-intentioned
people find this hard to believe.)

We are dealing, apparently, with a sort of habit which is a by-
product of the learning process. When Dr. Mead tells us that we
should leave off thinking in terms of blue-prints and should instead
evaluate our planned acts in terms of their immediate implicit
value, she is saying that in the upbringing and education of
children, we ought to try to inculcate a sort of by-product habit
rather different from that which we acquired and which we daily
reinforce in ourselves in our contacts with science, politics,
newspapers, and so on.

She states perfectly clearly that this new shift in the emphasis
or gestalt of our thinking will be a setting forth into uncharted
waters. We cannot know what manner of human beings will result
from such a course, nor can we be sure that we ourselves would
feel at home in the world of 1980. Dr. Mead can only tell us that if
we proceed on the course which would seem most natural,
planning our applications of social science as a means of attaining
a defined goal, we shall surely hit a rock. She has charted the rock
for us, and advises that we embark on a course in a direction where
the rock is not; but in a new, still uncharted direction. Her paper
raises the question of how we are to chart this new direction.

Actually, science can give us-something approaching a chart. I
indicated above that we might see a mixed bunch of abstract terms
—free will, predestination, responsibility, constructiveness,
passivity, dominance, etc.—as all of them descriptive of
apperceptive habits, habitual ways of looking at the stream of
events of which our own behavior is a part, and further that these
habits might all be, in some sense, by-products of the learning

process. Our next task, if we are to achieve some sort of chart, is
clearly to get something better than a random list of these possible
habits. We must reduce this list to a classification which shall show
how each of these habits is systematically related to the others.

We meet in common agreement that a sense of individual
autonomy, a habit of mind somehow related to what I have called
“free will,” is an essential of democracy, but we are still not
perfectly clear as to how this autonomy should be defined
operationally. What, for example, is the relation between
“autonomy” and compulsive negativism? The gas stations which
refuse to conform to the curfew—are they or are they not showing
a fine democratic spirit? This sort of “negativism” is undoubtedly
of the same degree of abstraction as “free will” or “determinism”;
like them it is an habitual way of apperceiving contexts, event
sequences and own behavior; but it is not clear whether this
negativism is a “subspecies” of individual autonomy; or is it rather
some entirely different habit? Similarly, we need to know how the
new habit of thought which Dr. Mead advocates is related to the

Evidently our need is for something better than a random list of
these habits of mind. We need some systematic frame-work or
classification which shall show how each of these habits is related to
the others, and such a classification might provide us with
something approaching the chart we lack. Dr. Mead tells us to sail
into as yet uncharted waters, adopting a new habit of thought; but if
we knew how this habit is related to others, we might be able to
judge of the benefits and dangers, the possible pitfalls of such a
course. Such a chart might provide us with the answers to some of
the questions which Dr. Mead raises—as to how we are to judge of
the “direction” and value implicit in our planned acts.

You must not expect the social scientist to produce such a chart
or classification at a moment’s notice, like a rabbit out of a hat, but I
think we can take a first step in this direction; we can suggest some
of the basic themes—the cardinal points, if you like—upon which
the final classification must be built.

We have noted that the sorts of habit with which we are
concerned are, in some sense, by-products of . the learning
processes, and it is therefore natural that we look first to the


phenomena of simple learning as likely to provide us with a clue.
We are raising questions one degree more abstract than those chiefly
studied by the experimental psychologists, but it is still to their
laboratories that we must look for our answers.

Now it so happens that in the psychological laboratories there is
a common phenomenon of a somewhat higher degree of abstraction
or generality than those which the experiments are planned to
elucidate. It is a commonplace that the experimental subject—
whether animal or man, becomes a better subject after repeated
experiments. He not only learns to salivate at the appropriate
moments, or to recite the appropriate nonsense syllables; he also, in
some way, learns to learn. He not only solves the problems set him
by the experimenter, where each solving is a piece of simple
learning; but, more than this, he becomes more and more skilled in
the solving of problems.

In semigestalt or semianthropomorphic phraseology, we might
say that the subject is learning to orient himself to certain types of
contexts, or is acquiring “insight” into the contexts of problem
solving. In the jargon of this paper, we may say that the subject has
acquired a habit of looking for contexts and sequences of one type
rather than another, a habit of “punctuating” the stream of events to
give repetitions of a certain type of meaningful sequence.

The line of argument which we have followed has brought us to
a point at which statements about simple learning meet statements
about gestalt and contextual structure, and we have reached the
hypothesis that “learning to learn” is a synonym for the acquisition
of that class of abstract habits of thought with which this paper is
concerned; that the states of mind which we call “free will,”
instrumental thinking, dominance, passivity, etc., are acquired by a
process which we may equate with “learning to learn.”

This hypothesis is to some extent new2 to psychologists as well
as to laymen, and therefore I must digress at this point to supply
technical readers with a more precise statement of my meaning. I
must demonstrate at least my willingness to state this bridge
between simple learning and gestalt in operational terms.

Let us coin two words, “proto-learning” and “deuterolearning,” to
avoid the labor of defining operationally all the other terms in the
field (transfer of learning, generalization, etc., etc.). Let us say that
there are two sorts of gradient discernible in all continued learning.
The gradient at any point on a simple learning curve (e.g., a curve of
rote learning) we will say chiefly represents rate of proto-learning.
If, however, we inflict a series of similar learning experiments on
the same subject, we shall find that in each successive experiment
the subject has a somewhat steeper proto-learning gradient, that he
learns somewhat more rapidly. This progessive change in rate of
proto-learning we will call “deutero-learning.”

From this point we can easily go on to represent deuterolearning
graphically with a curve whose gradient shall represent rate of
deutero-learning. Such a representation might be obtained, for
example, by intersecting the series of protolearning curves at some
arbitarily chosen number of trials, and noting what proportion of
successful responses occurred in each experiment at this point. The
curve of deutero-learning would then be obtained by plotting these
numbers against the serial numbers of the experiments.3

2 Psychological papers bearing upon this problem of the relationship between
gestalt and simple learning are very numerous, if we include all who have worked
on the concepts of transfer of learning, generalization, irradiation, reaction
threshold (Hull), insight, and the like. Historically, one of the first to pose these
questions was Mr. Frank (L. K. Frank, “The Problems of Learning,” Psych. Review,
1926, 33: 329–51; and Professor Maier has recently introduced a concept of
“direction” which is closely related to the notion of “deutero-learning.” He says:
“direction … is the force which integrates memories in a particular manner without
being a memory itself.” (N. R. F. Maier, “The Behavior Mechanisms Concerned
with Problem Solving,”-Psych. Review, 1940, 47: 43–58.) If for “force” we
substitute “habit,” and for “memory” we substitute “experience of the stream of
events,” the concept of deutero-learning can be seen as almost synonymous with
Professor Maier’s concept of “direction.”

3 It will be noted that the operational definition of deutero-learning is
necessarily somewhat easier than that of proto-learning. Actually, no simple
learning curve represents proto-learning alone. Even within the duration of the


In this definition of proto-and deutero-learning, one phrase
remains conspicuously vague, the phrase “a series of similar
experiments.” For purposes of illustration, I imagined a series of
experiments in rote learning, each experiment similar to the last,
except for the substitution of a new series of nonsense syllables in

single learning experiment we must suppose that some deutero-learning will occur,
and this will make the gradient at any point somewhat steeper than the hypothetical
gradient of “pure” proto-learning.

place of those already learned. In this example, the curve of

•deutero-learning represented in-creasing proficiency in the business
of rote learning, and, as an experimental fact, such increase in rote
proficiency can be demonstrated.4
Apart from rote learning, it is much more difficult to de-fine
what we mean by saying that one learning context is “similar” to
another, unless we are content to refer the matter back to the
experimentalists by saying that learning contexts shall be considered
to be “similar” one to another whenever it can be shown
experimentally that experience of learning in one context does, as a
matter of fact, promote speed of learning in another, and asking the
experimentalists to find out for us what sort of classification they
can build up by use of this criterion. We may hope that they will do
this; but we cannot hope for immediate answers to our questions,
because there are very serious difficulties in the way of such
experimentation. Experiments in simple learning are already
difficult enough to control and to per-form with critical exactness,
and experiments in deuterolearning are likely to prove almost

There is, however, an alternative course open to us. When we
equated “learning to learn” with acquiring apperceptive habits, this
did not exclude the possibility that such habits might be acquired in
other ways. To suggest that the only method of acquiring one of
these habits is through repeated experience of learning contexts of
a given kind would be logically analogous to saying that the only
way to roast pig is by burning the house down. It is obvious that in
human education such habits are acquired in very various ways.
We are not concerned with a hypothetical isolated individual in
contact with an impersonal events stream, but rather with real
individuals who have complex emotional patterns of relationship
with other individuals. In such a real world, the individual will be
led to acquire or reject apperceptive habits by the very complex
phenomena of personal example, tone of voice, hostility, love, etc.
Many such habits, too, will be conveyed to him, not through his
own naked experience of the stream of events, for no human

4 C. Hull, Mathematico-Deductive Theory of Rote Learning, New Haven, Yale
University Press, 1940.


beings (not even scientists) are naked in this sense. The events
stream is mediated to them through language, art, technology, and
other cultural media which are structured at every point by
tramlines of apperceptive habit.

It therefore follows that the psychological laboratory is not the
only possible source of knowledge about these habits; we may turn
instead to the contrasting patterns implicit and explicit in the
various cultures of the world studied by the anthropologists. We
can amplify our list of these obscure habits by adding those which
have been developed in cultures other than our own.

Most profitably, I believe, we can combine the insights of the
experimental psychologists with those of the anthropologists,
taking the contexts of experimental learning in the laboratory and
asking of each what sort of apperceptive habit we should expect to
find associated with it; then looking around the world for human
cultures in which this habit has been developed. Inversely, we may
be able to get a more definite—more operational—definition of
such habits as “free will” if we ask about each, “What sort of
experimental learning context would we devise in order to
inculcate this habit?” “How would we rig the maze or problem-box
so that the anthropomorphic rat shall obtain a repeated and
reinforced impression of his own free will?”

The classification of contexts of experimental learning is as yet
very incomplete, but certain definite advances have been made.5 It

5 Various classifications have been devised for purposes of exposition. Here I
follow that of Hilgard and Marquis (E. R. Hilgard and D. G. Marquis,
Conditioning and Learning, New York, Appleton Century Co., 1940). These
authors subject their own classification to a brilliant critical analysis, and to
this analysis I am indebted for one of the formative ideas upon which this
paper is based. They insist that any learning context can be described in
terms of any theory of learning, if we are willing to stretch and
overemphasize certain aspects of the context to fit onto the Procrustean bed
of the theory. I have taken this notion as a cornerstone of my thinking, substituting
“apperceptive habits” for “theories of learning,” and arguing that
almost any sequence of events can be stretched and warped and punctuated
to fit in with any type of apperceptive habit. (We may suppose that experimental
neurosis is what happens when the subject fails to achieve this

I am also indebted to Lewin’s topological analysis of the contexts of
reward and punishment. (K. Lewin, A Dynamic Theory of Personality, New


is possible to classify the principal contexts of positive
learning (as distinct from negative learning or inhibition,
learning not to do things) under four heads, as follows:

(1)Classical Pavlovian contexts

These are characterized by a rigid time sequence in which the
conditioned stimulus (e.g., buzzer) always pre-cedes the
unconditioned stimulus (e.g., meat powder) by a fixed interval of
time. This rigid sequence of events is not altered by anything that
the animal may do. In these con-texts, the animal learns to respond
to the conditioned stimulus with behavior (e.g., salivation) which
was formerly evoked only by the unconditioned stimulus.

(2)Contexts of instrumental reward or escape

These are characterized by a sequence which depends upon the
animal’s behavior. The unconditioned stimulus in these contexts is
usually vague (e.g., the whole sum of circumstances in which the
animal is put, the problem-box) and may be internal to the animal
(e.g., hunger). If and when, under these circumstances, the animal
performs some act within its behavioral repertoire and previously
selected by the experimenter (e.g., lifts its leg), it is immediately

(3)Contexts of instrumental avoidance

These are also characterized by a conditional sequence. The
unconditioned stimulus is usually definite (e.g., a warning buzzer)
and this is followed by an unpleasant experience (e.g., electric
shock) unless in the interval the animal per-forms some selected
act (e.g., lifts leg).

(4)Contexts of serial and rote learning

These are characterized by the predominant conditioned stimulus
being an act of the subject. He learns, for example, always to give
the conditioned response (nonsense syllable B) after he has himself
uttered the conditioned stimulus (nonsense syllable A),.

York, McGraw-Hill Book Co., 1936.)


This small beginning of a classification6 will be sufficient to
illustrate the principles with which we are concerned and we can
now go on to ask about the occurrence of the appropriate
apperceptive habits among men of various cultures. Of greatest
interest—because least familiar—are the Pavlovian pat-terns and the
patterns of rote. It is a little hard for members of Western civilization
to believe that whole systems of behavior can be built on premises
other than our own mixture of instrumental reward and instrumental
avoidance. The Trobriand Islanders, however, appear to live a life
whose coherence and sense is based upon looking at events through
Pavlovian spectacles, only slightly tinted with the hope of
instrumental reward, while the life of the Balinese is sensible if we

Many people feel that the contexts of experimental learning are so
oversimplified as to have no bearing upon the phenomena of the real world.
Actually, expansion of this classification will give means of defining systematically
many hundreds of possible contexts of learning with their associated apperceptive
habits. The scheme may be expanded in the following ways:

a.Inclusion of contexts of negative learning (inhibition).
b.Inclusion of mixed types (e.g., cases in which salivation, with its
physiological relevance to meat powder, is also instrumental in obtaining
the meat powder).
c.Inclusion of the cases in which the subject is able to deduce some sort of
relevance (other than the physiological) between some two or more
elements in the sequence. For this to be true, the subject must have
experience of contexts differing systematically one from another, e.g.,
contexts in which some type of change in one element is constantly
accompanied by a constant type of change in another element. These cases
can be spread out on a lattice of possibilities, according to which pair of
elements the subject sees as interrelated. There are only five elements (conditioned
stimulus, conditioned response, reward or punishment, and two
time intervals), but any pair of these may be interrelated, and of the
interrelated pair, either may be seen by the subject as determining the other.
These possibilities, multiplied for our four basic contexts, give forty-eight

d.The list of basic types may be extended by including those cases (not as yet
investigated in learning experiments but common in interpersonal relationships)
in which the roles of subject and experimenter are reversed. In these, the
learning partner provides the initial and final elements, while some other person
(or circumstance) provides the middle term. In these types, we see the buzzer and
the meat powder as the behavior of a person and ask: “What does this person
learn?” A great part of the gamut of apperceptive habits associated with
authority and parenthood is based on contexts of this general type.

accept premises based upon combining rote with instrumental

Clearly, to the “pure” Pavlovian, only a very limited fatal-ism
would be possible. He would see all events as preordained and he
would see himself as fated only to search for omens, not able to
influence the course of events—able, at most, from his reading of
the omens, to put himself in the properly receptive state, e.g., by
salivation, before the inevitable occurred. Trobriand culture is not so
purely Pavlovian as this, but Dr. Lee,7 analyzing Professor
Malinowski’s rich observations, has shown that Trobriand phrasings
of purpose, cause, and effect are profoundly different from our own;
and though Dr. Lee does not use the sort of classification here
proposed, it appears from Trobriand magic that these people
continually exhibit a habit of thinking that to act as if a thing were
so will make it so. In this sense, we may describe them as semi-
Pavlovians who have decided that “salivation” is instrumental to
obtaining “meat powder.” Malinowski, for example, gives us a
dramatic description of the almost physiological extremes of rage8
which the Trobriand black magician practices in his incantations,
and we may take this as an illustration of the semi-Pavlovian frame
of mind in contrast with the very various types of magical procedure
in other parts of the world, where, for example, the efficacy of a
spell may be associated not with the intensity but with the extreme
rote accuracy of the recitation.

Among the Balinese9 we find another pattern which contrasts
sharply both with our own and with that of the Trobrianders. The

7 Dorothy Lee, “A Primitive System of Values,” Journal Philos. of Science,
1940, 7: 355-78.

8 A It is possible that semi-Pavlovian phrasings of the stream of events tend, like the
experiments which are their prototypes, to hinge particularly upon autonomic reactions
—that those who see events in these terms tend to see these reactions, which are only
partially subject to voluntary control, as peculiarly effective and powerful causes of
outside events. There may be some ironical logic in Pavlovian fatalism which
predisposes us to believe that we can alter the course of events only by means of those
behaviors which we are least able to control.

9 The Balinese material collected by Dr. Mead and my-self has not yet been published in
extenso, but a brief out-line of the theory here suggested is available—cf. G. Bateson,
“The Frustration-Aggression Hypothesis and Culture,” Psychological Review, 1941,

48: 350-55.

treatment of children is such that they learn not to see life as
composed of connative sequences ending in satisfaction, but rather
to see it as composed of rote sequences inherently satisfying in
themselves—a pattern which is to some extent related to that pattern
which Dr. Mead has recommended, of looking for value in the act
itself rather than regarding the .act as a means to an end. There is,
how-ever, one very important difference between the Balinese
pattern and that recommended by Dr. Mead. The Balinese pattern is
essentially derivative from contexts of instrumental avoidance; they
see the world as dangerous, and themselves as avoiding, by the
endless rote behavior of ritual and courtesy, the ever-present risk of
faux pas. Their life is built upon fear, albeit that in general they
enjoy fear. The positive value with which they endow their
immediate acts, not looking for a goal, is somehow associated with
this enjoyment of fear. It is the acrobat’s enjoyment both of the thrill
and of his own virtuosity in avoiding disaster.

We are now, after a somewhat long and technical excursion into
psychological laboratories and foreign cultures, in a position to
examine Dr. Mead’s proposal in somewhat more concrete terms.
She advises that when we apply the social sciences we look for
“direction” and “value” in our very acts, rather than orient
ourselves to some blueprinted goal. She is not telling us that we
ought to be like the Balinese, except in our time orientation, and
she would be the first to disparage any suggestion that fear (even
enjoyed fear) should be our basis for assigning value to our acts.
Rather, as I understand it, this basis should be some sort of hope—
not looking to some far-off future, but still some sort of hope or
optimism. In fact, we might summarize the recommended attitude
by saying that it ought to be formally related to instrumental
reward, as the Balinese attitude is related to instrumental

Such an attitude is, I believe, feasible. The Balinese attitude
might be defined as a habit of rote sequences inspired by a thrilling
sense of ever-imminent but indefinite danger, and I think that what
Dr. Mead is urging us toward might be defined in like terms, as a
habit of rote sequences inspired by a thrilling sense of ever-
imminent but undefined reward.

As to the rote component, which is almost certainly a necessary
concomitant of the peculiar time orientation advocated by Dr.
Mead, 1, personally, would welcome it, and I believe that it would
be infinitely preferable to the compulsive type of accuracy after
which we strive. Anxious taking-care and automatic, rote caution
are alternative habits which perform the same function. We can
either have the habit of automatically looking before we cross the
street, or the habit of carefully remembering to look. Of the two I
prefer the automatic, and I think that, if Dr. Mead’s recommendation
implies as increase in rote automatism, we ought to
accept it. Already, indeed, our schools are inculcating more and
more automatism in such processes as reading, writing, arithmetic,
and languages.

As to the reward component, this, too, should not be beyond our
reach. If the Balinese is kept busy and happy by a nameless,
shapeless fear, not located in space or time, we might be kept on our
toes by a nameless, shapeless, unlocated hope of enormous
achievement. For such a hope to be effective, the achievement need
scarcely be defined.

All we need to be sure of is that, at any moment, achievement
may be just around the corner, and, true or false, this can never be
tested. We have got to be like those few artists and scientists who
work with this urgent sort of inspiration, the urgency that comes
from feeling that great discovery, the answer to all our problems,
or great creation, the perfect sonnet, is always only just beyond our
reach, or like the mother of a child who feels that, provided she
pay constant enough attention, there is a real hope that her child
may be that infinitely rare phenomenon, a great and happy person.


A Theory of Play and Fantasy*

This research was planned and started with an hypothesis to
guide our investigations, the task of the investigators being to collect
relevant observational data and, in the process, to amplify and
modify the hypothesis.

The hypothesis will here be described as it has grown in

our thinking.

Earlier fundamental work of Whitehead, Russell,10 Witt-
genstein,11 Carnap,12 Whorf,13 etc., as well as my own at-tempt14 to
use this earlier thinking as an epistemological base

for psychiatric theory, led to a series of generalizations:

(1) That human verbal communication can operate and always
does operate at many contrasting levels of abstraction. These range
in two directions from the seemingly simple denotative level (“The
cat is on the mat”). One range or set of these more abstract levels
includes those explicit or implicit messages where the subject of
discourse is the language. We will call these metalinguistic (for
example, “The verbal sound `cat’ stands for any member of such and
such class of objects,” or “The word, `cat,’ has no fur and cannot
scratch”). The other set of levels of abstraction we will call
metacommunicative (e.g., “My telling you where to find the cat was
friendly,” or “This is play”). In these, the subject of discourse is the
relationship between the speakers.
* This essay was read (by Jay Haley) at the A.P.A. Regional Research
Conference in Mexico City, March 11, 1954. It is here reprinted from A.P.A.
Psychiatric Research Reports, II, 1955, by permission of the American
Psychiatric Association.
10 A. N. Whitehead and B. Russell, Principia Mathematica, 3 vols., 2nd ed.,
Cambridge, Cambridge University Press, 1910-13.
11 L. Wittgenstein, Tractatus Logico-Philosophicus, Lon-don, Harcourt Brace,
12 R. Carnap, The Logical Syntax of Language, New York, Harcourt Brace,
13 B. L. Whorf, “Science and Linguistics,” Technology Review, 1940, 44: 229

14 J. Ruesch and G. Bateson, Communication: The Social Matrix of Psychiatry,
New York, Norton, 1951.


It will be noted that the vast majority of both metalinguistic and
metacommunicative messages remain implicit; and also that,
especially in the psychiatric interview, there occurs a further class of
implicit messages about how metacommunicative messages of
friendship and hostility are to be interpreted.

(2) If we speculate about the evolution of communication, it is
evident that a very important stage in this evolution occurs when the
organism gradually ceases to respond quite “automatically” to the
mood-signs of another and becomes able to recognize the sign as a
signal: that is, to recognize that the other individual’s and its own
signals are only signals, which can be trusted, distrusted, falsified,
denied, amplified, corrected, and so forth.
Clearly this realization that signals are signals is by no means
complete even among the human species. We all too often respond
automatically to newspaper headlines as though these stimuli were
direct object-indications of events in our environment instead of
signals concocted and transmitted by creatures as complexly
motivated as ourselves. The nonhuman mammal is automatically
excited by the sexual odor of another; and rightly so, inasmuch as
the secretion of that sign is an “involuntary” mood-sign; i.e., an
outwardly perceptible event which is a part of the physiological
process which we have called a mood. In the human species a more
complex state of affairs begins to be the rule. Deodorants mask the
involuntary olfactory signs, and in their place the cosmetic industry
provides the individual with perfumes which are not involuntary
signs but voluntary signals, recognizable as such. Many a man has
been thrown off balance by a whiff of perfume, and if we are to
believe the advertisers, it seems that these signals, voluntarily worn,
have sometimes an automatic and autosuggestive effect even upon
the voluntary wearer.

Be that as it may, this brief digression will serve to illustrate a
stage of evolution—the drama precipitated when organisms, having
eaten of the fruit of the Tree of Knowledge, discover that their
signals are signals. Not only the characteristically human invention
of language can then follow, but also all the complexities of
empathy, identification, projection, and so on. And with these comes
the possibility of communicating at the multiplicity of levels of
abstraction mentioned above.


(3) The first definite step in the formulation of the hypothesis
guiding this research occurred in January, 1952, when I went to the
Fleishhacker Zoo in San Francisco to look for behavioral criteria
which would indicate whether any given organism is or is not able
to recognize that the signs emitted by itself and other members of
the species are signals. In theory, I had thought out what such
criteria might look like—that the occurrence of metacommunicative
signs (or signals) in the stream of interaction between the animals
would indicate that the animals have at least some awareness
(conscious or unconscious) that the signs about which they
metacommunicate are signals.
I knew, of course, that there was no likelihood of finding
denotative messages among nonhuman mammals, but I was still not
aware that the animal data would require an almost total revision of
my thinking. What I encountered at the zoo was a phenomenon well
known to everybody: I saw two young monkeys playing, i.e.,
engaged in an interactive sequence of which the unit actions or
signals were similar to but not the same as those of combat. It was
evident, even to the human observer, that the sequence as a whole
was not combat, and evident to the human observer that to the
participant monkeys this was “not combat.”

Now, this phenomenon, play, could only occur if the participant
organisms were capable of some degree of meta-communication,
i.e., of exchanging signals which would carry the message “this is

(4) The next step was the examination of the message “This is
play,” and the realization that this message contains those elements
which necessarily generate a paradox of the Russellian or
Epimenides type -a negative statement containing an implicit
negative metastatement. Expanded, the statement “This is play”
looks something like this: “These actions in which we now engage
do not denote what those actions for which they stand would
We now ask about the italicized words, “for which they stand.”
We say the word “cat” stands for any member of a certain class. That
is, the phrase “stands for” is a near synonym of “denotes.” If we now
substitute “which they denote” for the words “for which they stand”
in the expanded definition of play, the result is: “These actions, in

which we now engage, do not denote what would be de-noted by
those actions which these actions denote.” The playful nip denotes
the bite, but it does not denote what would be denoted by the bite.

According to the Theory of Logical Types such a message is of
course inadmissable, because the word “denote” is being used in two
degrees of abstraction, and these two uses are treated as
synonymous. But all that we learn from such a criticism is that it
would be bad natural history to expect the mental processes and
communicative habits of mammals to conform to the logician’s
ideal. Indeed, if human thought and communication always
conformed to the ideal, Russell would n o t in fact could n o t
have formulated the ideal.

(5) A related problem in the evolution of communication
concerns the origin of what Korzybski15 has called the map-territory
relation: the fact that a message, of whatever kind, does not consist
of those objects which it denotes (“The word `cat’ cannot scratch
us”). Rather, language bears to the objects which it denotes a
relationship comparable to that which a map bears to a territory.
Denotative communication as it occurs at the human level is only
possible after the evolution of a complex set of metalinguistic (but
not verbalized)16 rules which govern how words and sentences shall
be related to objects and events. It is therefore appropriate to look
for the evolution of such metalinguistic and/or meta-communicative
rules at a prehuman and preverbal level.
It appears from what is said above that play is a phenomenon in
which the actions of “play” are related to, or denote, other actions of
“not play.” We therefore meet in play with an instance of signals
standing for other events, and it appears, therefore, that the
evolution of play may have been an important step in the evolution
of communication.

(6) Threat is another phenomenon which resembles play in that
actions denote, but are different from, other actions. The clenched
fist of threat is different from the punch, but it refers to a possible
15 A. Korzybski, Science and Sanity, New York, Science Press, 1941.

16 The verbalization of these metalinguistic rules is a much later
achievement which can only occur after the evolution of a
nonverbalized meta-metalinguistics.


future (but at present nonexistent) punch. And threat also is
commonly recognizable among non-human mammals. Indeed it has
lately been argued that a great part of what appears to be combat
among members of a single species is rather to be regarded as threat
(Tinbergen,17 Lorenz18).

(7) Histrionic behavior and deceit are other examples of the
primitive occurrence of map-territory differentiation. And there is
evidence that dramatization occurs among birds: a jackdaw may
imitate her own mood-signs (Lorenz19), and deceit has been
observed among howler monkeys (Carpenter20).
(8) We might expect threat, play, and histrionics to be three
independent phenomena all contributing to the evolution of the
discrimination between map and territory. But it seems that this
would be wrong, at least so far as mammalian communication is
concerned. Very brief analysis of childhood behavior shows that
such combinations as histrionic play, bluff, playful threat, teasing
play in response to threat, histrionic threat, and so on form together
a single total complex of phenomena. And such adult phenomena as
gambling and playing with risk have their roots in the combination
of threat and play. It is evident also that not only threat but the
reciprocal of threat—the behavior of the threatened individual—are
a part of this complex. It is probable that not only histrionics but
also spectatorship should be included within this field. It is also
appropriate to mention self-pity.
(9) A further extension of this thinking leads us to include ritual
within this general field in which the discrimination is drawn, but
not completely, between denotative action and that which is to be
denoted. Anthropological studies of peace-making ceremonies, to
cite only one example, sup-port this conclusion.
In the Andaman Islands, peace is concluded after each side has
been given ceremonial freedom to strike the other. This example,
however, also illustrates the labile nature of the frame “This is play,”

17 N. Tinbergen, Social Behavior in Animals with Special Reference to
Vertebrates, London, Methuen, 1953.

18 K. Z. Lorenz, King Solomon’s Ring, New York, Crowell, 1952.

19 Ibid.

20 C. R. Carpenter, “A Field Study of the Behavior and Social Relations of
Howling Monkeys,” Comp. Psychol. Monogr., 1934, 10: 1-168.


or “This is ritual.” The discrimination between map and territory is
always liable to break down, and the ritual blows of peace-making
are always liable to be mistaken for the “real” blows of combat. In
this event, the peace-making ceremony becomes a battle (RadcliffeBrown21)

(10) But this leads us to recognition of a more complex form of
play; the game which is constructed not upon the premise “This is
play” but rather around the question “Is this play?” And this type of
interaction also has its ritual forms, e.g., in the hazing of initiation.
(11) Paradox is doubly present in the signals which are
exchanged within the context of play, fantasy, threat, etc. Not only
does the playful nip not denote what would be denoted by the bite
for which it stands, but, in addition, the bite itself is fictional. Not
only do the playing animals not quite mean what they are saying
but, also, they are usually communicating about something which
does not exist. At the human level, this leads to a vast variety of
complications and inversions in the fields of play, fantasy, and art.
Conjurers and painters of the trompe l’oeil school concentrate upon
acquiring a virtuosity whose only reward is reached after the viewer
detects that he has been deceived and is forced to smile or marvel at
the skill of the deceiver. Hollywood film-makers spend millions of
dollars to increase the realism of a shadow. Other artists, perhaps
more realistically, insist that art be nonrepresentational; and poker
players achieve a strange addictive realism by equating the chips for
which they play with dollars. They still insist, however, that the
loser accept his loss as part of the game.
Finally, in the dim region where art, magic, and religion meet and
overlap, human beings have evolved the “metaphor that is meant,”
the flag which men will die to save, and the sacrament that is felt to
be more than “an outward and visible sign, given unto us.” Here we
can recognize an attempt to deny the difference between map and
territory, and to get back to the absolute innocence of
communication by means of pure mood-signs.

(12) We face then two peculiarities of play: (a) that the messages
or signals exchanged in play are in a certain sense untrue or not
21 A. R. Radcliffe-Brown, The Andaman Islanders, Cambridge, Cambridge
University Press, 1922.


meant; and (b) that that which is denoted by these signals is
nonexistent. These two peculiarities sometimes combine strangely to
a reverse a conclusion reached above. It was stated (4) that the
playful nip de-notes the bite, but does not denote that which would
be denoted by the bite. But there are other instances where an
opposite phenomenon occurs. A man experiences the full intensity
of subjective terror when a spear is flung at him out of the 3D screen
or when he falls headlong from some peak created in his own mind
in the intensity of nightmare. At the moment of terror there was no
questioning of “reality,” but still there was no spear in the movie
house and no cliff in the bedroom. The images did not denote that
which they seemed to denote, but these same images did really
evoke that terror which would have been evoked by a real spear or a
real precipice. By a similar trick of self-contradiction, the filmmakers
of Hollywood are free to offer to a puritanical public a vast
range of pseudosexual fantasy which otherwise would not be
tolerated. In David and Bathsheba, Bathsheba can be a Troilistic
link between David and Uriah. And in Hans Christian Andersen, the
hero starts out accompanied by a boy. He tries to get a woman, but
when he is defeated in this attempt, he returns to the boy. In all of
this, there is, of course, no homosexuality, but the choice of these
symbolisms is associated in these fantasies with certain
characteristic ideas, e.g., about the hopelessness of the heterosexual
masculine position when faced with certain sorts of women or with
certain sorts of male authority. In sum, the pseudohomosexuality of
the fantasy does not stand for any real homosexuality, but does stand
for and express attitudes which might accompany a real
homosexuality or feed its etiological roots. The symbols do not
denote homosexuality, but do denote ideas for which homosexuality
is an appropriate symbol. Evidently it is necessary to re-examine the
precise semantic validity of the interpretations which the psychiatrist
offers to a patient, and, as preliminary to this analysis, it will be
necessary to examine the nature of the frame in which these
interpretations are offered.

(13) What has previously been said about play can be used as an
introductory example for the discussion of frames and contexts. In
sum, it is our hypothesis that the message “This is play” establishes

a paradoxical frame comparable to Epimenides’ paradox. This frame
may be diagrammed thus:

The first statement within this frame is a self-contradictory
proposition about itself. If this first statement is true, then it must be
false. If it be false, then it must be true. But this first statement
carries with it all the other statements in the frame. So, if the first
statement be true, then all the others must be false; and vice versa, if
the first statement be untrue then all the others must be true.

(14) The logically minded will notice a non-sequitur. It could be
urged that even if the first staement is false, there remains a logical
possibility that some of the other statements in the frame are untrue.
It is, however, a characteristic of unconscious or “primary-process”
thinking that the thinker is unable to discriminate between “some”
and “all,” and unable to discriminate between “not all” and “none.” It
seems that the achievement of these discriminations is performed by
higher or more conscious mental processes which serve in the
nonpsychotic individual to correct the black-and-white thinking of
the lower levels. We assume, and this seems to be an orthodox
assumption, that primary process is continually operating, and that
the psychological validity of the paradoxical play frame depends
upon this part of the mind.
(15) But, conversely, while it is necessary to invoke the primary
process as an explanatory principle in order to delete the notion of
“some” from between “all” and “none,” this does not mean that play
is simply a primary-process phenomenon. The discrimination
between “play” and “nonplay,” like the discrimination between
fantasy and nonfantasy, is certainly a function of secondary process,
or “ego.” Within the dream the dreamer is usually unaware that he is
dreaming, and within “play” he must often be reminded that “This is

Similarly, within dream or fantasy the dreamer does not operate
with the concept “untrue.” He operates with all sorts of statements
but with a curious inability to achieve meta-statements. He cannot,
unless close to waking, dream a statement referring to (i.e., framing)
his dream.

It therefore follows that the play frame as here used as an
explanatory principle implies a special combination of primary and
secondary processes. This, however, is related to what was said
earlier, when it was argued that play marks a step forward in the
evolution of communication—the crucial step in the discovery of
map-territory relations. In primary process, map and territory are
equated; in secondary process, they can be discriminated. In play,
they are both equated and discriminated.

(16) Another logical anomaly in this system must be mentioned:
that the relationship between two propositions which is commonly
described by the word “premise” has become intransitive. In general,
all asymmetrical relationships are transitive. The relationship
“greater than” is typical in this respect; it is conventional to argue
that if A is greater than B, and B is greater than C, then A is greater
than C. But in psychological processes the transitivity of
asymmetrical relations is not observed. The proposition P may be a
premise for Q; Q may be a premise for R; and R may be a premise
for P. Specifically, in the system which we are considering, the circle
is still more contracted. The message, “All statements within this
frame are untrue” is itself to be taken as a premise in evaluating its
own truth or untruth. (Cf. the in-transitivity of psychological
preference discussed by McCulloch.22 The paradigm for all
paradoxes of this general type is Russell’s23 “class of classes which
are not members of them-selves.” Here Russell demonstrates that
paradox is generated by treating the relationship, “is a member of,”
as an in-transitive.) With this caveat, that the “premise” relation in
psychology is likely to be intransitive, we shall use the word
“premise” to denote a dependency of one idea or message upon
another comparable to the dependency of one proposition upon
22 W. S. McCulloch, “A Heterarchy of Values, etc.,” Bulletin of Math.
Biophys., 1945, 7: 89-93.

23 Whitehead and Russell, op. cit.


another which is referred to in logic by saying that the proposition P
is a premise for Q.

(17) All this, however, leaves unclear what is meant by “frame”
and the related notion of “context.” To clarify these, it is necessary
to insist first that these are psychological concepts. We use two sorts
of analogy to discuss these notions: the physical analogy of the
picture frame and the more abstract, but still not psychological,
analogy of the mathematical set. In set theory the mathematicians
have developed axioms and theorems to discuss with rigor the
logical implications of membership in overlapping categories or
“sets.” The relationships between sets are commonly illustrated by
diagrams in which the items or members of a larger universe are
represented by dots, and the smaller sets are delimited by imaginary
lines enclosing the members of each set. Such diagrams then
illustrate a topological approach to the logic of classification. The
first step in defining a psychological frame might be to say that it is
(or delimits) a class or set of messages (or meaningful actions). The
play of two individuals on a certain occasion would then be defined
as the set of all messages exchanged by them within a limited period
of time and modified by the paradoxical premise system which we
have described. In a set-theoretical diagram these messages might be
represented by dots, and the “set” enclosed by a line which would
separate these from other dots representing nonplay messages. The
mathematical analogy breaks down, however, because the
psychological frame is not satisfactorily represented by an
imaginary line. We assume that the psychological frame has some
degree of real existence. In many instances, the frame is consciously
recognized and even represented in vocabulary (“play,” “movie,”
“interview,” “job,” “language,” etc.). In other cases, there may be no
explicit verbal reference to the frame, and the subject may have no
consciousness of it. The analyst, however, finds that his own
thinking is simplified if he uses the notion of an unconscious frame
as an explanatory principle; usually he goes further than this and
infers its existence in the subject’s unconscious.
But while the analogy of the mathematical set is perhaps over
abstract, the analogy of the picture frame is excessively concrete.
The psychological concept which we are trying to define is neither
physical nor logical. Rather, the actual physical frame is, we believe,


added by human beings to physical pictures because these human
beings operate more easily in a universe in which some of their
psychological characteristics are externalized. It is these
characteristics which. we are trying to discuss, using the externalization
as an illustrative device.

(18) The common functions and uses of psychological frames
may now be listed and illustrated by reference to the analogies
whose limitations have been indicated in the previous paragraph:
(a)Psychological frames are exclusive, i.e., by including certain
messages (or meaningful actions) within a frame, certain other
messages are excluded.

(b)Psychological frames are inclusive, i.e., by excluding certain
messages certain others are included. From the point of view of set
theory these two functions are synonymous, but from the point of
view of psychology it is necessary to list them separately. The frame
around a picture, if we consider this frame as a message intended to
order or organize the perception of the viewer, says, “Attend to what
is within and do not attend to what is outside.” Figure and ground, as
these terms are used by gestalt psychologists, are not symmetrically
related as are the set and nonset of set theory. Perception of the
ground must be positively inhibited and perception of the figure (in
this case the picture) must be positively enhanced.

(c)Psychological frames are related to what we have called
“premises.” The picture frame tells the viewer that he is not to use
the same sort of thinking in interpreting the picture that he might use
in interpreting the wallpaper outside the frame. Or, in terms of the
analogy from set theory, the messages enclosed within the imaginary
line are defined as members of a class by virtue of their sharing
common premises or mutual relevance. The frame itself thus
becomes a part of the premise system. Either, as in the case of the
play frame, the frame is involved in the evaluation of the messages
which it contains, or the frame merely assists the mind in
understanding the contained messages by reminding the thinker that
these messages are mutually relevant and the messages outside the
frame may be ignored.

(d)In the sense of the previous paragraph, a frame is
metacommunicative. Any message, which either explicitly or
implicitly defines a frame, ipso facto gives the receiver instructions

or aids in his attempt to understand the messages included within the

(e)The converse of (d) is also true. Every meta-communicative or
metalinguistic message defines, either explicitly or implicitly, the set
of messages about which it communicates, i.e., every
metacommunicative message is or de-fines a psychological frame.
This, for example, is very evident in regard to such small
metacommunicative signals as punctuation marks in a printed
message, but applies equally to such complex metacommunicative
messages as the psychiatrist’s definition of his own curative role in
terms of which his contributions to the whole mass of messages in
psychotherapy are to be understood.

(f)The relation between psychological frame and perceptual
gestalt needs to be considered, and here the analogy of the picture
frame is useful. In a painting by Roualt or Blake, the human figures
and other objects represented are outlined. “Wise men see outlines
and therefore they draw them.” But outside these lines, which
delimit the perceptual gestalt or “figure,” there is a background or
“ground” which in turn is limited by the picture frame. Similarly, in
set-theoretical diagrams, the larger universe within which the
smaller sets are drawn is itself enclosed in a frame. This double
framing is, we believe, not merely a matter of “frames within
frames” but an indication that mental processes resemble logic in
needing an outer frame to delimit the ground against which the
figures are to be perceived. This need is often unsatisfied, as when
we see a piece of sculpture in a junk shop window, but this is
uncomfortable. We suggest that the need for this outer limit to the
ground is related to a preference for avoiding the paradoxes of
abstraction. When a logical class or set of items is defined—for
example, the class of matchboxes—it is necessary to delimit the set
of items which are to be excluded, in this case, all those things
which are not matchboxes. But the items to be included in the
background set must be of the same degree of abstraction, i.e., of the
same “logical type” as those within the set itself. Specifically, if
paradox is to be avoided, the “class of matchboxes” and the “class of
nonmatchboxes” (even though both these items are clearly not
matchboxes) must not be regarded as members of the class of
nonmatchboxes. No class can be a member, of itself. The picture


frame then, because it delimits a background, is here regarded as an
external representation of a very special and important type of
psycho-logical frame—namely a frame whose function is to delimit
a logical type. This, in fact, is what was indicated above when it was
said that the picture frame is an instruction to the viewer that he
should not extend the premises which obtain between the figures
within the picture to the., wall paper behind it.

But, it is precisely this sort of frame that precipitates paradox.
The rule for avoiding paradoxes insists that the items outside any
enclosing line be of the same logical type as those within, but the
picture frame, as analyzed above, is a line dividing items of one
logical type from those of another. In passing, it is interesting to note
that Russell’s rule cannot be stated without breaking the rule. Russell
insists that all items of inappropriate logical type be exluded (i.e., by
an imaginary line) from the background of any class, i.e., he insists
upon the drawing of an imaginary line of precisely the sort which he

(19) This whole matter of frames and paradoxes may be
illustrated in terms of animal behavior, where three types of message
may be recognized or deduced: (a) Messages of the sort which we
here call mood-signs; (b) messages which simulate mood-signs (in
play, threat, histrionics, etc.) ; and (c) messages which enable the
receiver to discriminate between mood-signs and those other signs
which resemble them. The message “This is play” is of this third
type. It tells the receiver that certain nips and other meaningful
actions are not messages of the first type.
The message “This is play” thus sets a frame of the sort which is
likely to precipitate paradox: it is an attempt to discriminate
between, or to draw a line between, categories of different logical

(20) This discussion of play and psychological frames establishes
a type of triadic constellation (or system of relationships) between
messages. One instance of this constellation is analyzed in
paragraph 19, but it is evident that constellations of this sort occur
not only at the nonhuman level but also in the much more complex
communication of human beings. A fantasy or myth may simulate a
denotative narrative, and, to discriminate between these types of
discourse, people use messages of the frame-setting type, and so on.

(21) In conclusion, we arrive at the complex task of applying this
theoretical approach to the particular phenomena of psychotherapy.
Here the lines of our thinking may most briefly be summarized by
presenting and partially answering these questions:
(a)Is there any indication that certain forms of psycho-pathology
are specifically characterized by abnormalities in the patient’s
handling of frames and paradoxes?

(b)Is there any indication that the techniques of psycho-therapy
necessarily depend upon the manipulation of frames and paradoxes?

(c)Is it possible to describe the process of a given psychotherapy
in terms of the interaction between the patient’s abnormal use of
frames and the therapist’s manipulation of them?

(22) In reply to the first question, it seems that the “word salad”
of schizophrenia can be described in terms of ,the patient’s failure to
recognize the metaphoric nature of his fantasies. In what should be
triadic constellations of messages., the frame-setting message (e.g.,
the phrase “as if”) is omitted, and the metaphor or fantasy is narrated
and acted upon in a manner which would be appropriate if the
fantasy were a message of the more direct kind. The absence of
metacommunicative framing which was noted in the case of dreams
(15) is characteristic of the waking communications of the
schizophrenic. With the loss of the ability to set metacommunicative
frames, there is also a loss of ability to achieve the more primary or
primitive message. The metaphor is treated directly as a message of
the more primary type. (This matter is discussed at greater length
in the paper given by Jay Haley at this Conference.)
(23) The dependence of psychotherapy upon the manipulation
of frames follows from the fact that therapy is an attempt to change
the patient’s metacommunicative habits. Before therapy, the patient
thinks and operates in terms of a certain set of rules for the making
and understanding of messages. After successful therapy; he
operates in terms of a different set of such rules. (Rules of this sort
are in general, unverbalized, and unconscious both before and
after.) It follows that, in the process of therapy, there must have
been communication at a level meta to these rules. There must
have been communication about a change in rules.
But such a communication about change could not conceivably
occur in messages of the type permitted by the patient’s


metacommunicative rules as they existed either be-fore or after

It was suggested above that the paradoxes of play are
characteristic of an evolutionary step. Here we suggest that similar
paradoxes are a necessary ingredient in that process of change
which we call psychotherapy.

The resemblance between the process of therapy and the
phenomenon of play is, in fact, profound. Both occur within a
delimited psychological frame, a spatial and temporal bounding of
a set of interactive messages. In both play and therapy, the
messages have a special and peculiar relationship to a more
concrete or basic reality. Just as the pseudocombat of play is not
real combat, so also the pseudolove and pseudohate of therapy are
not real love and hate. The “transfer” is discriminated from real
love and hate by signals invoking the psychological frame; and
indeed it is this frame which permits the transfer to reach its full
intensity and to be discussed between patient and therapist.

The formal characteristics of the therapeutic process may be
illustrated by building up a model in stages. Imagine first two
players who engage in a game of canasta according to a standard set
of rules. So long as these rules govern and are unquestioned by both
players, the game is unchanging, i.e:, no therapeutic change will
occur. (Indeed many at-tempts at psychotherapy fail for this reason.)
We may imagine, however, that at a certain moment the two canasta
players cease to play canasta and start a discussion of the rules.
Their discourse is now of a different logical type from that of their
play. At the end of this discussion, we can imagine that they return
to playing but with modified rules.

This sequence-of events is, however, still an imperfect model
of therapeutic interaction, though it illustrates our contention that
therapy necessarily involves a combination of discrepant logical
types of discourse. Our imaginary players avoided paradox by
separating their discussion of the rules from their play, and it is
precisely this separation that is impossible in psychotherapy. As we
see it, the process of psychotherapy is a framed interaction
between two persons, in which the rules are implicit but subject to
change. Such change can only be proposed by experimental action,
but every such experimental action, in which a proposal to change

the rules is implicit, is itself a part of the ongoing game. It is this
combination of logical types within the single meaningful act that
gives to therapy the character not of a rigid game like canasta but,
instead, that of an evolving system of interaction. The play of
kittens or otters has this character.

(24) In regard to the specific relationship between the way in
which the patient handles frames and the way in which the therapist
manipulates them, very little can at present be said. It is, however,
suggestive to observe that the psychological frame of therapy is an
analogue of the frame-setting message which the schizophrenic is
unable to achieve. To talk in “word salad” within the psychological
frame of therapy is, in a sense, not pathological. Indeed the neurotic
is specifically encouraged to do precisely this, narrating his dreams
and free associations so that patient and therapist may achieve an
understanding of this material. By the process of interpretation, the
neurotic is driven to insert an “as if” clause into the productions of
his primary process thinking, which productions he had previously
deprecated or re-pressed. He must learn that fantasy contains truth.
For the schizophrenic the problem is somewhat different. His
error is in treating the metaphors of primary process with the full
intensity of literal truth. Through the discovery of what these
metaphors stand for he must discover that they are only metaphors.

(25) From the point of view of the project, however,
psychotherapy constitutes only one of the many fields which we are
attempting to investigate. Our central thesis may be summed up as a
statement of the necessity of the paradoxes of abstraction. It is not
merely bad natural history to suggest that people might or should
obey the Theory of Logical Types in their communications; their
failure to do this is not due to mere carelessness or ignorance.
Rather, we believe that the paradoxes of abstraction must make their
appearance in all communication more complex than that of mood-
signals, and that without these paradoxes the evolution of
communication would be at an end. Life would then be an endless
interchange of stylized messages, a game with rigid rules,
unrelieved by change or humor.

Epidemiology of a Schizophrenia*

If we are to discuss the epidemiology of mental conditions, i.e.,
conditions partly induced by experience, our first task is to pinpoint
a defect of an ideational system sufficiently so that we can go on
from that pinpointing to postulate what sort of contexts of learning
might induce this formal defect.

It is conventionally said that schizophrenics have “ego
weakness.” I now define ego weakness as trouble in identifying and
interpreting those signals which should tell the individual what sort
of a message a message is, i.e., trouble with the signals of the same
logical type as the signal “This is play.” For example, a patient
comes into the hospital can-teen and the girl behind the counter
says, “What can I do for you?” The patient is in doubt as to what
sort of a message this is—is it a message about doing him in? Is it
an indication that she wants him to go to bed with her? Or is it an
offer of a cup of coffee? He hears the message and does not know
what sort or order of a message it is. He is unable to pick up the
more abstract labels which we are most of us able to use
conventionally but are most of us unable to identify in the sense
that we don’t know what told us what sort of a message it was. It is
as if we some-how make a correct guess. We are actually quite
unconscious of receiving these messages which tell us what sorts
of message we receive.

Difficulty with signals of this sort seems to be the center of a
syndrome which is characteristic for a group of schizophrenics, so
therefore we can reasonably look for an etiology starting from this
symptomatology as formally defined.

When you begin thinking in this way, a great deal of what the
schizophrenic says falls into place as a description of his

* This is an edited version of a talk, “How the Deviant Sees His Society,” given in
May, 1955, at a conference on “The Epidemiology of Mental Health” held at
Brighton, Utah, sponsored by the Departments of Psychiatry and Psychology of the
University of Utah, and the Veterans Administration Hospital, Fort Douglas
Division, of Salt Lake City, Utah. A rough transcript of the talks at this conference
was mimeographed and circulated by the organizers.

experience. That is, we have a second lead toward the theory of
etiology or transmission. The first lead is from the symptom. We
ask, “How does a human individual acquire an imperfect ability to
discriminate these specific signals?” and when we look at his
speeches, we find that, in that peculiar language which is
schizophrenic salad, he is de-scribing a :traumatic situation which
involves a metacommunicative tangle.

A patient, for example, has a central notion, that “some-thing
moved in space,” and that that is why he cracked up. I somehow,
from the way he spoke about “space,” got an idea that space is his
mother and said so. He said, “No, space is the mother.” I suggested
to him that she might be in some way a cause of his troubles. He
said, “I never condemned her.” At a certain point he got angry, and
he said—this is verbatim—”If we say she had movement in her
because of what she caused, we are only condemning ourselves.”
Something moved in space that made him crack up. Space is not
his mother, it is the mother. But now we focus upon his mother
whom he says he never condemned. And he now says, “If we say
that she had movement in her because of what she caused, we are
only condemning our-selves.”

Look very carefully at the logical structure of that last
quotation. It is circular. It implies a way of interaction and chronic
cross-purposes with the mother such that for the child to make
those moves which might straighten out the misunderstanding was
also prohibited.

On another occasion he had skipped his therapy session in the
morning, and I went over to the dining hall at supper time to see
him and assure him that he would see me next day. He refused to
look at me. He looked away. I made some remark about 9.30 the
next morning—no answer. Then, with great difficulty, he said,
“The judge disapproves.” Be-fore I left him, I said, “You need a
defense attorney,” and when I found him on the grounds next
morning I said, “Here is your defense attorney,” and we went into
session together. I started out by saying, “Am I right in supposing
that the judge not only disapproves of your talking to me but also
disapproves of your telling me that he disapproves?” He said,
“Yes!” That is, we are dealing with two levels here. The “judge”
disapproves of the attempt to straighten out the confusions and


disapproves of communicating the fact of his (the judge’s)

We have to look for an etiology involving multiple levels of

I am not talking at all about the content of these traumatic
sequences, whether they be sexual, or oral. Nor am I talking about
the age of the subject at the time of trauma, nor about which parent
is involved. That is all episodic as far as I’m concerned. I’m only
building up toward the statement that the trauma must have had
formal structure in the sense that multiple logical types were
played against each other to generate this particular pathology in
this individual.

Now, if you look at our conventional communication with one
another, what you find is that we weave these logical types with
incredible complexity and quite surprising facility. We even make
jokes, and these may be difficult for a foreigner to understand.
Most jokes, both canned and spontaneous, and nearly anywhere,
are weavings of multiple logical types. Kidding and hazing
similarly depend upon the unresolved question whether the kid-ee
can identify that this is kidding. In any culture, the individuals
acquire quite extraordinary skill in handling not only the flat
identification of what sort of a message a message is but in dealing
in multiple identifications of what sort of a message a message is.
When we meet these multiple identifications we laugh, and we
make new psychological discoveries about what goes on inside
ourselves, which is perhaps the reward of real humor.

But there are people who have the utmost difficulty with this
problem of multiple levels, and it seems to me that this unequal
distribution of ability is a phenomenon which we can approach
with the questions and terms of epidemiology. What is needed for a
child to acquire, or to not acquire, a skill in the ways of
interpreting these signals?

There is not only the miracle that any of them acquire the skills
—and a lot of them do—there is also the other side, that a great
many people have difficulty. There are people, for example, who,
when Big Sister in the soap opera suffers from a cold, will send a
bottle of aspirin to the radio station or recommend a cure for Big
Sister’s cold, in spite of the fact that Big Sister is a fictitious

character within a radio soap opera. These particular members of
the audience are apparently a little bit askew in their identification
of what sort of a communication this is that is coming from their

We all make errors of that kind at various times. I’m not sure
that I’ve ever met anybody that doesn’t suffer from “schizophrenia
P” more or less. We all have some difficulty in deciding sometimes
whether a dream was a dream or not, and it would not be very easy
for most of us to say how we know that a piece of our own fantasy
is fantasy and not experience. The ability to place an experience in
time is one of the important cues, and referring it to a sense organ
is another.

When you look at the mothers and fathers of patients for an
answer to this etiological question, you meet with several sorts of

First of all there are answers connected with what we may call
the intensifying factors. Any disease is made worse or more
probable by various circumstances, such as fatigue, cold, the
number of days of combat, the presence of other diseases, etc.
These seem to have a quantitative effect upon the incidence of
almost any pathology. Then there are those factors which I
mentioned—the hereditary characteristics and potentialities. To get
confused about the logical types, one presumably has to be
intelligent enough to know that there is something wrong, and not
so intelligent as to be able to see what it is that is wrong. I presume
that these characteristics are hereditarily determined.

But the nub of the problem, it seems to me, is to identify what
real circumstances lead to the specific pathology. I acknowledge that
the bacteria are not really by any means the sole determinant of a
bacterial disease, and grant also therefore that the occurrence of
such traumatic sequences or contexts is not by any means the sole
determinant of mental illness. But still it seems to me that the
identification of those contexts is the nub of understanding the
disease, as identifying the bacteria is essential to understanding a
bacterial disease.

I have met the mother of the patient whom I mentioned earlier.
The family is not badly off. They live in a nice tract house. I went
there with the patient, and when we arrived nobody was home. The

newspaper boy had tossed the evening paper out in the middle of
the lawn, and my patient wanted to get that paper from the middle
of that perfect lawn. He came to the edge of the lawn and started to

The house looks like what is called a “model” home—a house
which has been furnished by the real estate people in order to sell
other houses to the public. Not a house furnished to live in, but
rather furnished to look like a furnished house.

I discussed his mother with him one day, and suggested that
perhaps she was a rather frightened person. He said, “Yes.” I said,
“What is she frightened of?” He said, “The appeariential

There is a beautiful, perfectly centered mass of artificial, plastic
vegetation on the middle of the mantle. A china pheasant here and
a china pheasant there, symmetrically arranged. The wall-to-wall
carpet is exactly as it should be.

After his mother arrived, I felt a little uncomfortable, intruding in
this house. He had not visited there for about five years, but things
seemed to be going all right, so I decided to leave him there and to
come back when it was time to go back to the hospital. That gave
me an hour in the streets with absolutely nothing to do, and I began
to think what I would like to do to this setup. What and how could I
communicate? I decided that I would like to put into it something
that was both beautiful and untidy. In trying to implement that
decision, I decided that flowers were the answer, so I bought some
gladioluses. I took the gladioluses, and, when I went to get him, I
presented them to the mother with a speech that I wanted her to have
in her house something that was “both beautiful and untidy.” “Ohl”
she said, “Those are not untidy flowers. As each one withers, you
can snip it off.”

Now, as I see it, what is interesting is not so much the castrative
statement in that speech, but the putting me in the position of
having apologized when in fact I had not. That is, she took my
message and reclassified it. She changed the label which indicated
what sort of a message it was, and that is, I believe, what she does
all the time. An endless taking of the other person’s message and
replying to it as if it were either a statement of weakness on the

part of the speaker or an attack on her which should be turned into
a weakness on the part of the speaker; and so on.

What the patient is up against today—and was up against in
childhood—is the false interpretation of his messages. If he says,
“The cat is on the table,” she replies with some reply which makes
out that his message is not the sort of message that he thought it
was when he gave it. His own message identifier is obscured or
distorted by her when the message comes back. And her own
message identifier she continually contradicts. She laughs when
she is saying that which is least funny to her, and so on.

Now there is a regular maternal dominance picture in this
family, but I am not concerned at the moment to say that this is the
necessary form of the trauma. I am only concerned with the purely
formal aspects of this traumatic constellation; and I presume the
constellation could be made up with father taking certain parts of
it, mother taking certain other parts of it, and so forth.

I am trying to make only one point: that there is here a
probability of trauma which will contain certain formal characteristics.
It will propagate a specific syndrome in the patient
because the trauma itself has impact upon a certain element in the
communicational process. That which is at-tacked is the use of
what I have called the “message-identifying signals”—those
signals without which the “ego” dare not discriminate fact from
fantasy or the literal from the metaphoric.

What I tried to do was pinpoint a group of syndromata, namely
those syndromata related to an inability to know what sort of a
message a message is. At one end of the classification of those,
there will be more or less hebephrenic individuals for whom no
message is of any particular definite type but who live in a sort of
chronic shaggy-dog story. At the other end are those who try to
overidentify, to make an overly rigid identification of what sort of
a message every message is. This will give a much more paranoid
type of picture. Withdrawal is another possibility.

Finally, it seems to me that with a hypothesis of this kind, one
could look for the determinants in a population which might lead
to the occurrence of that sort of constellation. This would seem to
me an appropriate matter for epidemiological study.

Toward a Theory of Schizophrenia*

Schizophrenia—its nature, etiology, and the kind of therapy to
use for it—remains one of the most puzzling of the mental illnesses.
The theory of schizophrenia presented here is based on
communications analysis, and specifically on the Theory of Logical
Types. From this theory and from observations of schizophrenic
patients is derived a description, and the necessary conditions for, a
situation called the “double bind”—a situation in which no matter
what a person does, he “can’t win.” It is hypothesized that a person
caught in the double bind may develop schizophrenic symptoms.
How and why the double bind may arise in a family situation is discussed,
together with illustrations from clinical and experimental

This is a report24 on a research project which has been
formulating and testing a broad, systematic view of the nature,
etiology, and therapy of schizophrenia. Our research in this field
has proceeded by discussion of a varied body of data and
ideas, with all of us contributing according to our varied
experience in anthropology, communications analysis,
psychotherapy, psychiatry, and psychoanalysis. We have now

* This paper by Gregory Bateson, Don D. Jackson, Jay Haley, and. John H.
Weakland is here reproduced from Behavioral Science, Vol. I, No. 4, 1956, by
permission of Behavioral Science
24 This paper derives from hypotheses first developed in a research project
financed by the Rockfeller Foundation from 1952-54, administered by the
Department of Sociology and Anthropology at Stanford University and directed by
Gregory Bateson. Since 1954 the project has financed by the Josiah Macy, Jr.
Foundation. To Jay Haley is due credit for recognizing that the symptoms of
schizophrenia are suggestive of an inability to discriminate the Logical
Types, and this was amplified by Bateson, who added the notion that the
symptoms and etiology could be formally described in terms of a double bind
hypothesis. The hypothesis was communicated to D. D. Jackson and found to fit
closely with his ideas of family homeostasis. Since then Dr. Jackson has
worked closely with the project. The study of the formal analogies between
hypnosis and schizophrenia has been the work of John H. Weakland and Jay


reached common agreement on the broad outlines of a
communicational theory of the origin and nature of
schizophrenia; this paper is a preliminary report on our continuing

The Base in Communications Theory

Our approach is based on that part of communications theory
which Russell has called the Theory of Logical Types.25 The central
thesis of this theory is that there is a discontinuity between a class
and its members. The class cannot be a member of itself nor can one
of the members be the class, since the term used for the class is of a
different level of abstraction—a different Logical Type—from
terms used for members. Although in formal logic there is an attempt
to maintain this discontinuity between a class and its
members, we argue that in the psychology of real communications
this discontinuity is continually and inevitably breached,26 and that a
priori we must expect a pathology to occur in the human organism
when certain formal pat-terns of the breaching occur in the
communication between mother and child. We shall argue that this
pathology at its extreme will have symptoms whose formal
characteristics would lead the pathology to be classified as a

Illustrations of how human beings handle communication
involving multiple Logical Types can be derived from the following

1. The use of various communicational modes in human
communication. Examples are play, nonplay, fantasy, sacrament,
metaphor, etc. Even among the lower mammals there appears to be
an exchange of signals which identify certain meaningful behavior
as “play,” etc.27 These signals are evidently of higher Logical Type
than the messages they classify. Among human beings this framing
25 A. N. Whitehead and B. Russell, Principia Mathematica, Cambridge,
Cambridge University Press, 1910.

26 G. Bateson, “A Theory of Play and Fantasy,” Psychiatric Research Reports,
1955, 2: 39-51.

and labeling of messages and meaningful actions reaches
considerable complexity, with the peculiarity that our vocabulary for
such discrimination is still very poorly developed, and we rely
preponderantly upon nonverbal media of posture, gesture, facial expression,
intonation, and the context for the communication of these
highly abstract, but vitally important, labels.

2. Humor. This seems to be a method of exploring the implicit
themes in thought or in a relationship. The method of exploration
involves the use of messages which are characterized by a
condensation of Logical Types or communicational modes. A
discovery, for example, occurs when it suddenly becomes plain that
a message was not only metaphoric but also more literal, or vice
versa. That is to say, the explosive moment in humor is the moment
when the labeling of the mode undergoes a dissolution and
resynthesis. Commonly, the punch line compels a re-evaluation of
earlier signals which ascribed to certain messages a particular mode
(e.g., literalness or fantasy). This has the peculiar effect of
attributing mode to those signals which had previously the status of
that higher Logical Type which classifies the modes.
3. The falsification of mode-identifying signals. Among human
beings mode identifiers can be falsified, and we have the artificial
laugh, the manipulative simulation of friendliness, the confidence
trick, kidding, and the like. Similar falsifications have been recorded
among mammals.28 Among human beings we meet with a strange
phenomenon—the unconscious falsification of these signals. This
may occur within the self—the subject may conceal from himself
his own real hostility under the guise of metaphoric play—or it may
occur as an unconscious falsification of the subject’s understanding
of the other person’s mode-identifying signals. He may mistake
shyness for contempt, etc. Indeed most of the errors of self-reference
fall under this head.
4. Learning. The simplest level of this phenomenon is
exemplified by a situation in which a subject receives a message and
27 A film prepared by this project, “The Nature of Play; Part I, River Otters,” is

28 C. R. Carpenter, “A Field Study of the Behavior and Social Relations of
Howling Monkeys,” Comp. Psychol. Monogr., 1934, 10: 1–168; also K. Z. Lorenz,
King Solomon’s Ring, New York, Crowell, 1952.


acts appropriately on it: “I heard the clock strike and knew it was
time for lunch. So I went to the table.” In learning experiments the
analogue of this sequence of events is observed by the experimenter
and commonly treated as a single message of a higher type. When
the dog salivates between buzzer and meat powder, this sequence is
accepted by the experimenter as a message indicating that “The dog
has learned that buzzer means meat powder.” But this is not the end
of the hierarchy of types involved. The experimental subject may
become more skilled in learning. He may learn to learn,29 and it is
not inconceivable that still higher orders of learning may occur in
human beings.

5. Multiple levels of learning and the Logical Typing of signals.
These are two inseparable sets of phenomena—inseparable because
the ability to handle the multiple types of signals is itself a learned
skill and therefore a function of the multiple levels of learning.
According to our hypothesis, the term “ego function” (as this
term is used when a schizophrenic is described as having “weak ego
function”) is precisely the process of discriminating
communicational modes either within the self or between the
self and others. The schizophrenic exhibits weakness in three areas
of such function: (a) He has difficulty in assigning the correct
communicational mode to the messages he receives from other
persons. (b) He has difficulty in assigning the correct
communicational mode to those messages which he himself utters or
emits nonverbally. (c) He has difficulty in assigning the correct
communicational mode to his own thoughts, sensations, and

At this point it is appropriate to compare what was said in the
previous paragraph with von Domarus’7 approach to the systematic
description of schizophrenic utterance. He suggests that the
messages (and thought) of the schizophrenic are deviant in
syllogistic structure. In place of structures which derive from the
syllogism, Barbara, the schizophrenic, according to this theory, uses

29 G. Bateson, “Social Planning and the Concept of Deutero-Learning,”
Conference on Science, Philosophy and Religion, Second Symposium, New York,
Harper, 1942. (See above, p. 159) ; also H. F. Harlow, “The Formation of Learning
Sets,” Psychol. Review, 1949, 56: 51–65; also C. L. Hull, et al., Mathematicodeductive
Theory of Rote Learning, New Haven, Yale University Press, 1940.

structures which identify predicates. An example of such a distorted

syllogism is:

Men die.

Grass dies.

Men are grass.

But as we see it, von Domarus30 formulation is only a more
precise—and therefore valuable—way of saying that schizophrenic
utterance is rich in metaphor. With that generalization we agree. But
metaphor is an indispensable tool of thought and expression—a
characteristic of all human communication, even of that of the
scientist. The conceptual models of cybernetics and the energy
theories of psycho-analysis are, after all, only labeled metaphors.
The peculiarity of the schizophrenic is not that he uses metaphors,
but that he uses unlabeled metaphors. He has special difficulty in
handling signals of that class whose members assign Logical Types
to other signals.

If our formal summary of the symptomatology is correct and if
the schizophrenia of our hypothesis is essentially a result of family
interaction, it should be possible to arrive a priori at a formal
description of these sequences of experience which would induce
such a symptomatology. What is known of learning theory combines
with the evident fact that human beings use context as a guide for
mode discrimination. Therefore, we must look not for some specific
traumatic experience in the infantile etiology but rather for
characteristic sequential patterns. The specificity for which we
search is to be at an abstract or formal level. The sequences must
have this characteristic: that from them the patient will acquire the
mental habits which are exemplified in schizophrenic
communication. That is to say, he must live in a universe where the
sequences of events are such that his unconventional
communicational habits will be in some sense appropriate. The
hypothesis which we offer is that sequences of this kind in the
external experience of the patient are responsible for the inner

30 E. von Domarus, “The Specific Laws of Logic in Schizophrenia,” Language
and Thought in Schizophrenia, J. S. Kasanin, ed., Berkeley, University of
California Press, 1944.

conflicts of Logical Typing. For such unresolvable sequences of
experiences, we use the term “double bind.”

The Double Bind

The necessary ingredients for a double bind situation, as we see
it, are:

1. Two or more persons. Of these, we designate one, for
purposes of our definition, as the “victim.”. We do not assume that
the double bind is inflicted by the mother alone, but that it may be
done either by mother alone or by some combination of mother,
father, and/or siblings.
2. Repeated experience. We assume that the double bind is a
recurrent theme in the experience of the victim. Our hypothesis does
not invoke a single traumatic experience, but such repeated
experience that the double bind structure comes to be an habitual
3. A primary negative injunction. This may have either of two
forms: (a) “Do not do so and so, or I will punish you,” or (b) “If you
do not do so and so, I will punish you.” Here we select a context of
learning based on avoidance of punishment rather than a context of
reward seeking. There is perhaps no formal reason for this selection.
We assume that the punishment may be either the withdrawal of
love or the expression of hate or anger—or most devastating—the
kind of abandonment that results from the parent’s expression of
extreme helplessness.31
4. A secondary injunction conflicting with the first at amore
abstract level, and like the first enforced by punishments or signals
which threaten survival. This secondary injunction is more difficult
to describe than the primary for two reasons. First, the secondary
injunction is commonly communicated to the child by nonverbal
means. Posture, gesture, tone of voice, meaningful action, and the
implications concealed in verbal comment may all be used to
31 Our concept of punishment is being refined at present. It appears to us to
involve perceptual experience in a way that cannot be encompassed by the notion of

convey this more abstract message. Second, the secondary injunction
may impinge upon any element of the primary prohibition.
Verbalization of the secondary injunction may, there-fore, include a
wide variety of forms; for example, “Do not see this as punishment”;
“Do not see me as the punishing agent”; “Do not submit to my
prohibitions”; “Do not think of what you must not do”; “Do not
question my love of which the primary prohibition is (or is not) an
example”; and so on. Other examples become possible when the
double bind is inflicted not by one individual but by two. For example,
one parent may negate at a more abstract level the
injunctions of the other.

5. A tertiary negative injunction prohibiting the victim from
escaping from the field. In a formal sense it is perhaps unnecessary
to list this injunction as a separate item since the reinforcement at
the other two levels involves a threat to survival, and if the double
binds are imposed during infancy, escape is naturally impossible.
However, it seems that in some cases the escape from the field is
made impossible by certain devices which are not purely negative,
e.g., capricious promises of love, and the like.
6. Finally, the complete set of ingredients is no longer necessary
when the victim has learned to perceive his universe in double bind
patterns. Almost any part of a double bind sequence may then be
sufficient to precipitate panic or rage.
The pattern of conflicting injunctions may even be taken over by
hallucinatory voices.32

The Effect of the Double Bind

In the Eastern religion, Zen Buddhism, the goal is to achieve
enlightenment. The Zen master attempts to bring about

32 J. Perceval, A Narrative of the Treatment Experienced by a Gentleman During
a State of Mental Derangement, Designed to Explain the Causes and Nature of
Insanity, etc., London, Effingham Wilson, 1836 and 1840. (See bibliographic item,
1961 a.)

enlightenment in his pupil in various ways. One of the things he
does is to hold a stick over the pupil’s head and say fiercely, “If you
say this stick is real, I will strike you with it. If you say this stick is
not real, I will strike you with it. If you don’t say anything, I will
strike you with it.” We feel that the schizophrenic finds himself
continually in the same situation as the pupil, but he achieves
something like disorientation rather than enlightenment. The Zen
pupil might reach up and take the stick away from the master—who
might accept this response, but the schizophrenic has no such choice
since with him there is no not caring about the relationship, and his
mother’s aims and awareness are not like the master’s.

We hypothesize that there will be a breakdown in any
individual’s ability to discriminate between Logical Types whenever
a double bind situation occurs. The general characteristics of this
situation are the following:

(A) When the individual is involved in an intense relationship;
that is, a relationship in which he feels it is vitally important that he
discriminate accurately what sort of message is being communicated
so that he may respond appropriately.
(B) And, the individual is caught in a situation in which the other
person in the relationship is expressing two orders of message and
one of these denies the other.
(C) And, the individual is unable to comment on the messages
being expressed to correct his discrimination of what order of
message to respond to, i.e., he cannot make a metacommunicative
We have suggested that this is the sort of situation which occurs
between the preschizophrenic and his mother, but it also occurs in
normal relationships. When a person is caught in a double bind
situation, he will respond defensively in a manner similar to the
schizophrenic. An individual will take a metaphorical statement
literally when he is in a situation where he must respond, where he
is faced with contradictory messages, and when he is unable to
comment on the contradictions. For example, one day an employee
went home during office hours. A fellow employee called him at his
home, and said lightly, “Well, how did you get there?” The
employee replied, “By automobile.” He responded literally because
he was faced with a message which asked him what he was doing at


home when he should have been at the office, but which denied that
this question was being asked by the way it was phrased. (Since the
speaker felt it wasn’t really his business, he spoke metaphorically.)
The relationship was intense enough so that the victim was in doubt
how the information would be used, and he therefore responded
literally. This is characteristic of anyone who feels “on the spot,” as
demonstrated by the careful literal replies of a witness on the stand
in a court trial. The schizophrenic feels so terribly on the spot at all
times that he habitually responds with a defensive insistence on the
literal level when it is quite inappropriate, e.g., when someone is

Schizophrenics also confuse the literal and metaphoric in their
own utterance when they feel themselves caught in a double bind.
For example, a patient may wish to criticize his therapist for being
late for an appointment, but he may be unsure what sort of a
message that act of being late was—particularly if the therapist has
anticipated the patient’s reaction and apologized for the event. The
patient cannot say, “Why were you late? Is it because you don’t want
to see me today?” This would be an accusation, and so he shifts to a
metaphorical statement. He may then say, “I knew a fellow once
who missed a boat, his name was Sam and the boat almost sunk, . . .
etc.,” Thus he develops a metaphorical story and the therapist may
or may not discover in it a comment on his being late. The
convenient thing about a metaphor is that it leaves it up to the
therapist (or mother) to see an accusation in the statement if he
chooses, or to ignore it if he chooses. Should the therapist accept the
accusation in the metaphor, then the patient can accept the statement
he has made about Sam as metaphorical. If the therapist points out
that this doesn’t sound like a true statement about Sam, as a way of
avoiding the accusation in the story, the patient can argue that there
really was a man named Sam. As an answer to the double bind
situation, a shift to a metaphorical statement brings safety. However,
it also prevents the patient from making the accusation he wants to
make. But instead of getting over his accusation by indicating that
this is a metaphor, the schizophrenic patient seems to try to get over
the fact that it is a metaphor by making it more fantastic. If the
therapist should ignore the accusation in the story about Sam, the
schizophrenic may then tell a story about going to Mars in a rocket

ship as a way of putting over his accusation. The indication that it is
a metaphorical statement lies in the fantastic aspect of the metaphor,
not in the signals which usually accompany metaphors to tell the
listener that a metaphor is being used.

It is not only safer for the victim of a double bind to shift to a
metaphorical order of message, but in an impossible situation it is
better to shift and become somebody else, or shift and insist that he
is somewhere else. Then the double bind cannot work on the victim,
because it isn’t he and besides he is in a different place. In other
words, the statements which show that a patient is disoriented can be
interpreted as ways of defending himself against the situation he is
in. The pathology enters when the victim himself either does not
know that his responses are metaphorical or cannot say so. To
recognize that he was speaking metaphorically he would need to be
aware that he was defending himself and therefore was afraid of the
other person. To him such an awareness would be an indictment of
the other person and therefore provoke disaster.

If an individual has spent his life in the kind of double bind
relationship described here, his way of relating to people after a
psychotic break would have a systematic pat-tern. First, he would
not share with normal people those signals which accompany
messages to indicate what a person means. His metacommunicative
system—the communications about communication—would have
broken down, and he would not know what kind of message a
message was. If a person said to him, “What would you like to do
today?” he would be unable to judge accurately by the context or by
the tone of voice or gesture whether he was being condemned for
what he did yesterday, or being offered a sexual invitation, or just
what was meant. Given this in-ability to judge accurately what a
person really means and an excessive concern with what is really
meant, an individual might defend himself by choosing one or more
of several alternatives. He might, for example, assume that behind
every statement there is a concealed meaning which is detrimental to
his welfare. He would then be excessively concerned with hidden
meanings and determined to demonstrate that he could not be
deceived—as he had been all his life. If he chooses this alternative,
he will be continually searching for meanings behind what people


say and behind chance occurrences in the environment, and he will
be characteristically suspicious and defiant.

He might choose another alternative, and tend to accept literally
everything people say to him; when their tone or gesture or context
contradicted what they said, he might establish a pattern of laughing
off these metacommunicative signals. He would give up trying to
discriminate between levels of message and treat all messages as
unimportant or to be laughed at.

If he didn’t become suspicious of metacommunicative messages
or attempt to laugh them off, he might choose to try to ignore them.
Then he would find it necessary to see and hear less and less of what
went on around him, and do his utmost to avoid provoking a
response in his environment. He would try to detach his interest
from the external world and concentrate on his own internal
processes and, therefore, give the appearance of being a withdrawn,
perhaps mute, individual.

This is another way of saying that if an individual doesn’t know
what sort of message a message is, he may defend himself in ways
which have been described as paranoid, hebephrenic, or catatonic.
These three alternatives are not the only ones. The point is that he
cannot choose the one alternative which would help him to discover
what people mean; he cannot, without considerable help, discuss the
messages of others. Without being able to do that, the human being
is like any self-correcting system which has lost accept the
accusation in the metaphor, then the patient can accept the
statement he has made about Sam as metaphorical. If the therapist
points out that this doesn’t sound like a true statement about Sam,
as a way of avoiding the accusation in the story, the patient can
argue that there really was a man named Sam. As an answer to the
double bind situation, a shift to a metaphorical statement brings
safety. However, it also prevents the patient from making the
accusation he wants to make. But instead of getting over his
accusation by indicating that this is a metaphor, the schizophrenic
patient seems to try to get over the fact that it is a metaphor by
making it more fantastic. If the therapist should ignore the
accusation in the story about Sam, the schizophrenic may then tell
a story about going to Mars in a rocket ship as a way of putting
over his accusation. The indication that it is a metaphorical

statement lies in the fantastic aspect of the metaphor, not in the
signals which usually accompany metaphors to tell the listener that
a metaphor is being used.

It is not only safer for the victim of a double bind to shift to a
metaphorical order of message, but in an impossible situation it is
better to shift and become somebody else, or shift and insist that he
is somewhere else. Then the double bind cannot work on the victim,
because it isn’t he and besides he is in a different place. In other
words, the statements which show that a patient is disoriented can be
interpreted as ways of defending himself against the situation he is
in. The pathology enters when the victim himself either does not
know that his responses are metaphorical or cannot say so. To
recognize that he was speaking metaphorically he would need to be
aware that he was defending himself and therefore was afraid of the
other person. To him such an awareness would be an indictment of
the other person and therefore provoke disaster.

If an individual has spent his life in the kind of double bind
relationship described here, his way of relating to people after a
psychotic break would have a systematic pat-tern. First, he would
not share with normal people those signals which accompany
messages to indicate what a person means. His metacommunicative
system—the communications about communication—would have
broken down, and he would not know what kind of message a
message was. If a person said to him, “What would you like to do
today?” he would be unable to judge accurately by the context or by
the tone of voice or gesture whether he was being condemned for
what he did yesterday, or being offered a sexual invitation, or just
what was meant. Given this in-ability to judge accurately what a
person really means and an excessive concern with what is really
meant, an individual might defend himself by choosing one or more
of several alternatives. He might, for example, assume that behind
every statement there is a concealed meaning which is detrimental to
his welfare. He would then be excessively concerned with hidden
meanings and determined to demonstrate that he could not be
deceived—as he had been all his life. If he chooses this alternative,
he will be continually searching for meanings behind what people
say and behind chance occurrences in the environment, and he will
be characteristically suspicious and defiant.


He might choose another alternative, and tend to accept literally
everything people say to him; when their tone or gesture or context
contradicted what they said, he might establish a pattern of laughing
off these metacommunicative signals. He would give up trying to
discriminate between levels of message and treat all messages as
unimportant or to be laughed at.

If he didn’t become suspicious of metacommunicative messages
or attempt to laugh them off, he might choose to try to ignore them.
Then he would find it necessary to see and hear less and less of what
went on around him, and do his utmost to avoid provoking a
response in his environment. He would try to detach his interest
from the external world and concentrate on his own internal
processes and, therefore, give the appearance of being a withdrawn,
perhaps mute, individual.

This is another way of saying that if an individual doesn’t know
what sort of message a message is, he may defend himself in ways
which have been described as paranoid, hebephrenic, or catatonic.
These three alternatives are not the only ones. The point is that he
cannot choose the one alternative which would help him to discover
what people mean; he cannot, without considerable help, discuss the
messages of others. Without being able to do that, the human being
is like any self-correcting system which has lost its governor; it
spirals into never-ending, but always systematic, distortions.

A Description of the Family Situation

The theoretical possibility of double bind situations stimulated us
to look for such communication sequences in the schizophrenic
patient and in his family situation. Toward this end we have studied
the written and verbal reports of psychotherapists who have treated
such patients intensively; we have studied tape recordings of
psychotherapeutic inter-views, both of our own patients and others;
we have inter-viewed and taped parents of schizophrenics; we have
had two mothers and one father participate in intensive psychotherapy;
and we have interviewed and taped parents and patients
seen conjointly.

On the basis of these data we have developed a hypothesis about
the family situation which ultimately leads to an individual suffering
from schizophrenia. This hypothesis has not been statistically tested;
it selects and emphasizes a rather simple set of interactional
phenomena and does not attempt to describe comprehensively the
extraordinary complexity of a family relationship.
We hypothesize that the family situation of the schizophrenic hasthe following general characteristics:

(1) A child whose mother becomes anxious and with-draws if the
child responds to her as a loving mother. That is, the child’s very
existence has a special meaning to the mother which arouses her
anxiety and hostility when she is in danger of intimate contact with
the child.
(2) A mother to whom feelings of anxiety and hostility toward
the child are not acceptable, and whose way of denying them is to
express overt loving behavior to persuade the child to respond to her
as a loving mother and to with-draw from him if he does not.
“Loving behavior” does not necessarily imply “affection”; it can, for
example, be set in a framework of doing the proper thing, instilling
“goodness,” and the like.
(3) The absence of anyone in the family, such as a strong and
insightful father, who can intervene in the relationship between the
mother and child and support the child in the face of the
contradictions involved.
Since this is a formal description we are not specifically
concerned with why the mother feels this way about the child, but
we suggest that she could feel this way for various reasons. It may
be that merely having a child arouses anxiety about herself and her
relationships to her own family; or it may be important to her that
the child is a boy or a girl, or that the child was born on the
anniversary of one of her own siblings,33 or the child may be in the
same sibling position in the family that she was, or the child may be
special to her for other reasons related to her own emotional

33 R. Hilgard, “Anniversary Reactions in Parents Precipitated by Children,”
Psychiatry, 1953, 16: 73-80.

Given a situation with these characteristics, we hypothesize that
the mother of a schizophrenic will be simultaneously expressing at
least two orders of message. (For simplicity in this presentation we
shall confine ourselves to two orders.) These orders of message can
be roughly characterized as (a) hostile or withdrawing behavior
which is aroused when-ever the child approaches her, and (b)
simulated loving or approaching behavior which is aroused when
the child responds to her hostile and withdrawing behavior, as a way
of denying that she is withdrawing. Her problem is to control her
anxiety by controlling the closeness and distance between herself
and her child. To put this another way, if the mother begins to feel
affectionate and close to her child, she begins to feel endangered and
must withdraw from him; but she cannot accept this hostile act and
to deny it must simulate affection and closeness with her child. The
important point is that her loving behavior is then a comment on
(since it is compensatory for) her hostile behavior and consequently
it is of a different order of message than the hostile behavior—it is a
message about a sequence of messages. Yet by its nature it denies
the existence of those messages which it is about, i.e., the hostile

The mother uses the child’s responses to affirm that her behavior
is loving, and since the loving behavior is simulated, the child is
placed in a position where he must not accurately interpret her
communication if he is to maintain his relationship with her. In
other words, he must not discriminate accurately between orders of
message, in this case the difference between the expression of
simulated feelings (one Logical Type) and real feelings (another
Logical Type). As a result the child must systematically distort his
perception of metacommunicative signals. For ex-ample, if mother
begins to feel hostile (or affectionate) to-ward her child and also
feels compelled to withdraw from him, she might say, “Go to bed,
you ‘re very tired and I want you to get your sleep.” This overtly
loving statement is intended to deny a feeling which could be.
verbalized as “Get out of my sight because I’m sick of you.” If the
child correctly discriminates her metacommunicative signals, he
would have to face the fact that she both doesn’t want him and is
deceiving him by her loving behavior. He would be “punished” for
learning to discriminate orders of messages accurately. He

therefore would tend to accept the idea that he is tired rather than
recognize his mother’s deception. This means that he must deceive
himself about his own internal state in order to support mother in
her deception. To survive with her he must falsely discriminate his
own internal messages as well as falsely discriminate the messages
of others.

The problem is compounded for the child because the mother is
“benevolently” defining for him how he feels; she is expressing
overt maternal concern over the fact that he is tired. To put it another
way, the mother is controlling the child’s definitions of his own
messages, as well as the definition of his responses to her (e.g., by
saying, “You don’t really mean to say that,” if he should criticize
her) by insisting that she is not concerned about herself but only
about him. Consequently, the easiest path for the child is to accept
mother’s simulated loving behavior as real, and his desires to
interpret what is going on are undermined. Yet the result is that the
mother is withdrawing from him and defining this withdrawal as the
way a loving relationship should be.

However, accepting mother’s simulated loving behavior as real
also is no solution for the child. Should he make this false
discrimination, he would approach her; this move to-ward closeness
would provoke in her feelings of fear and helplessness, and she
would be compelled to withdraw. But if he then withdrew from her,
she would take his withdrawal as a statement that she was not a
loving mother and would either punish him for withdrawing or
approach him to bring him closer. If he then approached, she would
respond by putting him at a distance. The child is punished for
discriminating accurately what she is expressing, and he is punished
for discriminating inaccurately—he is caught in a double bind.

The child might try various means of escaping from this
situation. He might, for example, try to lean on his father or some
other member of the family. However, from our preliminary
observations we think it is likely that the fathers of schizophrenics
are not substantial enough to lean on. They are also in the awkward
position where if they agreed with the child about the nature of
mother’s deceptions, they would need to recognize the nature of
their own relation-ships to the mother, which they could not do and
remain attached to her in the modus operandi they have worked out.


The need of the mother to be wanted and loved also prevents the
child from gaining support from some other person in the
environment, a teacher, for example. A mother with these
characteristics would feel threatened by any other attachment of the
child and would break it up and bring the child back closer to her
with consequent anxiety when the child became dependent on her.

The only way the child can really escape from the situation is to
comment on the contradictory position his mother has put him in.
However, if he did so, the mother would take this as an accusation
that she is unloving and both punish him and insist that his
perception of the situation is distorted. By preventing the child from
talking about the situation, the mother forbids him using the
metacommunicative level—the level we use to correct our
perception of communicative behavior. The ability to communicate
about communication, to comment upon the meaningful actions of
oneself and others, is essential for successful social inter-course. In
any normal relationship there is a constant inter-change of
metacommunicative messages such as “What do you mean?” or
“Why did you do that?” or “Are you kidding me?” and so on. To
discriminate accurately what people are really expressing, we must
be able to comment directly or indirectly on that expression. This
metacommunicative level the schizophrenic seems unable to use
successfully.34 Given these characteristics of the mother, it is
apparent why. If she is denying one order of message, then any
statement about her statements endangers her and she must forbid it.
Therefore, the child grows up unskilled in his ability to communicate
about communication and, as a result, unskilled in
determining what people really mean and unskilled in expressing
what he really means, which is essential for normal relationships.

In summary, then, we suggest that the double bind nature of the
family situation of a schizophrenic results in placing the child in a
position where, if he responds to his mother’s simulated affection,
her anxiety will be aroused and she will punish him (or insist, to
protect herself, that his overtures are simulated, thus confusing him
about the nature of his own messages) to defend herself from close

34 G. Bateson, “. . . Play and Fantasy,” op. cit.

ness with him. Thus the child is blocked off from intimate and
secure associations with his mother. However, if he does not make
overtures of affection, she will feel that this means she is not a
loving mother and her anxiety will be aroused. Therefore, she will
either punish him for with-drawing or make overtures toward the
child to insist that he demonstrate that he loves her. If he then
responds and shows her affection, she will not only feel endangered
again, but she may resent the fact that she had to force him to
respond. In either case in a relationship, the most important in his
life and the model for all others, he is punished if he indicates love
and affection and punished if he does not; and his escape routes
from the situation, such as gaining support from others, are cut off.
This is the basic nature of the double bind relationship between
mother and child. This description has not depicted, of course, the
more complicated interlocking gestalt that is the “family” of which
the “mother” is one important part.35

Illustrations from Clinical Data

An analysis of an incident occurring between a schizophrenic
patient and his mother illustrates the double bind situation. A young
man who had fairly well recovered from an acute schizophrenic
episode was visited in the hospital by his mother. He was glad to see
her and impulsively put his arm around her shoulders, whereupon
she stiffened. He withdrew his arm and she asked, “Don’t you love
me any more?” He then blushed, and she said, “Dear, you must not
be so easily embarrassed and afraid of your feelings.” The patient
was able to stay with her only a few minutes more and following her
departure he assaulted an aide and was put in the tubs.

Obviously, this result could have been avoided if the young man
had been able to say, “Mother, it is obvious that you become
uncomfortable when I put my arm around you, and that you have

35 D. D. Jackson, “The Question of Family Homeostasis,” presented at the American
Psychiatric Association Meeting, St. Louis, May 7, 1954; also Jackson, “Some Factors
Influencing the Oedipus Complex,” Psychoanalytic Quarterly, 1954, 23: 566-81.

difficulty accepting a gesture of affection from me.” However, the
schizophrenic patient doesn’t have this possibility open to him. His
intense dependency and training prevents him from commenting
upon his mother’s communicative behavior, though she comments
on his and forces him to accept and to attempt to deal with the complicated
sequence. The complications for the patient include the

(1) The mother’s reaction of not accepting her son’s affectionate
gesture is masterfully covered up by her condemnation of him for
withdrawing, and the patient denies his perception of the situation
by accepting her condemnation.
(2) The statement “Don’t you love me any more” in this context
seems to imply:
(a) “I am lovable.”
(b) “You should love me and if you don’t you are bad or at fault.”
(c) “Whereas you did love me previously you don’t any longer,”
and thus focus is shifted from his expressing affection to his
inability to be affectionate. Since the patient has also hated her, she
is on good ground here, and he responds appropriately with guilt,
which she then attacks.
(d) “What you just expressed was not affection,” and in order to
accept this statement, the patient must deny what she and the culture
have taught him about how one expresses affection. He must also
question the times with her, and with others, when he thought he
was experiencing affection and when they seemed to treat the
situation as if he had. He experiences here loss-of-support phenomena
and is put in doubt about the reliability of past experience.
(3) The statement, “You must not be so easily embarrassed and
afraid of your feelings,” seems to imply:
(a) “You are not like me and are different from other nice or
normal people because we express our feelings.”
(b) “The feelings you express are all right, it’s only that you can’t
accept them.” However, if the stiffening on her part had indicated
“These are unacceptable feelings,” then the boy is told that he
should not be embarrassed by unacceptable feelings. Since he has
had a long training in what is and is not acceptable to both her and
society, he again comes into conflict with the past. If he is unafraid
of his own feelings (which mother implies is good), he should be

unafraid of his affection and would then notice it was she who was
afraid, but he must not notice that be-cause her whole approach is
aimed at covering up this short-coming in herself.

The impossible dilemma thus becomes: “If I am to keep my tie to
mother, I must not show her that I love her, but if I do not show her
that I love her, then I will lose her.”

The importance to the mother of her special method of control is
strikingly illustrated by the interfamily situation of a young woman
schizophrenic who greeted the therapist on their first meeting with
the remark, “Mother had to get married and now I’m here.” This
statement meant to the therapist that:

(1) The patient was the result of an illegitimate pregnancy.
(2) This fact was related to her present psychosis (in her
(3) “Here” referred to the psychiatrist’s office and to the patient’s
presence on earth for which she had to be eternally indebted to her
mother, especially since her mother had sinned and suffered in order
to bring her into the world.
(4) “Had to get married” referred to the shotgun nature of
mother’s wedding and to the mother’s response to pressure that she
must marry, and the reciprocal, that she resented the forced nature of
the situation and blamed the patient for it.
Actually, all these suppositions subsequently proved to be
factually correct and were corroborated by the mother during an
abortive attempt at psychotherapy. The flavor of the mother’s
communications to the patient seemed essentially this: “I am
lovable, loving, and satisfied with myself. You are lovable when you
are like me and when you do what I say.” At the same time the
mother indicated to the daughter both by words and behavior: “You
are physically delicate, unintelligent, and different from me (`not
normal’). You need me and me alone because of these handicaps,
and I will take care of you and love you.” Thus the patient’s life was
a series of beginnings, of attempts at experience, which would result
in failure and withdrawal back to the maternal hearth and bosom
because of the collusion between her and her mother.

It was noted in collaborative therapy that certain areas important
to the mother’s self-esteem were especially conflictual situations for
the patient. For example, the mother needed the fiction that she was


close to her family and that a deep love existed between her and her
own mother. By analogy the relationship to the grandmother served
as the prototype for the mother’s relationship to her own daughter.
On one occasion when the daughter was seven or eight years old,
the grandmother in a rage threw a knife which barely missed the
little girl. The mother said nothing to the grandmother but hurried
the little girl from the room with the words, “Grandmommy really
loves you.” It is significant that the grandmother took the attitude toward
the patient that she was not well enough controlled, and she
used to chide her daughter for being too easy on the child. The
grandmother was living in the house during one of the patient’s
psychotic episodes, and the girl took great delight in throwing
various objects at the mother and grandmother while they cowered
in fear.

Mother felt herself very attractive as a girl, and she felt that her
daughter resembled her rather closely, although by damning with
faint praise, it was obvious that she felt the daughter definitely ran
second. One of the daughter’s first acts during a psychotic period
was to announce to her mother that she was going to cut off all her
hair. She proceeded to do this while the mother pleaded with her to
stop. Subsequently the mother would show a picture of herself as a
girl and explain to people how the patient would look if she only
had her beautiful hair.

The mother, apparently without awareness of the significance of
what she was doing, would equate the daughter’s illness with not
being very bright and with some sort of organic brain difficulty.
She would invariably contrast this with her own intelligence as
demonstrated by her own scholastic record. She treated her
daughter with a completely patronizing and placating manner
which was insincere. For example, in the psychiatrist’s presence
she promised her daughter that she would not allow her to have
further shock treatments, and as soon as the girl was out of the
room she asked the doctor if he didn’t feel she should be
hospitalized and given electric shock treatments. One clue to this
deceptive behavior arose during the mother’s therapy. Although the
daughter had had three previous hospitalizations, the mother had
never mentioned to the doctors that she herself had had a psychotic
episode when she discovered that she was pregnant. The family

whisked her away to a small sanitarium in a nearby town, and she
was, according to her own statement, strapped to a bed for six
weeks. Her family did not visit her during this time, and no one
except her parents and her sister knew that she was hospitalized.

There were two times during therapy when the mother showed
intense emotion. One was in relating her own psychotic
experience; the other was on the occasion of her last visit when she
accused the therapist of trying to drive her crazy by forcing her to
choose between her daughter and her husband. Against medical
advice, she took her daughter out of therapy.

The father was as involved in the homeostatic aspects of the
intrafamily situation as the mother. For example, he stated that he
had to quit his position as an important attorney in order to bring his
daughter to an area where competent psychiatric help was available.
Subsequently, acting on cues from the patient (e.g., she frequently
referred to a character named “Nervous Ned”), the therapist was able
to elicit from him that he had hated his job and for years had been
trying to “get out from under.” However, the daughter was made to
feel that the move was initiated for her.

On the basis of our examination of the clinical data, we have
been impressed by a number of observations including:

(1) The helplessness, fear, exasperation, and rage which a double
bind situation provokes in the patient, but which the mother may
serenely and un-understandingly pass over. We have noted reactions
in the father that both create double bind situations, or extend and
amplify those created by the mother, and we have seen the father,
passive and outraged, but helpless, become ensnared in a similar
manner to the patient.
(2) The psychosis seems, in part, a way of dealing with double
bind situations to overcome their inhibiting and con-trolling effect.
The psychotic patient may make astute, pithy, often metaphorical
remarks that reveal an insight into the forces binding him.
Contrariwise, he may become rather expert in setting double bind
situations himself.
(3) According to our theory, the communication situation
described is essential to the mother’s security, and by inference to
the family homeostasis. If this be so, then when psychotherapy of
the patient helps him become less vulnerable to mother’s attempts at

control, anxiety will be produced in the mother. Similarly, if the
therapist interprets to the mother the dynamics of the situation she is
setting up with the patient, this should produce an anxiety response
in her. Our impression is that when there is a perduring con-tact
between patient and family (especially when the patient lives at
home during psychotherapy), this leads to a disturbance (often
severe) in the mother and sometimes in both mother and father and
other siblings.36

Current Position and Future Prospects

Many writers have treated schizophrenia in terms of the most
extreme contrast with any other form of human thinking and
behavior. While it is an isolable phenomenon, so much emphasis on
the differences from the normal—rather like the fearful physical
segregation of psychotics—does not help in understanding the
problems. In our approach we assume that schizophrenia involves
general principles which are important in all communication and
therefore many in-formative similarities can be found in “normal”
communication situations.

We have -been particularly interested in various sorts of
communication which involve both emotional significance and the
necessity of discriminating between orders of message. Such
situations include play, humor, ritual, poetry, and fiction. Play,
especially among animals, we have studied at some length.37 It is a
situation which strikingly illustrates the occurrence of metamessages
whose correct discrimination is vital to the cooperation of the
individuals involved; for ex-ample, false discrimination could easily
lead to combat. Rather closely related to play is humor, a continuing
subject of our -research. It involves sudden shifts in Logical Types
as well as discrimination of those shifts. Ritual is a field in which
unusually real or literal ascriptions of Logical Type are made and

36 D. D. Jackson, “An Episode of Sleepwalking,” Journal of the American
Psychoanalytic Association, 1954, 2: 503—508; also Jackson, “Some Factors . . . ,”
Psycho-analytic Quarterly, 1954, 23: 566—581.

37 Bateson, ” A Theory of Play …” op. cit.


defended as vigorously as the schizophrenic defends the “reality” of
his delusions. Poetry exemplifies the communicative power of
metaphor—even very unusual metaphor—when labeled as such by
various signs, as contrasted to the obscurity of unlabeled
schizophrenic metaphor. The entire field of fictional communication,
defined as the narration or depiction of a series of events with more
or less of a label of actuality, is most relevant to the investigation of
schizophrenia. We are not so much concerned with the content
interpretation of fiction—although analysis of oral and destructive
themes is illuminating to the student of schizophrenia—as with the
formal problems involved in simultaneous existence of multiple
levels of message in the fictional presentation of “reality.” The
drama is especially interesting in this respect, with both performers
and spectators responding to messages about both the actual and
the theatrical reality.

We are giving extensive attention to hypnosis. A great array of
phenomena that occur as schizophrenic symptoms—hallucinations,
delusions, alterations of personality, amnesias, and so on—can be
produced temporarily in normal subjects with hypnosis. These need
not be directly suggested as specific phenomena, but can be the
“spontaneous” result of an arranged communication sequence. For
example, Erickson38 will produce a hallucination by first inducing
catalepsy in a subject’s hand and then saying, “There is no
conceivable way in which your hand can move, yet when I give the
signal, it must move.” That is, he tells the subject his hand will
remain in place, yet it will move, and in no way the subject can
consciously conceive. When Erickson gives the signal, the subject
hallucinates the hand moved, or hallucinates himself in a different
place and therefore the hand was moved. This use of hallucination to
resolve a problem posed by contradictory commands which cannot
be discussed seems to us to illustrate the solution of a double bind
situation via a shift in Logical Types. Hypnotic responses to direct
suggestions or statements also commonly involve shifts in type, as
in accepting the words “Here’s a glass of water” or “You feel tired”
as external or internal reality, or in literal response to metaphorical

38 M. H. Erickson, Personal communication, 1955.


statements, much like schizophrenics. We hope that further study of
hypnotic induction, phenomena, and waking will, in this controllable
situation, help sharpen our view of the essential
communicational sequences which produce phenomena like those of

Another Erickson experiment seems to isolate a double bind
communicational sequence without the specific use of hypnosis.
Erickson arranged a seminar so as to have a young chain smoker sit
next to him and to be without cigarettes; other participants were
briefed on what to do. All was ordered so that Erickson repeatedly
turned to offer the young man a cigarette, but was always
interrupted by a question from someone so that he turned away,
“inadvertently” withdrawing the cigarettes from the young man’s
reach. Later another participant asked this young man if he had
received the cigarette from Dr. Erickson. He re-plied, “What
cigarette?”, showed clearly that he had forgot-ten the whole
sequence, and even refused a cigarette offered by another member,
saying that he was too interested in the seminar discussion to
smoke. This young man seems to us to be in an experimental
situation paralleling the schizophrenic’s double bind situation with
mother: an important relationship, contradictory messages (here of
giving and taking away), and comment blocked—because there
was a seminar going on, and anyway it was all “inadvertent.” And
note the similar outcome: amnesia for the double bind sequence
and reversal from “He doesn’t give” to “I don’t want.”

Although we have been led into these collateral areas, our main
field of observation has been schizophrenia itself. All of us have
worked directly with schizophrenic patients and much of this case
material has been recorded on tape for detailed study. In addition,
we are recording interviews held jointly with patients and their
families, and we are taking sound motion pictures of mothers and
disturbed, presumably preschizophrenic, children. Our hope is that
these operations will provide a clearly evident record of the continuing,
repetitive double binding which we hypothesize goes on
steadily from infantile beginnings in the family situation of
individuals who become schizophrenic. This basic family situation,
and the overtly communicational characteristics of schizophrenia,
have been the major focus of this paper. However, we expect our

concepts and some of these data will also be useful in future work
on other problems of schizophrenia, such as the variety of other
symptoms, the character of the “adjusted state” before schizophrenia
becomes manifest, and the nature and circumstances of the
psychotic break.

Therapeutic Implications of this Hypothesis

Psychotherapy itself is a context of multilevel communication,
with exploration of the ambiguous lines between the literal and
metaphoric, or reality and fantasy, and indeed, various forms of play,
drama, and hypnosis have been used extensively in therapy. We
have been interested in therapy, and in addition to our own data we
have been collecting and examining recordings, verbatim
transcripts, and personal accounts of therapy from other therapists.
In this we prefer exact records since we believe that how a
schizophrenic talks depends greatly, though often subtly, on how
another person talks to him; it is most difficult to estimate what
was really occurring in a therapeutic interview if one has only a
description of it, especially if the description is already in
theoretical terms.

Except for a few general remarks and some speculation,
however, we are not yet prepared to comment on the relation of the
double bind to psychotherapy. At present we can only note:

(1) Double bind situations are created by and within the
psychotherapeutic setting and the hospital milieu. From the point of
view of this hypothesis, we wonder about the effect of medical
“benevolence” on the schizophrenic patient. Since hospitals exist for
the benefit of personnel as well as—as much as—more than—for
the patient’s benefit, there will be contradictions at times in
sequences where actions are taken “benevolently” for the patient
when actually they are intended to keep the staff more comfortable.
We would assume that whenever the system is organized for hospital
purposes and it is announced to the patient that the actions are for
his benefit, then the schizophrenogenic situation is being

perpetuated. This kind of deception will provoke the patient to
respond to it as a double bind situation, and his response will be
“schizophrenic” in the sense that it will be indirect and the patient
will be unable to comment on the fact that he feels that he is being
deceived. One vignette, fortunately amusing, illustrates such a
response. On a ward with a dedicated and “benevolent” physician in
charge there was a sign on the physician’s door which said “Doctor’s
Office. Please Knock.” The doctor was driven to distraction and
finally capitulation by the obedient patient who carefully knocked
every time he passed the door.

(2) The understanding of the double bind and its communicative
aspects may lead to innovations in therapeutic technique. Just what
these innovations may be is difficult to say, but on the basis of our
investigation we are assuming that double bind situations occur
consistently in psychotherapy. At times these are inadvertent in the
sense that the therapist is imposing a double bind situation similar
to that in the patient’s history, or the patient is imposing a double
bind situation on the therapist. At other times therapists seem to
impose double binds, either deliberately or intuitively, which force
the patient to respond differently than he has in the past.
An incident from the experience of a gifted psychotherapist
illustrates the intuitive understanding of a double bind
communicational sequence. Dr. Frieda Fromm-Reichmann39 was
treating a young woman who from the age of seven had built a
highly complex religion of her own replete with powerful gods. She
was very schizophrenic and quite hesitant about entering into a
therapeutic situation. At the be-ginning of the treatment she said,
“God R says I shouldn’t talk with you.” Dr. Fromm-Reichmann
replied, “Look, let’s get something into the record. To me God R
doesn’t exist, and that whole world of yours doesn’t exist. To you it
does, and far be it from me to think that I can take that away from
you, I have no idea what it means. So I’m willing to talk with you in
terms of that world, if only you know I do it so that we have an
understanding that it doesn’t exist for me. Now go to God R and tell
him that we have to talk and he should give you permission. Also
you must tell him that I am a doctor and that you have lived with

39 F. Fromm-Reichmann, Personal communication, 1956

him in his kingdom now from seven to sixteen—that’s nine years —
and he hasn’t helped you. So now he must permit me to try and see
whether you and I can do that job. Tell him that I am a doctor and
this is what I want to try.”

The therapist has her patient in a “therapeutic double bind.” If the
patient is rendered doubtful about her belief in her god, then she is
agreeing with Dr. Fromm-Reichmann, and is admitting her
attachment to therapy. If she insists that God R is real, then she must
tell him that Dr. Fromm-Reichmann is “more powerful” than he—
again admitting her involvement with the therapist.

The difference between the therapeutic bind and the original
double bind situation is in part the fact that the therapist is not
involved in a life and death struggle himself. He can therefore set up
relatively benevolent binds and gradually aid the patient in his
emancipation from them. Many of the uniquely appropriate
therapeutic gambits arranged by therapists seem to be intuitive. We
share the goal of most psychotherapists who strive toward the day
when such strokes of genius will be well enough understood to be
systematic and commonplace.

Additional References

J. Haley, “Paradoxes in Play, Fantasy, and Psychotherapy,” Psychiatric
Research Reports, 1955, 2: 52-8.
J. Ruesch and G. Bateson, Communication: The Social Matrix of Psychiatry, New
York, Norton, 1951.

The Group Dynamics of Schizophrenia*

First, I intend to attach very specific meaning to the title of this
paper. An essential notion attached to the word “group” as I shall use
it is the idea of relatedness between members. Our concern is not
with the sort of phenomena which occur in experimentally formed
groups of graduate students who have no previously determined
habits of communication—no habitual differentiations of role. The
group to which I mostly refer is the family; in general, those families
in which the parents maintain an adjustment to the world around
them without being recognized as grossly deviant, while one or
more of their offspring differ conspicuously from the normal
population in the frequency and obvious nature of their responses. I
shall also be thinking of other groups analogous to these, i.e., ward
organizations, which work in such a way as to promote
schizophrenic or schizophrenoid behavior in some of the members.

The word “dynamics” is loosely and conventionally used for all
studies of personal interaction and especially when they stress
change or learning exhibited by the subjects. De-spite our following
its conventional use, this word is a misnomer. It evokes analogies
with physics which are totally false.

“Dynamics” is principally a language devised by physicists and
mathematicians for the description of certain events. In this strict
sense, the impact of one billiard ball upon an-other is subject matter
for dynamics, but it would be an error of language to say that
billiard balls “behave.” Dynamics appropriately describe those
events whose descriptions can be checked by asking whether they
contravene the First Law of Thermodynamics, the Law of the

* The ideas in this lecture represent the combined thinking of the staff of The
Project for the Study of Schizophrenic Communication. The staff consisted of
Gregory Bateson, Jay Haley, John H. Weakland, Don D. Jackson, M.D., and William F.
Fry, M.D
The article is reprinted from Chronic Schizophrenia: Explorations in Theory
and Treatment, edited by L. Appleby, J. M. Scher, and J. Cumming, The Free Press,
Glencoe, Illinois, 1960; reprinted by permission.

Conservation of Energy. When one billiard ball strikes another, the
motion of the second is energized by the impact of the first, and
such transferences of energy are the central subject matter of
dynamics. We, however, are not concerned with event sequences
which have this characteristic. If I kick a stone, the movement of the
stone is energized by the act, but if I kick a dog, the behavior of the
dog may indeed be partly conservative—he may travel along a
Newtonian trajectory if kicked hard enough, but this is mere
physics. What is important is that he may exhibit responses which
are energized not by the kick but by his metabolism; he may turn
and bite.

This, I think, is what people mean by magic. The realm of
phenomena in which we are interested is always characterized by
the fact that “ideas” may influence events. To the physicist, this is a
grossly magical hypothesis. It is one which cannot be tested by
asking questions about the conservation of energy.

All this, however, has been better and more rigorously said by
Bertalanffy, which makes it easier for me to further explore this
realm of phenomena in which communication occurs. We shall settle
for the conventional term “dynamics” provided it is clearly
understood that we are not talking about dynamics in the physical

Robert Louis Stevenson40 in “The Poor Thing” has achieved
perhaps the most vivid characterization of this magical realm:

“In my thought one thing is as good as another in this world; and
a shoe of a horse will do.” The word “yes” or a whole performance
of Hamlet, or an injection of epinephrine in the right place on the
surface of the brain may be interchangeable objects. Any one of
them may, ac-cording to the conventions of communication
established at that moment, be an affirmative (or a negative) answer
to any question. In the famous message, “One if by land; two if by
sea, ” the objects actually used were lamps, but from the point of
view of communications theory, they could have been anything from
aardvarks to zygomatic arches.

40 R. L. Stevenson, “The Poor Thing,” Novels and Tales of Robert Louis
Stevenson, Vol. 20, New York, Scribners, 1918, pp. 496-502.

It might well be sufficiently confusing to be told that, according
to the conventions of communication in use at the moment, anything
can stand for anything else. But this realm of magic is not that
simple. Not only can the shoe of a horse stand for anything else
according to the conventions of communication, it can also and
simultaneously be a signal which will alter the conventions of
communication. My fingers crossed behind my back may alter the
whole tone and implication of everything. I recall a schizophrenic
patient who, like many other schizophrenics, had difficulty with the
first person pronoun; in particular, he did not like to sign his name.
He had a number of aliases, alternative named aspects of self. The
ward organization, of which he was a part, required that he sign his
name to obtain a pass, and for one or two weekends he did not
receive a pass because he insisted on signing one of his aliases. One
day he remarked that he was going out the next weekend. I said,
“Oh, did you sign?” He said, “Yes,” with an odd grin. His real name,
we will say, was Edward W. Jones. What he had actually signed was
“W. Edward Jones.” The ward officials did not notice the difference.
It appeared to them that they had won a battle and had succeeded in
forcing him to act sanely. But to himself the message was, “He (the
real me) did not sign.” He had won the battle. It was as if his fingers
were crossed behind his back.

All communication has this characteristic—it can be magically
modified by accompanying communication. In this conference, we
have been discussing various ways of interacting with patients,
describing what we do and what our strategy seems to us to be. It
would have been more difficult to discuss our actions from the
patients’ point of view. How do we qualify our communications to
the patients, so that the experience which they receive will be

Appleby, for example, described a set of procedures on his
ward, and if I were a schizophrenic listening to him, I would have
been tempted to say, “It all sounds like occupational therapy to
me.” He tells us very convincingly and with figures that his
program is successful, and in documenting his success he is no
doubt telling the truth. If this is so, then his description of the
program must necessarily be incomplete. The experiences which
the program provides for the patients must be something a little

more alive than the dry bones of the program which he has
described. The whole series of therapeutic procedures must have
been qualified, possibly with enthusiasm or with humor, with some
set of signals which altered the mathematical sign—plus or minus
—of what was being done. Appleby has told us only about the shoe
of the horse, not about the multitude of realities which determined
for what that horseshoe stood.

It is as if he had related that a given musical composition was
set in the key of C major, and asked us to believe that this skeletal
statement was a sufficient description to enable us to understand
why this particular composition altered the mood of the listener in
a particular way. What is omitted in all such descriptions is the
enormous complexity of modulation of communication. It is this
modulation which is music.

Let me shift from a musical to a wide biological analogy in order
to examine further this magical realm of communication. All
organisms are partially determined by genetics, i.e., by complex
constellations of messages carried principally in the chromosomes.
We are products of a communicational process, modified and
qualified in various ways by environmental impact. It follows,
therefore, that the differences between related organisms, say, a crab
and a lobster, or between a tall pea and a short pea, must always be
the sort of differences that can be created by changes and
modulations in a constellation of messages. Sometimes these
changes in the message system will be relatively concrete—a shift
from “yes” to “no” in the answer to some question governing a
relatively superficial detail of the anatomy. The total picture of the
animal may be altered by as little as one spot in the whole halftone
block, or the change may be one which modifies or modulates the
whole system of genetic messages, so that every message in the
system takes on a different look while retaining its former relationship
to all neighboring messages. It is, I believe, this stability of the
relationship between messages under the impact of the change in
one part of the constellation that provides a basis for the French
aphorism “Plus get change, plus c’est la même chose.” It is a
recognized fact that the skulls of the various anthropoids can be
drawn upon diversely skewed coordinates ‘to demonstrate the


fundamental similarity of relations and the systematic nature of the
transformation from one species to another.41

My father was a geneticist, and he used to say, “It’s all
vibrations,”42 and to illustrate this he would point out that the
striping of the common zebra is an octave higher than that of
Grevy’s zebra. While it is true that in this particular case the
“frequency” is doubled, I don’t think that it is entirely a matter of
vibrations as he endeavored to ex-plain it. Rather, he was trying to
say that it is all a matter of the sort of modifications which could be
expected among systems whose determinants are not a matter of
physics in the crude sense, but a matter of messages and modulated
systems of messages.

It is worth noting, too, that perhaps organic forms are beautiful to
us and the systematic biologist can find aesthetic satisfaction in the
differences between related organisms simply because the
differences are due to modulations of communication, while we
ourselves are both organisms who communicate and whose forms
are determined by constellations of genetic messages. This is not the
place, however, for such a revision of aesthetic theory. An expert in
the theory of mathematical groups could make a major contribution
in this field.

All messages and parts of messages are like phrases or segments
of equations which a mathematician puts in brackets. Outside the
brackets there may always be a qualifier or multiplier which will
alter the whole tenor of the phrase. More-over, these qualifiers can
always be added, even years later.

They do not have to precede the phrase inside the brackets.
Otherwise, there could be no psychotherapy. The patient would be
entitled and even compelled to argue, “My mother slapped me down
in such and such ways, and, therefore, I am now sick; and because
those traumata occured in the past they cannot be altered, and I,
therefore, cannot get well.” In the realm of communication, the
events of the past constitute a chain of old horseshoes so that the

41 D. W. Thompson, On Growth and Form, Vol. 2, Ox-ford, Oxford University
Press, 1952.

42 Beatrice C. Bateson, William Bateson, Naturalist, Cambridge, Cambridge
University Press, 1928.


meaning of that chain can be changed and is continually being
changed. What exists today are only messages about the past which
we call memories, and these messages can al-ways be framed and
modulated from moment to moment.

Up to this point the realm of communication appears to be more
and more complex, more flexible, and less amenable to analysis.
Now the introduction of the group concept—the consideration of
many persons—suddenly simplifies this confused realm of slipping
and sliding meanings. If we shake up a number of irregular stones in
a bag, or subject them to an almost random beating by the waves on
the seashore, even at the crudely physical level, there will be a
gradual simplification of the system—the stones will resemble each
other. In the end, they will all become spherical, but in practice we
usually encounter them as partly rounded pebbles. Certain forms of
homogenization result from multiple impact even at the crude
physical level, and when the impacting entities are organisms
capable of complex learning and communication, the total system
operates rapidly to-ward either uniformity or toward systematic
differentiation—an increase of simplicity—which we call
organization. If there are differences between the impacting entities,
these differences will undergo change, either in the direction of
reducing the difference, or in the direction of achieving a mutual
fitting or complementarity. Among groups of people, whether the
direction of change is toward homogeneity or toward
complementarity, the achievement is a sharing of premises regarding
the meaning and appropriateness of messages and other acts in the
context of the relationship.

I shall not go into the complex problems of learning involved in
this process but shall proceed to the problem of schizophrenia. An
individual, i.e., the identified patient, exists within a family setting,
but when we view him singularly, certain pecularities of his
communicational habits are noted.

These peculiarities may be partly determined by genetics or
physiological accident, but it is still reasonable to question the
function of these peculiarities within the communicational system of
which they are a p a r t the family. A number of living creatures
have been, in a sense, shaken up together and one of them has come
out apparently different from the rest; we have to ask not only about


differences in the material of which this particular individual may be
made, but also how his particular characteristics were developed in
this family system. Can the peculiarities of the identified patient be
seen as appropriate, i.e., as either homogeneous with, or
complementary to, the characteristics of the other members of the
group? We do not doubt that a large part of schizophrenic.
symptomatology is, in some sense, learned or determined by
experience, but an organism can learn only that which it is taught by
the circumstances of living and the experiences of exchanging
messages with those around him. He cannot learn at random, but
only to be like or unlike those around him. We have, therefore, the
necessary task of looking at the experiential setting of

We shall outline briefly what we have been calling the double
bind hypothesis, which has been more fully described elsewhere.43
This hypothesis contains two parts; a formal description of the
communicational habits of the schizophrenic, and a formal
description of the sequences of experience which would
understandably train the individual in his peculiar distortions of
communication. Empirically we find that one description of the
symptoms is, on the whole, satisfactory, and that the families of
schizophrenics are characterized by the behavioral sequences
which are predicted by the hypothesis.

Typically, the schizophrenic will eliminate from his messages
everything that refers explicitly or implicitly to the relationship

43 G. Bateson, D. D. Jackson, J. Haley, and J. H. Weak-land, “Toward a Theory of
Schizophrenia,” Behavioral Science, 1956, 1: 251–64; also G. Bateson, “Language
and Psychotherapy, Frieda Fromm-Reichmann’s Last Project,” Psychiatry, 1958,

21: 96–100; also G. Bateson (moderator), “Schizophrenic Distortions of
Communication,” Psychotherapy of Chronic Schizophrenic Patients, C. A.
Whitacker, ed., Boston and Toronto, Little, Brown and Co., 1958, pp. 31–56; also G.
Bateson, “Analysis of Group Therapy in an Admission Ward, United States Naval
Hospital, Oakland, California,” Social Psychiatry in Action, H. A. Wilmer,
Springfield, Ill., Charles C. Thomas, 1958, pp. 334–49; also J. Haley, “The Art of
Psychoanalysis,” etc., 1958, .15: 190–200; also J. Haley, “An Interactional
Explanation of Hypnosis,” American Journal of Clinical Hypnosis, 1958, 1: 41–57;
also J. H. Weakland and D. D. Jackson, “Patient and Therapist Observations on the
Circumstances of a Schizophrenic Episode,” AMA Archives of Neurological
Psychiatry, 1958, 79: 554–74.

between himself and the person he is addressing. Schizophrenics
commonly avoid the first and second person pronouns. They avoid
telling you what sort of a message they are transmitting—whether it
be literal or metaphoric, ironic or d i r e c t and they are likely to
have difficulty with all messages and meaningful acts which imply
intimate contact between the self and some other. To receive food
may be almost impossible, but so also may be the repudiation of

When leaving for the A.P.A. meetings in Honolulu, I told my
patient that I would be away and where I was going. He looked out
the window and said, “That plane flies awfully slowly.” He could
not say, “I shall miss you,” because he would thus be identifying
himself in a relationship to me, or me in relationship to himself. To
say, “I shall miss you” would be to assert a basic premise about our
mutual relationship by defining the sorts of messages which should
be characteristic of that relationship.

Observably, the schizophrenic avoids or distorts anything which
might seem to identify either himself or the person whom he is
addressing. He may eliminate anything which implies that his
message refers to, and is a part of, a relationship between two
identifiable people, with certain styles and premises governing their
behavior in that relationship. He may avoid anything which might
enable the other to interpret what he says. He may obscure the fact
that he is speaking in metaphor or in some special code, and he is
likely to distort or omit all reference to time and place. If we use a
Western Union telegram form as an analogy, we might say that he
omits what would be put on the procedural parts of the telegraph
form and will modify the text of his message to distort or omit any
indication of these metacommunicative elements in the total normal
message. What remains is likely to be a metaphoric statement
unlabelled as to context. Or, in extreme cases, there may be nothing
left but a stolid acting out of the message, “There is no relationship
between us.”

This much is observable and may be summarized by saying that
the schizophrenic communicates as if he expected to be punished
every time he indicates that he is right in his view of the context of
his own message.


The “double bind,” which is central to the etiological half of our
hypothesis, may now simply be summarized by saying that it is an
experience of being punished precisely for being right in one’s own
view of the context. Our hypothesis assumes that repeated
experience of punishment in sequences of this kind will lead the
individual to behave habitually as if he expected such punishment.

The mother of one of our patients poured out blame upon her
husband for refusing for fifteen years to hand over control of the
family finances to her. The father of the patient said, “I admit that it
was a great mistake of me not to let you handle it, I admit that. I
have corrected that. My reasons for thinking it was a mistake are
entirely different from yours, but I admit that it was a very serious
error on my part.”

Mother: Now, you’re just being facetious.

Father: No, I am not being facetious.

Mother: Well, anyway I don’t care because when you come right
down to it the debts were incurred, still there is no reason why a
person would not be told of them. I think the woman should be told.

Father: It may be the same reason why when Joe (their psychotic
son) comes home from school and he has had trouble he doesn’t tell

Mother: Well, that’s a good dodge.

The pattern of such a sequence is simply the successive
disqualification of each of the father’s contributions to the
relationship. He is continuously being told that the messages are not
valid. They are received as if they were in some way different from
that which he thought he intended.

We may say that he is penalized either for being right about his
views of his own intentions, or he is penalized whenever his reply is
appropriate to what she said.

But, per contra, from her viewpoint, it seems that he is endlessly
misinterpreting her, and this is one of the most peculiar
characteristics of the dynamic system which surrounds—or is—
schizophrenia. Every therapist who has dealt with schizophrenics
will recognize the recurrent trap. The patient endeavors to put the
therapist in the wrong by his interpretation of what the therapist
said, and the patient does this because he expects the therapist to
misinterpret what he (the patient) said. The bind becomes mutual. A

stage is reached in the relationship in which neither person can
afford to receive or emit metacommunicative messages without

There is, however, usually, an asymmetry in such relationships.
This mutual doublebinding is a type of struggle and commonly one
or the other has the upper hand. We have deliberately chosen to
work with families where one of the offspring is the identified
patient, and, partly for this reason, in our data, it is the supposedly
normal parents who have the upper hand over an identifiably
psychotic younger member of the group. In such cases, the
asymmetry takes the curious form that the identified patient
sacrifices himself to maintain the sacred illusion that what the parent
says makes sense. To be close to that parent, he must sacrifice his
right to indicate that he sees any metacommunicative
incongruencies, even when his perception of these incongruencies is
correct. There is, therefore, a curious disparity in the distribution of
awareness of what is happening. The patient may know but must not
tell, and thereby enables the parent to not know what he or she is
doing. The patient is an accomplice in the parent’s unconscious
hypocrisy. The result may be very great unhappiness and very gross,
but al-ways systematic, distortions of communication.

Moreover, these distortions are always precisely those which
would seem appropriate when the victims are faced with a trap to
avoid which would be to destroy the very nature of the self. This
paradigm is neatly illustrated by a pas-sage which is worth quoting
in full from Festing Jones’ life of Samuel Butler.44

Butler went to dinner at Mr. Seebohm’s where he met Skertchley,
who told them about a rat-trap invented by Mr. Tylor’s coachman.


Mr. Dunkett found all his traps fail one after another, and was in
such despair at the way the corn got eaten that he resolved to invent
a rat-trap. He began by putting himself as nearly as possible in the
rat’s place.

44 H. F. Jones, Samuel Butler: A Memoir, Vol. 1, Lon-don, Macmillan,

“Is there anything,” he asked himself, “in which, if I were a rat, I
should have such complete confidence that I could not suspect it
without suspecting everything in the world and being unable
henceforth to move fearlessly in any direction?”

He pondered for a while and had no answer, till one night the
room seemed to become full of light and he hears a voice from
heaven saying:


Then he saw his way. To suspect a common drain-pipe would be
to cease to be a rat. Here Skertchley enlarged a little, explaining that
a spring was to be concealed inside, but that the pipe was to be open
at both ends; if the pipe were closed at one end, a rat would
naturally not like going into it, for he would not feel sure of being
able to get out again; on which I [Butler] interrupted and said:

“Ah, it was just this which stopped me from going in-to the

When he [Butler] told me this I [Jones] knew what was in his
mind, and that, if he had not been in such respectable company, he
would have said: “It was just this which stopped me from getting

Notice that Dunkett could only invent this double bind for rats by
way of an hallucinatory experience, and that both Butler and Jones
immediately regarded the trap as a paradigm for human relations.
Indeed, this sort of dilemma is not rare and is not confined to the
contexts of schizophrenia.

The question which we must face, therefore, is why these
sequences are either specially frequent or specially destructive in
those families which contain schizophrenics. I do not have the
statistics to assert this; however, from limited but intense
observation of a few of these families, I can offer an hypothesis
about the group dynamics which would deter-mine a system of
interaction, such that double bind experiences must recur ad
nauseam. The problem is to construct a model which will necessarily
cycle to recreate these patterned sequences over and over again.

Such a model is provided in Von Neumann’s and Mor


genstern’theory of games, presented here not, indeed, with its
full mathematical rigor, but at least in terms some-what technical.

Von Neumann was concerned with mathematical study of the
formal conditions under which entities, with total intelligence and a
preference for gain, would form coalitions among themselves in
order to maximize the profits which coalition members might
receive at the expense of the non-members. He imagined these
entities as engaged in some-thing like a game and proceeded to ask
about the formal characteristics of the rules which would compel the
totally intelligent but gain-oriented players to form coalitions. A
very curious conclusion emerged, and it is this conclusion which I
would propose as a model.

Evidently, coalition between players can only emerge when
there are at least three of them. Any two may then get together to
exploit the third, and if such a game be symmetrically devised, it
evidently has three solutions which we may represent as

AB vs. C
BC vs. A
AC vs. B

For this three-person system, Von Neumann demonstrates that
once formed, any one of these coalitions will be stable. If A and B
are in alliance, there is nothing C can do about it. And, interestingly
enough, A and B will necessarily develop conventions
(supplementary to the rules) which will, for example, forbid them
from listening to C’s approaches.

In the five-person game, the position becomes quite different;
there will be a variety of possibilities. It may be that four players
contemplate a combination against one, illustrated in the following
five patterns:

A vs. BCDE
B vs. ACDE
C vs. ABDE

45 J. Von Neumann and O. Morgenstern, Theory of Games and Economic
Behavior, Princeton, Princeton University Press, 1944.


D vs. ABCE
E vs. ABCD

But none of these would be stable. The four players within the
coalition must, necessarily, engage in a subgame in which they
maneuver against each other to achieve an unequal division of the
gains which the coalition could squeeze out of the fifth player. This
must lead to a coalition pattern which we may describe as 2 vs. 2 vs.
1, i.e., BC vs. DE vs. A. In such a situation, it would become possible
for A to approach and join one of these pairs, so that the coalition
system will become 3 vs. 2.

But in the system 3 vs. 2, it would be advantageous for the three
to recruit over to their side one of the two, in order to make their
gains more certain. Now we are back to a 4 vs. 1 system—not
necessarily the particular line-up that we started from but at any rate
a system having the same general properties. It, in turn, must break
down into 2 vs. 2 vs. 1, and so on.

In other words, for every possible pattern of coalitions, there is at
least one other pattern which will “dominate” it—to use Von
Neumann’s term—and the relationship of domination between
solutions is intransitive. There will al-ways be a circular list of
alternative solutions so that the system will never cease from
passing on from solution to solution, always selecting another
solution which is preferable to that which preceded it. This means,
in fact, that the robots (owing to their total intelligence) will be
unable to decide upon a single “play” of the game.

I offer this model as being reminiscent of what happens in
schizophrenic families. No two members seem able to get together
in a coalition stable enough to be decisive at the given moment.
Some other member or members of the family will always
intervene. Or, lacking such intervention, the two members who
contemplate a coalition will feel guilty vis-a-vis what the third might
do or say, and will draw back from the coalition.

Notice that it takes five hypothetical entities with total
intelligence to achieve this particular sort of instability or oscillation
in a Von Neumannian game. But three human beings seem to be
enough. Perhaps they are not totally intelligent or perhaps they are

systematically inconsistent regarding the sort of “gain” in terms of
which they are motivated.

I want to stress that in such a system, the experience of each
separate individual will be of this kind: every move which he makes
is the common-sense move in the situation as he correctly sees it at
that moment, but his every move is subsequently demonstrated to
have been wrong by the moves which other members of the system
make in response to his “right” move. The individual is thus caught
in a perpetual sequence of what we have called double bind

I do not know how valid this model may be, but I offer it for two
reasons. First, it is proposed as a sample of trying to talk about the
larger system—the family—instead of talking, as we habitually do,
about the individual. If we are to understand the dynamics of
schizophrenia, we must devise a language adequate to the
phenomena which are emergent in this larger system. Even if my
model is inappropriate, it is still worthwhile to try to talk in the sort
of language which we shall need for describing these emergent
phenomena. Secondly, conceptual models, even when incorrect, are
useful to the extent that criticism of the model may point to new
theoretical developments.

Let me, therefore, point out one criticism of this model, and
consider to what ideas it will lead. There is no theorem in Von
Neumann’s book which would indicate that his entities or robots,
engaged in this infinite dance of changing coalitions, would ever
become schizophrenic. According to the abstract theory, the entities
simply remain totally intelligent ad infinitum.

Now, the major difference between people and von Neumann’s
robots lies in the fact of learning. To be infinitely intelligent implies
to be infinitely flexible, and the players in the dance which I have
described could never experience the pain which human beings
would feel if continuallyproven wrong whenever they had been
wise. Human beings have a commitment to the solutions which
they discover, and it is this psychological commitment that makes
it possible for them to be hurt in the way members of a
schizophrenic family are hurt.

It appears then, from consideration of the model, that the
double bind hypothesis, to be explanatory of schizophrenia, must


depend upon certain psychological assumptions about the nature of
the human individual as a learning organism. For the individual to
be prone to schizophrenia, individuation must comprise two
contrasting psychological mechanisms. The first is a mechanism of
adaptation to demands of the personal environment; and the
second, a process or mechanism whereby the individual becomes
either briefly or enduringly committed to the adaptations which the
first process has discovered.

I think that what l am calling a brief commitment to an
adaptation is what Bertalanffy called the immanent state of action;
and that the more enduring commitment to adaptation is simply
what we usually call “habit.”

What is a person? What do I mean when I say “I?” Perhaps what
each of us means by the “self” is in fact an aggregate of habits of
perception and adaptive action plus, from moment to moment, our
“immanent states of action.” If somebody attacks the habits and
immanent states which characterize me at the given moment of
dealing with that somebody—that is, if they attack the very habits
and immanent states which have been called into being as part of my
relationship to them at that moment—they are negating me. If I care
deeply about that other person, the negation of me will be still more

What we have said so far is enough to indicate the sorts of
strategy—or perhaps we should say symptoms—which are to be
expected in that strange institution, the schizophrenic family. But it
is still surprising to observe how these strategies may be continually
and habitually practiced without friends and neighbors noticing that
something is wrong. From theory we may predict that every
participant member of such an institution must be defensive of his or
her own immanent states of action and enduring adaptive habits;
protective, that is, of the self.

To illustrate with one example: a colleague had been working for
some weeks with one of these families, particularly with the father,
the mother, and their adult schizophrenic son. His meetings were on
the conjoint pattern—the members of the family being present
together. This apparently provoked some anxiety in the mother and
she requested face-to-face interviews with me. This move was
discussed at the next conjoint meeting and in due course she came to

her first session. Upon arrival she made a couple of conversational
remarks, and then opened her purse and from it handed me a piece
of paper, saying, “It seems my husband wrote this.” I unfolded the
paper and found it to be a single sheet of single-spaced typescript,
starting with the words, “My husband and I much appreciate the
opportunity of discussing our problems with you,” etc. The
document then went on to outline certain specific questions which “I
would like to raise.

It appeared that the husband had, in fact, sat down at his
typewriter the night before and had written this letter to me as
though it were written by his wife, and in it he outlined the questions
for her to discuss with me.

In normal daily life this sort of thing is common enough; it
passes muster. When attention is focused upon the characteristic
strategies, however, these self-protecting and self-destroying
maneuvers become conspicuous. One suddenly discovers that in
such families these strategies seem to pre-dominate over all others.
It becomes hardly surprising that the identified patient exhibits
behavior which is almost a caricature of that loss of identity which is
characteristic of all the family members.

I believe that this is the essence of the matter, that the
schizophrenic family is an organization with great ongoing
stability whose dynamics and inner workings are such that each
member is continually undergoing the experience of negation of


Minimal Requirements for a Theory ofSchizophrenia*

Every science, like every person, has a duty toward its neighbors,
not perhaps to love them as itself, but still to lend them its tools, to
borrow tools from them, and, generally, to keep the neighboring
sciences straight. We may perhaps judge of the importance of an
advance in any one science in terms of the changes which this
advance compels the neigh-boring sciences to make in their methods
and in their thinking. But always there is the rule of parsimony. The
changes which we in the behavioral sciences may ask for in
genetics, or in philosophy, or in information theory must always be
minimal. The unity of science as a whole is achieved by this system
of minimal demands imposed by each science upon its neighbors,
and—not a little—by the lending of conceptual tools and patterns
which occurs among the various sciences.

My purpose, therefore, in the present lecture is not so much to
discuss the particular theory of schizophrenia which we have been
developing at Palo Alto. Rather, I want to indicate to you that this
theory and others like it have impact upon ideas about the very
nature of explanation. I have used the title “Minimal Requirements
for a Theory of Schizophrenia,” and what I had in mind in choosing
this title was a discussion of the implications of the double bind
theory for the wider field of behavioral science and even, beyond
that, its effect upon evolutionary theory and biological epistemology.
What minimal changes does this theory demand in related

I want to deal with questions about the impact of an experiential
theory of schizophrenia upon that triad of related sciences, learning
theory, genetics, and evolution.

* Second Annual Albert D. Lasker Memorial Lecture, delivered at the Institute for
Psychosomatic and Psychiatric Research and Training of the Michael Reese
Hospital, Chicago, April 7, 1959. This lecture is here reprinted by permission of the
A.M.A. Archives of General Psychiatry where it appeared in 1960, Vol. 2, pp.

The hypothesis may first be briefly described. In its essentials,
the idea appeals only to everyday experience, and elementary
common sense. The first proposition from which the hypothesis is
derived is that learning occurs always in some context which has
formal characteristics. You may think, if you will, of the formal
characteristics of an instrumental avoidance sequence, or of the
formal characteristics of a Pavlovian experiment. To learn to lift a
paw in a Pavlovian context is different from learning the same
action in a context of instrumental reward.

Further, the hypothesis depends upon the idea that this structured
context also occurs within a wider context—a metacontext if you
will—and that this sequence of contexts is an open, and conceivably
infinite, series.

The hypothesis also assumes that what occurs within the narrow
context (e.g., instrumental avoidance) will be affected by the wider
context within which this smaller one has its being. There may be
incongruence or conflict between context and metacontext. A
context of Pavlovian learning may, for example, be set within a
metacontext which would punish learning of this kind, perhaps by
insisting upon insight. The organism is then faced with the dilemma
either of being wrong in the primary context or of being right for the
wrong reasons or in a wrong way. This is the so-called double bind.
We are investigating the hypothesis that schizophrenic
communication is learned and be-comes habitual as a result of
continual traumata of this kind.

That is all there is to it.

But even these “common-sense” assumptions break away from
the classical rules of scientific epistemology. We have learned from
the paradigm of the freely falling body—and from many similar
paradigms in many other sciences—to approach scientific problems
in a peculiar way: the problems are to be simplified by ignoring—or
postponing consideration of—the possibility that the larger context
may influence the smaller. Our hypothesis runs counter to this rule,
and is focused precisely upon the determining relations between
larger and smaller contexts.

Even more shocking is the fact that our hypothesis suggests —
but does not stand or fall with the suggestion—that there may be an
infinite regress of such relevant contexts.


In all of this, the hypothesis requires and reinforces that revision
in scientific thought which has been occurring in many fields, from
physics to biology. The observer must be included within the focus
of observation, and what can be studied is always a relationship or
an infinite regress of relationships. Never a “thing.”

An example will make clear the relevance of the larger contexts.
Let us consider the larger context within which a learning
experiment might be conducted using a schizophrenic as a subject.
The schizophrenic is what is called a patient, vis-a-vis a member of
a superior and unloved organization, the hospital staff. If the patient
were a good pragmatic Newtonian, he would be able to say to
himself: “The cigarettes which I can get by doing what this fellow
expects me to do are after all only cigarettes, and as an applied
scientist I will go ahead and do what he wants me to do. I will solve
the experimental problem and obtain the cigarettes.” But human
beings, and especially schizophrenics, do not always see the matter
this way. They are affected by the circumstance that the experiment
is being conducted by somebody whom they would rather not
please. They may even feel that there would be a certain
shamelessness about seeking to please some one whom they dislike.
It thus comes about that the sign of the signal which the
experimenter emits, giving or withholding cigarettes, is reversed.
What the experimenter thought was a reward turns out to be a
message of partial indignity, and what the experimenter thought was
a punishment becomes in part a source of satisfaction.

Consider the acute pain of the mental patient in a large hospital
who is momentarily treated as a human being by a member of the

To explain the observed phenomena we always have to consider
the wider context of the learning experiment, and every transaction
between persons is a context of learning.

The double bind hypothesis, then, depends upon attributing
certain characteristics to the learning process. If this hypothesis is
even approximately true, room must be made for it within the theory
of learning. In particular, learning theory must be made
discontinuous so as to accommodate the discontinuities of the
hierarchy of the contexts of learning to which I have referred.

Moreover, these discontinuities are of a peculiar nature. I have
said that the larger context may change the sign of the
reinforcement proposed by a given message, and evidently the
larger context may also change the mode—may place the message
in the category of humor, metaphor, etc. The setting may make the
message inappropriate. The message may be out of tune with the
larger context, and so on. But there are limits to these
modifications. The context may tell the recipient anything about
the message, but it cannot ever destroy or directly contradict the
latter. “I was lying when I said `The cat is on the mat’ ” tells the
vis-a-vis nothing about the location of the cat. It tells him only
something about the reliability of his previous information. There
is a gulf between context and message (or between metamessage
and message) which is of the same nature as the gulf between a
thing and the word or sign which stands for it, or between the
members of a class and the name of the class. The context (or
metamessage) classifies the message, but can never meet it on
equal terms.

In order to fit these discontinuities into learning theory, it is
necessary to enlarge the scope of what is to be included within the
concept of learning. What the experimenters have described as
“learning” are in general changes in what an organism does in
response to a given signal. The experimenter observes, for
example, that at first the buzzer evokes no regular response, but
that after repeated trials in which the buzzer has been followed by
meat powder, the animal will begin to salivate whenever it hears
the buzzer. We may say loosely that the animal has begun to attach
significance or meaning to the buzzer.

A change has occurred. In order to construct a hierarchic series,
we pick on the word “change.” Series such as we are interested in
are in general built in two ways. Within the field of pure
communications theory, the steps of an hierarchic series may be
constructed by successive use of the word “about,” or “meta.” Our
hierarchic series will then consist of message, metamessage, metametamessage,
and so on. Where we deal with phenomena marginal
to communications theory, similar hierarchies may be constructed
by the piling up of “change” upon “change.” In classical physics, the
sequence: position; velocity (i.e., change in position); acceleration


(i.e., change in velocity or change in change of position); change of
acceleration, etc., is an example of such a hierarchy.

Further complications are added—rarely in classical physics but
commonly in human communication—by noting that messages may
be about (or “meta” to) the relationship between messages of
different levels. The smell of the experimental harness may tell the
dog that the buzzer will mean meat powder. We will then say that
the message of the harness is meta to the message of the buzzer. But
in human relations another sort of complexity may be generated;
e.g., messages may be emitted forbidding the subject to make the
meta connection. An alcoholic parent may punish a child for
showing that he knows that he should look out for storms whenever
the parent gets the bottle out of the cupboard. The hierarchy of
messages and contexts thus becomes a complex branching structure.

So we can construct a similar hierarchic classification within
learning theory in substantially the same way as the physicists. What
the experimenters have investigated is change in the receipt of a
signal. But, clearly, to receive a signal already denotes change—a
change of a simpler or lower order than that which the
experimenters have investigated. This gives us the two first steps in
a hierarchy of learning, and above these an infinite series can be
imagined. This hierarchy46 can now be laid out as follows :

(1) The Receipt of a Signal I am working at my desk on which
there is a paper bag, containing my lunch. I hear the hospital
whistle, and from this I know that it is twelve o’clock. I reach out
and take my lunch. The whistle may be regarded as an answer to a
question laid down in my mind by previous learning of the second
order; but the single event—the receiving of this piece of
information—is a piece of learning, and is demonstrated to be so
by the fact that having received it, I am now changed and respond
in a special way to the paper bag.
(2) Those Learnings Which Are Changes in (1) These are
exemplified by the classical learning experiments of various kinds:
46 ‘ 1971. In my final version of this hierarchy of orders of learning, published in
this volume as “The Logical Categories of Learning and Communication,” (see p.
283) I have used a different system of numbering. The receipt of a signal is there
called “Zero Learning”; changes in Zero Learning are called Learning I; “deuterolearning”
is called Learning II, etc


Pavlovian, instrumental reward, instrumental avoidance, rote, and so

(3) Those Learnings Which Constitute Changes in Second-Order
Learning I have in the past, unfortunately, called these phenomena
“deutero-learning,” and have translated this as “learning to learn.” It
would have been more correct to coin the word trito-learning and to
translate it as “learning to learn to receive signals.” These are the
phenomena in which the psychiatrist is preponderantly interested,
namely, the changes whereby an individual comes to expect his
world to be structured in one way rather than an-other. These are the
phenomena which underlie “transference”—the expectation on a
patient’s part that the relationship with the therapist will contain the
same sorts of contexts of learning that the patient has previously met
with in dealing with his parents.
(4) Changes in Those Processes of Change Referred to in (3)
Whether learning of this fourth order occurs in human beings is
unknown. What the psychotherapist attempts to produce in his
patient is usually a third-order learning, but it is possible, and
certainly conceivable, that some of the slow and unconscious
changes may be shifts in sign of some higher derivative in the
learning process.
At this point it is necessary to compare three types of hierarchy
with which we are faced: (a) the hierarchy of orders of learning; (b)
the hierarchy of contexts of learning, and (c) hierarchies of circuit
structure which we may—indeed, must—expect to find in a
telencephalized brain.

It is my contention that (a) and (b) are synonymous in the sense
that all statements made in terms of contexts of learning could be
translated (without loss or gain) into statements in terms of orders of
learning, and, further, that the classification or hierarchy of contexts
must be isomorphic with the classification or hierarchy of orders of
learning. Beyond this, I believe that we should look forward to a
classification or hierarchy of neurophysiological structures which
will be isomorphic with the other two classifications.

This synonymy between statements about context and statements
about orders of learning seems to me to be self-evident, but
experience shows that it must be spelled out. “The truth cannot be


said so as to be understood, and not be believed,” but, conversely, it
cannot be believed until it is said so as to be understood.

It is necessary first to insist that in the world of communication
the only relevant entities or “realities” are messages, including in
this term parts of messages, relations between messages, significant
gaps in messages, and so on. The perception of an event or object or
relation is real. It is a neurophysiological message. But the ‘event
itself or the object itself cannot enter this world and is, therefore,
irrelevant and, to that extent, unreal. Conversely, a message has no
reality or relevance qua message, in the Newtonian world: it there is
reduced to sound waves or printer’s ink.

By the same token, the “contexts” and “contexts of con-texts”
upon which I am insisting are only real or relevant insofar as they
are communicationally effective, i.e., function as messages or
modifiers of messages.

The difference between the Newtonian world and the world of
communication is simply this: that the Newtonian world ascribes
reality to objects and achieves its simplicity by excluding the
context of the context—excluding indeed all metarelationships—a
fortiori excluding an infinite regress of such relations. In contrast,
the theorist of communication insists upon examining the
metarelationships while achieving its simplicity by excluding all

This world, of communication, is a Berkeleyan world, but the
good bishop was guilty of understatement. Relevance or reality
must be denied not only to the sound of the tree which falls unheard
in the forest but also to this chair which I can see and on which I am
sitting. My perception of the chair is communicationally real, and
that on which I sit is, for me, only an idea, a message in which I put
my trust.

“In my thought, one thing is as good as another in this world, and
the shoe of a horse will do,” because in thought and in experience
there are no things, but only messages and the like.

In this world, indeed, I, as a material object, have no relevance
and, in this sense, no reality. “I,” however, exist in the
communicational world as an essential element in the syntax of my
experience and in the experience of others, and the communications

of others may damage my identity, even to the point of breaking up
the organization of my experience.

Perhaps one day, an ultimate synthesis will be achieved to
combine the Newtonian and the communicational worlds. But that
is not the purpose of the present discussion. Here I am concerned
to make clear the relation between the con-texts and the orders of
learning, and to do this it was first necessary to bring into focus the
difference between Newtonian and communicational discourse.

With this introductory statement, however, it becomes clear that
the separation between contexts and orders of learning is only an
artifact of the contrast between these two sorts of discourse. The
separation is only maintained by saying that the contexts have
location outside the physical individual, while the orders of
learning are located inside. But in the communicational world, this
dichotomy is irrelevant and meaningless. The contexts have
communicational reality only insofar as they are effective as
messages, i.e., insofar as they are represented or reflected
(correctly or with distortion) in multiple parts of the
communicational system which we are studying; and this system is
not the physical individual but a wide network of pathways of
messages. Some of these pathways happen to be located outside
the physical individual, others inside; but the characteristics of the
system are in no way dependent upon any boundary lines which
we may superpose upon the communicational map. It is not
communicationally meaningful to ask whether the blind man’s
stick or the scientist’s microscope are “parts” of the man who uses
them. Both stick and microscope are important pathways of
communication and, as such, are parts of the network in which we
are interested; but no boundary line—e.g., halfway up the stick—
can be relevant in a description of the topology of this net.

However, this discarding of the boundary of the physical
individual does not imply (as some might fear) that communicational
discourse is necessarily chaotic. On the contrary, the
proposed hierarchic classification of learning and/or context is an
ordering of what to the Newtonian looks like chaos, and it is this
ordering that is demanded by the double-bind hypothesis.


Man must be the sort of animal whose learning is characterized
by hierarchic discontinuities of this sort, else he could not become
schizophrenic under the frustrations of the double bind.

On the evidential side, there is beginning to be a body of
experiment demonstrating the reality of third-order learning47; but
on the precise point of discontinuity between these orders of
learning there is, so far as I know, very little evidence. The
experiments of John Stroud are worth quoting. These were tracking
experiments. The subject is faced with a screen on which a spot
moves to represent a moving target. A second spot, representing
the aim of a gun, can be controlled by the subject, who operates a
pair of knobs. The subject is challenged to maintain coincidence
between the target spot and the spot over which he has control. In
such an experiment the target can be given various sorts of motion,
characterized by second-, third-, or higher-order derivatives.
Stroud showed that, as there is a discontinuity in the orders of the
equations which a mathematician might use to describe the
movements of the target spot, so also there is a discontinuity in the
learning of the experimental subject. It is as if a new learning
process were involved with each step to a higher order of
complexity in the movement of the target.

It is to me fascinating to find that what one had supposed was a
pure artifact of mathematical description is also apparently an
inbuilt characteristic of the human brain, in spite of the fact that
this brain certainly does not operate by means of mathematical
equations in such a task.

There is also evidence of a more general nature which would
support the notion of discontinuity between the orders of learning.
There is, for example, the curious fact that psychologists have not
habitually regarded what I call learning of the first order, the receipt
of a meaningful signal, as learning at all; and the other curious fact,
that psychologists have until recently shown very little appreciation
of that third order of learning, in which the psychiatrist is predominantly
interested. There is a formidable gulf between the thinking of

47 C. L. Hull, et al., Mathematico-deductive Theory of Rote Learning: A Study
in Scientific Methodology, (Yale University Institute of Human Relations), New
Haven, Yale University Press, 1940; also H. F. Harlow, “The Formation of
Learning Sets,” Psychol. Review, 1949, 56: 51-65.

the experimental psychologist and the thinking of the psychiatrist or
anthropologist. This gulf I believe to be due to the discontinuity in
the hierarchic structure.

Learning, Genetics, and Evolution

Before we consider the impact of the double bind hypothesis
upon genetics and evolutionary theory, it is necessary to examine the
relationship between theories of learning and these two other bodies
of knowledge. I referred earlier to the three subjects together as a
triad. The structure of this triad we must now consider.

Genetics, which covers the communicational phenomena of
variation, differentiation, growth, and heredity, is commonly
recognized as the very stuff of which evolutionary theory is made.
The Darwinian theory, when purged of Lamarckian ideas, consisted
of a genetics in which variation was presumed to be random,
combined with a theory of natural selection would impart adaptive
direction to the accumulation of changes. But the relation between
learning and this theory has been a matter of violent controversy
which has raged over the so-called “inheritance of acquired

Darwin’s position was acutely challenged by Samuel Butler, who
argued that heredity should be compared with—even identified with
—memory. Butler proceeded from this premise to argue that the
processes of evolutionary change, and especially adaptation, should
be regarded as the achievements of a deep cunning in the ongoing
flow of life, not as fortuitous bonuses conferred by luck. He drew a
close analogy between the phenomena of invention and the phenomena
of evolutionary adaptation, and was perhaps the first to
point out the existence of residual organs in machines. The curious
homology whereby the engine is located in the front of an
automobile, where the horse used to be, would have delighted him.
He also argued very cogently that there is a process whereby the
newer inventions of adaptive behavior are sunk deeper into the
biological system of the organism. From planned and conscious
actions they become habits, and the habits become less and less


conscious and less and less subject to voluntary control. He
assumed, with-out evidence, that this habitualization, or sinking
process, could go so deep as to contribute to the body of memories,
which we would call the genotype, and which determine the
characteristics of the next generation.

The controversy about the inheritance of acquired characteristics
has two facets. On the one hand, it appears to be an argument which
could be settled by factual material. One good case of such
inheritance might settle the matter for the Lamarckian side. But the
case against such inheritance, being negative, can never be proved
by evidence and must rely upon an appeal to theory. Usually those
who take the negative view argue from the separation between germ
plasm and somatic tissue, urging that there can be no systematic
communication from the soma to the germ plasm in the light of
which the genotype might revise itself.

The difficulty looks like this: conceivably a biceps muscle
modified by use or disuse might secrete specific metabolites into the
circulation, and these might conceivably serve as chemical
messengers from muscle to gonad. But (a) it is difficult to believe
that the chemistry of biceps is so different from that of, say, triceps
that the message could be specific, and (b) it is difficult to believe
that the gonad tissue could be equipped to be appropriately affected
by such messages. After all, the receiver of any message must know
the code of the sender, so that if the germ cells are able to receive
the messages from the somatic tissue, they must already be carrying
some version of the somatic code. The directions which
evolutionary change could take with the aid of such messages from
the soma would have to be prefigured in the germ plasm.

The case against the inheritance of acquired characteristics thus
rests upon a separation, and the difference between the schools of
thought crystallizes around philosophic reactions to such a
separation. Those who are willing to think of the world as organized
upon multiple and separable principles will accept the notion that
somatic changes induced by environment may be covered by an
explanation which could be totally separate from the explanation of
evolutionary change. But those who prefer to see a unity in nature
will hope that these two bodies of explanation can somehow be

Moreover, the whole relationship between learning and
evolution has undergone a curious change since the days when
Butler maintained that evolution was a matter of cunning rather
than luck, and the change which has taken place is certainly one
which neither Darwin nor Butler could have foreseen. What. has
happened is that many theorists now assume learning to be
fundamentally a stochastic or probabilistic affair, and indeed, apart
from nonparsimonious theories which would postulate some
entelechy at the console of the mind, the stochastic approach is
perhaps the only organized theory of the nature of learning. The
notion is that random changes occur, in the brain or else-where,
and that the results of such random change are selected for survival
by processes of reinforcement and extinction. In basic theory,
creative thought has come to resemble the evolutionary process in
its fundamentally stochastic nature. Reinforcement is seen as
giving direction to the accumulation of random changes of the
neural system, just as natural selection is seen as giving direction
to the accumulation of random changes of variation.

In both the theory of evolution and the theory of learning,
however, the word “random” is conspicuously undefined, and the
word is not an easy one to define. In both fields, it is assumed that
while change may be dependent upon probabilistic phenomena, the
probability of a given change is determined by something different
from probability. Underlying both the stochastic theory of
evolution and that of learning, there are unstated theories regarding
the determinants of the probabilities in question.48 If, however, we
ask about change in these determinants, we shall again be given
stochastic answers, so that the word “random,” up-on which all of
these explanations turn, appears to be a word whose meaning is
hierarchically structured, like the meaning of the word “learning,”
which was discussed in the first part of this lecture.

Lastly, the question of the evolutionary function of acquired
characteristics has been reopened by Waddington’s work on
phenocopies in Drosophila. At the very least, this work indicates
that the changes of phenotype which can be achieved by the

48 In this sense, of course, all the theories of change assume that the next change
is in some degree prefigured in the system which is to undergo that change.


organism under environmental stress are a very important part of the
machinery by which the species or hereditary line maintains its
place in a stressful and competitive environment, pending the later
appearance of some mutation or other genetic change which may
make the species or line better able to deal with the ongoing stress.
In this sense at least, the acquired characteristics have important
evolutionary function. However, the actual experimental story
indicates something more than this and is worth reproducing briefly.

What Waddington works with is a phenocopy of the phenotype
brought about by the gene bithorax. This gene has very profound
effects upon the adult phenotype. In its presence the third segment
of the thorax is modified to resemble the second, and the little
balancing organs, or halteres, on this third segment become wings.
The result is a four-winged fly. This four-winged characteristic can
be produced artificially in flies which do not carry the gene bithorax
by subjecting the pupae to a period of intoxication with ethyl ether.
Waddington works with large populations of Drosophila flies
derived from a wild strain believed to be free of the gene bithorax.
He subjects the pupae of this population in successive generations to
the ether treatment, and from the resulting adults selects for
breeding those which show the best approximation to bithorax. He
has continued this experiment over many generations, and already in
the twenty-seventh generation he finds that the bithorax appearance
is achieved by a limited number of flies whose pupae were
withdrawn from the experimental treatment and not subjected to
ether. Upon breeding from these, it turns out that their bithorax
appearance is not due to the presence of the specific gene, bithorax,
but is due to a constellation of genes which work together to give
this effect.

These very striking results can be read in various ways. We can
say that in selecting the best phenocopies, Wadding-ton was in fact
selecting for a genetic potentiality for achieving this phenotype. Or
we can say that he was selecting to reduce the threshold of ether
stress necessary to produce this result.

Let me suggest a possible model for the description of these
phenomena. Let us suppose that the acquired characteristic is
achieved by some process of fundamentally stochastic nature—
perhaps some sort of somatic learning—and the mere fact that

Waddington is able to select the “best” phenocopies would lend
support to this assumption. Now, it is evident that any such process
is, in the nature of the case, wasteful. To achieve a result by trial and
error which could have been achieved in any more direct way
necessarily consumes time and effort in some sense of these words.
Insofar as we think of adaptability as achieved by stochastic process,
we let in the notion of an economics of adaptability.

In the field of mental process, we are very familiar with this
sort of economics, and in fact a major and necessary saving is
achieved by the familiar process of habit formation. We may, in the
first instance, solve a given problem by trial and error; but when
similar problems recur later, we tend to deal with them more and
more economically by taking them out of the range of stochastic
operation and handing over the solutions to a deeper and less
flexible mechanism, which we call “habit.” It is, therefore,
perfectly conceivable that some analogous phenomenon may
obtain in regard to the production of bithorax characteristics. It
may be more economical to produce these by the rigid mechanism
of genetic determination rather than by the more wasteful, more
flexible (and perhaps less predictable) method of somatic change.

This would mean that in Waddington’s population of flies there
would be a selective benefit for any hereditary line of flies which
might contain appropriate genes for the whole—or for some part—
of the bithorax phenotype. It is also possible that such flies would
have an extra advantage in that their somatic adaptive machinery
might then be available for dealing with stresses of other kinds. It
would appear that in learning, when the solution of the given
problem has been passed on to habit, the stochastic or exploratory
mechanisms are set free for the solution of other problems, and it
is quite conceivable that a similar advantage is achieved by passing
on the business of determining a somatic characteristic to the genescript49

49 These considerations alter somewhat the old problem of the evolutionary effect
of use and disuse. Orthodox theory could only suggest that a mutation reducing the
(potential) size of a disused organ had survival value in terms of the resulting
economy of tissue. The present theory would suggest that atrophy of an organ,
occurring at the somatic level, may constitute a drain upon the total available
adaptability of the organism, and that this waste of adaptability might be saved if


It may be noted that such a model would be characterized by two
stochastic mechanisms: first, the more superficial mechanism by
which the changes are achieved at the somatic level, and, second,
the stochastic mechanism of mutation (or the shuffling of gene
constellations) at the chromosomal level. These two stochastic
systems will, in the long run under selective conditions, be
compelled to work together, even though no message can pass from
the more superficial somatic system to the germ plasm. Samuel Butler’s
hunch that something like “habit” might be crucial in evolution
was perhaps not too wide of the mark.

With this introduction we can now proceed to look at the
problems which a double bind theory of schizophrenia would pose
for the geneticist.

Genetic Problems Posed by Double Bind Theory

If schizophrenia be a modification or distortion of the learning
process, then when we ask about the genetics of schizophrenia, we
cannot be content just with genealogies upon which we discriminate
some individuals who have been committed to hospitals, and others
who have not. There is no a priori expectation that these distortions
of the learning process, which are highly formal and abstract in their
nature, will necessarily appear with that appropriate content which
would result in hospital commitment. Our task as geneticists will not
be the simple one upon which the Mendelians concentrated,
assuming a one-to-one relation between phenotype and genotype.
We cannot simply assume that the hospitalized members carry a
gene for schizophrenia and that the others do not. Rather, we have
to expect that several genes or constellations of genes will alter
patterns and potentialities in the learning process, and that certain
of the resultant patterns, when confronted by appropriate forms of
environmental stress, will lead to overt schizophrenia.

In the most general terms, any learning, be it the absorption of
one bit of information or a basic change in the character structure

reduction of the organ could be achieved more directly by genetic determinants.

of the whole organism, is, from the point of view of genetics, the
acquisition of an “acquired characteristic.” It is a change in the
phenotype, of which that phenotype was capable thanks to a whole
chain of physiologic and embryologic processes which lead back
to the genotype. Every step in this backward leading series may
(conceivably) be modified or interrupted by environmental
impacts; but, of course, many of the steps will be rigid in the sense
that environmental impact at that point would destroy the
organism. We are concerned only with those points in the hierarchy
at which environment can take effect and the organism still be
viable. How many such points there may be we are far from
knowing. And ultimately, when we reach the genotype, we are
concerned to know whether the genotypic elements in which we
are interested are or are not variable. Do differences occur from
genotype to genotype which will affect the modifiability of the
processes leading to the phenotypic behaviors which we observe?

In the case of schizophrenia we deal evidently with a relatively
long and complex hierarchy; and the natural history of the disease
indicates that the hierarchy is not merely a chain of causes and
effects from gene-script to phenotype, which chain becomes at
certain points conditional upon environmental factors. Rather, it
seems that in schizophrenia the enviromental factors themselves
are likely to be modified by the subject’s behavior whenever
behavior related to schizophrenia starts to appear.

To illustrate these complexities, it is perhaps worthwhile to
consider for a moment the genetic problems presented by other
forms of communicational behavior—humor, mathematical skill, or
musical composition. Perhaps in all these cases, there are
considerable genetic differences between individuals in those factors
which make for an ability to acquire the appropriate skills. But the
skills themselves and their particular expression also depend largely
upon environmental circumstances and even upon specific training.
In addition, however, to these two components of the situation, there
is the fact that the individual who shows ability, e.g., in musical
composition, is likely to mold his environment in a direction which
will favor his developing his ability, and that he will, in turn, create
an environment for others which will favor their development in the
same direction.


In the case of humor, the situation may even be one degree
more complicated. It is not clear that in this case the relationship
between humorist and his human environment will necessarily be
symmetrical. Granted that in some cases the humorist promotes
humor in others, in many other cases there occurs the well-known
complementary relationship between humorist and “straight” man.
Indeed, the humorist, insofar as he hogs the center of the stage,
may reduce others to the position of receiving humor but not
themselves contributing.

These considerations can be applied unchanged to the problem
of schizophrenia. Anybody watching the trans-actions which occur
between the members of a family containing an identified
schizophrenic will perceive immediately that the symptomatic
behavior of the identified patient fits with this environment and,
indeed, promotes in the other members those characteristics
which.evoke the schizophrenic behavior. Thus, in addition to the
two stochastic mechanisms outlined in the previous section, we
now face a third, namely the mechanism of those changes whereby
the family, perhaps gradually, becomes organized (i.e., limits the
behaviors of the component individuals) in such a way as to fit the

A question which is frequently asked is this: “If this family is
schizophrenogenic, how does it happen that all of the siblings are
not diagnosable as schizophrenic patients?” Here it is necessary to
insist that the family, like any other organization, creates and
depends upon differentiation among its members. As in many
organizations, there is room only for one boss, in spite of the fact
that the organization operates upon those premises which would
induce administrative skill and ambition in its members; so also in
the schizophrenogenic family there may be room for only one
schizophrenic. The case of the humorist is quite comparable. The
organization of the Marx family, which could create four
professional humorists, must have been quite exceptional. More
usually one such individual would suffice to re-duce the others to
more commonplace behavioral roles. Gene-tics may play a role in
deciding which of several siblings shall be the schizophrenic—or
which shall be the clown—but it is by no means clear that such

hereditary factors could completely determine the evolution or roles
within the family organization.

A second question—to which we have no final answer—
concerns the degree of schizophrenia (genetic and/or acquired)
which must be assigned to the schizophrenogenic parent. Let me,
for purposes of the present inquiry, define two degrees of
schizophrenic symptomatology, and note that the so-called
“psychotic break” sometimes divides these two degrees.

The more serious and conspicuous degree of symptomatology is
what is conventionally called schizophrenia. I will call it “overt
schizophrenia.” The persons so afflicted be-have in ways which are
grossly deviant from the cultural-environment. In particular, their
behavior seems characterized by conspicuous or exaggerated errors
and distortions regarding the nature and typing of their own
messages (internal and external), and of the messages which they
receive from others. Imagination is seemingly confused with
perception. The literal is confused with the metaphoric. Internal
messages are confused with external. The trivial is confused with
the vital. The originator of the message is confused with the
recipient and the perceiver with the thing perceived. And so on. In
general, these distortions boil down to this: that the patient behaves
in such a way that he shall be responsible for no
metacommunicative aspect of his messages. He does this,
moreover, in a manner which makes his condition conspicuous: in
some cases, flooding the environment with messages whose logical
typing is either totally obscure or misleading; in other cases,
overtly withdrawing to such a point that he commits himself to no
overt message.

In the “covert” case the behavior of the identified patient is
similarly but less conspicuously characterized by a continual
changing of the logical typing of his or her messages, and a
tendency to respond to the messages of others (especially to those of
other family members) as though these were of logical type,
different from that which the speaker intended. In this system of
behavior the messages of the vis-a-vis are continually disqualified,
either by indicating that they are inappropriate replies to what the
covert schizophrenic has said or by indicating that they are the
product of some fault in the character or motivation of the speaker.


Moreover, this destructive behavior is in general maintained in such
a way as to be undetected. So long as the covert schizophrenic can
succeed in putting the other in the wrong, his or her pathology is
obscured and the blame falls else-where. There is some evidence
that these persons fear col-lapse into overt schizophrenia when faced
by circumstances which would force them to recognize the pattern
of their operations. They will even use the threat, “You are driving
me crazy,” as a defense of their position.

What I am here calling covert schizophrenia is characteristic of
the parents of schizophrenics in the families which we have
studied. This behavior, when it occurs in the mother, has been
extensively caricatured; so I shall use here an example of which
the central figure is the father. Mr. and Mrs. P. had been married
some eighteen years and have a near-hebephrenic son of sixteen.
Their marriage is difficult and is characterized by almost continual
hostility. However, she is a keen gardener, and on a certain Sunday
afternoon they worked together planting roses in what was to be
her rose garden. She recalls that this was an unusually pleasant
occasion. On Monday morning, the husband went to work as usual,
and while he was gone Mrs. P. received a phone call from a
complete stranger inquiring, rather apologetically, when Mrs. P.
was going to leave the house. This came as somewhat of a
surprise. She did not know that from her husband’s point of view
the messages of shared work on the rose garden were framed
within the larger context of his having agreed during the previous
week to sell the house.

In some cases, it almost looks as though the overt schizophrenic
were a caricature of the covert.

If we assume that both the grossly schizophrenic symptoms of
the identified patient and the “covert schizophrenia” of the parents
are in part determined by genetic factors, i.e., that, given the
appropriate experiential setting, genetics in some degree renders the
patient more liable to develop these particular patterns of behavior,
then we have to ask how these two degrees of pathology might be
related in a genetic theory.

Certainly, no answer to this question is at present avail-able, but
it is clearly possible we here face two quite distinct problems. In the
case of the overt schizophrenic, the geneticist will have to identify

those formal characteristics of the patient which will render him
more likely to be driven to a psychotic break by the covertly
inconsistent behavior of his parents (or by this in conjunction and
contrast with the more consistent behavior of people outside the
family). It is too early to make a specific guess at these
characteristics, but we may reasonably assume that they would
include some sort of rigidity. Perhaps the person prone to overt
schizophrenia would be characterized by some extra strength of
psychological commitment to the status quo as he at the moment
sees it, which commitment would be hurt or frustrated by the
parents’ rapid shifts of frame and context. Or perhaps this patient
might be characterized by the high value of some parameter
determining the relationship between problem solving and habit
formation. Perhaps it is the person who too readily hands over the
solutions to habit who is hurt by those changes in context which
invalidate his solutions just at the moment when he has incorporated
them into his habit structure.

In the case of covert schizophrenia, the problem for the geneticist
will be different. He will have to identify those formal
characteristics which we observe in the parents of the schizophrenic.
Here what is required would seem to be a flexibility rather than a
rigidity. But, having had some experience in dealing with these
people, ‘I must confess to feeling that they are rigidly committed to
their patterns of inconsistency.

Whether the two questions which the geneticist must answer can
simply be lumped together by regarding the covert patterns as
merely a milder version of the overt, or can be brought under a
single head by suggesting that in some sense the same rigidity
operates at different levels in the two cases, I do not know.

Be that as it may, the difficulties which we here face are entirely
characteristic of any attempt to find a genetic base for any
behavioral characteristic. Notoriously, the sign of any message or
behavior is subject to reversal, and this generalization is one of the
most important. contributions of psycho-analysis, to our thinking. If
we find that a sexual exhibitionist is the child of a prudish parent,
are we justified in going to the geneticist to ask him to trace out the
genetics of some basic characteristic which will find its phenotypic
expression both in the prudishness of the parent and in the exhibi


tionism of the offspring? The phenomena of suppression and
overcompensation lead continually to the difficulty that an excess of
something at one level (e.g., in the genotype) may lead to a
deficiency of the direct expression of that something at , some more
superficial level (e.g,. in the phenotype). And conversely.

We are very far, then, from being able to pose specific questions
for the geneticist; but I believe that the wider implications of what I
have been saying modify somewhat the philosophy of genetics. Our
approach to the problems of schizophrenia by way of a theory of
levels or logical types has disclosed first that the problems of
adaptation and learning and their pathologies must be considered in
terms of a hierarchic system in which stochastic change occurs at
the boundary points between the segments of the hierarchy. We have
considered three such regions of stochastic change —the level of
genetic mutation, the level of learning, and the level of change in
family organization. We have disclosed the possibility of a
relationship of these levels which orthodox genetics would deny,
and we have disclosed that at least in human societies the
evolutionary system consists not merely in the selective survival of
those persons who happen to select appropriate environments but
also in the modification of family environment in a direction which
might enhance the phenotypic and genotypic characteristics of the
individual members.

What Is Man?

If I had been asked fifteen years ago what I understood by the
word materialism, I think I should have said that materialism is a
theory about the nature of the universe, and I would have accepted
as a matter of course the notion that this theory is in some sense
nonmoral. I would have agreed that the scientist is an expert who
can provide himself and others with insights and techniques, but that
science could have nothing to say about whether these techniques
should be used. In this, I would have been following the general
trend of scientific philosophy associated with such names at

Democritus, Galileo, Newton,50 Lavoisier, and Darwin. I would have
been discarding the less respectable views of such men as
Heraclitus, the alchemists, William Blake, Lamarck, and Samuel
Butler. For these, the motive for scientific inquiry was the desire to
build a comprehensive view of the universe which should show
what Man is and how he is related to the rest of the universe. The
picture which these men were trying to build was ethical and

There is this much connection certainly between scientific truth,
on the one hand, and beauty and morality, on the other: that if a man
entertain false opinions regarding his own nature, he will be led
thereby to courses of action which will be in some profound sense
immoral or ugly.

Today, if asked the same question regarding the meaning of
materialism, I would say that this word stands in my thinking for a
collection of rules about what questions should be asked regarding
the nature of the universe. But I would not suppose that this set of
rules has any claim to be uniquely right.

The mystic “sees the world in a grain of sand,” and the world
which he sees is either moral or aesthetic, or both. The Newtonian
scientist sees a regularity in the behavior of falling bodies and
claims to draw from this regularity no normative conclusions
whatsoever. But his claim ceases to be consistent at the moment
when he preaches that this is the right way to view the universe. To
preach is possible only in terms of normative conclusions.

I have touched upon several matters in the course of this lecture
which have been foci of controversy in the long battle between a
nonmoral materialism and a more romantic view of the universe.
The battle between Darwin and Samuel Butler may have owed some
of its bitterness to what looked like personal affronts, but behind all
this the argument concerned a question which had religious status.

50 The name of Newton certainly belongs in this list. But the man himself was of
a different kidney. His mystical preoccupation with alchemy and apocalyptic writings,
and his secret theological monism indicate that he was not the first objective
scientist but, rather, the “last of the magicians” (see J. M. Keynes, “Newton, the
Man,” Tercentenary Celebrations, London, Cambridge University Press, 1947, pp.
27-34). Newton and Blake were alike in devoting much time and thought to the
mystical works of Jacob Boehme.


The battle was really about “vitalism.” It was a question of how
much life and what order of life could be assigned to organisms; and
Darwin’s victory amounted to this, that while he had not succeeded
in detracting from the mysterious liveliness of the individual
organism, he had at least demonstrated that the evolutionary picture
could be reduced to natural “law.”

It was, therefore, very important to demonstrate that the as yet
unconquered territory—the life of the individual organism—could
not contain anything which would recapture this evolutionary
territory. It was still mysterious that living organisms could achieve
adaptive change during their individual lives, and at all costs these
adaptive changes, the famous acquired characteristics, must not have
influence up-on the evolutionary tree. The “inheritance of acquired
characteristics” threatened always to recapture the field of evolution
for the vitalist side. One part of biology must be separate from the
other. The objective scientists claimed, of course, to believe in a
unity in nature—that ultimately the whole of natural phenomena
would prove susceptible to their analysis, but for about a hundred
years it was convenient to set up an impermeable screen between the
biology of the individual and the theory of evolution. Samuel
Butler’s “inherited memory” was an attack upon this screen.

The question with which I am concerned in this concluding
section of the lecture could be put in various ways. Is the battle
between nonmoral materialism and the more mystical view of the
universe affected by a change in the function assigned to the
“acquired characteristics?” Does the older materialist thesis really
depend upon the premise that contexts are isolable? Or is our view
of the world changed when we admit an infinite regress of contexts,
linked to each other in a complex network of metarelations? Does
the possibility that the separate levels of stochastic change (in
phenotype and genotype) may be connected in the larger context of
the ecological system alter our allegiance in the battle?

In breaking away from the premise that contexts are al-ways
conceptually isolable, I have let in the notion of a universe much
more unified—and in that sense much more mystical—than the
conventional universe of nonmoral materialism. Does the new
position so achieved give us new grounds for hope that science
might answer moral or aesthetic questions?

I believe that the position is significantly changed, and perhaps
I can best make this clear by considering a matter which you as
psychiatrists have thought about many times. I mean the matter of
“control” and the whole related complex suggested by such words
as manipulation, spontaneity, free will, and technique. I think you
will agree with me that there is no area in which false premises
regarding the nature of the self and its relation to others can be so
surely productive of destruction and ugliness as this area of ideas
about control. A human being in relation with another has very
limited control over what happens in that relationship. He is a
part of a two-person unit, and the. control which any part can
have over any whole is strictly limited.

The infinite regress of contexts which I have talked about is only
another example of the same phenomenon. What I have contributed
to this discussion is the notion. that the contrast between part and
whole, whenever this contrast appears in the realm of
communication, is simply a contrast in logical typing. The whole is
always in a metarelationship with its parts. As in logic the
proposition can never determine the meta proposition, so also in
matters of control the smaller context can never determine the larger.
I have remarked (e.g., when discussing the phenomena of
phenotypic compensation) that in hierarchies of logical typing there
is often some sort of change of sign at each level, when the levels
are related to each other in such a way as to create a self-corrective
system. This appears in a simple diagrammatic form in the initiatory
hierarchy which I studied in a New Guinea tribe. The initiators are
the natural enemies of the novices, because it is their task to bully
the novices into shape. The men who initiated the present initiators
now have a role of criticizing what is now being done in the initiation
ceremonies, and this makes them the natural allies of the
present novices. And so on. Something of the same sort also occurs
in American college fraternities, where juniors tend to be allied with
freshmen and seniors with sophomores.

This gives us a view of the world which is still almost
unexplored. But some of its complexities may be suggested by a
very crude and imperfect analogy. I think that the functioning of
such hierarchies may be compared with the business of trying to
back a truck to which one or more trailers are attached. Each

segmentation of such a system denotes a reversal of sign, and each
added segment denotes a drastic decrease in the amount of control
that can be exerted by the driver of the truck. If the system is
parallel to the right-hand side of the road, and he wants the trailer
immediately behind him to approach the right-hand side, he must
turn his front wheels to the left. This will guide the rear of the
truck away from the right-hand side of the road so that the front of
the trailer is pulled over to its left. This will now cause the rear of
the trailer to point toward the right. And so on.

As anybody who has attempted this will know, the amount of
available control falls off rapidly. To back a truck with one trailer
is already difficult because there is only a limited range of angles
within which the control can be exerted. If the trailer is in line, or
almost in line, with the truck, the control is easy, but as the angle
between trailer and truck diminishes, a point is reached at which
control is lost and the attempt to exert it only results in jackknifing
of the system. When we consider the problem of controlling a
second trailer, the threshold for jackknifing is drastically reduced,
and control becomes, therefore, almost negligible.

As I see it, the world is made up of a very complex net-work
(rather than a chain) of entities which have this sort of relationship
to each other, but with this difference, that many of the entities
have their own supplies of energy and perhaps even their own
ideas of where they would like to go.

In such a world the problems of control become more akin to
art than to science, not merely because we tend to think of the
difficult and the unpredictable as contexts for art but also because
the results of error are likely to be ugliness.

Let me then conclude with a warning that we social scientists
would do well to hold back our eagerness to control that world
which we so imperfectly understand. The fact of our imperfect
understanding should not be allowed to feed our anxiety and so
increase the need to control. Rather, our studies could be inspired by
a more ancient, but today less honored, motive: a curiosity about the
world of which we are part. The rewards of such work are not power
but beauty.

It is a strange fact that every great scientific advance—not least
the advances which Newton achieved—has been elegant.

Additional References

W. R. Ashby, Design for a Brain, New York, John Wiley & Sons,
Inc., 1952.
, Introduction to Cybernetics, New York and
London, John Wiley & Sons, Inc., 1956.
G. Bateson, D. D. Jackson, J. Haley, and J. H. Weakland, “Toward
a Theory of Schizophrenia,” Behavioral Science, 1956, 1:
G. Bateson, “Cultural Problems Posed by a Study of Schizophrenic
Process,” Symposium on Schizophrenia, an Integrated
Approach, by Alfred Auerback, M. D., ed., American
Psychiatric Association, Symposium of the Hawaiian Divisional
Meeting, 1958, New York, Ronald Press, 1959.
, “The New Conceptual Frames for Behavioral Re-search,”
Proceedings of the Sixth Annual Psychiatric Conference at
the New Jersey Neuro-Psychiatric Institute, Princeton,
1958, pp. 54-71.
, “The Group Dynamics of Schizophrenia,” Chronic
Schizophrenia, L. Appleby, J. M. Scher, and J. H.
Cummings, eds., Glencoe, Ill., The Free Press, 1960.
, “Social Planning and the Concept of Deutero-
Learning,” Relation to the Democratic Way of Life, Conference
on Science, Philosophy and Religion, Second
Symposium, led by L. Bryson and L. Finkelstein, New
York, Harper & Bros., 1942.
, Naven, a Survey of Problems Suggested by a Composite
Picture of Culture of a New Guinea Tribe Drawn from
Three Points of View, Ed. 2, Stanford, Calif., Stanford
University Press, 1958.
Butler, Thought and Language, 1890, published in the
Shrewsbury Edition of the works of Samuel Butler, 1925,
vol. xix.
, Luck or Cunning as the Main Means of Organic
Modification, London, Trubner, 1887.
D. Darlington, “The Origins of Darwinism,” Scientific
American, 1959, 200: 60-65.

Darwin, On the Origin of Species, by Means of Natural
Selection, London, Murray, 1859.
C. C. GiIlispie, “Lamarck and Darwin in the History of Science,”
American Scientist, 1958, 46: 388-409.
Stroud, “Psychological Moment in Perception-Discussion,”
Cybernetics: Circular Causal and Feedback Mechanisms in
Biological and Social Systems, Transactions of the Sixth
Conference, H. Von Foerster, et al:, eds., New York, Josiah
Macy; Jr. Foundation, 1949, pp. 27-63.
H. Waddington, The Strategy of the Genes, London, George
Allen & Unwin, Ltd., 1957.
, “The Integration of Gene-Controlled Processes and Its
Bearing on Evolution,” Caryologia, Supplement, 1954, pp.
, “Genetic Assimilation of an Acquired Character,”
Evolution, 1953, 7: 118-26.
A. Weismann, Essays upon Heredity, authorized translation, E. B.
Poulton, et al., eds., Oxford, Clarendon Press, 1889.

Double Bind, 1969*

Double bind theory was, for me, an exemplification of how tothink about such matters and, in this aspect at least, the wholebusiness is worth some re-examination.

Sometimes—often in science and always in art—one does not
know what the problems were till after they have been solved. So
perhaps it will be useful to state retrospectively what problems were
solved for me by double bind theory.

First there was the problem of reification.

Clearly there are in the mind no objects or events—no pigs, no
coconut palms, and no mothers. The mind contains only transforms,
percepts, images, etc., and rules for making these transforms,
percepts, etc. In what form these rules exist we do not know, but
presumably they are embodied in the very machinery which creates
the transforms. The rules are certainly not commonly explicit as
conscious “thoughts.”

In any case, it is nonsense to say that a man was frightened by a
lion, because a lion is not an idea. The man makes an idea of the

The explanatory world of substance can invoke no differences
and no ideas but only forces and impacts. And, per contra, the world
of form and communication invokes no things, forces, or impacts
but only differences and ideas. (A difference which makes a
difference is an idea. It is a “bit,” a unit of information.)

But these things I learned only later—was enabled to learn them
by double bind theory. And yet, of course, they are implicit in the
theory which could hardly have been created without them.

Our original paper on the double bind contains numerous errors
due simply to our having not yet articulately examined the

* This paper was given in August, 1969, at a Symposium on the Double Bind;
Chairman, Dr. Robert Ryder; sponsored by the American Psychological Association.
It was prepared under Career Development Award (MH-21,931) of the National
Institute of Mental Health.

reification problem. We talk in that paper as though a double bind
were a something and as though such some-things could be counted.

Of course that’s all nonsense. You cannot count the bats in an
inkblot because there are none. And yet a man—if he be “batminded”—
may “see” several.

But are there double binds in the mind? The question is not
trivial. As there are in the mind no coconuts but only percepts and
transforms of coconuts, so also, when I perceive (consciously or
unconsciously) a double bind in my boss’ behavior, I acquire in my
mind no double bind but only a percept or transform of a double
bind. And that is not what the theory is about.

We are talking then about some sort of tangle in the rules for
making the transforms and about the acquisition or cultivation of
such tangles. Double bind theory asserts that there is an experiential
component in the determination or etiology of schizophrenic
symptoms and related behavioral patterns, such as humor, art,
poetry, etc. Notably the theory does not distinguish between these
subspecies. Within its terms there is nothing to determine whether a
given individual shall become a clown, a poet, a schizophrenic, or
some combination of these. We deal not with a single syndrome but
with a genus of syndromes, most of which are not conventionally
regarded as pathological.

Let me coin the word “transcontextual” as a general term for
this genus of syndromes.

It seems that both those whose life is enriched by trans-
contextual gifts and those who are impoverished by transcontextual
confusions are alike in one respect: for them there is always or often
a “double take.” A falling leaf, the greeting of a friend, or a
“primrose by the river’s brim” is not “just that and nothing more.”
Exogenous experience may be framed in the contexts of dream, and
internal thought may be projected into the contexts of the external
world. And so on. For all this, we seek a partial explanation in
learning and experience.

There must, of course, also be genetic components in the etiology
of the transcontextual syndromes. These would expectably operate
at levels more abstract than the experiential. For example, genetic
components might determine skill in learning to be transcontextual
or (more abstractly) the potentialities for acquiring this skill. Or,

conversely, the genome might determine skills in resisting
transcontextual pathways, or the potentiality for acquiring this latter
skill. (Geneticists have paid very little attention to the necessity of
defining the logical typing of messages carried by DNA.)

In any case, the meeting point where the genetic determination
meets the experiential is surely quite abstract, and this must be true
even though the embodiment of the genetic message be a single
gene. (A single bit of information—a single difference—may be the
yes-or-no answer to a question of any degree of complexity, at any
level of abstraction. )

Current theories which propose (for “schizophrenia”) a single
dominant gene of “low penetrance” seem to leave the field open for
any experiential theory which would indicate what class of
experiences might cause the latent potentiality to appear in the

I must confess however that these theories seem to me of little
interest until the proponents try to specify what components of the
complex process of determining “schizophrenia” are provided by the
hypothetical gene. To identify these components must be a
subtractive process. Where the contribution of environment is
large, the genetics cannot be investigated until the environmental
effect has been identified and can be controlled.

But sauce for the goose is also sauce for the gander, and what is
said above about geneticists places an obligation upon me to make
clear what components of transcontextual process could be provided
by double bind experience. It is appropriate therefore to re-examine
the theory of deuterolearning upon which double bind theory is

All biological systems (organisms and social or ecological
organizations of organisms) are capable of adaptive change. But
adaptive change takes many forms, such as response, learning,
ecological succession, biological evolution, cultural evolution, etc.,
according to the size and complexity of the system which we
choose to consider.

Whatever the system, adaptive change depends upon feedback
loops, be it those provided by natural selection or those of
individual reinforcement. In all cases, then, there must be a process
of trial and error and a mechanism of comparison.


But trial and error must always involve error, and error is
always biologically and/or psychically expensive. It follows
therefore that adaptive change must always be hierarchic.

There is needed not only that first-order change which suits the
immediate environmental (or physiological) demand but also
second-order changes which will reduce the amount of trial and
error needed to achieve the first-order change. And so on. By
superposing and interconnecting many feedback loops, we (and all
other biological systems) not only solve particular problems but
also form habits which we apply to the solution of classes of

We act as though a whole class of problems could be solved in
terms of assumptions or premises, fewer in number than the
members of the class of problems. In other words, we (organisms)
learn to learn, or in the more technical phrase, we deutero-learn.

But habits are notoriously rigid and their rigidity follows as a
necessary corollary of their status in the hierarchy of adaptation.
The very economy of trial and error which is achieved by habit
formation is only possible because habits are comparatively “hard
programmed,” in the engineers’ phrase. The economy consists
precisely in not re-examining or rediscovering the premises of
habit every time the habit is used. We may say that these premises
are partly “unconscious”, or—if you please—that a habit of not
examining them is developed.

Moreover, it is important to note that the premises of habit are
almost necessarily abstract. Every problem is in some degree
different from every other and its description or representation in the
mind will therefore contain unique propositions. Clearly to sink
these unique propositions to the level of premises of habit would be
an error. Habit can deal successfully only with propositions which
have general or repetitive truth, and these are commonly of a
relatively high order of abstraction.51

51 What is important, however, is that the proposition be constantly true, rather than
that it be abstract. It just so happens—coincidentally—that abstractions, if well chosen,
have a constancy of truth. For human beings it is rather constantly true that air is
present around the nose; the reflexes which control respiration can therefore be hard-
programmed in the medulla. For the porpoise, the proposition “air around the
blowhole” is only intermittently true, and therefore respiration must be con-trolled

Now the particular propositions which I believe to be important
in the determination of the transcontextual syndromes are those
formal abstractions which describe and determine interpersonal

I say “describe and determine,” but even this is inadequate.
Better would be to say that the relationship is the exchange of these
messages; or that the relationship is immanent in these messages.

Psychologists commonly speak as if the abstractions of relationship
(“dependency,” “hostility,” “love,” etc.) were real things
which are to be described or “expressed” by messages. This is
epistemology backwards: in truth, the messages constitute the
relationship, and words like `.`dependency” are verbally coded
descriptions of patterns immanent in the combination of exchanged

As has already been mentioned, there are no “things” in the mind
—not even “dependency.”

We are so befuddled by language that we cannot think straight,
and it is convenient, sometimes, to remember that we are really
mammals. The epistemology of the “heart” is that of any nonhuman
mammal. The cat does not say “milk”; she simply acts out (or is) her
end of an interchange, the pattern of which we in language would
call “dependency.”

But to act or be one end of a pattern of interaction is to propose
the other end. A context is set for a certain class of response.

This weaving of contexts and of messages which propose context
—but which, like all messages whatsoever, have “meaning” only by
virtue of context—is the subject matter of the so-called double bind

The matter may be illustrated by a famous and formally correct52
botanical analogy. Goethe pointed out 150 years ‘ ago that there is a
sort of syntax or grammar in the anatomy of flowering plants. A
“stem” is that which bears “leaves”; a “leaf” is that which has a bud
in its axil; a bud is a stem which originates in the axil of a leaf; etc.
The formal (i.e., the communicational) nature of each organ is

in a more flexible manner from some higher center.

52 Formally correct because morphogenesis, like behavior, is surely a matter of
messages in contexts. (See G. Bateson, “A Re-examination of ‘Bateson’s Rule,'”
Journal of Genetics, in press.)


determined by its contextual status—the context in which it occurs
and the context which it sets for other parts.

I said above that double bind theory is concerned with the
experiential component in the genesis of tangles in the rules or
premises of habit. I now go on to assert that experienced breaches in
the weave of contextual structure are in fact “double binds” and
must necessarily (if they contribute at all to the hierarchic processes
of learning and adaptation) promote what I am calling
transcontextual syndromes.

Consider a very simple paradigm: a female porpoise (Steno
bredanensis) is trained to accept the sound of the trainer’s whistle
as a “secondary reinforcement.” The whistle is expectably followed
by food, and if she later repeats what she was doing when the
whistle blew, she will expectably again hear the whistle and receive

This porpoise is now used by the trainers to demonstrate “operant
conditioning” to the public. When she enters the exhibition tank, she
raises her head above surface, hears the whistle and is fed. She then
raises her head again and is again reinforced. Three repetitions of
this sequence is enough for the demonstration and the porpoise is
then sent off-stage to wait for the next performance two hours later.
She has learned some simple rules which relate her actions, the
whistle, the exhibition tank, and the trainer into a pattern—a
contextual structure, a set of rules for how to put the in-formation

But this pattern is fitted only for a single episode in the
exhibition tank. She must break that pattern to deal with the class of
such episodes. There is a larger context of contexts which will put
her in the wrong.

At the next performance, the trainer again wants to demonstrate
“operant conditioning,” but to do this she must pick on a different
piece of conspicuous behavior.

When the porpoise comes on stage, she again raises her head.
But she gets no whistle. The trainer waits for the next piece of
conspicuous behavior—likely a tail flap, which is a common
expression of annoyance. This behavior is then rein-forced and

But the tail flap was, of course, not rewarded in the third

Finally the porpoise learned to deal with the context of contexts
—by offering a different or new piece of conspicuous behavior
whenever she came on stage.

All this had happened in the free natural history of the
relationship between porpoise and trainer and audience. The
sequence was then repeated experimentally with a new porpoise and
carefully recorded.53

Two points from this experimental repeat of the sequence must
be added:

First, that it was necessary (in the trainer’s judgment) to break the
rules of the experiment many times. The experience of being in the
wrong was so disturbing to the porpoise that in order to preserve the
relationship between porpoise and trainer (i.e., the context of context
of context) it was necessary to give many reinforcements to which
the porpoise was not entitled.

Second, that each of the first fourteen sessions was characterized
by many futile repetitions of whatever behavior had been reinforced
in the immediately previous session. Seemingly only by “accident”
did the animal provide a piece of different behavior. In the time-out
between the fourteenth and fifteenth sessions, the porpoise appeared
to be much excited, and when she came on stage for the fifteenth
session she put on an elaborate performance including eight
conspicuous pieces of behavior of which four were entirely new—
never before observed in this species of animal.

The story illustrates, I believe, two aspects of the genesis of a
transcontextual syndrome:

First, that severe pain and maladjustment can be induced by
putting a mammal in the wrong regarding its rules for making sense
of an important relationship with another mammal.

And second, that if this pathology can be warded off or resisted,
the total experience may promote creativity.

53 K. Pryor, R. Haag, and J. O’Rielly, “Deutero-Learning in a
Roughtooth Porpoise (Steno bredanensis),” U. S. Naval Ordinance
Test Station, China Lake, NOTS TP 4270


Bateson, “Social Planning and the Concept of Deutero-
Learning,” Science, Philosophy and Religion; Second
Symposium, L. Bryson and L. Finkelstein, eds., New York,
Conference on Science, Philosophy and Religion in their
Relation to the Democratic Way of Life, Inc., 1942.
“Minimal Requirements for a Theory of Schizophrenia,”
A.M.A. Archives of General Psychiatry, 1960, 2: 477-91.
, Perceval’s Narrative, A Patient’s Account of his Psychosis,
1830-1832, edited and with an introduction by Gregory
Bateson, Stanford, Calif., Stanford University Press, 1961.
, “Exchange of Information about Patterns of Hu-man
Behavior,” Information Storage and Neural Control; Tenth
Annual Scientific Meeting of the Houston Neurological
Society, W. S. Fields and W. Abbott, eds., Springfield, Ill.,
Charles C. Thomas, 1963.
, “The Role of Somatic Change in Evolution,” Evolution,
1963, 17: 529-39.

The Logical Categories of Learning andCommunication*

All species of behavioral scientists are concerned with “learning”
in one sense or another of that word. Moreover, since “learning” is a
communicational phenomenon, all are affected by that cybernetic
revolution in thought which has occurred in the last twenty-five
years. This revolution was triggered by the engineers and
communication theorists but has older roots in the physiological
work of Cannon and Claude Bernard, in the physics of Clarke
Maxwell, and in the mathematical philosophy of Russell and
Whitehead. Insofar as behavioral scientists still ignore the problems
of F incipia Mathematica,54 they can claim approximately sixty
years of obsolescence.

It appears, however, that the barriers of misunderstanding which
divide the various species of behavioral scientists can be illuminated
(but not eliminated) by an application of Russell’s Theory of Logical
Types to the concept of “learning” with which all are concerned. To
attempt this illumination will be a purpose of the present essay.

The Theory of Logical Types

First, it is appropriate to indicate the subject matter of the Theory
of Logical Types: the theory asserts that no class can, in formal
logical or mathematical discourse, be a member of itself; that a class
of classes cannot be one of the classes which are its members; that a
name is not the thing named; that “John Bateson” is the class of

* This essay was written in 1964 while the author was employed by the
Communications Research Institute, under a Career Development Award (K3-NH21,
931) from the National Institute of Mental Health. It was submitted as a position
paper to the “Conference on World Views” sponsored by the Wenner-Gren Foundation,
August 2-11, 1968. The section on “Learning III” was added in 1971.
54 A. N. Whitehead and B. Russell, Principia Mathematica, 3 vols., 2nd ed.,
Cambridge, Cambridge University Press, 1910-13.


which that boy is the unique member; and so forth. These assertions
may seem trivial and even obvious, but we shall see later that it is
not at all unusual for the theorists of behavioral science to commit
errors which are precisely analogous to the error of classifying the
name with the thing named—or eating the menu card instead of the
dinner—an error of logical typing.

Somewhat less obvious is the further assertion of the theory: that
a class cannot be one of those items which are correctly classified as
its nonmembers. If we classify chairs together to constitute the class
of chairs, we can go on to note that tables and lamp shades are
members of a large class of “nonchairs,” but we shall commit an
error in formal discourse if we count the class of chairs among the
items within the class of nonchairs.

Inasmuch as no class can be a member of itself, the class of
nonchairs clearly cannot be a nonchair. Simple considerations of
symmetry may suffice to convince the nonmathematical reader: (a)
that the class of chairs is of the same order of abstraction (i.e., the
same logical type) as the class of nonchairs; and further, (b) that if
the class of chairs is not a chair, then, correspondingly, the class of
nonchairs is not a nonchair.

Lastly, the theory asserts that if these simple rules of formal
discourse are contravened, paradox will be generated and the
discourse vitiated.

The theory, then, deals with highly abstract matters and was first
derived within the abstract world of logic. In that world, when a
train of propositions can be shown to generate a paradox, the entire
structure of axioms, theorems, etc., involved in generating that
paradox is thereby negated and reduced to nothing. It is as if it had
never been. But in the real world (or at least in our descriptions of
it), there is always time, and nothing which has been can ever be
totally negated in this way. The computer which encounters a paradox
(due to faulty programming) does not vanish away.

The “if … then …” of logic contains no time. But in the computer,
cause and effect are used to simulate the “if .. . then . . .” of logic;
and all sequences of cause and effect necessarily involve time.
(Conversely, we may say that in scientific explanations the “if …
then …” of logic is used to simulate the “if … then …” of cause and

The computer never truly encounters logical paradox, but only
the simulation of paradox in trains of cause and effect. The
computer therefore does not fade away. It merely oscillates.

In fact, there are important differences between the world of
logic and the world of phenomena, and these differences must be
allowed for whenever we base our arguments upon the partial but
important analogy which exists between them.

It is the thesis of the present essay that this partial analogy can
provide an important guide for behavioral scientists in their
classification of phenomena related to learning. Precisely in the field
of animal and mechanical communication something like the theory
of types must apply.

Questions of this sort, however, are not often discussed in
zoological laboratories, anthropological field camps, or psychiatric
conventions, and it is necessary therefore to demonstrate that these
abstract considerations are important to behavioral scientists.

Consider the following syllogism:

Changes in frequency of items of mammalian behavior
can be described and predicted in terms of various
“laws” of reinforcement.
“Exploration” as observed in rats is a category, or
class, of mammalian behavior.
Therefore, changes in frequency of “exploration”
should be describable in terms of the same “laws” of
Be it said at once: first, that empirical data show that the
conclusion (c) is untrue; and second, that if the conclusion (c) weredemonstrably true, then either (a) or (b) would be untrue.55

55 is conceivable that the same words might be used in describing
both a class and its members and be true in both cases. The word
“wave” is the name of a class of movements of particles. We can
also say that the wave itself “moves,” but we shall be referring to a
movement of a class of movements. Under friction, this metamovement
will not lose velocity as would the movement of a particle.


Logic and natural history would be better served by an expanded
and corrected version of the conclusion (c) some-what as follows:

(c) If,
as asserted in (b), “exploration” is not an item of
mammalian behavior but is a category of such items,
then no descriptive statement which is true of items of
behavior can be true of “exploration.” If, however,
descriptive statements which are true of items of
behavior are also true of “exploration,” then
“exploration” is an item and not a category of items.
The whole matter turns on whether the distinction between a
class and its members is an ordering principle in the behavioral
phenomena which we study.

In less formal language: you can reinforce a rat (positively

or negatively) when he investigates a particular strange object,
and he will appropriately learn to approach or avoid it. But the very
purpose of exploration is to get information about which objects
should be approached and which avoided. The discovery that a
given object is dangerous is therefore a success in the business of
getting information. The success will not discourage the rat from
future exploration of other strange objects.

A priori it can be argued that all perception and all response, all
behavior and all classes of behavior, all learning and all genetics, all
neurophysiology and endocrinology, all organization and all
evolution—one entire subject matter must be regarded as
communicational in nature, and there-fore subject to the great
generalizations or “laws” which apply to communicative
phenomena. We therefore are warned to expect to find in our data
those principles of order which fundamental communication theory
would pro-pose. The Theory of Logical Types, Information Theory,
and so forth, are expectably to be our guides.

The “Learning” of Computers, Rats, and Men

The word “learning” undoubtedly denotes change of some kind.
To say what kind of change is a delicate matter.

However, from the gross common denominator, “change,” we can
deduce that our descriptions of “learning” will have to make the
same sort of allowance for the varieties of logical type which has
been routine in physical science since the days of Newton. The
simplest and most familiar form of change is motion, and even if
we work at that very simple physical level we must structure our
descriptions in terms of “position or zero motion,” “constant
velocity,” “acceleration,” “rate of change of acceleration,” and so on



Change denotes process. But processes are themselves subject to
“change.” The process may accelerate, it may slow down, or it may
undergo other types of change such that we shall say that it is now a
“different” process.

These considerations suggest that we should begin the ordering
of our ideas about “learning” at the very simplest level.

Let us consider the case of specificity of response, or zero
learning. This is the case in which an entity shows minimal change
in its response to a repeated item of sensory input. Phenomena
which approach this degree of simplicity occur in various contexts:

In experimental settings, when “learning” is complete
and the animal gives approximately 100 per cent correct
responses to the repeated stimulus.
In cases of habituation, where the animal has ceased to
give overt response to what was formerly a disturbing
In cases where the pattern of the response is minimally
determined by experience and maximally determined by
genetic factors.
In cases where the response is now highly stereo-typed.
In simple electronic circuits, where the circuit
structure is not itself subject to change resulting
from the passage of impulses within the circuit—i.e.,
56 The Newtonian equations which describe the motions of a “particle” stop at the
level of “acceleration.” Change of acceleration can only happen with deformation
of the moving body, but the Newtonian “particle” was not made up of “parts”
and was therefore (logically) incapable of deformation or any other internal change. It
was therefore not subject to rate of change of acceleration.

where the causal links between “stimulus” and
“response” are as the engineers say “soldered in.”

In ordinary, nontechnical parlance, the word “learn” is often
applied to what is here called “zero learning,” i.e., to the simple
receipt of information from an external event, in such a way that a
similar event at a later (and appropriate) time will convey the same
information: I “learn” from the factory whistle that it is twelve

It is also interesting to note that within the frame of our
definition many very simple mechanical devices show at least the
phenomenon of zero learning. The question is not, “Can machines
learn?” but what level or order of learning does a given machine
achieve? It is worth looking at an extreme, if hypothetical, case:

The “player” of a Von Neumannian game is a mathematical
fiction, comparable to the Euclidean straight line in geometry or the
Newtonian particle in physics. By definition, the “player” is capable
of all computations necessary to solve whatever problems the events
of the game may present; he is incapable of not performing these
computations whenever they are appropriate; he always obeys the
findings of his computations. Such a “player” receives information
from the events of the game and acts appropriately upon that information.
But his learning is limited to what is here called zero

An examination of this formal fiction will contribute to our
definition of zero learning.

The “player” may receive, from the events of the game,
information of higher or lower logical type, and he may use this
information to make decisions of higher or lower type. That is, his
decisions may be either strategic or tactical, and he can identify and
respond to indications of both the tactics and the strategy of his
opponent. It is, how-ever, true that in Von Neumann’s formal
definition of a “game,” all problems which the game may present are
conceived as computable, i.e., while the game may contain problems
and information of many different logical types, the hierarchy of
these types is strictly finite.

It appears then that a definition of zero learning will not depend
upon the logical typing of the information received by the organism

nor upon the logical typing of the adaptive decisions which the
organism may make. A very high (but finite) order of complexity
may characterize adaptive behavior based on nothing higher than
zero learning.

(1) The “player” may compute the value of information which
would benefit him and may compute that it will pay him to acquire
this information by engaging in “exploratory” moves. Alternatively,
he may make delaying or tentative moves while he waits for needed
It follows that a rat engaging in exploratory behavior might do so
upon a basis of zero learning.

(2) The “player” may compute that it will pay him to make
random moves. In the game of matching pennies, he will compute
that if he selects “heads” or “tails” at random, he will have an even
chance of winning. If he uses any plan or pattern, this will appear as
a pattern or redundancy in the sequence of his moves and his
opponent will thereby receive information. The “player” will
therefore elect to play in a random manner.
(3) The “player” is incapable of “error.” He may, for good reason,
elect to make random moves or exploratory moves, but he is by
definition incapable of “learning by trial and error.”
If we assume that, in the name of this learning process, the word
“error” means what we meant it to mean when we said that the
“player” is incapable of error, then “trial and error” is excluded from
the repertoire of the Von Neumannian player. In fact, the Von
Neumannian “player” forces us to a very careful examination of
what we mean by “trial and error” learning, and indeed what is
meant by “learning” of any kind. The assumption regarding the
meaning of the word “error” is not trivial and must now be

There is a sense in which the “player” can be wrong. For
example, he may base a decision upon probabilistic considerations
and then make that move which, in the light of the limited available
information, was most probably right. When more information
becomes available, he may discover that that move was wrong. But
this discovery can contribute nothing to his future skill. By
definition, the player used correctly all the available information.
He estimated the probabilities correctly and made the move which


was most probably correct. The discovery that he was wrong in the
particular instance can have no bearing upon future in-stances.
When the same problem returns at a later time, he will correctly go
through the same computations and reach the same decision.
Moreover, the set of alternatives among which he makes his choice
will be the same set—and correctly so.

In contrast, an organism is capable of being wrong in a number
of ways of which the “player” is incapable. These wrong choices are
appropriately called “error” when they are of such a kind that they
would provide information to the organism which might contribute
to his future skill. These will all be cases in which some of the
available information was either ignored or incorrectly used. Various
species of such profitable error can be classified.

Suppose that the external event system contains details which
might tell the organism: (a) from what set of alternatives he should
choose his next move; and (b) which member of that set he should
choose. Such a situation permits two orders of error:

The organism may use correctly the information which tells
him from what set of alternatives he should choose, but choose the
wrong alternative within this set; or

He may choose from the wrong set of alternatives. (There is
also an interesting class of cases in which the sets of alternatives
contain common members. It is then possible for the organism to
be “right” but for the wrong reasons. This form of error is
inevitably self-reinforcing.)

If now we accept the overall notion that all learning (other than
zero learning) is in some degree stochastic (i.e., contains
components of “trial and error”), it follows that an ordering of the
processes of learning can be built upon an hierarchic classification
of the types of error which are to be corrected in the various learning
processes. Zero learning will then be the label for the immediate
base of all those acts (simple and complex) which are not subject to
correction by trial and error. Learning I will be an appropriate label
for the revision of choice within an unchanged set of alternatives;
Learning II will be the label for the revision of the set from which
the choice is to be made; and so on.

Learning I

Following the formal analogy provided by the “laws” of motion
(i.e., the “rules” for describing motion), we now look for the class of
phenomena which are appropriately described as changes in zero
learning (as “motion” describes change of position). These are the
cases in which an entity gives at Time 2 a different response from
what it gave at Time 1, and again we encounter a variety of cases
variously related to experience, physiology, genetics, and
mechanical process:

(a)There is the phenomenon of habituation—the change from
responding to each occurrence of a repeated event to not overtly
responding. There is also the extinction or loss of habituation,
which may occur as a result of a more or less long gap or other
interruption in the sequence of repetitions of the stimulus event.
(Habituation is of especial interest. Specificity of response, which
we are calling zero learning, is characteristic of all protoplasm,
but it is interesting to note that “habituation” is perhaps the only
form of Learning I which living things can achieve without a
neural circuit.)

(b)The most familiar and perhaps most studied case is that of
the classical Pavlovian conditioning. At Time 2 the dog salivates
in response to the buzzer; he did not do this at Time 1.

(c)There is the “learning” which occurs in contexts of
instrumental reward and instrumental avoidance.

(d)There is the phenomenon of rote learning, in which an

item in the behavior of the organism becomes a stimulus for

another item of behavior.

(e)There is the disruption, extinction, or inhibition of

“completed” learning which may follow change or absence of


In a word, the list of Learning I contains those items which are
most commonly called “learning” in the psycho-logical laboratory.


Note that in all cases of Learning I, there is in our description an
assumption about the “context.” This assumption must be made
explicit. The definition of Learning I assumes that the buzzer (the
stimulus) is somehow the “same” at Time 1 and at Time 2. And this
assumption of “sameness” must also delimit the “context,” which
must (theoretically) be the same at both times. It follows that the
events which occurred at Time 1 are not, in our description; included
in our definition of the context at Time 2, because to include them
would at once create a gross difference between “con-text at Time 1”
and “context at Time 2.” (To paraphrase Heraclitus: “No man can go
to bed with the same girl for the first time twice.”)

The conventional assumption that context can be repeated, at
least in some cases, is one which the writer adopts in this essay as a
cornerstone of the thesis that the study of behavior must be ordered
according to the Theory of Logical Types. Without the assumption of
repeatable context (and the hypothesis that for the organisms which
we study the sequence of experience is really somehow punctuated
in this manner), it would follow that all “learning” would be of one
type: namely, all would be zero learning. Of the Pavlovian
experiment, we would simply say that the dog’s neural circuits
contain “soldered in” from the beginning such characteristics that in
Context A at Time 1 he will not salivate, and that in the totally
different Context B at Time 2 he will salivate. What previously we
called “learning” we would now describe as “discrimination”
between the events of Time 1 and the events of Time 1 plus Time 2.
It would then follow logically that all questions of the type, “Is this
behavior `learned’ or `innate’?” should be answered in favor of

We would argue that without the assumption of repeat-able
context, our thesis falls to the ground, together with the whole
general concept of “learning.” If, on the other hand, the assumption
of repeatable context is accepted as somehow true of the organisms
which we study, then the case for logical typing of the phenomena
of learning necessarily stands, because the notion “context” is itself
subject to logical typing.

Either we must discard the notion of “context,” or we retain this
notion and, with it, accept the hierarchic series—stimulus, context of

stimulus, context of context of stimulus, etc. This series can be
spelled out in the form of a hierarchy of logical types as follows:

Stimulus is an elementary signal, internal or external. Context of
stimulus is a metamessage which classifies the elementary signal.

Context of context of stimulus is a meta-metamessage which
classifies the metamessage.

And so oil.

The same hierarchy could have been built up from the notion of
“response” or the notion of “reinforcement.”

Alternatively, following up the hierarchic classification of errors
to be corrected by stochastic process or “trial and error,” we may
regard “context” as a collective term for all those events which tell
the organism among what set of alternatives he must make his next

At this point it is convenient to introduce the term “con-text
marker.” An organism responds to the “same” stimulus differently in
differing contexts, and we must therefore ask about the source of the
organisms’s information. From what percept does he know that
Context A is different from Con-text B?

In many instances, there may be no specific signal or label
which will classify and differentiate the two contexts, and the
organism will be forced to get his information from the actual
congeries of events that make up the context in each case. But,
certainly in human life and probably in that of many other
organisms, there occur signals whose major function is to classify
contexts. It is not unreasonable to sup-pose that when the harness is
placed upon the dog, who has had prolonged training in the
psychological laboratory, he knows from this that he is now
embarking upon a series of contexts of a certain sort. Such a source
of information we shall call a “context marker,” and note
immediately that, at least at the human level, there are also
“markers of contexts of contexts.” For example: an audience is
watching Hamlet on the stage, and hears the hero discuss suicide
in the con-text of his relationship with his dead father, Ophelia,
and the rest. The audience members do not immediately telephone
for the police because they have received information about the
context of Hamlet’s context. They know that it is a “play” and have
received this information from many “markers of context of


context”—the playbills, the seating arrangements, the curtain, etc.,
etc. The “King,” on the other hand, when he lets his conscience be
pricked by the play within the play, is ignoring many “markers of
context of context.”

At the human level, a very diverse set of events falls within the
category of “context markers.” A few examples are here listed:

The Pope’s throne from which he makes announcements
ex cathedra, which announcements are there.. by
endowed with a special order of validity.
The placebo, by which the doctor sets the stage for achange in the patient’s subjective experience.
The shining object used by some hypnotists in “inducingtrance.”
The air raid siren and the “all clear.”
The handshake of boxers before the fight.
The observances of etiquette.
These, however, are examples from the social life of a highly
complex organism, and it is more profitable at this stage to ask about
the analogous phenomena at the pre-verbal level.

A dog may see the leash in his master’s hand and act as if he
knows that this indicates a walk; or he may get in-formation from
the sound of the word “walk” that this type of context or sequence is

When a rat starts a sequence of exploratory activities, does he do
so in response to a “stimulus?” Or in response to a context? Or in
response to a context marker?

These questions bring to the surface formal problems about the
Theory of Logical Types which must be discussed. The theory in its
original form deals only with rigorously digital communication, and
it is doubtful how far it may be applied to analogue or iconic
systems. What we are here calling “context markers” may be digital
(e.g., the word “walk” mentioned above) ; or they may be analogue
signals —a briskness in the master’s movements may indicate that a
walk is pending; or some part of the coming context may serve as a
marker (the leash as a part of the walk) ; or in the extreme case, the

walk itself in all its complexity may stand for itself, with no label or
marker between the dog and the experience. The perceived event
itself may communicate its own occurrence. In this case, of course,
there can be no error of the “menu card” type. Moreover, no paradox
can be generated because in purely analogue or iconic communication
there is no signal for “not.”

There is, in fact, almost no formal theory dealing with analogue
communication and, in particular, no equivalent of Information
Theory or Logical Type Theory. This gap in formal knowledge is
inconvenient when we leave the rarified world of logic and
mathematics and come face to face with the phenomena of natural
history. In the natural world, communication is rarely either purely
digital or purely analogic. Often discrete digital pips are combined
together to make analogic pictures as in the printer’s halftone block;
and sometimes, as in the matter of context markers, there is a
continuous gradation from the ostensive through the iconic to the
purely digital. At the digital end of this scale all the theorems of
information theory have their full force, but at the ostensive and
analogic end they are meaningless.

It seems also that while much of the behavioral communication
of even higher mammals remains ostensive or analogic, the internal
mechanism of these creatures has become digitalized at least at the
neuronal level. It would seem that analogic communication is in
some sense more primitive than digital and that there is a broad
evolutionary trend toward the substitution of digital for analogic
mechanisms. This trend seems to operate faster in the evolution of
internal mechanisms than in the evolution of external behavior.

Recapitulating and extending what was said above:

(a)The notion of repeatable context is a necessary premise forany theory which defines “learning” as change.

(b)This notion is not a mere tool of our description but contains
the implicit hypothesis that for the organisms which we study, the
sequence of life experience, action, etc., is somehow segmented or
punctuated into subsequences or “contexts” which may be equated
or differentiated by the organism.


(c)The distinction which is commonly drawn between
perception and action, afferent and efferent, input and out-put, is
for higher organisms in complex situations not valid. On the one
hand, almost every item of action may be re-ported either by
external sense or endoceptive mechanism to the C.N.S., and in this
case the report of this item be-comes an input. And, on the other
hand, in higher organisms, perception is not by any means a
process of mere passive receptivity but is at least partly determined
by efferent control from higher centers. Perception, notoriously,
can be changed by experience. In principle, we must allow both for
the possibility that every item of action or output may create an
item of input; and that percepts may in some cases par-take of the
nature of output. It is no accident that almost all sense organs are
used for the emission of signals between organisms. Ants
communicate by their antennae; dogs by the pricking of their ears;
and so on.

(d)In principle, even in zero learning, any item of experience or
behavior may be regarded as either “stimulus” or “response” or as
both, according to how the total sequence is punctuated. When the
scientist says that the buzzer is the “stimulus” in a given sequence,
his utterance implies an hypothesis about how the organism
punctuates that sequence. In Learning I, every item of perception
or behavior may be stimulus or response or reinforcement
according to how the total sequence of interaction is punctuated.

Learning II

What has been said above has cleared the ground for the
consideration of the next level or logical type of “learning” which
we shall here call Learning II. Various terms have been proposed in
the literature for various phenomena of this order. “Deutero

learning,”4 57″set learning,”58 “learning to learn,” and “transfer of

learning” may be mentioned.

We recapitulate and extend the definitions so far given:

Zero learning is characterized by specificity of response, which
—right or wrong—is not subject to correction.

Learning I is change in specificity of response by correction of
errors of choice within a set of alternatives.

Learning II is change in the process of Learning I, e.g., a
corrective change in the set of alternatives from which choice is
made, or it is a change in how the sequence of experience is

Learning III is change in the process of Learning II, e.g., a
corrective change in the system of sets of alternatives from which
choice is made. (We shall see later that to demand this level of
performance of some men and some mammals is sometimes

Learning IV would be change in Learning III, but probably does
not occur in any adult living organism on this earth. Evolutionary
process has, however, created organisms whose ontogeny brings
them to Level III. The combination of phylogenesis with
ontogenesis, in fact, achieves Level IV.

Our immediate task is to give substance to the definition of
Learning II as “change in Learning I,” and it is for this that the
ground has been prepared. Briefly, I believe that the phenomena of
Learning II can all be included under the rubric of changes in the
manner in which the stream of action and experience is segmented
or punctuated into contexts together with changes in the use of
context markers.

The list of phenomena classified under Learning I includes a
considerable (but not exhaustive) set of differently structured
contexts. In classical Pavlovian contexts, the contingency pattern
which describes the relation between “stimulus” (CS), animal’s
action (CR), and reinforcement. (UCS ) is profoundly different from

57 ‘G. Bateson, “Social Planning and the Concept of Deutero-Learning,”

Conference on Science, Philosophy and Religion, Second Symposium,

New York, Harper, 1942.
58 H. E. Harlow, “The Formation of Learning Sets,” Psycho!. Review, 1949,

56: 51-65.

the contingency pattern characteristic of instrumental contexts of

In the Pavlovian case: If stimulus and a certain lapse of time:
then reinforcement.

In the Instrumental Reward case: If stimulus and a particular item
of behavior: then reinforcement.

In the Pavlovian case, the reinforcement is not contingent upon
the animal’s behavior, whereas in the instrumental case, it is. Using
this contrast as an example, we say that Learning II has occurred if it
can be shown that experience of one or more contexts of the
Pavlovian type results in the animal’s acting in some later context as
though this, too, had the Pavlovian contingency pattern. Similarly, if
past experience of instrumental sequences leads an animal to act in
some later context as though expecting this also to be an
instrumental context, we shall again say that Learning II has

When so defined, Learning II is adaptive only if the animal
happens to be right in its expectation of a given contingency pattern,
and in such a case we shall expect to see a measurable learning to
learn. It should require fewer trials in the new context to establish
“correct” behavior. If, on the other hand, the animal is wrong in his
identification of the later contingency pattern, then we shall expect a
delay of Learning I in the new context. The animal who has had
prolonged experience of Pavlovian contexts might never get around
to the particular sort of trial-and-error behavior necessary to
discover a correct instrumental response.

There are at least four fields of experimentation where Learning
II has been carefully recorded:

(a) In human rote learning. Hull59 carried out very careful
quantitative studies which revealed this phenomenon, and
constructed a mathematical model which would simulate or explain
the curves of Learning I which he recorded. He also observed a
second-order phenomenon which we may call “learning to rote
learn” and published the curves for this phenomenon in the
Appendix to his book. These curves were separated from the main
59 E. L. Hull, et al., Mathematico-deductive Theory of Rote Learning,
New Haven, Yale University, Institute of Human Relations, 1940


body of the book because, as he states, his mathematical model (of
Rote Learning I) did not cover this aspect of the data.

It is a corollary of the theoretical position which we here take
that no amount of rigorous discourse of a given logical type can
“explain” phenomena of a higher type. Hull’s model acts as a
touchstone of logical typing, automatically excluding from
explanation phenomena beyond its logical scope. That this was so—
and that Hull perceived it—is testimonial both to his rigor and to his

What the data show is that for any given subject, there is an
improvement in rote learning with successive sessions,
asymptotically approaching a degree of skill which varied from
subject to subject.

The context for this rote learning was quite complex and no
doubt appeared subjectively different to each learner. Some may
have been more motivated by fear of being wrong, while others
looked rather for the satisfactions of being right. Some would be
more influenced to put up a good record as compared with the other
subjects; others would be fascinated to compete in each session with
their own previous showing, and so on. All must have had ideas
(correct or incorrect) about the nature of the experimental setting, all
must have had “levels of aspiration,” and all must have had previous
experience of memorizing various sorts of material. Not one of
Hull’s subjects could have come into the learning context
uninfluenced by previous Learning II.

In spite of all this previous Learning II, and in spite of genetic
differences which might operate at this level, all showed
improvement over several sessions. This improvement cannot have
been due to Learning I because any recall of the specific sequence of
syllables learned in the previous session would not be of use in
dealing with the new sequence. Such recall would more probably be
a hindrance. I submit, therefore, that the improvement from session
to session can only be accounted for by some sort of adaptation to
the context which Hull provided for rote learning.

It is also worth noting that educators have strong opinions about
the value (positive or negative) of training in rote learning.
“Progressive” educators insist on training in “insight,” while the
more conservative insist on rote and drilled recall.


(b) The second type of Learning II which has been experimentally
studied is called “set learning.” The concept and term
are derived from Harlow and apply to a rather special case of
Learning II. Broadly, what Harlow did was to present rhesus
monkeys with more or less complex gestalten or “problems.”
These the monkey had to solve to get a food reward. Harlow
showed that if these problems were of similar “set,” i.e., contained
similar types of logical complexity, there was a carry-over of
learning from one problem to the next. There were, in fact, two
orders of contingency patterns involved in Harlow’s experiments:
first the overall pattern of instrumentalism (if the monkey solves
the problem, then reinforcement); and second, the contingency
patterns of logic within the specific problems.
(c) Bitterman and others have recently set a fashion in
experimentation with “reversal learning.” Typically in these
experiments the subject is first taught a binary discrimination. When
this has been learned to criterion, the meaning of the stimuli is
reversed. If X initially “meant” R1, and Y initially meant R2, then
after reversal X comes to mean R2, and, Y comes to mean R1. Again
the trials are run to criterion when again the meanings are reversed.
In these experiments, the crucial question is: Does the subject learn
about the reversal? I.e., after a series of reversals, does the subject
reach criterion in fewer trials than he did at the beginning of the
In these experiments, it is conspicuously clear that the question
asked is of logical type higher than that of questions about simple
learning. If simple learning is based upon a set of trials, then
reversal learning is based upon a set of such sets. The parallelism
between this relation and Russell’s relation between “class” and
“class of classes” is direct.

(d) Learning II is also exemplified in the well-known phenomena
of “experimental neurosis.” Typically an animal is trained, either in
a Pavlovian or instrumental learning con-text, to discriminate
between some X and some Y; e.g., between an ellipse and a circle.
When this discrimination has been learned, the task is made more
difficult: the ellipse is made progressively fatter and the circle is
flattened. Finally a stage is reached at which discrimination is

impossible. At this stage the animal starts to show symptoms of
severe disturbance.

Notably, (a) a naive animal, presented with a situation in which
some X may (on some random basis) mean either A or B, does not
show disturbance; and (b) the disturbance does not occur in
absence of the many context markers characteristic of the laboratory

It appears, then, that Learning II is a necessary preparation for
the behavioral disturbance. The information, “This is a context for
discrimination,” is communicated at the beginning of the sequence
and underlined in the series of stages in which discrimination is
made progressively more difficult. But when discrimination
becomes impossible, the structure of the context is totally changed.
The context markers (e.g., the smell of the laboratory and the
experimental harness) now become misleading because the animal is
in a situation which demands guesswork or gambling, not
discrimination. The en-tire experimental sequence is, in fact, a
procedure for putting the animal in the wrong at the level of
Learning 11.

In my phrase, the animal is placed in a typical “double bind,”
which is expectably schizophrenogenic.61

In the strange world outside the psychological laboratory,
phenomena which belong to the category Learning II are a major
preoccupation of anthropologists, educators, psychiatrists, animal
trainers, human parents, and children. All who think about the
processes which determine the character of the individual or the
processes of change in human (or animal) relationship must use in
their thinking a variety of assumptions about Learning II. From time
to time, these people call in the laboratory psychologist as a
consultant, and then are confronted with a linguistic barrier. Such
barriers must always result when, for example, the psychiatrist is
talking about Learning II, the psychologist is talking about Learning
I, and neither recognizes the logical structure of the difference.

60 H . S. Liddell, “Reflex Method and Experimental Neurosis,” Personality
and Behavior Disorders, New York, Ronald Press, 1944

61 G. Bateson, et al., “Toward a Theory of Schizophrenia,” Behavioral
Science, 1956, 1: 251-64.


Of the multitudinous ways in which Learning II emerges in
human affairs, only three will be discussed in this essay:

(a) In describing individual human beings, both the scientist and
the layman commonly resort to adjectives descriptive of “character.”
It is said that Mr. Jones is dependent, hostile, fey, finicky, anxious,
exhibitionistic, narcissistic, passive, competitive, energetic, bold,
cowardly, fatalistic, humorous, playful, canny, optimistic,
perfectionist, careless, careful, casual, etc. In the light of what has
already been said, the reader will be able to assign all these
adjectives to their appropriate logical type. All are descriptive of
(possible) results of Learning II, and if we would define these
words more carefully, our definition will consist in laying down
the contingency pattern of that context of Learning I which would
expectably bring about that Learning II which would make the
adjective applicable.
We might say of the “fatalistic” man that the pattern of his
transactions with the environment is such as he might have
acquired by prolonged or repeated experience as subject of
Pavlovian experiment; and note that this definition of “fatalism” is
specific and precise. There are many other forms of “fatalism”
besides that which is defined in terms of this particular context of
learning. There is, for example, the more complex type
characteristic of classical Greek tragedy where a man’s own action
is felt to aid the inevitable working of fate.

(b) In the punctuation of human interaction. The critical reader
will have observed that the adjectives above which purport to
describe individual character are really not strictly applicable to
the individual but rather describe transactions between the
individual and his material and human environment. No man is
“resourceful” or “dependent” or “fatalistic” in a vacuum. His
characteristic, whatever it be, is not his but is rather a
characteristic of what goes on between him and something (or
somebody) else.
This being so, it is natural to look into what goes on between
people, there to find contexts of Learning I which are likely to lend
their shape to processes of Learning II. In such systems, involving
two or more persons, where most of the important events are
postures, actions, or utterances of the living creatures, we note

immediately that the stream of events is commonly punctuated into
contexts of learning by a tacit agreement between the persons
regarding the nature of their relationship—or by context markers
and tacit agreement that these context markers shall “mean” the
same for both parties. It is instructive to attempt analysis of an
ongoing interchange between A and B. We ask about any particular
item of A’s behavior: Is this item a stimulus for B? Or is it a
response of A to something B said earlier? Or is it a reinforcement
of some item provided by B? Or is A, in this item, consummating a
reinforcement for himself? Etc.

Such questions will reveal at once that for many items of A’s
behavior the answer is often quite unclear. Or if there be a clear
answer, the clarity is due only to a tacit (rarely fully explicit)
agreement between A and B as to the nature of their mutual roles,
i.e., as to the nature of the contextual structure which they will
expect of each other.

If we look at such an exchange in the abstract:

a1b1a2b2a3b3a4b4a5b5 where the a’s refer to items
of A’s behavior, and the b’s to items of B’s behavior, we can take any
ai and construct around it three simple contexts of learning. These
will be:

(ai bi a,+ 1) , in which ai is the stimulus for bi.

(b1_1 ai bi) , in which ai is the response to b.-1, which response B
reinforces with bi.

(a1_1 bi _1 ai) , in which ai is now A’s reinforcement of B’s bi-1,
which was response to ai_1.

It follows that ai may be a stimulus for B or it may be A’s
response to B, or it may be A’s reinforcement of B.

Beyond this, however, if we consider the ambiguity of the
notions “stimulus” and “response,” “afferent” and “efferent”—as
discussed above—we note that any ai may also be a stimulus for A;
it may be A’s reinforcement of self; or it may be A’s response to
some previous behavior of his own, as is the case in sequences of
rote behavior.

This general ambiguity means in fact that the ongoing sequence
of interchange between two persons is structured only by the
person’s own perception of the sequence as a series of contexts, each
context leading into the next. The particular manner in which the


sequence is structured by any particular person will be determined
by that person’s previous Learning II (or possibly by his genetics).

In such a system, words like “dominant” and “submissive,”
“succoring” and “dependent” will take on definable meaning as
descriptions of segments of interchange. We shall say that “A
dominates B” if A and B show by their behavior that they see their
relationship as characterized by sequences of the type a1b1a2, where
a1 is seen (by A and B) as a signal defining conditions of
instrumental reward or punishment; b1 as a signal or act obeying
these conditions; and a2 as a signal reinforcing b1.

Similarly we shall say that “A is dependent on B” if their
relationship is characterized by sequences a1b1a2,, where al is seen as
a signal of weakness; b1 as a helping act; and a2 as an
acknowledgement of b1.

But it is up to A and B to distinguish (consciously or unconsciously
or not at all) between “dominance” and “dependence.” A
“command” can closely resemble a cry for “help.”

(c) In psychotherapy, Learning II is exemplified most conspicuously
by the phenomena of “transference.” Orthodox Freudian
theory asserts that the patient will inevitably bring to the therapy
room inappropriate notions about his relation-ship to the therapist.
These notions (conscious or unconscious) will be such that he will
act and talk in a way which would press the therapist to respond in
ways which would resemble the patient’s picture of how some
important other person (usually a parent) treated the patient in the
near or distant past. In the language of the present paper, the patient
will try to shape his interchange with the therapist according to the
premises of his (the patient’s) former Learning II.
It is commonly observed that much of the Learning II which
determines a patient’s transference patterns and, in-deed, determines
much of the relational life of all human beings, (a ) dates from early
infancy, and (b) is unconscious. Both of these generalizations seem
to be correct and both need some explanation.

It seems probable that these two generalizations are true because
of the very nature of the phenomena which we are discussing. We
suggest that what is learned in Learning II is a way of punctuating
events. But a way of punctuating is not true or false. There is
nothing contained in the propositions of this learning that can be

tested against reality. It is like a picture seen in an inkblot; it has
neither correctness nor incorrectness. It is only a way of seeing the

Consider the instrumental view of life. An organism with this
view of life in a new situation will engage in trial-and-error behavior
in order to make the situation provide a positive reinforcement. If he
fails to get this reinforcement, his purposive philosophy is not
thereby negated. His trial-and-error behavior will simply continue.
The premises of “purpose” are simply not of the same logical type
as the material facts of life, and therefore cannot easily be
contradicted by them.

The practitioner of magic does not unlearn his magical view of
events when the magic does not work. In fact, the propositions
which govern punctuation have the general characteristic of being
self-validating.62 What we term “con-text” includes the subject’s
behavior as well as the external events. But this behavior is
controlled by former Learning II and therefore it will be of such a
kind as to mold the total context to fit the expected punctuation. In
sum, this self-validating characteristic of the content of Learning II
has the effect that such learning is almost ineradicable. It follows
that Learning II acquired in infancy is likely to persist through life.
Conversely, we must expect many of the important characteristics
of an adult’s punctuation to have their roots in early infancy.

In regard to the unconsciousness of these habits of punctuation,
we observe that the “unconscious” includes not only repressed
material but also most of the processes and habits of gestalt
perception. Subjectively we are aware of our “dependency” but
unable to say clearly how this pattern was constructed nor what
cues were used in our creation of it.

62 J. Ruesch and G. Bateson, Communication: The Social Matrix of
Psychiatry, New York, Norton, 1951.

Learning III

What has been said above about the self-validating character of
premises acquired by Learning II indicates that Learning III is likely
to be difficult and rare even in human beings. Expectably, it will
also be difficult for scientists, who are only human, to imagine or
describe this process. But it is claimed that something of the sort
does from time to time occur in psychotherapy, religious conversion,
and in other sequences in which there is profound reorganization of

Zen Buddhists, Occidental mystics, and some psychiatrists
assert that these matters are totally beyond the reach of language.
But, in spite of this warning, let me begin to speculate about what
must (logically) be the case.

First a distinction must be drawn: it was noted above that the
experiments in reversal learning demonstrate Learning II whenever
there is measurable learning about the fact of reversal. It is
possible to learn (Learning I) a given premise at a given time and
to learn the converse premise at a later time without acquiring the
knack of reversal learning. In such a case, there will be no
improvement from one reversal to the next. One item of Learning I
has simply re-placed another item of Learning I without any
achievement of Learning II. If, on the other hand, improvement
occurs with successive reversals, this is evidence for Learning II.

If we apply the same sort of logic to the relation between
Learning II and Learning III, we are led to expect that there might
be replacement of premises at the level of Learning II without the
achievement of any Learning III.

Preliminary to any discussion of Learning III, it is there-fore
necessary to discriminate between mere replacement without
Learning III and that facilitation of replacement which would be
truly Learning III.

That psychotherapists should be able to aid their patients even
in a mere replacement of premises acquired by Learning II is
already no mean feat when we consider the self-validating

character of such premises and their more or less unconscious
nature. But that this much can be done there is no doubt.

Within the controlled and protected setting of the therapeutic
relationship, the therapist may attempt one or more of the
following maneuvers:

(a)to achieve a confrontation between the premises of the

patient and those of the therapist—who is carefully trained

not to fall into the trap of validating the old premises;

(b)to get the patient to act, either in the therapy room or

outside, in ways which will confront his own premises;

(c)to demonstrate contradiction among the premises which

currently control the patient’s behavior;

(d)to induce in the patient some exaggeration or caricature

(e.g., in dream or hypnosis) of experience based on his old


As William Blake noted, long ago, “Without Contraries is no
progression.” (Elsewhere I have called these contradictions at level
II “double binds.”)

But there are always loopholes by which the impact of
contradiction can be reduced. It is a commonplace of learning
psychology that while the subject will learn (Learning I) more
rapidly if he is reinforced every time he responds correctly, such
learning will disappear rather rapidly if reinforcement ceases. If, on
the other hand, reinforcement is only occasional, the subject will
learn more slowly but the resulting learning will not easily be
extinguished when reinforcement ceases altogether. In other words,
the subject may learn (Learning 11) that the context is such that
absence of reinforcement does not indicate that his response was
wrong or inappropriate. His view of the context was, in fact, correct
until the experimenter changed his tactics.

The therapist must certainly so support or hedge the contraries by
which the patient is driven that loopholes of this and other kinds are
blocked. The Zen candidate who has been assigned a paradox
(koan) must labor at his task “like a mosquito biting on an iron bar.”

I have argued elsewhere (“Style, Grace, and Information in
Primitive Art,” see p. 128) that an essential and necessary function
of all habit formation and Learning I1 is an economy of the thought
processes (or neural pathways) which are used for problem-solving


or Learning I. The premises of what is commonly called
“character”—the definitions of the “self” —save the individual from
having to examine the abstract, philosophical, aesthetic, and ethical
aspects of many sequences of life. “I don’t know whether it’s good
music; I only know whether I like it.”

But Learning III will throw these unexamined premises open to
question and change.
Let us, as was done above for Learning I and II, list some of the
changes which we shall be willing to call Learning III.

(a)The individual might learn to form more readily those habits
the forming of which we call Learning II.

(b)He might learn to close for himself the “loopholes” which
would allow him to avoid Learning III.

(c)He might learn to change the habits acquired by Learning II.

(d)He might learn that he is a creature which can and doesunconsciously achieve Learning II.

(e)He might learn to limit or direct his Learning II.

(f)If Learning II is a learning of the contexts of Learning I,
then Learning III should be a learning of the contexts of those

But the above list proposes a paradox. Learning III (i.e., learning
about Learning II) may lead either to an increase in Learning II or
to a limitation and perhaps a reduction of that phenomenon.
Certainly it must lead to a greater flexibility in the premises
acquired by the process of Learning II —a freedom from their

I once heard a Zen master state categorically: “To become
accustomed to anything is a terrible thing.”

But any freedom from the bondage of habit must also denote a
profound redefinition of the self. If I stop at the level of Learning II,
“I” am the aggregate of those characteristics which I call my
“character.” “I” am my habits of acting in context and shaping and
perceiving the contexts in which I act. Selfhood is a product or
aggregate of Learning II. To the degree that a man achieves


Learning III, and learns to perceive and act in terms of the contexts
of contexts, his “self” will take on a sort of irrelevance. The concept
of “self” will no longer function as a nodal argument in the
punctuation of experience.

This matter needs to be examined. In the discussion of Learning
II, it was asserted that all words like “dependency,” “pride,”
“fatalism,” refer to characteristics of the self which are learned
(Learning II) in sequences of relationship. These words are, in fact,
terms for “roles” in relationships and refer to something artificially
chopped out of interactive sequences. It was also suggested that
the correct way to assign rigorous meaning to any such words is to
spell out the formal structure of the sequence in which the named
characteristic might have been learned. Thus the interactive
sequence of Pavlovian learning was proposed as a paradigm for a
certain sort of “fatalism,” etc.

But now we are asking about the contexts of these con-texts of
learning, i.e., about the larger sequences within which such
paradigms are embedded.

Consider the small item of Learning II which was mentioned
above as providing a “loophole” for escape from Learning III. A
certain characteristic of the self—call it “persistence”—is generated
by experience in multiple sequences among which reinforcement is
sporadic. We must now ask about the larger context of such
sequences. How are such sequences generated?

The question is explosive. The simple stylized experimental
sequence of interaction in the laboratory is generated by and partly
determines a network of contingencies which goes out in a hundred
directions leading out of the laboratory into the processes by which
psychological research is designed, the interactions between
psychologists, the economics of re-search money, etc., etc.

Or consider the same formal sequence in a more “natural”
setting. An organism is searching for a needed or missing object. A
pig is rooting for acorns, a gambler is feeding a slot machine hoping
for a jackpot, or a man must find the key to his car. There are
thousands of situations where living things must persist in certain
sorts of behavior precisely because reinforcement is sporadic or
improbable. Learning II will simplify the universe by handling these
instances as a single category. But if Learning III be concerned with


the contexts of these instances, then the categories of Learning II
will be burst open.

Or consider what the word “reinforcement” means at the
various levels. A porpoise gets a fish from the trainer when he does
what the trainer wants. At level I, the fact of the fish is linked with
the “rightness” of the particular action. At level II, the fact of the
fish confirms the porpoise’s under-standing of his (possibly
instrumental or dependent) relationship with the trainer. And note
that at this level, if the porpoise hates or fears the trainer, pain
received from the latter may be a positive reinforcement
confirming that hate. (“If it’s not the way I want it, I’ll prove it.”)

But what of “reinforcement” at level III (for porpoise or for

If, as I have suggested above, the creature is driven to level III by
“contraries” generated at level II, then we may expect that it is the
resolving of these contraries that will constitute positive
reinforcement at level III. Such resolution can take many forms.

Even the attempt at level III can be dangerous, and some fall by
the wayside. These are often labeled by psychiatry as psychotic, and
many of them find themselves inhibited from using the first person

For others, more successful, the resolution of the contraries may
be a collapsing of much that was learned at level II, revealing a
simplicity in which hunger leads directly to eating, and the
identified self is no longer in charge of organizing the behavior.
These are the incorruptible innocents of the world.

For others, more creative, the resolution of contraries reveals a
world in which personal identity merges into all the processes of
relationship in some vast ecology or aesthetics of cosmic interaction.
That any of these can survive seems almost miraculous, but some
are perhaps saved from being swept away on oceanic feeling by
their ability to focus in on the minutiae of life. Every detail of the
universe is seen as proposing a view of the whole. These are the
people for whom Blake wrote the famous advice in the “Auguries of

To see the World in a Grain of Sand, And a
Heaven in a Wild Flower,

Hold Infinity in the palm of your hand, And
Eternity in an hour.

The Role of Genetics in Psychology

Whatever can be said about an animal’s learning or in-ability to
learn has bearing upon the genetic make-up of the animal. And what
has been said here about the levels of learning has bearing upon the
whole interplay between genetic make-up and the changes which
that individual can and must achieve.

For any given organism, there is an upper limit beyond which all
is determined by genetics. Planarians can probably not go beyond
Learning I. Mammals other than man are probably capable of
Learning II but incapable of Learning III. Man may sometimes
achieve Learning III.

This upper limit for any organism is (logically and presumably)
set by genetic phenomena, not perhaps by individual genes or
combinations of genes, but by whatever factors control the
development of basic phylar characteristics.

For every change of which an organism is capable, there is the
fact of that capability. This fact may be genetically determined; or
the capability may have been learned. If the latter, then genetics
may have determined the capability of learning the capability. And
so on.

This is in general true of all somatic changes as well as of those
behavioral changes which we call learning. A man’s skin tans in the
sun. But where does genetics enter this picture? Does genetics
completely determine his ability to tan? Or can some men increase
their ability to tan? In the latter case, the genetic factors evidently
have effect at a higher logical level.

The problem in regard to any behavior is clearly not “Is it
learned or is it innate?” but “Up to what logical level is learning
effective and down to what level does genetics play a
determinative or partly effective role?”


The broad history of the evolution of learning seems to have
been a slow pushing back of genetic determinism to levels of
higher logical type.



A Note on Hierarchies

The model discussed in this paper assumes, tacitly, that the
logical types can be ordered in the form of a simple, unbranching
ladder. I believe that it was wise to deal first with the problems
raised by such a simple model.

But the world of action, experience, organization, and learning
cannot be completely mapped onto a model which excludes
propositions about the relation between classes of different logical

If CI is a class of propositions, and C2 is a class of propositions
about the members of C1; C3 then being a class of propositions about
the members of C2; how then shall we classify propositions about
the relation between these classes? For example, the proposition “As
members of C1 are to members of C2, so members of C2 are to
members of C3” cannot be classified within the unbranching ladder
of types.

The whole of this essay is built upon the premise that the relation
between C2 and C3 can be compared with the relation between C1
and C2. I have again and again taken a stance to the side of my
ladder of logical types to discuss the structure of this ladder. The
essay is therefore itself an example of the fact that the ladder is not

It follows that a next task will be to look for examples of
learning which cannot be classified in terms of my hierarchy of
learning but which fall to the side of this hierarchy as learning
about the relation between steps of the hierarchy. I have suggested
elsewhere (“Style, Grace, and Information in Primitive Art”) that
art is commonly concerned with learning of this sort, i.e., with
bridging the gap between the more or less unconscious premises
acquired by Learning II and the more episodic content of
consciousness and immediate action.

It should also be noted that the structure of this essay is inductive
in the sense that the hierarchy of orders of learning is presented to
the reader from the bottom upward, from level zero to level III. But
it is not intended that the explanations of the phenomenal world
which the model affords shall be unidirectional. In explaining the
model to the reader, a unidirectional approach was necessary, but
within the model it is assumed that higher levels are explanatory of
lower levels and vice versa. It is also assumed that a similar
reflexive relation—both inductive and deductive—obtains among
ideas and items of learning as these exist in the lives of the creatures
which we study.

Finally, the model remains ambiguous in the sense that while it is
asserted that there are explanatory or determinative relations
between ideas of adjacent levels both upward and downward, it is
not clear whether direct explanatory relations exist between
separated levels, e.g., between level III and level I or between level
zero and level II.

This question and that of the status of propositions and ideas
collateral to the hierarchy of types remains unexamined.


The Cybernetics of “Self”: A Theory ofAlcoholism*

The “logic” of alcoholic addiction has puzzled psychiatrists no
less than the “logic” of the strenuous spiritual regime whereby the
organization Alcoholics Anonymous is able to counteract the
addiction. In the present essay it is suggested: (1) that an entirely
new epistemology must come out of cybernetics and systems theory,
involving a new understanding of mind, self, human relationship,
and power; (2) that the addicted alcoholic is operating, when sober,
in terms of an epistemology which is conventional in Occidental
culture but which is not acceptable to systems theory; (3) that
surrender to alcoholic intoxication provides a partial and subjective
short cut to a more correct state of mind; and (4) that the theology of
Alcoholics Anonymous coincides closely with an epistemology of

The present essay is based upon ideas which are, perhaps all of
them, familiar either to psychiatrists who have had dealings with
alcoholics, or to philosophers who have thought about the
implications of cybernetics and systems theory. The only novelty
which can be claimed for the thesis here offered derives from
treating these ideas seriously as premises of argument and from the
bringing together of commonplace ideas from two too separate
fields of thought.

In its first conception, this essay was planned to be a systems-
theoretic study of alcoholic addiction, in which I would use data
from the publications of Alcoholics Anonymous, which has the only
outstanding record of success in dealing with alcoholics. It soon
became evident, however, that the religious views and the
organizational structure of AA presented points of great interest to
systems theory, and that the correct scope of my study should
include not only the premises of alcoholism but also the premises of
the AA system of treating it and the premises of AA organization.

* This article appeared in Psychiatry, Vol. 34, No. 1, pp. 1-18, 1971. Copyright
© 1971 by the William Alanson White Psychiatric Foundation. Reprinted by
permission of Psychiatry

My debt to AA will be evident throughout—also, I hope, my
respect for that organization and especially for the extraordinary
wisdom of its cofounders, Bill W. and Dr. Bob.

In addition, I have to acknowledge a debt to a small sample of
alcoholic patients with whom I worked intensively for about two
years in 1949-52, in the Veterans Administration Hospital, Palo Alto,
California. These men, it should be mentioned, carried other
diagnoses—mostly of “schizophrenia”—in addition to the pains of
alcoholism. Several were members of AA. I fear that I helped them
not at all.

The Problem

It is rather generally believed that “causes” or “reasons” for
alcoholism are to be looked for in the sober life of the alcoholic.
Alcoholics, in their sober manifestations, are commonly dubbed
“immature,” “maternally fixated,” “oral,” “homosexual,” “passiveaggressive,”
“fearful of success,” “oversensitive,” “proud,” “affable,”
or simply “weak.” But the logical implications of this belief are
usually not examined:

(1)If the sober life of the alcoholic somehow drives him to drink
or proposes the first step toward intoxication, it is not to be
expected that any procedure which reinforces his particular style of
sobriety will reduce or control his alcoholism.

(2)If his style of sobriety drives him to drink, then that style must
contain error or pathology; and intoxication must provide some—at
least subjective—correction of this error. In other words, compared
with his sobriety, which is in some way “wrong,” his intoxication
must be in some way “right.” The old tag In vino veritas may
contain a truth more profound than is usually attributed to it.

(3)An alternative hypothesis would suggest that when sober,
the alcoholic is somehow more sane than the people around him,
and that this situation is intolerable. I have heard alcoholics argue

in favor of this possibility, but I shall ignore it in this essay. I
think that Bernard Smith, the non-alcoholic legal representative of
AA, came close to the mark when he said, “the [AA] member was
never enslaved by alcohol. Alcohol simply served as an escape
from personal enslavement to the false ideals of a materialistic
society.”63 It is not a matter of revolt against insane ideals around
him but of escaping from his own insane premises, which are
continually reinforced by the surrounding society. It is possible,
however, that the alcoholic is in some way more vulnerable or
sensitive than the normal to the fact that his insane (but
conventional) premises lead to unsatisfying results.

(4)The present theory of alcoholism, therefore, will pro-vide a
converse matching between the sobriety and the intoxication, such
that the latter may be seen as an appropriate subjective correction
for the former.
(5)There are, of course, many instances in which people resort
to alcohol and even to extreme intoxication as an anesthetic
giving release from ordinary grief, resentment, or physical pain.
It might be argued that the anesthetic action of alcohol provides a
sufficient converse matching for our theoretical purposes. I shall,
however, specifically exclude these cases from consideration as
being not relevant to the problem of addictive or repetitive
alcoholism; and this in spite of the undoubted fact that “grief,”
“resentment,” and “frustration” are commonly used by addicted
alcoholics as excuses for drinking.
I shall demand, therefore, a converse matching between sobriety
and intoxication more specific than that provided by mere

63 [Alcoholics Anonymous], Alcoholics Anonymous Comes of Age, New
York, Harper, 1957, p. 279. (Italics added.)


The friends and relatives of the alcoholic commonly urge him to
be “strong,” and to “resist temptation.” What they mean by this is
not very clear, but it is significant that the alcoholic himself—while
sober—commonly agrees with their view of his “problem.” He
believes that he could be, or, at least, ought to be “the captain of his
soul.”64 But it is a cliche of alcoholism that after “that first drink,”
the motivation to stop drinking is zero. Typically the whole matter is
phrased overtly as a battle between “self” and “John Barleycorn.”
Covertly the alcoholic may be planning or even secretly laying in
supplies for the next binge, but it is almost impossible (in the
hospital setting) to get the sober alcoholic to plan his next binge in
an overt manner. He cannot, seemingly, be the “captain” of his soul
and overtly will or command his own drunkenness. The “captain”
can only command sobriety —and then not be obeyed.

Bill W., the cofounder of Alcoholics Anonymous, himself an
alcoholic, cut through all this mythology of conflict in the very first
of the famous “Twelve Steps” of AA. The first step demands that the
alcoholic agree that he is powerless over alcohol. This step is.
usually regarded as a “surrender” and many alcoholics are either
unable to achieve it or achieve it only briefly during the period of
remorse following a binge. AA does not regard these cases as
promising: they have not yet “hit bottom”; their despair is
inadequate and after a more or less brief spell of sobriety they will
again attempt to use “self-control” to fight the “temptation.” They
will not or cannot accept the premise that, drunk or sober, the total
personality of an alcoholic is an alcoholic personality which cannot
conceivably fight alcoholism. As an AA leaflet puts it, “trying to use
will power is like trying to lift yourself by your bootstraps.”

The first two steps of AA are as follows:

64 ‘ This phraseis usedbyAAinderisionofthealcoholic whotriestousewillpower
against the bottle. The quotation, along with the line, “My head is bloody but unbowed,”
comes from the poem “Invictus” by William Ernest Henley, who was a
cripple but not an alcoholic. The use of the will to conquer pain and physical disability
is probably not comparable to the alcoholic’s use of will.

1.We admitted we were powerless over alcohol—that ourlives had become unmanageable.
2.Came to believe that a Power greater than our-selves
could restore us to sanity.65

Implicit in the combination of these two steps is an extraordinary
—and I believe correct—idea: the experience of defeat not only
serves to convince the alcoholic that change is necessary; it is the
first step in that change. To be defeated by the bottle and to know it
is the first “spiritual experience.” The myth of self-power is thereby
broken by the demonstration of a greater power.

In sum, I shall argue that the “sobriety” of the alcoholic is
characterized by an unusually disastrous variant of the Cartesian
dualism, the division between Mind and Matter, or, in this case,
between conscious will, or “self,” and the remainder of the
personality. Bill W.’s stroke of genius was to break up with the first
“step” the structuring of this dualism.

Philosophically viewed, this first step is not a surrender; it is
simply a change in epistemology, a change in how to know about
the personality-in-the-world. And, notably, the change is from an
incorrect to a more correct epistemology.

Epistemology and Ontology

Philosophers have recognized and separated two sorts of
problem. There are first the problems of how things are, what is a
person, and what sort of a world this is. These are the problems of
ontology. Second, there are the problems of how we know anything,
or more specifically, how we know what sort of a world it is and
what sort of creatures we are that can know something (or perhaps
nothing) of this matter. These are the problems of epistemology. To
these questions, both ontological and epistemological, philosophers
try to find true answers.

65 ‘[Alcoholics Anonymous], Alcoholics Anonymous, New York, Works
Publishing, 1939


But the naturalist, observing human behavior, will ask rather
different questions. If he be a cultural relativist, he may agree with
those philosophers who hold that a “true” ontology is conceivable,
but he will not ask whether the ontology of the people he observes is
“true.” He will expect their epistemology to be culturally determined
or even idiosyncratic, and he will expect the culture as a whole to
make sense in terms of their particular epistemology and ontology.

If, on the other hand, it is clear that the local epistemology is
wrong, then the naturalist should be alert to the possibility that the
culture as a whole will never really make “sense,” or will make sense
only under restricted circumstances, which contact with other
cultures and new technologies might disrupt.

In the natural history of the living human being, ontology and
epistemology cannot be separated. His (commonly unconscious)
beliefs about what sort of world it is will determine how he sees it
and acts within it, and his ways of perceiving and acting will
determine his beliefs about its nature. The living man is thus bound
within a net of epistemological and ontological premises which—
regardless of ultimate truth or falsity—become partially self-
validating for him66

It is awkward to refer constantly to both epistemology and
ontology and incorrect to suggest that they are separable in human
natural history. There seems to be no convenient word to cover the
combination of these two concepts. The nearest approximations are
“cognitive structure” or “character structure,” but these terms fail to
suggest that what is important is a body of habitual assumptions or
premises implicit in the relationship between man and environment,
and that these premises may be true or false. I shall there-fore use
the single term “epistemology” in this essay to cover both aspects of
the net of premises which govern adaptation (or maladaptation) to
the human and physical environment. In George Kelly’s vocabulary,
these are the rules by which an individual “construes” his

I am concerned especially with that group of premises upon
which Occidental concepts of the “self” are built, and conversely,

66 J. Ruesch and G. Bateson, Communications: The Social Matrix of Psychiatry,
New York, Norton, 1951.


with premises which are corrective to some of the more gross
Occidental errors associated with that concept.

The Epistemology of Cybernetics

What is new and surprising is that we now have partial answers
to some of these questions. In the last twenty-five years
extraordinary advances have been made in our knowledge of what
sort of thing the environment ‘is, what sort of thing an organism is,
and, especially, what sort of thing a mind is. These advances have
come out of cybernetics, systems theory, information theory, and
related sciences.

We now know, with considerable certainty, that the ancient
problem of whether the mind is immanent or transcendent can be
answered in favor of immanence, and that this answer is more
economical of explanatory entities than any transcendent answer: it
has at least the negative sup-port of Occam’s Razor.

On the positive side, we can assert that any ongoing ensemble of
events and objects which has the appropriate complexity of causal
circuits and the appropriate energy relations will surely show mental
characteristics. It will compare, that is, be responsive to difference
(in addition to being affected by the ordinary physical “causes” such
as impact or force). It will “process information” and will inevitably
be self-corrective either toward homeostatic optima or toward the
maximization of certain variables.

A “bit” of information is definable as a difference which makes a
difference. Such a difference, as it travels and undergoes successive
transformation in a circuit, is an elementary idea.

But, most relevant in the present context, we know that no part of
such an internally interactive system can have unilateral control over
the remainder or over any other part. The mental characteristics are
inherent or immanent in the ensemble as a whole.

Even in very simple self-corrective systems, this holistic
character is evident. In the steam engine with a “governor,” the very
word “governor” is a misnomer if it be taken to mean that this part
of the system has unilateral control. The governor is, essentially, a

sense organ or transducer which receives a transform of the
difference between the actual running speed of the engine and
some ideal or preferred speed. This sense organ transforms these
differences into differences in some efferent message, for example,
to fuel supply or to a brake. The behavior of the governor is determined,
in other words, by the behavior of the other parts of the
system, and indirectly by its own behavior at a previous time.

The holistic and mental character of the system is most clearly
demonstrated by this last fact, that the behavior of the governor
(and, indeed, of every part of the causal circuit) is partially
determined by its own previous behavior. Message material (i.e.,
successive transforms of difference) must pass around the total
circuit, and the time required for the message material to return to
the place from which it started is a basic characteristic of the total
system. The behavior of the governor (or any other part of the
circuit) is thus in some degree determined not only by its
immediate past, but by what it did at a time which precedes the
present by the interval necessary for the message to complete the
circuit. There is thus a sort of determinative memory in even the
simplest cybernetic circuit.

The stability of the system (i.e., whether it will act self-
correctively or oscillate or go into runaway) depends upon the
relation between the operational product of all the transformations
of difference around the circuit and upon this characteristic time.
The “governor” has no control over these factors. Even a human
governor in a social system is bound by the same limitations. He is
controlled by information from the system and must adapt his own
actions to its time characteristics and to the effects of his own past

Thus, in no system which shows mental characteristics can any
part have unilateral control over the whole. In other words, the
mental characteristics of the system are immanent, not in some part,
but in the system as a whole.

The significance of this conclusion appears when we ask, “Can a
computer think?” or, “Is the mind in the brain?” And the answer to
both questions will be negative unless the question is focused upon
one of the few mental characteristics which are contained within the
computer or the brain. A computer is self-corrective in regard to


some of its internal variables. It may, for example, include
thermometers or other sense organs which are affected by
differences in its working temperature, and the response of the sense
organ to these differences may affect the action of a fan which in
turn corrects the temperature. We may therefore say that the system
shows mental characteristics in regard to its internal temperature.
But it would be incorrect to say that the main business of the
computer—the transformation of input differences into output
differences—is “a mental process.” The computer is only an are of a
larger circuit which always includes a man and an environment from
which information is received and upon which efferent messages
from the computer have effect. This total system, or ensemble, may
legitimately be said to show mental characteristics. It operates by
trial and error and has creative character.

Similarly, we may say that “mind” is immanent in those circuits
of the brain which are complete within the brain. Or that mind is
immanent in circuits which are complete within the system, brain
plus body. Or, finally, that mind is immanent in the larger system—
man plus environment.

In principle, if we desire to explain or understand the mental
aspect of any biological event, we must take into account the system
—that is, the network of closed circuits, within which that
biological event is determined. But when we seek to explain the
behavior of a man or any other organism, this “system” will usually
not have the same limits as the “self”—as this term is commonly
(and variously) understood.

Consider a man felling a tree with an axe. Each stroke of the axe
is modified or corrected, according to the shape of the cut face of the
tree left by the previous stroke. This self-corrective (i.e., mental)
process is brought about by a total system, tree-eyes-brain-musclesaxe-
stroke-tree; and it is this total system that has the characteristics
of immanent mind.

More correctly, we should spell the matter out as: (differences in
tree) -(differences in retina) -(differences in brain) -(differences in
muscles) -(differences in movement of axe) -(differences in tree),
etc. What is transmitted around the circuit is transforms of
differences. And, as noted above, a difference which makes a
difference is an idea or unit of information.

But this is not how the average Occidental sees the event
sequence of tree felling. He says, “I cut down the tree” and he even
believes that there is a delimited agent, the “self,” which performed
a delimited “purposive” action upon a de-limited object.

It is all very well to say that “Billiard ball A hit billiard ball B
and sent it into the pocket”; and it would perhaps be all right (if we
could do it) to give a complete hard-science account of the events
all around the circuit containing the man and the tree. But popular
parlance includes mind in its utterance by invoking the personal
pronoun, and then achieves a mixture of mentalism and
physicalism by restricting mind within the man and reifying the
tree. Finally the mind itself becomes reified by the notion that,
since the “self” acted upon the axe which acted upon the tree, the
“self” must also be a “thing.” The parallelism of syntax between “I
hit the billiard ball” and “The ball hit another ball” is totally

If you ask anybody about the localization and boundaries of the
self, these confusions are immediately displayed. Or consider a
blind man with a stick. Where does the blind man’s self begin? At
the tip of the stick? At the handle of the stick? Or at some point
halfway up the stick? These questions are nonsense, because the
stick is a pathway along which differences are transmitted under
transformation, so that to draw a delimiting line across this pathway
is to cut off a part of the systemic circuit which determines the blind
man’s locomotion.

Similarly, his sense organs are transducers or pathways for
information, as also are his axons, etc. From a systems-theoretic
point of view, it is a misleading metaphor to say that what travels in
an axon is an “impulse.” It would be more correct to say that what
travels is a difference, or a transform of a difference. The metaphor
of “impulse” suggests a hard-science line of thought which will
ramify only too easily into nonsense about “psychic energy,” and
those who talk this kind of nonsense will disregard the information
content of quiescence. The quiescence of an axon differs as much
from activity as its activity does from quiescence. Therefore
quiescence and activity have equal informational relevance. The
message of activity can only be accepted as valid if the message of
quiescence can also be trusted.


It is even incorrect to speak of the “message of activity” and the
“message of quiescence.” Always the fact that in-formation is a
transform of difference should be remembered, and we might better
call the one message “activity —not quiescence” and the other
“quiescence—not activity.”

Similar considerations apply to the repentant alcoholic. He
cannot simply elect “sobriety.” At best he could only elect “sobriety
—not drunkenness,” and his universe remains polarized, carrying
always both alternatives.

The total self-corrective unit which processes information, or, as
I say, “thinks” and “acts” and “decides,” is a system whose
boundaries do not at all coincide with the boundaries either of the
body or of what is popularly called the “self” or “consciousness”;
and it is important to notice that there are multiple differences
between the thinking system and the “self” as popularly conceived:

The system is not a transcendent entity as the “self” is commonly
supposed to be.

The ideas are immanent in a network of causal path-ways along
which transforms of difference are conducted. The “ideas” of the
system are in all cases at least binary in structure. They are not
“impulses” but “information.”

This network of pathways is not bounded with consciousness but
extends to include the pathways of all unconscious mentation—both
autonomic and repressed, neural and hormonal.

The network is not bounded by the skin but includes all external
pathways along which information can travel. It also includes those
effective differences which are immanent in the “objects” of such
information. It includes the path-ways of sound and light along
which travel transforms of differences originally immanent in things
and other people —and especially in our own actions.

It is important to note that the basic—and I believe erroneous—
tenets of popular epistemology are mutually rein-forcing. If, for
example, the popular premise of transcendence is discarded, the
immediate substitute is a premise of immanence in the body. But
this alternative will be unacceptable because large parts of the
thinking network are located outside the body. The so-called “Body-
Mind” problem is wrongly posed in terms which force the argument
toward paradox: if mind be supposed immanent in the body, then it


must be transcendent. If transcendent, it must be immanent. And so

Similarly, if we exclude the unconscious processes from the
“self” and call them “ego-alien,” then these processes take on the
subjective coloring of “urges” and “forces”; and this pseudodynamic
quality is then extended to the conscious “self” which attempts to
“resist” the “forces” of the unconscious. The “self” thereby becomes
itself an organization of seeming “forces.” The popular notion which
would equate “self” with consciousness thus leads into the notion
that ideas are “forces”; and this fallacy is in turn supported by saying
that the axon carries “impulses.” To find a way out of this mess is by
no means easy.

We shall proceed by first examining the structure of the
alcoholic’s polarization. In the epistemologically unsound resolution,
“I will fight the bottle,” what is supposedly lined up against

Alcoholic “Pride”

Alcoholics are philosophers in that universal sense in which all
human beings (and all mammals) are guided by highly abstract
principles of which they are either quite unconscious, or unaware
that the principle governing their perception and action is
philosophic. A common misnomer for such principles is

This misnomer arises naturally from the Anglo-Saxon
epistemological tendency to reify or attribute to the body all mental
phenomena which are peripheral to consciousness. And the
misnomer is, no doubt, supported by the fact that the exercise and/or
frustration of these principles is often accompanied by visceral and
other bodily sensations. I believe, however, that Pascal was correct

67 ” R. G. Collingwood, The Idea of Nature, Oxford, Ox-ford University Press,

68 ” G. Bateson, “A Social Scientist Views the Emotions,” Expression of the
Emotions in Man, P. Knapp, ed., International University Press, 1963.

when he said, “The heart has its reasons which the reason does not
at all perceive.

But the reader must not expect the alcoholic to present a
consistent picture. When the underlying epistemology is full of
error, derivations from it are inevitably either self-contradictory or
extremely restricted in scope. A consistent corpus of theorems
cannot be derived from an inconsistent body of axioms. In such
cases, the attempt to be consistent leads either to the great
proliferation of complexity characteristic of psychoanalytic theory
and Christian theology or to the extremely narrow view
characteristic of contemporary behaviorism.

I shall therefore proceed to examine the “pride” which is
characteristic of alcoholics to show that this principle of their
behavior is derived from the strange dualistic epistemology
characteristic of Occidental civilization.

A convenient way of describing such principles as “pride,”
“dependency,” “fatalism,” and so forth, is to examine the principle as
if it were a result of deutero-learning69 and to ask what contexts of
learning might understandably inculcate this principle.

(1) It is clear that the principle of alcoholic life which AA calls
“pride” is not contextually structured around past achievement. They
do not use the word to mean pride in something accomplished. The
emphasis is not upon “I succeeded,” but rather upon “I can. . . .” It is
an obsessive acceptance of a challenge, a repudiation of the
proposition “I cannot.”
(2)After the alcoholic has begun to suffer from—or be blamed
for—alcoholism, this principle of “pride” is mobilized behind the
proposition, “I can stay sober.” But, noticeably, success in this
achievement destroys the “challenge.” The alcoholic becomes
69 This use of formal contextual structure as a descriptive device does not
necessarily assume that the principle discussed is wholly or in part actually
learned in contexts having the appropriate formal structure. The principle could
have been genetically determined, and it might still follow that the principle is best
described by the formal delineation of the contexts in which it is exemplified. It is
precisely this fitting of behavior to context that makes it difficult or impossible to
determine whether a principle of behavior was genetically determined or learned in
that context; see G. Bateson, “Social Planning and the Concept of Deutero-
Learning,” Conference on Science, Philosophy and Religion, Second
Symposium, New York, Harper, 1942.


“cocksure,” as AA says. He relaxes his determination, risks a
drink, and finds himself on a binge. We may say that the
contextual structure of sobriety changes with its achievement.
Sobriety, at this point, is no longer the appropriate contextual
setting for “pride.” It is the risk of the drink that now is
challenging and calls out the fatal “I can….

(3)AA does its best to insist that this change in con-textual
structure shall never occur. They restructure the whole context by
asserting over and over again that “Once an alcoholic, always an
alcoholic.” They try to have the alcoholic place alcoholism within
the self, much as a Jungian analyst tries to have the patient
discover his “psychological type” and to learn to live with the
strengths and weaknesses of that type. In contrast, the contextual
structure of alcoholic “pride” places the alcoholism outside the
self: “I can resist drinking.”
(4)The challenge component of alcoholic “pride” is linked with
risk-taking. The principle might be put in words: “I can do
something where success is improbable and failure would be
disastrous.” Clearly this principle will never serve to maintain
continued sobriety. As success begins to appear probable, the
alcoholic must challenge the risk of a drink. The element of “bad
luck” or “probability” of failure places failure beyond the limits of
the self. “If failure occurs, it is not mine.” Alcoholic “pride”
progressively narrows the concept of “self,” placing what happens
outside its scope.
(5)The principle of pride-in-risk is ultimately almost suicidal.
It is all very well to test once whether the universe is on your side,
but to do so again and again, with increasing stringency of proof,
is to set out on a project which can only prove that the universe
hates you. But, still and all, the AA narratives show repeatedly
that, at the very bottom of despair, pride sometimes prevents
suicide. The final quietus must not be delivered by the “self.”70
70 See Bill’s Story, Alcoholics Anonymous, op. cit.


Pride and Symmetry

The so-called pride of the alcoholic always presumes a real or
fictitious “other,” and its complete contextual definition therefore
demands that we characterize the real or imagined relationship to
this “other.” A first step in this task is to classify the relationship as
either “symmetrical” or “complementary.”71 To do this is not entirely
simple when the “other” is a creation of the unconscious, but we
shall see that the indications for such a classification are clear.

An explanatory digression is, however, necessary. The primary
criterion is simple:

If, in a binary relationship, the behaviors of A and B are regarded
(byAand B) as similar and are linked so that more of the given
behavior by A stimulates more of it in B, and vice versa, then the
relationship is “symmetrical” in regard to these behaviors.

If, conversely, the behaviors of A and B are dissimilar but
mutually fit together (as, for example, spectatorship fits exhibitionism),
and the behaviors are linked so that more of A’s
behavior stimulates more of B’s fitting behavior, then the
relationship is “complementary” in regard to these behaviors.

Common examples of simple symmetrical relationship are
armaments races, keeping up with the Joneses, athletic emulation,
boxing matches, and the like. Common examples of complementary
relationship are dominance-submission, sadism-masochism,
nurturance-dependency, spectatorship-exhibitionism, and the like.

More complex considerations arise when higher logical typing is
present. For example: A and B may compete in gift-giving, thus
superposing a larger symmetrical frame upon primarily
complementary behaviors. Or, conversely, a therapist might engage
in competition with a patient in some sort of play therapy, placing a
complementary nurturant frame around the primarily symmetrical
transactions of the game.

Various sorts of “double binds” are generated when A and B
perceive the premises of their relationship in different terms—A

71 G. Bateson, Naven, Cambridge, Cambridge University Press, 1936.

may regard B’s behavior as competitive when B thought he was
helping A. And so on.

With these complexities we are not here concerned, be-cause the
imaginary “other” or counterpart in the “pride” of the alcoholic does
not, I believe, play the complex games which are characteristic of
the “voices” of schizophrenics.

Both complementary and symmetrical relationships are liable to
progressive changes of the sort which I have called
“schismogenesis.”72 Symmetrical struggles and armaments races
may, in the current phrase, “escalate”; and the normal pattern of
succoring-dependency between parent and child may become
monstrous. These potentially pathological developments are due to
undamped or uncorrected positive feedback in the system, and may
—as stated—occur in either complementary or symmetrical
systems. However, in mixed systems schismogenesis is necessarily
reduced. The armaments race between two nations will be slowed
down by acceptance of complementary themes such as dominance,
de-pendency, admiration, and so forth, between them. It will be
speeded up by the repudiation of these themes.

This antithetical relationship between complementary and
symmetrical themes is, no doubt, due to the fact that each is the
logical opposite of the other. In a merely symmetrical armaments
race, nation A is motivated to greater efforts by its estimate of the
greater strength of B. When it estimates that B is weaker, nation A
will relax its efforts. But the exact opposite will happen if A’s
structuring of the relationship is complementary. Observing that B is
weaker than they, A will go ahead with hopes of conquest.73

This antithesis between complementary and symmetrical patterns
may be more than simply logical. Notably, in psychoanalytic
theory,74 the patterns which are called “libidinal” and which are
modalities of the erogenous zones are all complementary. Intrusion,

72 Ibid.

73 G. Bateson, “The Pattern of an Armaments Race–Part I: An Anthropological
Approach,” Bulletin of Atomic

Scientists, 1946, 2(5): 10–11: also L. F. Richardson, “Generalized Foreign
Politics,” British Journal of Psychology, Monograph Supplements, 1939.

74 E. H. Erikson, “Configurations in Play—Clinical

Notes,” Psychoanalytic Quarterly, 1937, 6: 139–214.

inclusion, exclusion, reception, retention, and the like—all of these
are classed as “libidinal.” Whereas rivalry, competition, and the
like fall under the rubric of “ego” and “defense.”

It is also possible that the two antithetical codes—symmetrical
and complementary—may be physiologically represented by
contrasting states of the central nervous system. The progressive
changes of schismogenesis may reach climactic discontinuities and
sudden reversals. Symmetrical rage may suddenly turn to grief; the
retreating animal with tail between his legs may suddenly “turn at
bay” in a desperate battle of symmetry to the death. The bully may
suddenly become the coward when he is challenged, and the wolf
who is beaten in a symmetrical conflict may suddenly give “surrender”
signals which prevent further attack.

The last example is of special interest. If the struggle between the
wolves is symmetrical—that is, if wolf A is stimulated to more
aggressive behavior by the aggressive behavior of B—then if B
suddenly exhibits what we may call “negative aggression,” A will
not be able to continue to fight unless he can quickly switch over to
that complementary state of mind in which B’s weakness would be a
stimulus for his aggression. Within the hypothesis of symmetrical
and complemetary modes, it becomes unnecessary to postulate a
specifically “inhibitory” effect for the surrender signal.

Human beings who possess language can apply the label
“aggression” to all attempts to damage the other, regardless of
whether the attempt is prompted by the other’s strength or
weakness; but at the prelinguistic mammalian level these two sorts
of “aggression” must appear totally different. We are told that from
the lion’s point of view, an “attack” on a zebra is totally different
from an “attack” on another lion.75

Enough has now been said so that the question can be posed: Is
alcoholic pride contextually structured in symmetrical or
complementary form?

First, there is a very strong tendency toward symmetry in the
normal drinking habits of Occidental culture. Quite apart from
addictive alcoholism, two men drinking together are impelled by
convention to match each other, drink for drink. At this stage, the

75 13 K. Z. Lorenz, On Aggression, New York, Harcourt, Brace & World, 1966.

“other” is still real and the symmetry, or rivalry, between the pair is

As the alcoholic becomes addicted and tries to resist drinking, he
begins to find it difficult to resist the social context in which he
should match his friends in their drinking. The AA says, “Heaven
knows, we have tried hard enough and long enough to drink like
other people!”

As things get worse, the alcoholic is likely to become a solitary
drinker and to exhibit the whole spectrum of response to challenge.
His wife and friends begin to suggest that his drinking is a
weakness, and he may respond, with symmetry, both by resenting
them and by asserting his strength to resist the bottle. But, as is
characteristic of symmetrical responses, a brief period of successful
struggle weakens his motivation and he falls off the wagon. Symmetrical
effort requires continual opposition from the opponent.

Gradually the focus of the battle changes, and the alcoholic finds
himself committed to a new and more deadly type of symmetrical
conflict. He must now prove that the bottle cannot kill him. His
“head is bloody but unbowed.” He is still the “captain of his soul”—
for what it’s worth.

Meanwhile, his relationships with wife and boss and friends have
been deteriorating. He never did like the complementary status of
his boss as an authority; and now as he deteriorates his wife is more
and more forced to take a complementary role. She may try to exert
authority, or she becomes protective, or she shows forbearance, but
all those provoke either rage or shame. His symmetrical “pride” can
tolerate no complementary role.

In sum, the relationship between the alcoholic and his real or
fictitious “other” is clearly symmetrical and clearly schismogenic. It
escalates. We shall see that the religious conversion of the alcoholic
when saved by AA can be de-scribed as a dramatic shift from this
symmetrical habit, or epistemology, to an almost purely
complementary view of his relationship to others and to the universe
or God.


Pride or Inverted Proof?

Alcoholics may appear to be stiff-necked, but they are not stupid.
The part of the mind in which their policy is decided certainly lies
too deep for the word “stupidity” to be applicable. These levels of
the mind are prelinguistic and the computation which goes on there
is coded in primary process.

Both in dream and in mammalian interaction, the only way to
achieve a proposition which contains its own negation (“I will not
bite you,” or “I am not afraid of him”) is by an elaborate imagining
or acting out of the proposition to be negated, leading to a reductio
ad absurdum. “I will not bite you” is achieved between two
mammals by an experimental combat which is a “not combat,”
sometimes called “play.” It is for this reason that “agonistic”
behavior commonly evolves into friendly greeting.76

In this sense, the so-called pride of the alcoholic is in some
degree ironic. It is a determined effort to test some-thing like “selfcontrol”
with an ulterior but unstateable purpose of proving that
“self-control” is ineffectual and absurd. “It simply won’t work.” This
ultimate proposition, since it contains a simple negation, is not to be
expressed in primary process. Its final expression is in an action—
the taking of a drink. The heroic battle with the bottle, that fictitious
“other,” ends up in a “kiss and make friends.”

In favor of this hypothesis, there is the undoubted fact that the
testing of self-control leads back into drinking. And, as I have
argued above, the whole epistemology of self-control which his
friends urge upon the alcoholic is monstrous. If this be so, then the
alcoholic is right in rejecting it. He has achieved a reductio ad
absurdum of the conventional epistemology.

But this description of achieving a reductio ad absurdum verges
upon teleology. If the proposition “It won’t work” can-not be
entertained within the coding of primary process, how then can the
computations of primary process direct the organism to try out

76 G. Bateson, “Metalogue: What Is an Instinct?,” Aproaches to Animal
Communication, T. Sebeok, Hague, Mouton, 1969.


those courses of action which will demonstrate that “It won ‘t

Problems of this general type are frequent in psychiatry and can
perhaps only be resolved by a model in which, under certain
circumstances, the organism’s discomfort activates a positive
feedback loop to increase the behavior which preceded the
discomfort. Such positive feedback would provide a verification that
it was really that particular behavior which brought about the
discomfort, and might in-crease the discomfort to some threshold
level at which change would become possible.

In psychotherapy such a positive feedback loop is commonly
provided by the therapist who pushes the patient in the direction of
his symptoms—a technique which has been called the “therapeutic
double bind.” An example of this technique is quoted later in this
essay, where the AA member challenges the alcoholic to go and do
some “controlled drinking” in order that he may discover for himself
that he has no control.

It is also usual that the symptoms and hallucinations of the
schizophrenic—like dreams—constitute a corrective experience, so
that the whole schizophrenic episode takes on the character of a self-
initiation. Barbara O’Brien’s account of her own psychosis77 is
perhaps the most striking example of this phenomenon, which has
been discussed elsewhere.78

It will be noted that the possible existence of such a positive
feedback loop, which will cause a runaway in the direction of
increasing discomfort up to some threshold (which might be on the
other side of death), is not included in conventional theories of
learning. But a tendency to verify the unpleasant by seeking
repeated experience of it is a common human trait. It is perhaps
what Freud called the “death instinct.”

77 B. O’Brien, Operators and Things: The Inner Life of a Schizophrenic,
Cambridge, Mass., Arlington Books, 1958.

78 G. Bateson, ed., Perceval’s Narrative, Stanford, Calif., Stanford University
Press, 1961, Introduction

The Drunken State

What has been said above about the treadmill of symmetrical
pride is only one half of the picture. It is the picture of the state of
mind of the alcoholic battling with the bottle. Clearly this state is
very unpleasant and clearly it is also unrealistic. His “others” are
either totally imaginary or are gross distortions of persons on whom
he is dependent and whom he may love. He has an alternative to this
uncomfortable state—he can get drunk. Or, “at least,” have a drink.

With this complementary surrender, which the alcoholic will
often see as an act of spite—a Barthian dart in a symmetrical
struggle—his entire epistemology changes. His anxieties and
resentments and panic vanish as if by magic. His self-control is
lessened, but his need to compare himself with others is reduced
even further. He feels the physiological warmth of alcohol in his
veins and, in many cases, a corresponding psychological warmth
toward others. He may be either maudlin or angry, but he has at least
become again a part of the human scene.

Direct data bearing upon the thesis that the step from sobriety
into intoxication is also a step from symmetrical challenge into
complementarity are scarce, and always confused both by the
distortions of recall and by the complex toxicity of the alcohol. But
there is strong evidence from song and story to indicate that the step
is of this kind. In ritual, partaking of wine has always stood for the
social aggregation of persons united in religious “communion” or
secular Gemütlichkeit. In a very literal sense, alcohol supposedly
makes the individual see himself as and act as a part of the group.
That is, it enables complementarity in the relationships which
surround him.

Hitting Bottom

AA attaches great importance to this phenomenon and regards
the alcoholic who has not hit bottom as a poor prospect for their
help. Conversely, they are inclined to explain their failure by saying
that the individual who goes back to his alcoholism has not yet “hit

Certainly many sorts of disaster may cause an alcoholic to hit
bottom. Various sorts of accidents, an attack of delirium tremens, a
patch of drunken time of which he has no memory, rejection by
wife, loss of job, hopeless diagnosis, and so on—any of these may
have the required effect. AA says that “bottom” is different for
different men and some may be dead before they reach it.79

It is possible, however, that “bottom” is reached many times by
any given individual; that “bottom” is a spell of panic which
provides a favorable moment for change, but not a moment at which
change is inevitable. Friends and relatives and even therapists may
pull the alcoholic out of his panic, either with drugs or reassurance,
so that he “re-covers” and goes back to his “pride” and alcoholism—
only to hit a more disastrous “bottom” at some later time, when he
will again be ripe for a change. The attempt to change the alcoholic
in a period between such moments of panic is unlikely to succeed.

The nature of the panic is made clear by the following
description of a “test.”

We do not like to pronounce any individual as alcoholic, but you
can quickly diagnose yourself. Step over to the nearest barroom and
try some controlled drinking. Try to drink and stop abruptly. Try it
more than once. It will not take long for you to decide, if you are
honest with yourself about it. It may be worth a bad case of jitters if
you get a full knowledge of your condition.80

We might compare the test quoted above to commanding a driver
to brake suddenly when traveling on a slippery road: he will
discover fast that his control is limited. (The metaphor “skid row”
for the alcoholic section of town is not inappropriate.)

79 Personal communication from a member.


The panic of the alcoholic who has hit bottom is the panic of the
man who thought he had control over a vehicle but suddenly finds
that the vehicle can run away with him. Suddenly, pressure on what
he knows is the brake seems to make the vehicle go faster. It is the
panic of discovering that it (the system, self plus vehicle) is bigger
than he is.

In terms of the theory here presented, we may say that hitting
bottom exemplifies systems theory at three levels:

(1)The alcoholic works on the discomforts of sobriety to a
threshold point at which he has bankrupted the epistemology of
“self-control.” He then gets drunk—because the “system” is bigger
than he is—and he may as well surrender to it.
(2)He works repeatedly at getting drunk until he proves that
there is a still larger system. He then encounters the panic of
“hitting bottom.”
(3)If friends and therapists reassure him, he may achieve a
further unstable adjustment—becoming addicted to their help—
until he demonstrates that this system won’t work, and “hits
bottom” again but at a lower level. In this, as in all cybernetic
systems, the sign (plus or minus) of the effect of any intrusion
upon the system depends upon timing.

(4)Lastly, the phenomenon of hitting bottom is complexly
related to the experience of double bind.81 Bill W. narrates that he
hit bottom when diagnosed as a hopeless alcoholic by Dr. William

D. Silkworth in 1939, and this event is regarded as the beginning
of AA history.82 Dr. Silkworth also “supplied us with the tools with
which to puncture the toughest alcoholic ego, those shattering
phrases by which he described our illness: the obsession of the mind
that compels us to drink and the allergy of the body that condemns us
to go mad or die.”83 This is a double bind correctly founded upon
the alcoholic’s dichotomous epistemology of mind versus body. He
is forced by these words back and back to the point at which only
80 Alcoholics Anonymous, op. cit., p. 43.

81 Bateson, et al., “Toward a Theory of Schizophrenia,” Behavioral Science,
1956, 1: 251-64.

82 A A Comes of Age, op. cit., p. vii

83 Ibid., p. 13. (Italics in the original)


an involuntary change in deep unconscious epistemology—a
spiritual experience—will make the lethal description irrelevant.

The Theology of Alcoholics Anonymous

Some outstanding points of the theology of AA are:

(1) There is a Power greater than the self. Cybernetics would go
somewhat further and recognize that the “self” as ordinarily
understood is only a small part of a much larger trial-and-error
system which does the thinking, acting, and deciding. This system
includes all the informational path-ways which are relevant at any
given moment to any given decision. The “self” is a false reification
of an improperly de-limited part of this much larger field of
interlocking processes. Cybernetics also recognizes that two or more
persons —any group of persons—may together form such a
thinkingand-acting system.
(2) This Power is felt to be personal and to be intimately linked
with each person. It is “God as you understand him to be.”
Cybernetically speaking, “my” relation to any larger system
around me and including other things and persons will be different
from “your” relation to some similar system around you. The
relation “part of” must necessarily and logically al-ways be
complementary but the meaning of the phrase “part of” will be
different for every person.84 This difference will be especially
important in systems containing more than one person. The system
or “power” must necessarily appear different from where each
person sits. Moreover, it is expect-able that such systems, when they
encounter each other, will recognize each other as systems in this
sense. The “beauty” of the woods through which I walk is my
recognition both of the individual trees and of the total ecology of
the woods as systems. A similar esthetic recognition is still more
striking when I talk with another person.

84 This diversity in styles of integration could account for the fact that some
persons become alcoholic while others do not.

(3) A favorable relationship with this Power is discovered
through “hitting bottom” and “surrender.”
(4) By resisting this Power, men and especially alcoholics bring
disaster upon themselves. The materialistic philosophy which sees
“man” as pitted against his environment is rapidly breaking down as
technological man becomes more and more able to oppose the
largest systems. Every battle that he wins brings a threat of disaster.
The unit of survival—either in ethics or in evolution—is not the
organism or the species but the largest system or “power” within
which the creature lives. If the creature destroys its environment, it
destroys it-self.
(5) But—and this is important—the Power does not re-ward and
punish. It does not have “power” in that sense. In the biblical phrase,
“All things work together for good to them that love God.” And,
conversely, to them that do not. The idea of power in the sense of
unilateral control is foreign to AA. Their organization is strictly
“democratic” (their word), and even their deity is still bound by
what we might call a systemic determinism. The same limitation
applies both to the relationship between the AA sponsor and the
drunk whom he hopes to help and to the relationship between AA
central office and every local group.
(6) The first two “steps” of Alcoholics Anonymous taken together
identify the addiction as a manifestation of this Power.
(7) The healthy relation between each person and this Power is
complementary. It is in precise contrast to the “pride” of the
alcoholic, which is predicated upon a symmetrical relationship to an
imagined “other.” The schismogenesis is always more powerful than
the participants in it.
(8) The quality and content of each person’s relation to the Power
is indicated or reflected in the social structure of AA. The secular
aspect of this system—its governance—is delineated in “Twelve
Traditions”85 which supplement the “Twelve Steps,” the latter
developing man’s relationship to the Power. The two documents
overlap in the Twelfth Step, which enjoins aid to other alcoholics as
a necessary spiritual exercise without which the member is likely to
relapse. The total system is a Durkheimian religion in the sense that
85 AA Comes of Age, op. cit. 24 Ibid., p. 288. 25 Ibid., pp. 286-94.

the relationship between man and his community parallels the
relationship between man and God. “AA is a power greater than any
of us.”86

In sum, the relationship of each individual to the “Power” is best
defined in the words is part of.”

(9) Anonymity. It must be understood that anonymity means
much more in AA thinking and theology than the mere protection of
members from exposure and shame. With increasing fame and
success of the organization as a whole, it has become a temptation
for members to use the fact of their membership as a positive asset
in public relations, politics, education, and many other fields. Bill
W., the co-founder of the organization, was himself caught by this
temptation in early days and has discussed the matter in a published
article.87 He sees first that any grabbing of the spotlight must be a
personal and spiritual danger to the member, who cannot affort such
self-seeking; and beyond this that it would be fatal for the
organization as a whole to become involved in politics, religious
controversy, and social reform. He states clearly that the errors of
the alcoholic are the same as the “forces which are today ripping the
world apart at its seams,” but that it is not the business of AA to save
the world. Their single purpose is “to carry the AA message to the
sick alcoholic who wants it.”88 He concludes that anonymity is “the
greatest symbol of self-sacrifice that we know.” Elsewhere the
twelfth of the “Twelve Traditions” states that “anonymity is the
spiritual foundation of our traditions, ever reminding us to place
principles before personalities.”
To this we may add that anonymity is also a profound statement
of the systemic relation, part-to-whole. Some systems theorists
would go even further, because a major temptation for systems
theory lies in the reification of theoretical concepts. Anatol Holt says
he wants a bumper sticker which would (paradoxically) say, “Stamp
out nouns.”89

86 Ibid, p. 288.

87 Ibid, pp.286-294

88 Ibid.

89 M. C. Bateson, ed., Our Own Metaphor, Wenner-Gren Foundation,
Conference on the Effects of Conscious Purpose on Human Adaptation, 1968; New
York, Knopf, in press.

(10) Prayer. The AA use of prayer similarly affirms the
complementarity of part-whole relationship by the very simple
technique of asking for that relationship. They ask for those-
personal characteristics, such as humility, which are in fact-
exercised in the very act of prayer. If the act of prayer be sincere
(which is not so easy), God cannot but grant the request. And this is
peculiarly true of “God, as you understand him.” This self-affirming
tautology, which contains its own beauty, is precisely the balm
required after the anguish of the double binds which went with
hitting bottom.
Somewhat more complex is the famous “Serenity Prayer”: “God
grant us the serenity to accept the things we cannot change, courage
to change the things we can, and wisdom to know the difference.”90

If double binds cause anguish and despair and destroy personal
epistemological premises at some deep level, then it follows,
conversely, that for the healing of these wounds and the growth of a
new epistemology, some converse of the double bind will be
appropriate. The double bind leads The Serenity Prayer explicitly
frees the worshipper from these maddening bonds.

to the conclusion of despair, “There are no alternatives.”

In this connection it is worth mentioning that the great
schizophrenic, John Perceval, observed a change in his “voices.” In
the beginning of his psychosis they bullied him with “contradictory
commands” (or as I would say, double binds), but later he began to
recover when they offered him choice of clearly defined

(11) In one characteristic, AA differs profoundly from such
natural mental systems as the family or the redwood forest. It has a
single purpose—”to carry the AA message to the sick alcoholic who
wants it”—and the organization is dedicated to the maximization of
that purpose. In this respect, AA is no more sophisticated than
General Motors or an Occidental nation. But biological systems,
other than those premised upon Occidental ideas (and especially
money), are multipurposed. There is no single variable in the red90
This was not originally an AA document and its authorship is unknown. Small
variations in the text occur. I have quoted the form which I personally prefer from AA
Comes of Age, op. cit., p. 196.

91 Bateson, Perceval . . . , op. cit.

wood forest of which we can say that the whole system is oriented
to maximizing that variable and all other variables are subsidiary to
it; and, indeed, the redwood forest works toward optima, not
maxima. Its needs are satiable, and too much of anything is toxic.

There is, however, this: that the single purpose of AA is directed
outward and is aimed at a noncompetitve relationship to the larger
world. The variable to be maximized is a complementarity and is of
the nature of “service” rather than dominance.

The Epistemological Status of Complementary and
Symmetrical Premises

It was noted above that in human interaction, symmetry and
complementarity may be complexly combined. It is therefore
reasonable to ask how it is possible to regard these themes as so
fundamental that they shall be called “epistemological,” even in a
natural history study of cultural and interpersonal premises.

The answer seems to hang upon what is meant by “fundamental”
in such a study of man’s natural history; and the word seems to carry
two sorts of meaning.

First, I call more fundamental those premises which are the more
deeply embedded in the mind, which are the more “hard
programmed” and the less susceptible to change. In this sense, the
symmetrical pride or hubris of the alcoholic is fundamental.

Second, I shall call more fundamental those premises of mind
which refer to the larger rather than the smaller systems or gestalten
of the universe. The proposition “Grass is green” is less fundamental
than the proposition “Color differences make a difference.”

But, if we ask about what happens when premises are changed, it
becomes clear that these two definitions of the “fundamental”
overlap to a very great extent. If a man achieves or suffers change in
premises which are deeply embedded in his mind, he will surely find
that the results of that change will ramify throughout his whole
universe. Such changes we may well call “epistemological.”

The question then remains regarding what is epistemologically
“right” and what is epistemologically “wrong.” Is the change from
alcoholic symmetrical “pride” to the AA species of complementarity


a correction of his epistemology? And is complementarity always
somehow better than symmetry?

For the AA member, it may well be true that complementarity is
always to be preferred to symmetry and that even the trivial rivalry
of a game of tennis or chess may be dangerous. The superficial
episode may touch off the deeply embedded symmetrical premise.
But this does not mean that tennis and chess propose
epistemological error for everybody.

The ethical and philosophic problem really concerns only the
widest universe and the deepest psychological levels. If we deeply
and even unconsciously believe that our relation to the largest
system which concerns us—the “Power greater than self”—is
symmetrical and emulative, then we are in error.

Limitations of the Hypothesis

Finally, the above analysis is subject to the following limitations
and implications:
(1)It is not asserted that all alcoholics operate according to thelogic which is here outlined. It is very possible that other types ofalcoholics exist and almost certain that alcoholic addiction in other
cultures will follow other lines.

(2)It is not asserted that the way of Alcoholics Anonymous is the

only way to live correctly or that their theology is the only correct

derivation from the epistemology of cybernetics and systems


(3)It is not asserted that all transactions between human beings

ought to be complementary, though it is clear that the relation

between the individual and the larger system of which he is a part

must necessarily be so. Relations between persons will (I hope)

always be complex.

(4)It is, however, asserted that the nonalcoholic world has many

lessons which it might learn from the epistemology of systems

theory and from the ways of AA. If we continue to operate in terms

of a Cartesian dualism of mind versus matter, we shall probably

also continue to see the world in terms of God versus man; elite

versus people; chosen race versus others; nation versus nation; and
man versus environment. It is doubtful whether a species having
both an advanced technology and this strange way of looking at its
world can endure.


Comment on Part III

In the essays collected in Part III, I speak of an action or
utterance as occurring “in” a context, and this conventional way of
talking suggests that the particular action is a “dependent” variable,
while the context is the “independent” or determining variable. But
this view of how an action is related to its context is likely to distract
the reader—as it has distracted me—from perceiving the ecology of
the ideas which together constitute the small subsystem which I call
“context. ”

This heuristic error—copied like so many others from the ways
of thought of the physicist and chemist—requires correction.

It is important to see the particular utterance or action as part of
the ecological subsystem called context and not as the product or
effect of what remains of the context after the piece which we want
to explain has been cut out from it.

The mistake in question is the same formal error as that
mentioned in the comment on Part II where I discuss the evolution
of the horse. We should not think of this process just as a set of
changes in the animal’s adaptation to life on the grassy plains but .as
a constancy in the relationship between animals and
environment. It is the ecology which survives and slowly evolves. In
this evolution, the relata—the animals and the grass—undergo
changes which are indeed adaptive from moment to moment. But if
the process of adaptation were the whole story, there could be no
systemic pathology. Trouble arises precisely because the “logic” of
adaptation is a different “logic” from that of the survival and
evolution of the ecological system.

In Warren Brodey’s phrase, the “time-grain” of the adaptation is
different from that of the ecology.

“Survival” means that certain descriptive statements about some
living system continue to be true through some period of time; and,
conversely, “evolution” refers to changes in the truth of certain
descriptive statements about some living system. The trick is to
define which statements about which systems remain true or
undergo change.

The paradoxes (and the pathologies) of systemic process arise
precisely because the constancy and survival of some larger system
is maintained by changes in the constituent subsystems.

The relative constancy—the survival—of the relationship
between animals and grass is maintained by changes in both relata.
But any adaptive change in either of the relata, if uncorrected by
some change in the other, will always jeopardize the relationship
between them. These arguments propose a new conceptual frame for
the “double bind” hypothesis, a new conceptual frame for thinking
about “schizophrenia,” and a new way of looking at context and
levels of learning.

In a word, schizophrenia, deutero-learning, and the double bind
cease to be matters of individual psychology and be-come part of
the ecology of ideas in systems or “minds” whose boundaries no
longer coincide with the skins of the participant individuals.


Part IV: Biology and Evolution

On Empty-Headedness Among Biologists
State Boards of Education*

My father, the geneticist William Bateson, used to read us
passages of the Bible at breakfast—lest we grow up to be empty-
headed atheists; and so I find it natural to wonder what broadening
of the mind may come from the strange anti-evolutionary ruling of
the State Board of Education in California.1

Evolution has long been badly taught. In particular, students—
and even professional biologists—acquire theories of evolution
without any deep understanding of what problem these theories
attempt to solve. They learn but little of the evolution of
evolutionary theory.

The extraordinary achievement of the writers of the first chapter
of Genesis was their perception of the problem: Where does order
come from? They observed that the land and the water were, in
fact, separate and that species were separate; they saw that such
separation and sorting in the universe presented a fundamental
problem. In modern terms, we may say that this is the problem
implicit in the Second Law of Thermodynamics: If random events
lead to things getting mixed up, by what nonrandom events did
things come to be sorted? And what is a “random” event?

This problem has been central to biology and to many other
sciences for the last 5000 years, and the problem is not trivial.

With what Word should we designate the principle of order
which seems to be immanent in the universe?

The California ruling suggests that students be told of other
attempts to solve this ancient problem. I myself collected one of
these among the Stone Age head-hunters of the Iatmul tribe in New
Guinea. They, too, note that the land and the water are separate even
in their swampy region. They say that in the beginning there was a

* This item in BioScience, Vol. 20, 1970, is reproduced by permission from
that journal.
1 See “California’s Anti-Evolution Ruling,” BioScience, March 1, 1970.


vast crocodile, Kavwokmali, who paddled with his front legs and
paddled with his back legs, and thereby kept the mud in suspension.
The culture hero, Kevembuangga, speared the crocodile, who then
ceased to paddle, causing the mud and the water to separate. The
result was dry land upon which Kevembuangga stamped his foot in
triumph. We might say he verified that “it was good.”

Our students might have their minds broadened somewhat if they
would look at other theories of evolution and consider how a man’s
spirit must take a different shape if he believes that all sorting in the
universe is due to an external agent, or if, like the Iatmul and
modern scientists, he sees that the potentiality for order and pattern
is immanent throughout this world.

And then the student may be forced by the new system to look at
the “Great Chain of Being,” with Supreme Mind at the top and the
protozoa at the bottom. He will see how Mind was invoked as an
explanatory principle all through the Middle Ages and how Mind
later became the problem. Mind became that which needed
explanation when Lamarck showed that the Great Chain of Being
should be inverted to give an evolutionary sequence from the
protozoa upward. The problem then was to explain Mind in terms of
what could be known of this sequence.

And when the student reaches the mid-nineteenth century, he
might be given as a textbook Philip Henry Gosse’s Creation
(Omphalos): An Attempt to Untie the Geological Knot. He will learn
from this extraordinary book things about the structure of animals
and plants which are today scarcely mentioned in many courses of
biology; notably, that all animals and plants show a time structure,
of which the rings of growth in trees are an elementary example and
the cycles of life history, a more complex one. Every plant and
animal is constructed upon the premise of its cyclic nature.

After all, there can be no harm in Gosse, who was a devout
fundamentalist—a Plymouth Brother—as well as a distinguished
marine biologist. His book was published in 1857, two years before
the Origin a f Species. He wrote it to show that the facts of the fossil
record as well as those of biological homology could be made to fit
with the principles of fundamentalism. It was to him inconceivable
that God could have created a world in which Adam had no navel;
the trees in the Garden of Eden, no rings of growth; and the rocks,

no strata. Therefore, God must have created the world as though it
had a past.

It will do the student no harm to wrestle with the paradoxes of
Gosse’s “Law of Prochronism”; if he listens care-fully to Gosse’s
groping generalizations about the biological world, he will hear an
early version of the “steady state” hypothesis.

Of course, everybody knows that biological phenomena are
cyclic-from egg, to hen, to egg, to hen, etc. But not all biologists
have examined the implications of this cyclic characteristic for
evolutionary and ecological theory. Gosse’s view of the biological
world might broaden their minds.

It is silly and vulgar to approach the rich spectrum of
evolutionary thought with questions only about who was right and
who was wrong. We might as well assert that the amphibia and
reptiles were “wrong” and the mammals and birds “right” in their
solutions to the problems of how to live.

By fighting the fundamentalists, we are led into an emptyheadedness
analogous to theirs. The truth of the matter is that “Other
men have laboured and ye are entered into their labours” (John
1:38), and this text is not only a reminder of the need for humility, it
is also an epitome of the vast evolutionary process into which we
organisms are willy-nilly entered.


The Role of Somatic Change in

All theories of biological evolution depend upon at least three
sorts of change: (a) change of genotype, either by mutation or by
redistribution of genes; (b) somatic change under pressure of
environment; and (c) changes in environmental conditions. The
problem for the evolutionist is to build a theory combining these
types of change into an ongoing process which, under natural
selection, will account for the phenomena of adaptation and

Certain conventional premises may be selected to govern such
theory building:

(a) The theory shall not depend upon Lamarckian inheritance.
August Weismann’s argument for this premise still stands. There is
no reason to believe that either somatic change or changes in
environment can, in principle, call (by physiological
communication) for appropriate genotypic change. Indeed, the little
that we know about communication within the multicellular2
individual indicates that such communication from soma to gene
script is likely to be rare and unlikely to be adaptive in effect.
However, it is appropriate to attempt to spell out in this essay what
this premise implies:

Whenever some characteristic of an organism is modifiable
under measurable environmental impact or under measurable impact
of internal physiology, it is possible to write an equation in which
the value of the characteristic in question is expressed as some
function of the value of the impacting circumstance. “Human skin
color is some function of exposure to sunlight,” “respiration rate is
some function of atmospheric pressure,” etc. Such equations are

* This essay appeared in the journal Evolution, Vol 17, 1963, and is reprinted
with the editor’s permission.
2 The problems of bacterial genetics are here deliberately excluded.

constructed to be true for a variety of particular observations, and
necessarily contain subsidiary propositions which are stable (i.e.,
continue to be true) over a wide range of values of impacting
circumstance and somatic characteristic. These subsidiary
propositions are of different logical type from the original
observations in the laboratory and are, in fact, descriptive not of the
data but of our equations. They are statements about the form of the
particular equation and about the values of the parameters
mentioned within it.

It would be simple, at this point, to draw the line between
genotype and phenotype by saying that the forms and parameters of
such equations are provided by genes, while the impacts of
environment, etc. determine the actual event within this frame.
This would amount to saying, e.g., that the ability to tan is
genotypically determined, while the amount of tanning in a
particular case depends upon exposure to sun-light.

In terms of this oversimplified approach to the overlapping
roles of genotype and environment, the proposition excluding
Lamarckian inheritance would read somewhat as follows: In the
attempt to explain evolutionary process, there shall be no
assumption that the achievement of a particular value of some
variable under particular circumstances will affect, in the gametes
produced by that individual, the form or parameters of the
functional equation governing the relationship between that
variable and its environmental circumstances.

Such a view is oversimplified, and parentheses must be added to
deal with more complex and extreme cases. First, it is important to
recognize that the organism, considered as a communicational
system, may itself operate at multiple levels of logical typing; i.e.,
that there will be instances in which what were above called
“parameters” are subject to change. The individual organism might
as a result of “training” change its ability to develop a tan under
sunlight. And this type of change is certainly of very great
importance in the field of animal behavior, where “learning to learn”
can never be ignored.

Second, the oversimplified view must be elaborated to cover
negative effects. An environmental circumstance may have such


impact upon an organism unable to adapt to it, that the individual in
question will in fact produce no gametes.

Third, it is expectable that some of the parameters in one
equation may be subject to change under impact from some
environmental or physiologic circumstance other than the
circumstance mentioned in that equation.

Be all that as it may, both Weismann’s objection to Lamarckian
theory and my own attempt to spell the matter out share a certain
parsimony: an assumption that the principles which order
phenomena shall not themselves be supposed changed by those
phenomena which they order. William of Occam’s razor might be
reformulated: in any explanation, logical types shall not be
multiplied beyond necessity.

(b) Somatic change is absolutely necessary for survival. Any
change of environment which requires adaptive change in the
species will be lethal unless, by somatic change, the organisms (or
some of them) are able to weather out a period of unpredictable
duration, until either appropriate genotypic change occurs (whether
by mutation or by redistribution of genes already available in the
population), or because the environment returns to the previous
normal. The premise is truistical, regardless of the magnitude of the
time span involved.
(c) Somatic change is also necessary to cope with any changes of
genotype which might aid the organism in its external struggle with
the environment. The individual organism is a complex
organization of interdependent parts. A mutational or other
genotypic change in any one of these (however externally valuable
in terms of survival) is certain to require change in many others—
which changes will probably not be specified or implicit in the
single mutational change of the genes. A hypothetical pregiraffe,
which had the luck to carry a mutant gene “long neck,” would have
to adjust to this change by complex modifications of the heart and
circulatory system. These collateral adjustments would have to be
achieved at the somatic level. Only those pregiraffes which are
(genotypically) capable of these somatic modifications would
(d) In this essay, it is assumed that the corpus of genotypic
messages is preponderantly digital in nature. In contrast, the soma is

seen as a working system in which the genotypic recipes are tried
out. Should it transpire that the genotypic corpus is also in some
degree analogic—a working model of the soma—premise c (above)
would be negated to that degree. It would then be conceivable .that
the mutant gene “long neck” might modify the message of those
genes which affect the development of the heart. It is, of course,
known that genes may have pleiotropic effect, but these phenomena
are relevant in the present connection only if it can be shown, e.g.,
that the effect of gene A upon the phenotype and its effect upon the
phenotypic expression of gene B are mutually appropriate in the
overall integration and adaptation of the organism.

These considerations lead to a classifying of both genotypic and
environmental changes in terms of the price which they exact of the
flexibility of the somatic system. A lethal change in either
environment or genotype is simply one which demands somatic
modifications which the organism cannot achieve.

But the somatic price of a given change must depend, not
absolutely upon the change in question, but upon the range of
somatic flexibility available to the organism at the given time. This
range, in turn, will depend upon how much of the organism’s
somatic flexibility is already being used up in adjusting to other
mutations or environmental changes. We face an economics of
flexibility which, like any other economics, will become
determinative for the course of evolution if and only if the organism
is operating close to the limits set by this economics.

However, this economics of somatic flexibility will differ in one
important respect from the more familiar economics of money or
available energy. In these latter, each new expenditure can simply be
added to the preceding expenditures and the economics becomes
coercive when the additive total approaches the limit of the budget.
In contrast, the combined effect of multiple changes, each of which
exacts a price in the soma, will be multiplicative. This point may
be stated as follows: Let S be the finite set of all possible living
states of the organism. Within S, let s1 be the smaller set of all states
compatible with a given mutation (ml), and let s2 be the set of states
compatible with a second mutation (m2). It follows that the two
mutations in combination will limit the organism to the logical


product of s1 and s2, i.e., to that usually smaller subset of states
which is composed only of members common to both s1 and s2. In
this way each successive mutation (or other genotypic change) will
fractionate the possibilities for the somatic adjustment of the
organism. And, should the one mutation require some somatic
change, the exact opposite of a change required by the other, the
possibilities for somatic adjustment may immediately be reduced to

The same argument must surely apply to multiple environmental
changes which demand somatic adjustments; and this will be true
even of those changes in environment which might seem to benefit
the organism. An improvement in diet, for example, will exclude
from the organism’s range of somatic adjustments those patterns of
growth which we would call “stunted” and which might be required
to meet some other exigency of the environment.

From these considerations it follows that if evolution proceeded
in accordance with conventional theory, its process would be
blocked. The finite nature of somatic change indicates that no
ongoing process of evolution can result only from successive
externally adaptive genotypic changes since these must, in
combination, become lethal, demanding combinations of internal
somatic adjustments of which the soma is incapable.

We turn therefore to a consideration of other classes of genotypic
change. What is required to give a balanced “theory of evolution is
the occurrence of genotypic changes which shall increase the
available range of somatic flexibility. When the internal organization
of the organisms of a species has been limited by environmental or
mutational pressure to some narrow subset of the total range of
living states, further evolutionary progress will require some sort of
genotypic change which will compensate for this limitation.

We note first that while the results of genotypic change are
irreversible within the life of the individual organism, the opposite is
usually true of changes which are achieved at the somatic level.
When the latter are produced in response to special environmental
conditions, a return of the environment to the previous norm is
usually followed by a diminution or loss of the characteristic. (We
may reasonably expect that the same would be true of those somatic
adjustments which must accompany an externally adaptive mutation

but, of course, it is impossible in this case to remove from the
individual the impact of the mutational change.)

A further point regarding these reversible somatic changes is of
special interest. Among higher organisms it is not unusual to find
that there is what we may call a “defense in depth” against
environmental demands. If a man is moved from sea level to 10,000
feet, he may begin to pant and his heart may race. But these first
changes are swiftly reversible: if he descends the same day, they will
disappear immediately. If, however, he remains at the high altitude,
a second line of defense appears. He will become slowly acclimated
as a result of complex physiological changes. His heart will cease to
race, and he will no longer pant unless he undertakes some special
exertion. If now he returns to sea level, the characteristics of the
second line of defense will disappear rather slowly and he may even
experience some discomfort.

From the point of view of an economics of somatic flexibility,
the first effect of high altitude is to reduce the organism to a limited
set of states (si) characterized by the racing of the heart and the
panting. The man can still survive, but only as a comparatively
inflexible creature. The later acclimation has precisely this value: it
corrects for the loss of flexibility. After the man is acclimated he can
use his panting mechanisms to adjust to other emergencies which
might otherwise be lethal.

A similar “defense in depth” is clearly recognizable in the field of
behavior. When we encounter a new problem for the first time, we
deal with it either by trial and error or possibly by insight. Later, and
more or less gradually, we form the “habit” of acting in the way
which earlier experience rewarded. To continue to use insight or trial
and error upon this class of problem would be wasteful. These
mechanisms can now be saved for other problems.3

Both in acclimation and in habit formation the economy of
flexibility is achieved by substituting a deeper and more enduring
change for a more superficial and more reversible one. In the terms
used above in discussing the anti-Lamarckian premise, a change has
occurred in the parameters of the functional equation linking rate of

3 Bateson, “Minimal Requirements for a Theory of Schizophrenia,” A.M.A.
Archives of General Psychiatry, 1960, 2: 447.

respiration to external atmospheric pressure. Here it seems that the
organism is behaving as we may expect any ultrastable system to
behave. Ashby4 has shown that it is a general formal characteristic
of such systems that those circuits con-trolling the more rapidly
fluctuating variables act as balancing mechanisms to protect the
ongoing constancy of those variables in which change is normally
slow and of small amplitude; and that any interference which fixes
the values of the changeful variables must have a disturbing effect
upon the constancy of the normally steady components of the
system. For the man who must constantly pant at high altitudes, the
respiration rate can no longer be used as a changeable quantity in the
maintaining of physiological balance. Conversely, if the respiration
rate is to become avail-able again as a rapidly fluctuating variable,
some change must occur among the more stable components of the
system. Such a change will, in the nature of the case, be achieved
comparatively slowly and be comparatively irreversible.

Even acclimation and habit formation are, however, still
reversible within the life of the individual, and this very reversibility
indicates a lack of communicational economy in these adaptive
mechanisms. Reversibility implies that the changed value of some
variable is achieved by means of homeostatic, error-activated
circuits. There must be a means of detecting an undesirable or
threatening change in some variable, and there must be a train of
cause and effect whereby corrective action is initiated. Moreover,
this entire circuit must, in some degree, be available for this
purpose for the entire time during which the reversible change is
maintained—a considerable using up of available message

The matter of communicational economics becomes still more
serious when we note that the homeostatic circuits of an organism
are not separate but complexly interlocked, e.g., hormonal
messengers which play a part in the homeostatic control of organ
A will also affect the states of organs B, C, and D. Any special
ongoing loading of the circuit controlling A will therefore
diminish the organism’s freedom to control B, C, and D.

4 W. R. Ashby, “The Effect of Controls on Stability,” Nature, 1945, 155: 242;
also Ashby, Design for a Brain, New York, John Wiley & Co., 1952.

In contrast, the changes brought about by mutation or other
genotypic change are presumably of a totally different nature. Every
cell contains a copy of the new genotypic corpus and therefore will
(when appropriate) behave in the changed manner, without any
change in the messages which it receives from surrounding tissues
or organs. If the hypothetical pregiraffes carrying the mutant gene
“long neck” could also get the gene “big heart,” their hearts would
be enlarged without the necessity of using the homeostatic pathways
of the body to achieve and maintain this enlargement. Such a
mutation will have survival value not be-cause it enables the
pregiraffe to supply its elevated head with sufficient blood, since
this was already achieved by somatic change but because it
increases the overall flexibility of the organism, enabling it to
survive other demands which may be placed upon it either by
environmental or, genotypic change.

It appears, then, that the process of biological evolution could be
continuous if there were a class of mutations or other genotypic
changes which would simulate Lamarckian inheritance. The
function of these changes would be to achieve by genotypic flat
those characteristics which the organism at the given time is already
achieving by the uneconomical method of somatic change. Such a
hypothesis, I believe, conflicts in no way with conventional theories
of genetics and natural selection. It does, however, somewhat alter
the current conventional picture of evolution as a whole, though
related ideas were put forward over sixty years ago. Baldwin5
suggested that we consider not only the operation of the external
environment in natural selection but also what he called “organic
selection” in which the fate of a given variation would depend upon
its physiologic viability. In the same article, Baldwin attributes to
Lloyd Morgan the suggestion that there might exist “coincident
variations” which would simulate Lamarckian inheritance (the so-
called “Baldwin effect”).

According to such a hypothesis, genotypic change in an organism
becomes comparable to legislative change in a society. The wise
legislator will only rarely initiate a new rule of behavior; more
usually he will confine himself to affirming in law that which has

5 J. M. Baldwin, “Organic Selection,” Science, 1897, 5: 634.


already become the custom of the people. An innovative rule can be
introduced only at the price of activating and perhaps overloading a
large number of homeostatic circuits in the society.

It is interesting to ask how a hypothetical process of evolution
would work if Lamarckian inheritance were the rule, i.e., if
characteristics achieved by somatic homeostasis were inherited. The
answer is simple: it would not work, for the following reasons:

(1)The question turns upon the concept of economy in the use
of homeostatic circuits, and it would be the reverse of economical
to fix by genotypic change all the variables which accompany a
given desirable and homeostatically achieved characteristic. Every
such characteristic is achieved by ancillary homeostatic changes all
around the circuits, and it is most undesirable that these ancillary
changes should be fixed by inheritance, as would logically happen
according to any theory involving an indiscriminate Lamarckian
inheritance. Those who would defend a Lamarckian theory must be
prepared to suggest how in the genotype an appropriate selection
can be achieved. Without such a selection, the inheritance of
acquired characteristics would merely in-crease the proportion of
nonviable genotypic changes.
(2)Lamarckian inheritance would disturb the relative timing of
the processes upon which evolution must—according to the present
hypothesis—depend. It is essential that there be a time lag between
the uneconomical but reversible somatic achievement of a given
characteristic and the economical but more enduring alterations of
the genotype. If we look upon every soma as a working model
which can be modified in various ways in the workshop, it is clear
that sufficient but not infinite time must be given for these
workshop trials before the results of these trials are incorporated
into the final blueprint for mass production. This delay is provided
by the indirection of stochastic process. It would be unduly
shortened by Lamarckian inheritance.

The principle involved here is general and by no means trivial. It
obtains in all homeostatic systems in which a given effect can be
brought about by means of a homeostatic circuit, which circuit can,
in turn, be modified in its characteristics by some-higher system of
control. In all such systems (ranging from the house thermostat to
systems of government and administration) it is important that the

higher system of control lag behind the event sequences in the
peripheral homeostatic circuit.

In evolution two control systems are present: the homeostases of
the body which deal with tolerable internal stress, and the action of
natural selection upon the (genetically) nonviable members of the
population. From an engineering point of view, the problem is to
limit communication from the lower, reversible somatic system to
the higher irreversible genotypic system.

Another aspect of the proposed hypothesis about which we can
only speculate is the probable relative frequency of the two classes
of genotypic change: those which initiate something new and those
which affirm some homeostatically achieved characteristic. In the
Metazoa and multicellular plants, we face complex networks of
multiple interlocking homeostatic circuits, and any given mutation
or gene recombination which initiates change will probably require
very various and multiple somatic characteristics to be achieved by
homeostasis. The hypothetical pregiraffe with the mutant gene “long
neck” will need to modify not only its heart and circulatory system
but also perhaps its semicircular canals, its intervertebral discs, its
postural reflexes, the ratio of length and thickness of many muscles,
its evasive tactics vis-a-vis predators, etc. This suggests that in such
complex organisms, the merely affirmative genotypic changes must
far outnumber those which initiate change, if the species is to avoid
that cul-de-sac in which the flexibility of the soma approaches zero.

Conversely, this picture suggests that most organisms, at any
given time, are probably in such a state that there are multiple
possibilities for affirmative genotypic change. If, as seems probable,
both mutation and gene redistribution are in some sense random
phenomena, at least the chances are considerable that one or other of
these multiple possibilities will be met.

Finally, it is appropriate to discuss what evidence is avail-able or
might be sought to support or disprove such a hypothesis. It is clear
at the outset that such a testing will be difficult. The affirmative
mutations upon which the hypothesis depends will usually be
invisible. From among the many members of a population which
are achieving a given adjustment to environmental circumstances by
somatic change, it will not be possible immediately to pick out those
few in which the same adjustment is provided by the genotypic


method. In such a case, the genotypically changed individuals will
have to be identified by breeding and raising the offspring under
more normal conditions.

A still greater difficulty arises in cases where we would
investigate those homeostatically acquired characteristics which are
achieved in response to some innovative genotypic change. It will
often be impossible, by mere inspection of the organism, to tell
which of its characteristics are the primary results of genotypic
change and which are secondary somatic adjustments to these. In the
imaginary case of the pregiraffe with a somewhat elongated neck
and an enlarged heart, it may be easy to guess that the modification
of the neck is genotypic while that of the heart is somatic. But all
such guesses will depend upon the very imperfect present knowledge
of what an organism can achieve in way of somatic adjustment.

It is a major tragedy that the Lamarckian controversy has
deflected the attention of geneticists away from the phenomenon of
somatic adaptability. After all, the mechanisms, thresholds, and
maxima of individual phenotypic change under stress must surely be
genotypically determined.

Another difficulty, of rather similar nature, arises at the
population level, where we encounter another “economics” of
potential change, theoretically distinguishable from that which
operates within the individual. The population of a wild species is
today conventionally regarded as genotypically heterogeneous in
spite of the high degree of superficial resemblance between the
individual phenotypes. Such a population expectably functions as a
storehouse of genotypic possibilities. The economic aspect of this
storehouse of possibilities has, for example, been stressed by
Simmonds.6 He points out that farmers and breeders who demand
100 per cent phenotypic uniformity in a highly select crop are in fact
throwing away most of the multiple genetic possibilities
accumulated through hundreds of generations in the wild
population. From this Simmonds argues that there is urgent need for
institutions which shall “conserve” this storehouse of variability by
maintaining unselected populations.

6 N. W. Simmonds, “Variability in Crop Plants, Its Use and Conservation,” Biol.
Review, 1962, 37: 422-62.


Lerner7 has argued that self-corrective or buffering mechanisms
operate to hold constant the composition of these mixtures of wild
genotypes and to resist the effects of artificial selection. There is
therefore at least a presumption that this economics of variability
within the population will turn out to be of the multiplicative kind.

Now, the difficulty of discriminating between a characteristic
achieved by somatic homeostasis and the same characteristic
achieved (more economically) by a genotypic short cut is clearly
going to be compounded when we come to consider populations
instead of physiologic individuals. All actual experimentation in the
field will inevitably work with populations, and, in this work, it will
be necessary to discriminate the effects of that economics of
flexibility which operates inside the individuals from the effects of
the economics of variability which operates at the population level.
These two orders of economics may be easy to separate in theory,
but to separate them in experimentation will surely be difficult.

Be all that as it may, let us consider what evidential sup-port may
be available for some of the propositions which are crucial to the

(1) That the phenomena of somatic adjustment are appropriately
described in terms of an economics of flexibility. In general, we
believe that the presence of stress A may reduce an organism’s
ability to respond to stress B and, guided by this opinion, we
commonly protect the sick from the weather. Those who have
adjusted to the office life may have difficulty in climbing
mountains, and trained mountain climbers may have difficulty
when confined to offices; the stresses of retirement from business
may be lethal; and so on. But scientific knowledge of these
matters, in man or other organisms, is very slight.
(2) That this economics of flexibility has the logical structure
described above—each successive demand upon flexibility
fractionating the set of available possibilities. The proposition is
expectable, but so far as I know there is no evidence for it. It is,
however, worthwhile to examine the criteria which determine
whether a given “economic” system is more appropriately
7 I. M. Lerner, Genetic Homeostasis, Edinburgh, Oliver and Boyd, 1954.


described in additive or multiplicative terms. There would seem to
be two such criteria:

(a)A system will be additive insofar as the units of its currency
are mutually interchangeable and, therefore, can-not meaningfully
be classified into sets such as were used earlier in this paper to
show that the economics of flexibility must surely be
multiplicative. Calories in the economics of energy are completely
interchangeable and unclassifiable, as are dollars in the individual
budget. Both these systems are therefore additive. The
permutations and combinations of variables which define the states
of an organism are classifiable and—to this extent—
noninterchangeable. The system is therefore multiplicative. Its
mathematics will resemble that of information theory or negative
entropy rather than that of money or energy conservation.

(b)A system will be additive insofar as the units of its currency
are mutually independent. Here there would seem to be a difference
between the economic system of the individual, whose budgetary
problems are additive (or sub-tractive) and those of society at large,
where the overall distribution or flow of wealth is governed by
complex (and perhaps imperfect) homeostatic systems. Is there,
perhaps, an economics of economic flexibility (a metaeconomics)
which is multiplicative and so resembles the economics of
physiological flexibility discussed above? Notice, however, that the
units of this wider economics will be not dollars but patterns of
distribution of wealth. Similarly, Lerner’s “genetic homeostasis,”
insofar as it is truly homeostatic, will have multiplicative character.

The matter is, however, not simple and we cannot expect that
every system will be either totally multiplicative or totally additive.
There will be intermediate cases which combine the two
characteristics. Specifically, where several independent alternative
homeostatic circuits control a single variable, it is clear that the
system may show additive characteristics—and even that it may pay
to incorporate such alternative pathways in the system provided they
can be effectively insulated from each other. Such systems of
multiple alternative controls may give survival advantage insofar as
the mathematics of addition and subtraction will pay better than the
mathematics of logical fractionation.

(3) That innovative genotypic change commonly makes demands
upon the adjustive ability of the soma. This proposition is
orthodoxly believed by biologists but cannot in the nature of the
case be verified by direct evidence.
(4) That successive genotypic innovations make multiplicative
demands upon the soma. This proposition (which involves both the
notion of multiplicative economics of flexibility and the notion
that each innovative genotypic change has its somatic price) has
several interesting and perhaps verifiable implications.
(a)We may expect that organisms in which numerous recent
genotypic changes have accumulated (e.g., as a result of selection,
or planned breeding) will be delicate, i.e., will need to be protected
from environmental stress. This sensitivity to stress is to be
expected in new breeds of domesticated animals and plants and
experimentally produced organisms carrying either several mutant
genes or unusual (i.e., recently achieved) genotypic combinations.

(b)We may expect that for such organisms further genotypic
innovation (of any kind other than the affirmative changes
discussed above) will be progressively deleterious.

(c)Such new and special breeds should become more resistant
both to environmental stress and to genotypic change, as selection
works upon successive generations to favor those individuals in
which “genetic assimilation of acquired characteristics” is achieved
(Proposition 5).

(5) That environmentally induced acquired characteristics may,
under appropriate conditions of selection, be replaced by similar
characteristics which are genetically determined. This phenomenon
has been demonstrated by Waddington8 for the bithorax phenotypes
of Drosophila.
He calls it the “genetic assimilation of acquired characteristics.”
Similar phenomena have also probably occurred in various
experiments when the experimenters set out to prove the inheritance
of acquired characteristics but did not achieve this proof through
failure to control the conditions of selection. We have, however, no
evidence at all as to the frequency of this phenomenon of genetic

8 C. H. Waddington, “Genetic Assimilation of an Acquired Character,”
Evolution, 1953, 7: 118; also Waddington, The Strategy of Genes, London,
Allen and Unwin, 1957.


assimilation. It is worth noting, however, that, according to the
arguments of this essay, it may be impossible, in principle, to
exclude the factor of selection from experiments which would test
“the inheritance of acquired characteristics.” It is precisely my thesis
that the simulation of Lamarckian inheritance will have survival
value under circumstance of undefined or multiple stress.

(6). That it is, in general, more economical of flexibility to
achieve a given characteristic by genotypic than by soma-tic change.
Here the Waddington experiments do not throw any light, because it
was the experimenter who did the selecting. To test this proposition,
we need experiments in which the population of organisms is placed
under double stress: (a) that stress which will induce the characteristic
in which we are interested, and (b) a second stress which will
selectively decimate the population, favoring, we hope, the survival
of those individuals whose flexibility is more able to meet this
second stress after adjusting to the first. According to the hypothesis,
such a system should favor those individuals which achieve their
adjustment to the first stress by genotypic process.

(7) Finally, it is interesting to consider a corollary which is the
converse of the thesis of this essay. It has been argued here that
simulated Lamarckian inheritance will have survival value when the
population must adjust to a stress which remains constant over
successive generations. This case is in fact the one which has been
examined by those who would demonstrate an inheritance of
acquired characteristics. A converse problem is presented by those
cases in which a population faces a stress which changes its
intensity unpredictably and rather often—perhaps every two or
three generations. Such situations are perhaps very rare in nature,
but could be produced in the laboratory.
Under such variable circumstances, it might pay the organisms
in survival terms to achieve the converse of the genetic
assimilation of acquired characteristics. That is, they might
profitably hand over to somatic homeostatic mechanisms the
control of some characteristic which had previously been more
rigidly controlled by the genotype.

It is evident, however, that such experimentation would be very
difficult. Merely to establish the genetic assimilation of such
characteristics as bithorax requires selection on an astronomical

scale, the final population in which the genetically determined
bithorax individuals can be found being a selected sample from a
potential population of something like 1050
. or 1060 individuals. It is
very doubtful whether, after this selective process, there would still
. exist in the sample enough genetic heterogeneity to undergo a
further converse selection favoring those individuals which still
achieve their bithorax phenotype by somatic means.

Nevertheless, though this converse corollary is possibly not
demonstrable in the laboratory, something of the sort seems to
operate in the broad picture of evolution. The matter may be
presented in dramatic form by considering the dichotomy between
“regulators” and “adjusters.”9 Prosser proposes that where internal
physiology contains some variable of the same dimensions as some
external environmental variable, it is convenient to classify
organisms according to the degree to which they hold the internal
variable constant in spite of changes in the external variable. Thus,
the homoiothermic animals are classified as “regulators” in regard
to temperature while the poikilothermic are “adjusters.” The same
dichotomy can be applied to aquatic animals according to how
they handle internal and external osmotic pressure.

We usually think of regulators as being in some broad
evolutionary sense “higher” than adjusters. Let us now consider
what this might mean. If there is a broad evolutionary trend in favor
of regulators, is this trend consistent with what has been said above
about the survival benefits which accrue when control is transferred
to genotypic mechanisms?

Clearly, not only the regulators but also the adjusters must rely
upon homeostatic mechanisms. If life is to go on, a large number of
essential physiological variables must be held within narrow limits.
It the internal osmotic pressure, for example, is allowed to change,
there must be mechanisms which will defend these essential
variables. It follows that the difference between adjusters and
regulators is a matter of where, in the complex network of
physiologic causes and effects, homeostatic process operates.

9 C. L. Prosser, “Physiological Variation in Animals,” Biol. Review, 1955, 30:

In the regulators, the homeostatic processes operate at or close to
the input and output points of that network which is the individual
organism. In the adjusters, the environmental variables are permitted
to enter the body and the organism must then cope with their effects,
using mechanisms which will involve deeper loops of the total

In terms of this analysis, the polarity between adjusters and
regulators can be extrapolated another: step to include what we may
call “extraregulators” which achieve homeostatic controls outside
the body by changing and controlling the environment—man being
the most conspicuous example of this class.

In the earlier part of this essay, it was argued that in adjusting to
high altitude there is a benefit to be obtained, in terms of an
economics of flexibility, by shifting from, e.g., panting to the more
profound and less reversible changes of acclimation; that habit is
more economical than trial and error; and that genotypic control
may be more economical than acclimation. These are all centripetal
changes in the location of control.

In the broad picture of evolution, however, it seems that the trend
is in the opposite direction: that natural selection, in the long run,
favors regulators more than adjusters, and extraregulators more than
regulators. This seems to indicate that there is a long time
evolutionary advantage to be gained by centrifugal shifts in the
locus of control.

To speculate about problems so vast is perhaps romantic, but it is
worth noting that this contrast between the overall evolutionary
trend and the trend in a population faced with constant stress is
what we might expect from the converse corollary here being
considered. If constant stress favors centripetal shift in the locus of
control, and variable stress favors centrifugal shift, then it should
follow that in the vast spans of time and change which determine
the broad evolutionary picture, centrifugal shift of control will be


In this essay the author uses a deductive approach. Starting from
premises of conventional physiology and evolutionary theory and
applying to these the arguments of cybernetics, he shows that there
must be an economics of somatic flexibility and that this economics
must, in the long run, be coercive upon the evolutionary process.
External adaptation by mutation or genotypic reshuffling, as
ordinarily thought of, will inevitably use up the available somatic
flexibility. It follows—if evolution is to be continuous—that there
must also be a class of genotypic changes which will confer a bonus
of somatic flexibility.

In general, the somatic achievement of change is uneconomical
because the process depends upon homeostasis, i.e., upon whole
circuits of interdependent variables. It follows that inheritance of
acquired characteristics would be lethal to the evolutionary system
because it would fix the values of these variables all around the
circuits. The organism or species would, however, benefit (in
survival terms) by genotypic change which would simulate
Lamarckian inheritance, i.e., would bring about the adaptive
component of somatic homeostasis without involving the whole
homeostatic circuit. Such a genotypic change (erroneously called the
“Bald-win effect”) would confer a bonus of somatic flexibility and
would therefore have marked survival value.

Finally, it is suggested that a contrary argument can be applied in
those cases where a population must acclimate to variable stress.
Here natural selection should favor an anti-Baldwin effect.


Problems in Cetacean and Other
Mammalian Communication*

The Communication of Preverbal Mammals*

Of the Cetacea I have had little experience. I once dissected in
the Cambridge Zoological Laboratories a specimen of Phocoena
bought from the local fishmonger, and did not really encounter
cetaceans again until this year, when I had an opportunity to meet
Dr. Lilly’s dolphins. I hope that my discussion of some of the
questions that are in my mind as I approach these peculiar mammals
will assist you in examining either these or related questions.

My previous work in the fields of anthropology, animal ethology,
and psychiatric theory provides a theoretical framework for the
transactional analysis of behavior. The premises of this theoretical
position may be briefly summarized: (1) that a relationship between
two (or more) organisms is, in-fact, a sequence of S-R sequences
(i.e.,. of contexts in which proto-learning occurs) ; (2) that deuterolearning
(i.e., learning to learn) is, in fact, the acquiring of
information about the contingency patterns of the contexts in which
proto-learning occurs; and (3) that the “character” of the organism is
the aggregate of its deutero-learning and therefore reflects the
contextual patterns of past protolearning.10

* This article appeared as Chapter 25, pp. 569-799, in Whales, Dolphins
and Porpoises, edited by Kenneth S. Norris, University of California Press, 1966.
Reprinted by permission of The Regents of the University of California.
10 J. Ruesch and G. Bateson, Communication: The Social Matrix of Psychiatry,
New York, Norton, 1951.

These premises are essentially a hierarchic structuring of learning
theory along lines related to Russell’s Theory of Logical Types.11
The premises, following the Theory of Types, are primarily
appropriate for the analysis of digital communication. To what
extent they may be applicable to analogic communication or to
systems that combine the digital with the analogic is problematic. I
hope that the study of dolphin communication will throw light on
these fundamental problems. The point is not either to discover that
dolphins have complex language or to teach them English, but to
close gaps in our theoretical knowledge of communication by
studying a system that, whether rudimentary or complex, is almost
certainly of a totally unfamiliar kind.

Let me start from the fact that the dolphin is a mammal. This fact
has, of course, all sorts of implications for anatomy and physiology,
but it is not with these that I am concerned. I am interested in his
communication, in what is called his “behavior,” looked at as an
aggregate of data perceptible and meaningful to other members of
the same species. It is meaningful, first, in the sense that it affects a
recipient animal’s behavior, and, second, in the sense that
perceptible failure to achieve appropriate meaning in the first sense
will affect the behavior of both animals. What I say to you may be
totally ineffective, but my ineffectiveness, if perceptible, will affect
both you and me. I stress this point because it must be remembered
that in all relationships between man and some other animal,
especially when that animal is a dolphin, a very large proportion of
the behavior of both organisms is determined by this kind of

When I view the behavior of dolphins as communication, the
mammalian label implies, for me, something very definite. Let me
illustrate what I have in mind by an example from Benson
Ginsburg’s wolf pack in the Brookfield Zoo.

Among the Canidae, weaning is performed by the mother. When
the puppy asks for milk, she presses down with her open mouth on
the back of his neck, crushing him down to the ground. She does
this repeatedly until he stops asking. This method is used by

11 A. N. Whitehead and B. Russell, Principia Mathematica, London, Cambridge
University Press, 1910.


coyotes, dingoes, and the domestic dog. Among wolves the system
is different. The puppies graduate smoothly from the nipple to
regurgitated food. The pack comes back to the den with their bellies
full. All regurgitate what they have got and all eat together. At some
point the adults start to wean the puppies from these meals, using
the method employed by the other Canidae; the adult crushes the
puppy down by pressing its open mouth on the back of the puppy’s
neck. In the wolf this function is not confined to the mother, but is
performed by adults of both sexes.

The pack leader of the Chicago pack is a magnificent male
animal who endlessly patrols the acre of land to which the pack is
confined. He moves with a beautiful trot that appears tireless,
while the other eight or nine members of the pack spend most of
their time dozing. When the females come in heat they usually
proposition the leader, bumping against him with their rear ends.
Usually, however, he does not respond, though he does act to
prevent other males from getting the females. Last year one of
these males succeeded in establishing coitus with a female. As in
the other Canidae, the male wolf is locked in the female, unable to
withdraw his penis, and this animal was helpless. Up rushed the
pack leader. What did he do to the helpless male who dared to
infringe the leader’s prerogatives? Anthropomorphism would
suggest that he would tear the helpless male to pieces. But no. The
film shows that he pressed down the head of the offending male
four times with his open jaws and then simply walked away.

What are the implications for research from this illustration?
What the pack leader does is not describable, or only insufficiently
described, in S-R terms. He does not “negatively reinforce” the other
male’s sexual activity. He asserts or affirms the nature of the
relationship between himself and the other. If we were to translate
the pack leader’s action into words, the words would not be “Don’t
do that.” Rather, they would translate the metaphoric action: “I am
your senior adult male, you puppy!” What I am trying to say about
wolves in particular, and about preverbal mammals in general, is
that their discourse is primarily about the rules and the contingencies
of relationship.

Let me offer a more familiar example to help bring home to you
the generality of this view, which is by no means orthodox among

ethologists. When your cat is trying to tell you to give her food,
how does she do it? She has no word for food or for milk. What
she does is to make movements and sounds that are
characteristically those that a kitten makes to a mother cat. If we
were to translate the cat’s message into words, it would not be
correct to say that she is crying “Milk!” Rather, she is saying
something like “Ma-ma!” Or, perhaps still more correctly, we
should say that she is asserting “Dependency! Dependency!” The
cat talks in terms of patterns and contingencies of relationship, and
from this talk it is up to you to take a deductive step, guessing that
it is milk that the cat wants. It is the necessity for this deductive
step which marks the difference between preverbal mammalian
communication and both the communication of bees and the
languages of men.

What was extraordinary—the great new thing—in the evolution
of human language was not the discovery of abstraction or
generalization, but the discovery of how to be specific about
something other than relationship. Indeed, this discovery, though it
has been achieved, has scarcely affected the behavior even of
human beings. If A says to B, “The plane is scheduled to leave at
6.30,” B rarely accepts this remark as simply and solely a
statement of fact about the plane. More often he devotes a few
neurons to the question, “What does A’s telling me this indicate for
my relationship to A?” Our mammalian ancestry is very near the
surface, despite recently acquired linguistic tricks.

Be that as it may, my first expectation in studying dolphin
communication is that it will prove to have the general mammalian
characteristic of being primarily about relationship. This premise is
in itself perhaps sufficient to account for the sporadic development
of large brains among mammals. We need not complain that, as
elephants do not talk and whales invent no mousetraps, these
creatures are not overtly intelligent. All that is needed is to suppose
that large-brained creatures were, at some evolutionary stage,
unwise enough to get into the game of relationship and that, once
the species was caught in this game of interpreting its members’
behavior toward one another as relevant to this complex and vital
subject, there was survival value for those individuals who could
play the game with greater ingenuity or greater wisdom. We may,


then, reasonably expect to find a high complexity of
communication about relationship among the Cetacea. Because
they are mammals, we may expect that their communication will
be about, and primarily in terms of, patterns and contingencies of
relationship. Be-cause they are social and large-brained, we may
expect a high degree of complexity in their communication.

Methodological Considerations

The above hypothesis introduces very special difficulties into the
problem of how to test what is called the “psychology” (e.g.,
intelligence, ingenuity, discrimination, etc.) of individual animals. A
simple discrimination experiment, such as has been run in the Lilly
laboratories, and no doubt elsewhere, involves a series of steps: (1)
The dolphin may or may not perceive a difference between the
stimulus objects, X and Y. (2) The dolphin may or may not perceive
that this difference is a cue to behavior. (3) The dolphin may or may
not perceive that the behavior in question has a good or bad effect
upon reinforcement, that is, that doing “right” is conditionally
followed by fish. (4) The dolphin may or may not choose to do
“right,” even after he knows which is right. Success in the first three
steps merely provides the dolphin with a further choice point. This
extra degree of freedom must be the first focus of our investigations.

It must be our first focus for methodological reasons. Consider
the arguments that are conventionally based upon experiments of
this kind. We argue always from the later steps in the series to the
earlier steps. We say, “If the animal was able to achieve step 2 in our
experiment, then he must have been able to achieve step 1.” If he
could learn to behave in the way that would bring him the reward,
then he must have had the necessary sensory acuity to discriminate
between X and Y, and so on.

Precisely because we want to argue from observation of the
animal’s success in the later steps to conclusions about the more
elementary steps, it becomes of prime importance to know whether
the organism with which we are dealing is capable of step 4. If it is
capable, then all arguments about steps 1 through 3 will be

invalidated unless appropriate methods of controlling step 4 are
built into the experimental design. Curiously enough, though
human beings are fully capable of step 4, psychologists working
with human subjects have been able to study steps 1 through 3
without taking special care to exclude the confusions introduced
by this fact. If the human subject is “cooperative and sane,” he
usually responds to the testing situation by repressing most of his
impulses to modify his behavior according to his personal view of
his relationship to the experimenter. The words cooperative and
sane imply a degree of consistency at the level of step 4. The
psychologist operates by a sort of petitio principii: if the subject is
cooperative and sane (i.e., if the relational rules are fairly
constant), the psychologist need not worry about changes in those

The problem of method becomes entirely different when the
subject is noncooperative, psychopathic, schizophrenic, a naughty
child, or a dolphin. Perhaps the most fascinating characteristic of
this animal is derived precisely from his ability to operate at this
relatively high level, an ability that is still to be demonstrated.

Let me now consider for a moment the art of the animal trainer.
From conversations with these highly skilled people —trainers of
both dolphins and guide dogs—my impression is that the first
requirement of a trainer is that he must be able to prevent the animal
from exerting choice at the level of step 4. It must continually be
made clear to the animal that, when he knows what is the right thing
to do in a given context, that is the only thing he can do, and no nonsense
about it. In other words, it is a primary condition of circus
success that the animal shall abrogate the use of certain higher levels
of his intelligence. The art of the hypnotist is similar.

There is a story told of Dr. Samuel Johnson. A silly lady made
her dog perform tricks in his presence. The Doctor seemed
unimpressed. The lady said, “But Dr. Johnson, you don’t know how
difficult it is for the dog.” Dr. Johnson re-plied, “Difficult, madam?
Would it were impossible!”

What is amazing about circus tricks is that the animal can
abrogate the use of so much of his intelligence and still have
enough left to perform the trick. I regard the conscious intelligence


as the greatest ornament of the human mind. But many authorities,
from the Zen masters to Sigmund Freud, have stressed the
ingenuity of the less conscious and perhaps more archaic level.

Communication About Relationship

As I said earlier, I expect dolphin communication to be of an
almost totally unfamiliar kind. Let me expand on this point. As
mammals, we are familiar with, though largely unconscious of, the
habit of communicating about our relationships. Like other
terrestrial mammals, we do most of our communicating on this
subject by means of kinesic and paralinguistic signals, such as
bodily movements, involuntary tensions of voluntary muscles,
changes of facial expression, hesitations, shifts in tempo of speech
or movement, overtones of the voice, and irregularities of
respiration. If you want to know what the bark of a dog “means,”
you look at his lips, the hair on the back of his neck, his tail, and so
on. These “expressive” parts of his body tell you at what object of
the environment he is barking, and what patterns of relationship to
that object he is likely to follow in the next few seconds. Above all,
you look at his sense organs: his eyes, his ears, and his nose.

In all mammals, the organs of sense become also organs for the
transmission of messages about relationship. A blind man makes us
uncomfortable, not because he cannot see that is his problem and we
are only dimly aware of it—but because he does not transmit to us
through the movement of his eyes the messages we expect and need
so that we may know and be sure of the state of our relationship to
him. We shall not know much about dolphin communication until
we know what one-dolphin can read in another’s use, direction,
volume, and pitch of echolocation.

Perhaps it is this lack in us which makes the communication of
dolphins seem mysterious and opaque, but I suspect a more

profound explanation. Adaptation to life in the ocean has stripped
the whales of facial expression. They have no external ears to flap
and few if any erectile hairs. Even the cervical vertebrae are fused
into a solid block in many species, and evolution has streamlined the
body, sacrificing the expressiveness of separate parts to the
locomotion of the whole. Moreover, conditions of life in the sea are
such that even if a dolphin had a mobile face, the details of his expression
would be visible to other dolphins only at rather short
range, even in the clearest waters.

It is reasonable, then, to suppose that in these animals
vocalization has taken over the communicative functions that most
animals perform by facial expression, wagging tails, clenched fists,
supinated hands, flaring nostrils, and the like. We might say that the
whale is the communicational opposite of the giraffe; it has no neck,
but has a voice. This speculation alone would make the
communication of dolphins a subject of great theoretical interest. It
would be fascinating, for example, to know whether or not, in an
evolutionary shift from kinesics to vocalization, the same general
structure of categories is retained.

My own impression—and it is only an impression unsupported
by testing—is that the hypothesis that dolphins have substituted
paralinguistics for kinesics does not quite fit in with my experience
when I listen to their sounds. We terrestrial mammals are familiar
with paralinguistic communication; we use it ourselves in grunts and
groans, laughter and sobbing, modulations of breath while speaking,
and so on. Therefore we do not find the paralinguistic sounds of
other mammals totally opaque. We learn rather easily to recognize in
them certain kinds of greeting, pathos, rage, persuasion, and
territoriality, though our guesses may often be wrong. But when we
hear the sounds of dolphins we cannot even guess at their
significance. I do not quite trust the hunch that would explain the
sounds of dolphins as merely an elaboration of the paralinguistics of
other mammals. (To argue thus from our inability is, however,
weaker than to argue from what we can do.)

I personally do not believe that the dolphins have any-thing that a
human linguist would call a “language.” I do not think that any
animal without hands would be stupid enough to arrive at so
outlandish a mode of communication.


To use a syntax and category system appropriate for the
discussion of things that can be handled, while really discussing
the patterns and contingencies of relationship, is fantastic. But that,
I submit, is what is happening in this room. I stand here and talk
while you listen and watch. I try to convince you, try to get you to
see things my way, try to earn your respect, try to indicate my
respect for you, challenge you, and so on. What is really taking
place is a discussion of the patterns of our relationship, all
according to the rules of a scientific conference about whales. So it
is to be human.

I simply do not believe that dolphins have language in this
sense. But I do believe that, like ourselves and other mammals,
they are preoccupied with the patterns of their relationships. Let us
call this discussion of patterns of relationship the t function of the
message. After all, it was the cat who showed us the great
importance of this function by her mewing. Preverbal mammals
communicate about things, when they must, by using what are
primarily µ-function signals. In contrast, human beings use
language, which is primarily oriented toward things, to discuss
relationships. The cat asks for milk by saying “Dependency,” and I
ask for your attention and perhaps respect by talking about whales.
But we do not know that dolphins, in their communication, resemble
either me or the cat. They may have a quite different

Analogic versus Digital Communication

There is another side of the problem. How does it happen that the
paralinguistics and kinesics of men from strange cultures, and even
the paralinguistics of other terrestrial mammals, are at least partly
intelligible to us, whereas the verbal languages of men from strange
cultures seem to be totally opaque? In this respect it would seem that
the vocalizations of the dolphin resemble human language rather
than the kinesics or paralinguistics of terrestrial mammals.

We know, of course, why gestures and tones of voice are partly
intelligible while foreign languages are unintelligible. It is because

language is digital and kinesics and paralinguistics are analogic.12
The essence of the matter is that in digital communication a number
of purely conventional signs -1 , 2, 3, X, Y, and so on—are pushed
around according to rules called algorithms. The signs themselves
have no simple connection (e.g., correspondence of magnitude) with
what they stand for. The numeral “5” is not bigger than the numeral
“3.” It is true that if we remove the crossbar from “7” we obtain the
numeral “1”; but the crossbar does not, in any sense, stand for “6.” A
name usually has only a purely conventional or arbitrary connection
with the class named. The numeral “5” is only the name of a
magnitude. It is non-sense to ask if my telephone number is larger
than yours, because the telephone exchange is a purely digital computer.
It is not fed with magnitudes, but only with names of positions
on a matrix.

In analogic communication, however, real magnitudes are used,
and they correspond to real magnitudes in the subject of discourse.
The linked range finder of a camera is a familiar example of an
analogue computer. This device is fed with an angle that has real
magnitude and is, in fact, the angle that the base of the range finder
subtends at some point on the object to be photographed. This angle
controls a cam that in turn moves the lens of the camera forward or
back. The secret of the device lies in the shape of the cam, which is
an analogic representation (i.e., a picture, a Cartesian graph) of the
functional relationship between distance of object and distance of

Verbal language is almost (but not quite) purely digital. The word
“big” is not bigger than the word “little”; and in general there is
nothing in the pattern (i.e., the system of interrelated magnitudes) in
the word “table” which would correspond to the system of
interrelated magnitudes in the object denoted. On the other hand, in
kinesic and paralinguistic communication, the magnitude of the
gesture, the loudness of the voice, the length of the pause, the

12 The difference between digital and analogic modes of communication may
perhaps be made clear by thinking of an English-speaking mathematician confronted
with a paper by a Japanese colleague. He gazes uncomprehendingly at the Japanese
ideographs, but he is able partly to understand the Cartesian graphs in the Japanese
publication. The ideographs, though they may originally have been analogic pictures,
are now purely digital; the Cartesian graphs are analogic.

tension of the muscle, and so forth—these magnitudes commonly
correspond (directly or inversely) to magnitudes in the relationship
that is the subject of discourse. The pattern of action in the
communication of the wolf pack leader is immediately intelligible
when we have data about the weaning practices of the animal, for
the weaning practices are themselves analogic kinesic signals.

It is logical, then, to consider the hypothesis that the
vocalization of dolphins may be a digital expression of µ
functions. It is this possibility that I especially have in mind in
saying that this communication may be of an almost totally
unfamiliar kind. Man, it is true, has a few words for µ functions,
words like “love,” “respect,” “dependency,” and so on. But these
words function poorly in the actual discussion of relationship
between participants in the relationship. If you say to a girl, “I love
you,” she is likely to pay more attention to the accompanying
kinesics and paralinguistics than to the words themselves.

We humans become very uncomfortable when somebody starts
to interpret our postures and gestures by translating them into
words about relationship. We much prefer that our messages on
this subject remain analogic, unconscious, and involuntary. We
tend to distrust the man who can simulate messages about
relationship. We therefore have no idea what it is like to be a
species with even a very simple and rudimentary digital system
whose primary subject matter would be µ functions. This system is
something we terrestrial mammals cannot imagine and for which
we have no empathy.

Research Plans

The most speculative part of my paper is the discussion of plans
for the testing and amplification of such a body of hypotheses. I
shall be guided by the following heuristic assumptions:

(1) The epistemology in whose terms the hypotheses are
constructed is itself not subject to testing. Derived from Whitehead

and Russell,13 it serves to guide our work. Should the work prove
rewarding, the success will be only a weak verification of the

(2)We do not even know what a primitive digital system for the
discussion of patterns of relationship might look like, but we can
guess that it would not look like a “thing” language. (It might, more
probably, resemble music.) I shall therefore not expect the
techniques for cracking human linguistic codes to be immediately
applicable to the vocalization of dolphins.
(3)The first requirement, then, is to identify and to classify the
varieties and the components of relationship existing among the
animals through detailed ethological study of their actions,
interactions, and social organization. The elements of which these
patterns are built are doubtless still present in the kinesics and
actions of the species. We there-fore begin with a listing of the
kinesic signals of individual dolphins, and then try to relate them to
the contexts in which they are used.
(4)No doubt, just as the pack leader’s behavior tells us that
“dominance” among wolves is metaphorically related to weaning,
so also the dolphins will tell us their kinesic metaphors for
“dominance,” “dependency,” and other µ functions. Gradually this
system of signals will fit together piece by piece to form a picture
of the varieties of relationship existing even among animals
arbitrarily confined together in a tank.
(5)As we begin to understand the metaphor system of the
dolphin, it will become possible to recognize and classify the
contexts of his vocalization. At this point the statistical techniques
for cracking codes may conceivably become useful.
The assumptions regarding the hierarchic structure of the
learning process—upon which this whole paper is based —provide
the basis for various kinds of experimentation. The contexts of
proto-learning may be variously constructed with a view to
observing in what types of contexts certain types of learning most
readily occur. We shall pay special attention to those contexts that
involve either relationships between two or more animals and one
person, or relationships between two or more people and one

13 ‘Whitehead and Russell, op. cit.

animal. Such contexts are miniature models of social organization
within which the animal may be expected to show characteristic
behaviors and to make characteristic attempts to modify the
context (i.e., to manipulate the humans).


Mr. Wood: In the course of twelve years in Marine Studios in
Florida, I spent a great deal of time watching what was perhaps the
most natural assemblage of Tursiops in captivity, including animals
of various ages, usually two or more of them in the process of
growing up, and I saw remarkably little of what you are going to
look for in a much more restricted group of animals in the Virgin

One time I saw something very interesting. Early one morning
about six or six-thirty, over a period of at least half an hour, the adult
male assumed a position next to one of the females in the tank who
was hanging motionless in the cur-rent. He would go up
occasionally and move away and then come back and assume a
position beside her, and he would stroke her side with his right
flipper repeatedly. There was no indication that this had sexual
significance. There was no erection on the part of the male, and no
observable response on the part of the female. But it was as clear-cut
a nonvocal signal as I ever observed in the tank.

Mr. Bateson: I would like to say that the amount of signaling that
goes on is much greater than is evident at first sight. There are, of
course, the rather specific kinds of signals which are very important.
I am not denying that. I mean the touching, and so on. But the shy
individual, the traumatized female, staying almost stationary three
feet be-low the surface while two other individuals fool around, is
getting a great deal of attention just by sitting there and staying. She
may not be actively transmitting, but in this business of bodily
communication, you don’t have to be actively transmitting in order
to have your signals picked up by other people. You can just be, and
just by being she attracts an enormous amount of attention from
these other two individuals who come over, pass by, pause a little as

they pass, and so on. She is, we would say, “withdrawn,” but she is
actually about as withdrawn as a schizophrenic who by being
withdrawn becomes the center of gravity of the family. All other
members of the group move around the fact of her withdrawal,
which she never lets them for-get.

Dr. Ray: I tend to agree with Mr. Bateson. We are working at the
New York Aquarium with the beluga whale, and I believe these
animals are much more expressive than we like to suspect. I think
one of the reasons they don’t do very much in captivity is that they
are bored to tears most of the time. There is nothing much of interest
in their tank environment, and I would like to suggest that we have
to manipulate their captivity much more cleverly than we do. I don’t
mean handling the whales. They don’t like that. But the introduction
of different types of animals, or clever little things that we might do
would get them to respond more. Captive cetaceans are like
monkeys in a cage. They are highly intelligent and highly
developed, and they are bored.

Another factor is our skill in observation, and in the beluga
whale, at least, we have been able to notice visually the sounds they
are making by watching the change in the shape of the melon, which
is extremely marked in this animal. It can swell on one side or the
other, or take several different shapes correlated with sound
production. So, by very careful observation and/or skilled
manipulation, I think a great deal can be done with these animals
rather simply.

Mr. Bateson: I had meant to point out that all sense organs
among mammals, and even among ants, become major organs for
the transmission of messages, such as, “Where are the other fellow’s
eyes focused?” and, “Are his pinnae focused in one direction or
another?” In this way sense organs become transmitting organs for

One of the things we must absolutely acquire if we are going to
understand dolphins is a knowledge of what one animal knows and
can read from another animals’ use of sonar. I suspect the presence
of all sorts of courtesy rules in this business; it probably isn’t polite
to sonar scan your friends too much, just as among human beings it
is not polite, really, to look at another’s feet in detail. We have many


taboos on observing one anothers’ kinesics, because too much
information can be got in that way.

Dr. Purves: It seems to me that the dolphin or the cetacean must
suffer from an even greater disadvantage than man has in the past,
because—I have forgotten the authority—it has been said that the
origin of human speech is an analogue language. In other words, if
you use the word “down,” you lower the hand and lower the lower
jaw at the same time. If you say “up,” you raise the hand and raise
the lower jaw. And if you use the word “table,” and, better still,
pronounce it in French, your mouth widens out and you make a
horizontal gesture. However complicated the human language is, it
has its origin in an analogue language. The poor porpoise has
nothing like this to start from. So he must have been highly
intelligent to have developed a communication system completely
de novo.

Mr. Bateson: What has happened to this creature is that the
information we get visually and the other terrestrial animals get
visually must have been pushed into voice. I still maintain that it is
appropriate for us to start by investigating what is left of the visual

A Re-examination of “Bateson’s Rule”*


Nearly eighty years ago, my father, William Bateson, be-came
fascinated by the phenomena of symmetry and metameric regularity
as exhibited in the morphology of animals and plants. It is difficult
today to define precisely what he was after, but, broadly, it is clear
that he believed that an entirely new concept of the nature of living
things would develop from the study of such phenomena. He held,
no doubt correctly, that natural selection could not be the only
determinant of the direction of evolutionary change and that the
genesis of variation could not be a random matter. He therefore set
out to demonstrate regularity and “lawfulness” among the
phenomena of variability.

In his attempt to demonstrate a sort of order which the biologists
of his day had largely ignored, he was guided by the notion, never
clearly formulated, that the place to look for regularity in variation
would be precisely where variation had its impact upon what was
already regular and repetitive. The phenomena of symmetry and
metamerism, themselves strikingly regular, must surely have been
brought about by regularities or “laws” within the evolutionary
process and, therefore, the variations of symmetry and metamerism
should precisely exemplify these laws at work.

In the language of today, we might say that he was groping for
those orderly characteristics of living things which illustrate the fact
that organisms evolve and develop with-in cybernetic,
organizational, and other communicational limitations.

It was for this study that he coined the word “genetics.”14

He set out to examine the material in the world’s museums,
private collections, and journals bearing upon the teratology of

* This essay has been accepted for publication in the Journal of Genetics, and
is here reproduced with the permission of that journal.

animal symmetry and metamerism. The de-tails of this survey were
published in a large book15 which is still of considerable interest.

To demonstrate regularity within the field of teratological
variation, he attempted a classification of the various sorts of
modification that he encountered. With this classification I am not
here concerned, except that in the survey he happened upon a
generalization which can be called a “discovery.” This discovery
came to be called “Bateson’s Rule” and remains one of the
unexplained mysteries of biology.

The purpose of the present note is to place Bateson’s Rule in a
new theoretical perspective determined by cybernetics, information
theory, and the like.

Briefly, Bateson’s Rule asserts in its simplest form that when an
asymmetrical lateral appendage (e.g., a right hand) is reduplicated,
the resulting reduplicated limb will be bilaterally symmetrical,
consisting of two parts each a mirror image of the other and so
placed that a plane of symmetry could be imagined between them.

He himself was, however, very doubtful whether such simple
reduplication ever occurs. He believed and accumulated evidence to
show that, in a very large proportion of such cases, one component
of the reduplicated system was it-self double. He asserted that in
such systems the three components are normally in one plane; that
the two components of the doublet are mirror images of each other;
and that that component of the doublet which is the nearer to the
primary appendage is a mirror image of the primary.

This generalization was shown by my father to hold for a very
large number of examples of reduplication in the vertebrates and in
arthropods, and for a few cases in other phyla where the museum
material was, of course, more scarce.

Ross Harrison16 believed that Bateson underestimated the
importance of simple reduplication.

14 1W. Bateson, “The Progress of Genetic Research,” In-augural Address, Royal
Horticultural Society Report, 1906.

15 ‘W. Bateson, Materials for the Study of Variation, London, Macmillan and Co.,

16 R. G. Harrison, “On Relations of Symmetry in Transplanted Limbs,”

Journal of Experimental Zoology, 1921, 32: 1-118.


Whether or not simple reduplication is a real and common
phenomenon, I shall begin this essay with a discussion of the logical
problems which it would present.

The Problem Redefined

In 1894, it appeared that the problem centered around the
question: What causes the development of bilateral symmetry in a
context where it does not belong?

But modern theory has turned all such questions upside down.
Information, in the technical sense, is that which excludes certain
alternatives. The machine with a governor does not elect the steady
state; it prevents itself from staying in any alternative state; and in
all such cybernetic systems, corrective action is brought about by
difference. In the jargon of the engineers, the system is “error
activated.” The difference between some present state and some
“preferred” state activates the corrective response.

The technical term.
“information” may be succinctly de-fined as
any difference which makes a difference in some later event. This
definition is fundamental for all analysis of cybernetic systems and
organization. The definition links such analysis to the rest of
science, where the causes of events are commonly not differences
but forces, impacts, and the like. The link is classically exemplified
by the heat engine, where available energy (i.e., negative entropy) is
a function of a difference between two temperatures. In this classical
instance, “information” and “negative entropy” overlap.

Moreover, the energy relations of such cybernetic systems are
commonly inverted. Because organisms are able to store energy, it is
usual that the energy expenditure is, for limited periods of time, an
inverse function of energy in-put. The amoeba is more active when
it lacks food, and the stem of a green plant grows faster on that side
which is turned away from the light.

Let us therefore invert the question about the symmetry of the
total reduplicated appendage: Why is this double appendage not
asymmetrical like the corresponding appendages of normal


To this question a formal and general (but not particular) answer
can be constructed on the following lines:

(1) An unfertilized frog’s egg is radially symmetrical, with animal
and vegetal poles but no differentiation of its equatorial radii. Such
an egg develops into a bilaterally symmetrical embryo, but how
does it select one meridian to be the plane of bilateral symmetry of
that embryo? The answer is known—that, in fact, the frog’s egg
receives information from the outside. The point of entry of the
spermatozoon (or the prick of a fine fiber) marks one meridian as
different from all others, and that meridian is the future plane of
bilateral symmetry.
Converse cases can also be cited. Plants of many families bear
bilaterally symmetrical flowers. Such flowers are all clearly derived
from triadic radial symmetry (as in orchids) or from pentadic
symmetry (as in Labiatae, Leguminosae, etc.) ; and the bilateral
symmetry is achieved by the differentiation of one axis (e.g., the
“standard” of the familiar sweet pea) of this radial symmetry. We
again ask how it is possible to select one of the similar three (or
five) axes. And again we find that each flower receives information
from the outside. Such bilaterally symmetrical flowers can only be
produced on branch stems, and the differentiation of the flower is
always oriented to the manner in which the flower-bearing branch
stem comes off from the main stem. Very occasionally a plant which
normally bears bilaterally symmetrical flowers will form a flower at
the terminus of a main stem. Such a flower is necessarily only radial
in its symmetry—a cup-shaped monstrosity. (The problem of bilaterally
asymmetrical flowers, e.g., in the Catasetum group of
orchids, is interesting. Presumably these must be borne, like the
lateral appendages of animals, upon branches from main stems
which are themselves already bilaterally symmetrical, e.g., dorsoventrally

(2) We note then that, in biological systems, the step from radial
symmetry to bilateral symmetry commonly requires a piece of
information from the outside. It is, however, conceivable that some
divergent process might be touched off by minute and randomly
distributed differences, e.g., among the radii of the frog’s egg. In this
case, of course, the selection of a particular meridian for special
development would itself be random and could not be oriented to

other parts of the organism as is the plane of bilateral symmetry in
sweet peas and labiate flowers.

(3) Similar considerations apply to the step from bilateral
symmetry to asymmetry. Again either the asymmetry (the
differentiation of one half from the other) must be achieved by a
random process or it must be achieved by information received from
the outside, i.e., from neighboring tissues and organs. Every lateral
appendage of a vertebrate or arthropod is more or less
asymmetrical17 and the asymmetry is never set randomly in relation
to the rest of the animal. Right limbs are not borne upon the left side
of the body, except under experimental circumstances. Therefore the
asymmetry must depend upon the outside information, presumably
derived from the neighboring tissues.
(3) But if the step from bilateral symmetry to asymmetry requires
additional information, then it follows that in absence of this
additional information, the appendage which should have been
asymmetrical can only be bilaterally symmetrical.
The problem of the bilateral symmetry of reduplicated limbs thus
becomes simply a problem of the loss of a piece of information.
This follows from the general logical rule that every reduction in
symmetry (from radial to bilateral or from bilateral to
asymmetrical) requires additional in-formation.

It is not claimed that the above argument is an explanation of all
the phenomena which illustrate Bateson’s Rule. Indeed, the
argument is offered only to show that there are simple ways of
thinking about these phenomena which have scarcely been explored.
What is proposed is a family of hypotheses rather than a single one.
A critical examination of what has been said above as if it were a
single hypothesis will, how-ever, provide a further illustration of the

17 In this connection, scales and feathers and hairs are of special interest. A
feather would seem to have a very clear bilateral symmetry in which the plane of
symmetry is related to the antero-posterior differentiation of the bird. Superposed on
this is an asymmetry like that of the individual bilateral limbs. As in the case of lateral
limbs, corresponding feathers on opposite sides of the body are mirror images of
each other. Every feather is, as it were, a flag whose shape and coloring denote the
values of determining variables at the point and time of its growth.

In any given case of reduplication, it will be necessary to decide
what particular piece of information has been lost, and the argument
so far given should make this decision easy. A natural first guess
would be that the developing appendage needs three sorts of
orienting information to en-able it to achieve asymmetry: proximo-
distal information; dorso-ventral information; and antero-posterior
information. The simplest hypothesis suggests that these might be
separately received and therefore that one of these sorts of
information will be lost or absent in any given case of reduplication.
It should then be easy to classify cases of reduplication ac-cording
to which piece of orienting information is missing. There should be
at most three such types of reduplication, and these should be clearly

Supernumerary Double Legs in Coleoptera

But in the only set of cases where this deduction can be tested,
facts clearly do not fit the hypothesis. The cases are those of
supernumerary pairs of appendages in beetles. About a hundred such
cases were known in 1894, and of these Bateson18 describes about
half and figures thirteen.

The formal relations are remarkably uniform and leave no doubt
that a single type of explanation should apply to the symmetry in all

18 W. Bateson, Materials . . . , op. cit., pp. 477-503.

Fig. 1 Carabus scheidleri, No. 736. The normal right fore leg, R, bearing an
extra pair of legs, SL and SR’, arising from the ventral surface of the coxa, C. Seen
from in front. (The property of Dr. Kraatz.) From Bateson, W., Materials for the
Study of Variation, London: Macmillan, 1894, p 483.

Fig. 2 Pterostichus muhlfeldii, No. 742. Semidiagrammatic representation of the
left middle tibia bearing the extra tarsi upon the antero-ventral border of the apex.
L, the normal tarsus; R, the extra right; L’ the extra left tarsus. ( The property of
Dr. Kraatz. ) From Bateson, W., Materials for the Study of Variation, London:
Macmillan, 1894, p 485.

Fig. 5 A mechanical device for showing the relations that extra legs in
Secondary Symmetry bear to each other and to the normal leg from which
they arise. The model R represents a normal right leg. SL and SR represent
respectively the extra right and extra left legs of the supernumerary pair. A
and P, the anterior and posterior spurs of the tibia. In each leg the
morphologically anterior surface is shaded, the posterior being white. R is
seen from the ventral aspect and SL and SR are in Position VP. From
Bateson, W., Materials for the Study of Variation, London: Macmillan,
1894, p. 480.

Typically19 one leg (rarely more than one) of a beetle is abnormal
in bearing a branch at some point in its length. This branch is

19 See Figures 1 and 2, pages 385 and 386.


regularly a doublet, consisting of two parts which may be fused at
the point of branching off from the primary leg but which are
commonly separate at their distal ends.

Distally from the point of branching there are thus three
components—a primary leg and two supernumerary legs. These
three lie in one plane and have the following symmetry: the two
components of the supernumerary doublet are a complementary pair
—one being a left and the other a right—as Bateson’s Rule would
suggest. Of these two, the leg nearest to the primary leg is
complementary to it.

These relations are represented in Figure 3. (See page 387.) Each
component is shown in diagrammatic cross section, and their dorsal,
ventral, anterior, and posterior faces are indicated by the letters D, V,
A, and P, respectively.

What is surprising about these abnormalities—in that it conflicts
with the hypothesis offered above—is that there is no clear
discontinuity by which the cases can be classified according to
which sort of orienting information has been lost. The
supernumerary doublet may be borne on any part of the
circumference of the primary leg.

Figure 3 illustrates the symmetry of a doublet occurring in the
dorsal region. Figure 4 (page 387) illustrates the symmetry of a
doublet in the dorso-anterior region.

It appears, then, that the planes of symmetry are parallel to a
tangent of the circumference of the primary leg at the point of
branching but, since the points of branching may be anywhere on
the circumference, a continuous series of possible bilateral
symmetries is generated.

Figure 5 (page 388) is a machine invented by W. Bateson to
demonstrate this continuous series of possible bilateral symmetries.

If the bilateral symmetry of the doublet is due to a loss of
orienting information, we should expect the plane of that bilateral
symmetry to be at right angles to the direction of the lost
information; i.e., if dorso-ventral information were lost, the resulting
limbs or doublet should contain a plane of symmetry which would
be at right angles to the dorso-ventral line.

(The argument for this expectation may be spelled out as follows:
a gradient in a lineal sequence creates a difference between the two

ends of the sequence. If this gradient is not present, then the ends of
the sequence will be similar, i.e., the sequence will be symmetrical
about a plane of symmetry transverse to itself. Or, consider the case
of the frog’s egg. The two poles and the point of entry of the
spermatozoon determine a plane of bilateral symmetry. To achieve
asymmetry, the egg requires information at right angles to this plane,
i.e., something which will make the right half different from the left.
If this something is lost, then the egg will revert to the original
bilateral symmetry, with the original plane of symmetry transverse
to the direction of the lost information.)

As noted above, the supernumerary doublets may originate from
any face of the primary leg, and therefore all intermediates occur
between the expectedly discontinuous types of loss of information.
It follows that if bilateral symmetry in these doublets is due to loss
of information, then the information lost cannot be classified as
antero-posterior, dorso-ventral, or proximo-distal.

The hypothesis must therefore be corrected.

Let us retain the general notion of lost information, and the
corollary of this that the plane of bilateral symmetry must be at right
angles to the direction of the information that was lost.

The next simplest hypothesis suggests that the lost information
must have been centro-peripheral. (I here retain this bipolar term
rather than use the simpler “radial.”)

Let us imagine, then, some centro-peripheral difference —
possibly a chemical or electrical g r a d i e n t within the cross section
of the primary leg; and suppose that the loss or blurring of this
difference at some point along the length of the primary leg
determines that any branch limb produced at this point shall fail to
achieve asymmetry.

It will follow, naturally, that such a branch limb (if produced)
will be bilaterally symmetrical and that its plane of bilateral
symmetry will be at right angles to the direction of the lost gradient
or difference.

But, clearly, a centro-peripheral difference or gradient is not a
primary component of that information system which determined
the asymmetry of the primary leg. Such a gradient might, however,
inhibit branching, so that its loss or blurring would result in
production of a supernumerary branch at the point of loss.


The matter becomes superficially paradoxical: the loss of a
gradient which might inhibit branching results in branch formation,
such that the branch cannot achieve asymmetry. It appears, then, that
the hypothetical Centro-peripheral gradient or difference may have
two sorts of command functions : (a) to inhibit branching; and (b)
to determine an asymmetry in that branch which can only come into
existence at all if the Centro-peripheral gradient is absent. If these
two sorts of message functions can be shown to overlap or be in
some sense synonymous, we shall have generated an economical
hypothetical description of the phenomena.

We therefore address ourselves to the question: Is there an a
priori case for expecting that the absence of a gradient which
would prohibit branching in the primary leg will permit the
formation of a branch which will lack the information necessary to
determine asymmetry across a plane at right angles to the missing

The question must be inverted to fit the upside-downness of all
cybernetic explanation. The concept “information necessary to
determine asymmetry” then becomes “information necessary to
prohibit bilateral symmetry.”

But anything which “prohibits bilateral symmetry” will also
“prohibit branching,” since the two components of a branching
structure constitute a symmetrical pair (even though the components
may be radially symmetrical).

It therefore becomes reasonable to expect that loss or blurring of
a Centro-peripheral gradient which prohibits branch formation will
permit the formation of a branch which will, however, itself be
bilaterally symmetrical about a plane parallel to the circumference
of the primary limb.

Meanwhile, within the primary limb, it is possible that a
Centro-peripheral gradient, by preventing branch formation, could
have a function in preserving a previously deter-mined asymmetry.

The above hypotheses provide a possible framework of
explanation of the formation of the supernumerary doublet and the
bilateral symmetry within it. It remains to consider the orientation of
the components of that doublet. According to Bateson’s Rule, the
component nearest to the primary leg is in bilateral symmetry with
it. In other words, that face of the supernumerary which is toward

the primary is the morphological counterpart of that face of the
periphery of the primary from which the branch sprang.

The simplest, and perhaps obvious, explanation of this regularity
is that in the process of branching there was a sharing of
morphologically differentiated structures between branch and
primary and that these shared structures are, in fact, the carriers of
the necessary information. However, since information carried this
way will clearly have proper-ties very different from those of
information carried by gradients, it is appropriate to spell the matter
out in some detail.

Consider a radially symmetrical cone with circular base. Such a
figure is differentiated in the axial dimension, as between apex and
base. All that is necessary to make the cone fully asymmetrical is to
differentiate on the circumference of the base two points which shall
be different from each other and shall not be in diametrically
opposite positions, i.e., the base must contain such differentiation
that to name its parts in clockwise order gives a result different from
the result of naming the parts in anticlockwise order.

Assume now that the supernumerary branch, by its very origin as
a unit growing out from a matrix, has proximo-distal differentiation,
and that this differentiation is analogous to the differentiation in the
axial dimension of the cone. To achieve complete asymmetry, it is
then only necessary that the developing limb receive directional
information in some arc of its circumference. Such information is
clearly immediately available from the circumstance that, at the
point of branching, the secondary limb must share some circumference
with the primary. But the shared points which are in
clockwise order on the periphery of the primary will be in
anticlockwise order on the periphery of the branch. The information
from the shared arc will therefore be such as to determine both that
the resulting limb will be a mirror image of the primary and that the
branch will face appropriately toward the primary.

It is now possible to construct a hypothetical sequence of events
for the reduplications in the legs of beetles:

(1)A primary leg develops asymmetry, deriving the
necessary information from surrounding tissues.
(2)This information, after it has had its effect, continues toexist, transformed into morphological differentiation.

(3)The asymmetry of the normal primary leg is hence-forthmaintained by a centro-peripheral gradient which normally
prevents branching.
(4)In the abnormal specimens, this centro-peripheralgradient is lost or blurred—possibly at some point of lesion or

(5)Following the loss of the centro-peripheral gradient,
branching occurs.
(6)The resulting branch is a doublet; lacking the gradientinformation which would have determined asymmetry, it musttherefore be bilaterally symmetrical.
(7)That component of the doublet which is next to the
primary is oriented to be a mirror image of the primary by thesharing of differentiated peripheral structures.
(8)Similarly each component of the doublet is itself
asymmetrical, deriving the necessary information from the
morphology of shared peripheries in the plane of the doublet.
The above speculations are intended to illustrate how the
explanatory principle of loss of information might be applied to
some of the regularities subsumed under Bateson’s Rule. But it will
be noted that the data on symmetry in the legs of beetles have, in
fact, been overexplained.

Two d i s t i n c t but not mutually exclusive—types of explanation
have been invoked: (a) the loss of information which
should have been derived from a centro-peripheral gradient, and (b)
information derived from shared peripheral morphology.

Neither of these types of explanation is sufficient by itself to
explain the phenomena, but when combined the two principles
overlap so that some details of the total picture can be referred
simultaneously to both principles.

Such redundancy is, no doubt, the rule rather than the exception
in biological systems, as it is in all other systems of organization,
differentiation, and communication. In all such systems, redundancy
is a major and necessary source of stability, predictability, and

Redundancy within the system will inevitably appear as
overlapping between our explanations of the system. Indeed,
without overlapping, our explanations will commonly be
insufficient, failing to explain the facts of biological integration.

We know little about how the pathways of evolutionary change
are influenced by such morphogenetic and physiological redundancies.
But certainly such internal redundancies must impose
nonrandom characteristics upon the phenomena of variation.20

Reduplicated Limbs in Amphibia

At this point it is interesting to turn from analysis of
reduplication in beetles’ legs to another body of data in which
reduplication commonly occurs and has been referred to Bateson’s
Rule.21 These are the data on reduplication in the experimentally
transplanted limbs of larval newts.

(1) There are some cases, mostly of heterotopic trans-plants in
which the grafted limb bud develops into a simple and apparently
equal binary system, in which the two components are in mirror
image symmetry. I was shown about three years ago a very striking
preparation by Dr. Emerson Hibbard of the California Institute of
Technology. In this specimen the limb bud had been rotated through
180°, so that the anterior edge of the bud faced toward the posterior
end of the host, and had been implanted in a median dorsal position
on the posterior region of the head of the host. This transplant had
developed into two remarkably complete legs in mirror image
relationship. This binary system was connected to the head of the
host only by a slender bridge of tissue.
Such preparations, where the product is binary and the parts
equal, certainly look like what would be expected from a simple loss
of one dimension of orienting information. (It was Dr. Hibbard’s
specimen that suggested to me that the hypothesis of lost
information might be applicable to the amphibian material.)

(2) However, apart from these instances of equal binary
reduplication, the amphibian material does not at all fit with any
hypothesis that would explain the reduplication as due to a simple
20 G. Bateson, “The Role of Somatic Change in Evolution,” Evolution, 1962,

17: 529 -39.
21 Harrison, op. cit.; also F. H. Swett, “On The Production of Double Limbs in
Amphibians,” Journal of Experimental Zoology, 1926, 44: 419-72.


loss of information. Indeed, if Bateson’s Rule were restricted to
cases where the explanation is formally analogous to that which fits
the reduplication in the beetles’ legs, then the amphibian cases
would probably not fall under this rubric.

The limitations of a hypothesis are, however, as important as its
applications, and I shall therefore summarize here the very complex
data on orthotopic transplants.

One schematic paradigm will suffice: if the right anterior limb
bud is excised, turned through 180° and replaced in the wound, it
will grow to be a left limb. But this primary limb may subsequently
form secondary limb buds at its base, usually either immediately
anterior or posterior to the point of insertion. The secondary will be
a mirror image of the primary, and may even later develop a tertiary
which will typically be formed outside the secondary, i.e., on that
side of the secondary which is farthest from the primary.

The formation of the left primary on the right side of the body is
explained22 by assuming that antero-posterior orientation is received
by the limb bud earlier than dorso-ventral information, and that,
once received, this antero-posterior information is irreversible. It is
supposed that the graft is already antero-posteriorly determined at
the time of grafting but later receives dorso-ventral information
from the tissues with which it is now in contact. The result is a limb
whose dorso-ventral orientation is correct for its new setting but
whose antero-posterior orientation is reversed. It is tacitly assumed
that the proximo-distal orientation of the bud is undisturbed. The
result is a limb which is reversed in regard to one of its three sorts
of asymmetry. Such a limb must logically be a left.

This explanation I accept and proceed to consider the

These differ in four important respects from the reduplications in
beetles’ legs discussed above:

(a)In the beetles, the reduplication is usually equal. The two
halves of the supernumerary doublet are equal in size, and are
usually approximately equal in size to the corresponding parts of
the primary leg. Such differences as do appear among the three

22 Swett, op. cit.; also Harrison, op. cit.

components are such as might expectably result from trophic
differences. But in the larval newts, great differences in size occur
between the components of the reduplicated system, and it appears
that these differences are determined by time. The secondaries are
smaller than the primaries because they are produced later and,
similarly, the rare tertiaries are later and smaller than the
secondaries. This spacing of events in time indicates clearly that
the primary limb received all the information necessary to
determine its own asymmetry. It received, in-deed, “wrong”
information and grew to be a left leg on the right side of the body
but it did not suffer from such a deficiency of information as
would make it immediately fail to achieve asymmetry. The
reduplication cannot simply be ascribed to loss of orienting
information in the primary.

(b)The reduplications in beetles’ legs may occur at any point
along the length of the leg. But those of amphibian larvae usually
arise from the region of attachment of the limb to the body. It is
not even sure that the secondary always shares tissue with the

(c)In the case of the beetles, the supernumerary doublets form a
continuous series, being given off from any portion of the
periphery of the primary. In contrast, the reduplication of limbs in
amphibian larvae is localized either anterior or posterior to the

In the beetles it is clear that the two supernumerary components
form together a single unit. In many cases there is actual
compounding of the two components (as in Figure 1). In no case23 is
that component of the doublet which is nearer to the primary
compounded with it rather than with the other supernumerary. In the
amphibian preparations, on the other hand, it is not clear that
secondary and tertiary form a subunit. The relation between tertiary
and secondary seems no closer than between secondary and primary.
Above all, the relation is asymmetric in the time dimension.

23 Bateson (Materials . . . , op. cit., p. 507) describes and figures one
doubtful exception to this statement. This is a reduplication in the left hind tarsus of
Platycerus caraboides.

These profound formal differences between the two bodies of
data indicate that the explanations for the amphibian data must be of
a different order. It would seem that the processes are located not in
the shaft of the limb but in its base and the tissues surrounding the
base. Tentatively we may guess that the primary in some way
proposes the later formation of a secondary by a reversal of gradient
information, and that the secondary similarly proposes a reversed
tertiary. Models for such systems are available in cybernetic theory
in those circuit structures which propose Russellian paradoxes.24 To
attempt to construct any such model at the present time would be


This essay on the symmetry of reduplicated lateral appendages
starts from an explanatory principle, viz., that any step of
ontogenetic differentiation which reduces the symmetry of an organ
(e.g., from radial to bilateral symmetry, or from bilateral symmetry
to asymmetry) requires additional orienting information. From this
principle it is argued that a normally asymmetrical lateral
appendage, lacking some necessary piece of orienting information,
will only be able to achieve bilateral symmetry, i.e., instead of a
normal asymmetrical appendage, the result will be a bilaterally
symmetrical doublet.

To examine this explanatory principle, the writer has at-tempted
to construct a hypothesis to explain Bateson’s Rule as this regularity
is exemplified in the rare supernumerary double legs of Coleoptera.
In the construction of this hypothesis, it was assumed that
morphogenetic orienting information may undergo transformation
from one type of coding to another, and that each transform or code
is subject to characteristic limitations:

(a) The information may be embodied in gradients (perhaps
biochemical). In this coding, the information can be diffused from
neighboring tissues and provide the first determinants of
24 G. Bateson, “Minimal Requirements for a Theory of Schizophrenia,” A.M.A.
Archives of General Psychiatry, 1960, 2: 477-91.

asymmetry in the developing appendage. It is suggested that
information coded in this way is only briefly available, and that
once the asymmetry of the limb is established, the information
continues to exist, but trans-formed into morphology.

(b) It is suggested that information coded as morphological
difference is essentially static. It cannot be diffused to neighboring
tissues and it cannot inhibit branching. It can, however, be used by a
branch which at its inception shares tissue with the primary limb
from which it branches off. In this case, the information passed on
by the method of shared periphery will be necessarily inverted: if
the primary be a right, the branch will be a left.
(c) The information in morphological form being (by hypothesis)
unable to inhibit branching, the asymmetry of a growing primary
must be preserved by a centro-peripheral gradient—not itself a
determinant of that asymmetry.
(d) It is suggested that the loss of such a centro-peripheral
gradient might have two effects: that of permitting branching and
that of depriving the resulting branch of one dimension of necessary
orienting information; so that the branch can only be a bilaterally
symmetrical unit with a plane of symmetry at right angles to the lost
centroperipheral gradient.
The data on reduplication in the experimentally trans-planted
limb buds of amphibia are also examined. It is argued that these data
are not to be explained by simple loss of orienting information.
Simple loss, it is suggested, will expectably result in equal and
synchronous bilateral symmetry. The amphibian reduplicates are, in
general, unequal and successive. In a few cases, synchronous and
equal re-duplication occurs in the amphibian experiments, especially
in heterotopic implants. Such cases could perhaps be regarded as
due to simple loss of orienting information.


Postscript, 1971

Compare the bilateral symmetry in the supernumerary doublet of
the beetle’s leg with the bilateral symmetry in the sweet pea or
orchid flower. Both in the plant and in the animal, the bilaterally
symmetrical unit comes off from a point of branching.

In the plant, the morphology of the fork provides information
enabling the flower to be not radially but bilaterally symmetrical,
i.e., information which will differentiate the “dorsal” standard from
the ventral lip of the flower.

In the doublet on the beetle’s leg, the plane of bilateral symmetry
is orthogonal to that in the flower.

We might say that the information which the beetle’s leg has lost
is precisely that information which the plant creates by the act of

Comment on Part IV

The papers placed together in this part are diverse in that while
each paper is a branch from the main stem of the argument of the
book, these branches come off from very different locations. “The
Role of Somatic Change in Evolution” is an expansion of the thought
behind “Minimal Requirements for a Theory of Schizophrenia,”
while “Problems in Cetacean and Other Mammalian
Communication” is an application of “The Logical Categories of
Learning and Communication” to a particular type of animal.

“A Re-examination of Bateson’s Rule” may seem to break new
ground, but is related to the remainder of the book in that it ex-tends
the notion of informational control to include the field of
morphogenesis and, by discussing what happens in absence of
needed information, brings out the importance of the context into
which information is received.

Samuel Butler, with uncanny insight, once commented upon the
analogy between dreams and parthenogenesis. We may say that the
monstrous double legs of the beetles share in this analogy: they are
the projection of the receptive context deprived of information
which should have come from an external source.

Message material, or information, comes out of a context into a
context, and in other parts of the book the focus has been on the
context out of which information came. Here the focus is rather
upon the internal state of the organism as a context into which the
information must be received.

Of course, neither focus is sufficient by itself for our understanding
of either animals or men. But it is perhaps not an accident
that in these papers dealing with non-human organisms the “context”
which is discussed is the obverse or complement of the “context”
upon which I have focussed attention in other parts of the book.

Consider the case of the unfertilized frog’s egg for which the entry
point of the spermatozoon defines the plane of bilateral symmetry of the
future embryo.

The prick of a hair from a camel’s hair brush can be substituted and
still carry the same message. From this it seems that the external context


out of which the message comes is relatively undefined. From the entry
point alone, the egg learns but little about the external world. But the
internal context into which the message comes must be exceedingly

The unfertilized egg, then, embodies an immanent question to
which the entry point of spermatozoon provides an answer; and this way
of stating the matter is the contrary or obverse of the conventional view,
which would see the external context of learning as a “question” to which
the “right” behavior of the organism is an answer.

We can even begin to list some of the components of the immanent
question. First there are the already existing poles of the egg and,
necessarily, some polarization of the intervening protoplasm towards
these poles. Without some such structural conditions for the receipt of the
prick of the spermatozoon, this message could have no meaning. The
message must come into an appropriate structure.

But structure alone is not enough. It seems probable that any meridian
of the frog’s egg can potentially become the plane of bilateral symmetry
and that, in this, all meridians are alike. It follows that there is, to this
extent, no structural difference between them. But every meridian must
be ready for the activating message, its “readiness” being given direction
but otherwise unrestricted by structure. Readiness, in fact, is precisely
not-structure. If and when the spermatozoon delivers its message, new
structure is generated.

In terms of the economics of flexibility, discussed in “The Role of
Somatic Change in Evolution” and later in “Ecology and Flexibility in
Urban Civilization” (Part VI), this “readiness” is uncommitted
potentiality for change, and we note here that this uncommitted
potentiality is not only always finite in quantity but must be appropriately
located in a structural matrix, which also must be quantitatively finite at
any given time.

These considerations lead naturally into Part V, which I have titled
“Epistemology and Ecology.” Perhaps “epistemology” is only another
word for the study of the ecology of mind.

Part V: Epistemology and

Cybernetic Explanation*

It may be useful to describe some of the peculiarities of
cybernetic explanation.

Causal explanation is usually positive. We say that billiard ball B
moved in such and such a direction because billiard ball A hit it at
such and such an angle. In contrast to this, cybernetic explanation is
always negative. We consider what alternative possibilities could
conceivably have occurred and then ask why many of the
alternatives were not followed, so that the particular event was one
of those few which could, in fact, occur. The classical example of
this type of explanation is the theory of evolution under natural
selection. Ac-cording to this theory, those organisms which were not
both physiologically and environmentally viable could not possibly
have lived to reproduce. Therefore, evolution always followed the
pathways of viability. As Lewis Carroll has pointed out, the theory
explains quite satisfactorily why there are no bread-and-butter-flies

In cybernetic language, the course of events is said to be subject
to restraints, and it is assumed that, apart from such restraints, the
pathways of change would be governed only by equality of
probability. In fact, the “restraints” upon which cybernetic
explanation depends can in all cases be regarded as factors which
determine inequality of probability. If we find a monkey striking a
typewriter apparently at random but in fact writing meaningful
prose, we shall look for restraints, either inside the monkey or inside
the typewriter. Perhaps the monkey could not strike inappropriate
letters; perhaps the type bars could not move if improperly struck;
perhaps incorrect letters could not survive on the paper. Somewhere
there must have been a circuit which could identify error and
eliminate it.

* This article is reprinted from the American Behavioral Scientist, Vol. 10, No. 8,
April 1967, pp. 29-32, by per-mission of the publisher, Sage Publications, Inc.

Ideally—and commonly—the actual event in any sequence or
aggregate is uniquely determined within the terms of the cybernetic
explanation. Restraints of many different kinds may combine to
generate this unique determination. For example, the selection of a
piece for a given position in a jigsaw puzzle is “restrained” by many
factors. Its shape must conform to that of its several neighbors and
possibly that of the boundary of the puzzle; its color must conform
to the color pattern of its region; the orientation of its edges must
obey the topological regularities set by the cutting machine in which
the puzzle was made; and so on. From the point of view of the man
who is trying to solve the puzzle, these are all clues, i.e., sources of
information which will guide him in his selection. From the point of
view of the cybernetic observer, they are restraints.

Similarly, from the cybernetic point of view, a word in a
sentence, or a letter within the word, or the anatomy of some part
within an organism, or the role of a species in an ecosystem, or the
behavior of a member within a family—these are all to be
(negatively) explained by an analysis of restraints.

The negative form of these explanations is precisely comparable
to the form of logical proof by reductio ad absurdum. In this species
of proof, a sufficient set of mutually exclusive alternative
propositions is enumerated, e.g., “P” and “not P,” and the process of
proof procedes by demonstrating that all but one of this set are
untenable or “absurd.” It follows that the surviving member of the
set must be tenable within the terms of the logical system. This is a
form of proof which the nonmathematical sometimes find
unconvincing and, no doubt, the theory of natural selection
sometimes seems unconvincing to nonmathematical persons for
similar reasons—whatever those reasons may be.

Another tactic of mathematical proof which has its counterpart in
the construction of cybernetic explanations is the use of “mapping”
or rigorous metaphor. An algebraic proposition may, for example, be
mapped onto a system of geometric coordinates and there proven by
geometric methods. In cybernetics, mapping appears as a technique
of explanation whenever a conceptual “model” is invoked or, more
concretely, when a computer is used to simulate a complex
communicational process. But this is not the only appearance of
mapping in this science. Formal processes of mapping, translation,


or transformation are, in principle, imputed to every step of any
sequence of phenomena which the cyberneticist is attempting to
explain. These mappings or trans-formations may be very complex,
e.g., where the output of some machine is regarded as a transform of
the input; or they may be very simple, e.g., where the rotation of a
shaft at a given point along its length is regarded as a transform
(albeit identical) of its rotation at some previous point.

The relations which remain constant under such transformation
may be of any conceivable kind.

This parallel, between cybernetic explanation and the tactics of
logical or mathematical proof, is of more than trivial interest.
Outside of cybernetics, we look for explanation, but not for anything
which would simulate logical proof. This simulation of proof is
something new. We can say, however, with hindsight wisdom, that
explanation by simulation of logical or mathematical proof was
expectable. After all, the subject matter of cybernetics is not events
and objects but the information “carried” by events and objects. We
consider the objects or events only as proposing facts, propositions,
messages, percepts, and the like. The subject matter being
propositional, it is expectable that explanation would simulate the

Cyberneticians have specialized in those explanations which
simulate reductio ad absurdum and “mapping.” There are perhaps
whole realms of explanation awaiting discovery by some
mathematician who will recognize, in the informational aspects of
nature, sequences which simulate other types of proof.

Because the subject matter of cybernetics is the propositional or
informational aspect of the events and objects in the natural world,
this science is forced to procedures rather different from those of the
other sciences. The differentiation, for example, between map and
territory, which the semanticists insist that scientists shall respect
in their writings must, in cybernetics, be watched for in the very
phenomena about which the scientist writes. Expectably,
communicating organisms and badly programmed computers will
mistake map for territory; and the language of the scientist must be
able to cope with such anomalies. In human behavioral systems,
especially in religion and ritual and wherever primary process

dominates the scene, the name often is the thing named. The bread
is the Body, and the wine is the Blood.

Similarly, the whole matter of induction and deduction —and our
doctrinaire preferences for one or the other—will take on a new
significance when we recognize inductive and deductive steps not
only in our own argument but in the relationships among data.

Of especial interest in this connection is the relationship between
context and its content. A phoneme exists as such only in
combination with other phonemes which make up a word. The word
is the context of the phoneme. But the word only exists as such—
only has “meaning”—in the larger context of the utterance, which
again has meaning only in a relationship.

This hierarchy of contexts within contexts is universal for the
communicational (or “emic”) aspect of phenomena and drives the
scientist always to seek for explanation in the ever larger units. It
may (perhaps) be true in physics that the explanation of the
macroscopic is to be sought in the microscopic. The opposite is
usually true in cybernetics: without context, there is no

In accord with the negative character of cybernetic ex-planation,
“information” is quantified in negative terms. An event or-object
such as the letter K in a given position in the text of a message
might have been any other of the limited set of twenty-six letters in
the English language. The actual letter excludes (i.e., eliminates by
restraint) twenty-five alternatives. In comparison with an English
letter, a Chinese ideograph would have excluded several thousand
alternatives. We say, therefore, that the Chinese ideograph carries
more information than the letter. The quantity of information is
conventionally expressed as the log to base 2 of the improbability of
the actual event or object.

Probability, being a ratio between quantities which have similar
dimensions, is itself of zero dimensions. That is, the central
explanatory quantity, information, is of zero dimensions.
Quantities of real dimensions (mass, length, time) and their
derivatives (force, energy, etc.) have no place in cybernetic

The status of energy is of special interest. In general in
communicational systems, we deal with sequences which resemble


stimulus-and-response rather than cause-and-effect. When one
billiard ball strikes another, there is an energy transfer such that the
motion of the second ball is energized by the impact of the first. In
communicational systems, on the other hand, the energy of the
response is usually provided by the respondent. If I kick a dog, his
immediately sequential behavior is energized by his metabolism, not
by my kick. Similarly, when one neuron fires another, or an impulse
from a microphone activates a circuit, the sequent event has its own
energy sources.

Of course, everything that happens is still within the limits
defined by the law of energy conservation. The dog’s metabolism
might in the end limit his response, but, in general, in the systems
with which we deal, the energy supplies are large compared with the
demands upon them; and, long before the supplies are exhausted,
“economic” limitations are imposed by the finite number of available
alternatives, i.e., there is an economics of probability. This
economics differs from an economics of energy or money in that
probability—being a ratio—is not subject to addition or subtraction
but only to multiplicative processes, such as fractionation. A
telephone exchange at a time of emergency may be “jammed” when
a large fraction of its alternative pathways are busy. There is, then, a
low probability of any given message getting through.

In addition to the restraints due to the limited economics of
alternatives, two other categories of restraint must be discussed:
restraints related to “feedback” and restraints related to

We consider first the concept of feedback:

When the phenomena of the universe are seen as linked together
by cause-and-effect and energy transfer, the resulting picture is of
complexly branching and interconnecting chains of causation. In
certain regions of this universe (notably organisms in environments,
ecosystems, thermostats, steam engines with governors, societies,
computers, and the like), these chains of causation form circuits
which are closed in the sense that causal interconnection can be
traced around the circuit and back through whatever position was
(arbitarily) chosen as the starting point of the description. In such a
circuit, evidently, events at any position in the circuit may be
expected to have effect at all positions on the circuit at later times.

Such systems are, however, always open: (a) in the sense that the
circuit is energized from some external source and loses energy
usually in the form of heat to the outside; and (b) in the sense that
events within the circuit may be influenced from the outside or may
influence outside events.

A very large and important part of cybernetic theory is concerned
with the formal characteristics of such causal circuits, and the
conditions of their stability. Here I shall consider such systems only
as sources of restraint.

Consider a variable in the circuit at any position and sup-pose
this variable subject to random change in value (the change perhaps
being imposed by impact of some event external to the circuit). We
now ask how this change will affect the value of this variable at that
later time when the sequence of effects has come around the circuit.
Clearly the answer to this last question will depend upon the
characteristics of the circuit and will, therefore, be not random.

In principle, then, a causal circuit will generate a non-random
response to a random event at that position in the circuit at which the
random event occurred.

This is the general requisite for the creation of cybernetic
restraint in any variable at any given position. The particular
restraint created in any given instance will, of course, depend upon
the characteristics of the particular circuit—whether its overall gain
be positive or negative, its time characteristics, its thresholds of
activity, etc. These will together determine the restraints which it
will exert at any given position.

For purposes of cybernetic explanation, when a machine is
observed to be (improbably) moving at a constant rate, even under
varying load, we shall look for restraints—e.g., for a circuit which
will be activated by changes in rate and which, when activated, will
operate upon some variable (e.g., the fuel supply) in such a way as
to diminish the change in rate.

When the monkey is observed to be (improbably) typing prose,
we shall look for some circuit which is activated whenever he makes
a “mistake” and which, when activated, will delete the evidence of
that mistake at the position where it occurred.

The cybernetic method of negative explanation raises the
question: Is there a difference between “being right” and “not being


wrong”? Should we say of the rat in a maze that he has “learned the
right path” or should we say only that he has learned “to avoid the
wrong paths”?

Subjectively, I feel that I know how to spell a number of English
words, and I am certainly not aware of discarding as unrewarding
the letter K when I have to spell the word “many.” Yet, in the first
level cybernetic explanation, I should be viewed as actively
discarding the alternative K when I spell “many.”

The question is not trivial and the answer is both subtle and
fundamental: choices are not all at the same level. I may have to avoid
error in my choice of the word “many” in a given context, discarding
the alternatives, “few,” “several,” “frequent,” etc. But if I can
achieve this higher level choice on a negative base, it follows that
the word “many” and its alternatives somehow must be conceivable
to me—must exist as distinguishable and possibly labeled or coded
patterns in my neural processes. If they do, in some sense, exist,
then it follows that, after making the higher level choice of what
word to use, I shall not necessarily be faced with alternatives at the
lower level. It may become unnecessary for me to exclude the letter
K from the word “many.” It will be correct to say that I know
positively how to spell “many”; not merely that I know how to avoid
making mistakes in spelling that word.

It follows that Lewis Carroll’s joke about the theory of natural
selection is not entirely cogent. If, in the communicational and
organizational processes of biological evolution, there be something
like levels—items, patterns, and possibly patterns of patterns—then
it is logically possible for the evolutionary system to make
something like positive choices. Such levels and patterning might
conceivably be in or among genes or elsewhere.

The circuitry of the above mentioned monkey would be required
to recognize deviations from “prose,” and prose is characterized by
pattern or—as the engineers call it—by redundancy.

The occurrence of the letter K in a given location in an English
prose message is not a purely random event in the sense that there
was ever an equal probability that any other of the twenty-five
letters might have occurred in that location. Some letters are more
common in English than others, and certain combinations of letters
are more common than others. There is, thus, a species of patterning

which partly determines which letters shall occur in which slots. As
a result: if the receiver of the message had received the entire rest of
the message but had not received the particular letter K which we
are discussing, he might have been able, with better than random
success, to guess that the missing letter was, in fact, K. To the extent
that this was so, the let-ter K did not, for that receiver, exclude the
other twenty-five letters because these were already partly excluded
by information which the recipient received from the rest of the
message. This patterning or predictability of particular events within
a larger aggregate of events is technically called “redundancy.”

The concept of redundancy is usually derived, as I have derived
it, by considering first the maximum of information which might be
carried by the given item and then considering how this total might
be reduced by knowledge of the surrounding patterns of which the
given item is a component part. There is, however, a case for
looking at the whole matter the other way round. We might regard
patterning or predictability as the very essence and raison d’etre of
communication, and see the single letter unaccompanied by
collateral clues as a peculiar and special case.

The idea that communication is the creation of redundancy or
patterning can be applied to the simplest engineering examples. Let
us consider an observer who is watching A send a message to B. The
purpose of the transaction (from the point of view of A and B) is to
create in B’s message pad a sequence of letters identical with the
sequence which formerly occurred in A’s pad. But from the point of
view of the observer this is the creation of redundancy. If he has
seen what A had on his pad, he will not get any new information
about the message itself from inspecting B’s pad.

Evidently, the nature of “meaning,” pattern, redundancy,
information and the like. depends upon where we sit. In the usual
engineers’ discussion of a message sent from A to B, it is customary
to omit the observer and to say that B received information from A
which was measurable in terms of the number of letters transmitted,
reduced by such redundancy in the text as might have permitted B to
do some guessing. But in a wider universe, i.e., that defined by the
point of view of the observer, this no longer appears as a
“transmission” of information but rather as a spreading of
redundancy. The activities of A and B have combined to make the


universe of the observer more predictable, more ordered, and more
redundant. We may say that the rules of the “game” played by A and
B explain (as “restraints”) what would otherwise be a puzzling and
improbable coincidence in the observer’s universe, namely the
conformity between what is written on the two message pads.

To guess, in essence, is to face a cut or slash in the sequence of
items and to predict across that slash what items might be on the
other side. The slash may be spatial or temporal (or both) and the
guessing may be either predictive or retrospective. A pattern, in fact,
is definable as an aggregate of events or objects which will permit in
some degree such guesses when the entire aggregate is not available
for inspection.

But this sort of patterning is also a very, general phenomenon,
outside the realm of communication between organisms. The
reception of message material by one organism is not fundamentally
different from any other case of perception. If I see the top part of a
tree standing up, I can predict —with better than random success—
that the tree has roots in the ground. The percept of the tree top is
redundant with (i.e., contains “information” about) parts of the
system which I cannot perceive owing to the slash provided by the
opacity of the ground.

If then we say that a message has “meaning” or is “about” some
referent, what we mean is that there is a larger universe of relevance
consisting of message-plus-referent, and that redundancy or pattern
or predictability is introduced into this universe by the message.

If I say to you “It is raining,” this message introduces redundancy
into the universe, message-plus-raindrops, so that from the message
alone you could have guessed—with better than random success—
something of what you would see if you looked out of the window.
The universe, message-plus-referent, is given pattern or form—in
the Shakespearean sense, the universe is informed by the message;
and the “form” of which we are speaking is not in the message nor is
it in the referent. It is a correspondence between message and

In loose talk, it seems simple to locate information. The letter K
in a given slot proposes that the letter in that particular slot is a K.
And, so long as all information is of this very direct kind, the

information can be “located”: the information about the letter K is
seemingly in that slot.

The matter is not quite so simple if the text of the message is
redundant but, if we are lucky and the redundancy is of low order,
we may still be able to point to parts of the text which indicate
(carry some of the information) that the letter K is expectable in that
particular slot.

But if we are asked: Where are such items of information as that:

(a) “This message is in English”; and (b) “In English, a letter K often
follows a letter C, except when the C begins a word”; we can only
say that such information is not localized in any part of the text but
is rather a statistical induction from the text as a whole (or perhaps
from an aggregate of “similar” texts). This, after all, is metainformation
and is of a basically different order—of different logical type—
from the information that “the letter in this slot is K.”
This matter of the localization of information has be-deviled
communication theory and especially neurophysiology for many
years and it is, therefore, interesting to consider how the matter
looks if we start from redundancy, pattern or form as the basic

It is flatly obvious that no variable of zero dimensions can be
truly located. “Information” and “form” resemble contrast,
frequency, symmetry, correspondence, congruence, conformity, and
the like in being of zero dimensions and, therefore, are not to be
located. The contrast between this white paper and that black coffee
is not somewhere between the paper and the coffee and, even if we
bring the paper and coffee into close juxtaposition, the contrast
between them is not thereby located or pinched between them. Nor
is that contrast located between the two objects and my eye. It is not
even in my head; or, if it be, then it must also be in your head. But
you, the reader, have not seen the paper and the coffee to which I
was referring. I have in my head an image or transform or name of
the contrast between them; and you have in your head a transform of
what I have in mine. But the conformity between us is not localizable.
In fact, information and form are not items which can be

It is, however, possible to begin (but perhaps not complete) a sort
of mapping of formal relations within a system containing


redundancy. Consider a finite aggregate of objects or events (say a
sequence of letters, or a tree) and an observer who is already
informed about all the redundancy rules which are recognizable (i.e.,
which have statistical significance) within the aggregate. It is then
possible to delimit regions of the aggregate within which the
observer can achieve better than random guessing. A further step
toward localization is accomplished by cutting across these regions
with slash marks, such that it is across these that the educated
observer can guess, from what is on one side of the slash, something
of what is on the other side.

Such a mapping of the distribution of patterns is, how-ever, in
principle, incomplete because we have not considered the sources of
the observer’s prior knowledge of the redundancy rules. If, now, we
consider an observer with no prior knowledge, it is clear that he
might discover some of the relevant rules from his perception of
less than the whole aggregate. He could then use his discovery in
predicting rules for the remainder—rules which would be correct
even though not exemplified. He might discover that “H often
follows T” even though the remainder of the aggregate contained no
example of this combination. For this order of phenomenon a
different order of slash mark—metaslashes —will be necessary.

It is interesting to note that metaslashes which demarcate what is
necessary for the naive observer to discover a rule are, in principle,
displaced relative to the slashes which would have appeared on the
map prepared by an observer totally informed as to the rules of
redundancy for that aggregate. (This principle is of some importance
in aesthetics.

To the aesthetic eye, the form of a crab with one claw bigger than
the other is not simply asymmetrical. It first pro-poses a rule of
symmetry and then subtly denies the rule by proposing a more
complex combination of rules.)

When we exclude all things and all real dimensions from our
explanatory system, we are left regarding each step in a
communicational sequence as a transform of the previous step. If
we consider the passage of an impulse along an axon, we shall
regard the events at each point along the pathway as a transform
(albeit identical or similar) of events at any previous point. Or if we
consider a series of neurons, each firing the next, then the firing of

each neuron is a transform of the firing of its predecessor. We deal
with event sequences which do not necessarily imply a passing on of
the same energy.

Similarly, we can consider any network of neurons, and
arbitrarily transect the whole network at a series of different
positions, then we shall regard the events at each transection as a
transform of events at some previous transection.

In considering perception, we shall not say, for example, “I see a
tree,” because the tree is not within our explanatory system. At best,
it is only possible to see an image which is a complex but systematic
transform of the tree. This image, of course, is energized by my
metabolism and the nature of the transform is, in part, determined by
factors within my neural circuits: “I” make the image, under various
restraints, some of which are imposed by my neural circuits, while
others are imposed by the external tree. An hallucination or dream
would be more truly “mine” insofar as it is produced without
immediate external restraints.

All that is not information, not redundancy, not form and not
restraints—is noise, the only possible source of new patterns.


Redundancy and Coding*

Discussion of the evolutionary and other relationships between
the communication systems of men and those of other animals has
made it very clear that the coding devices characteristic of verbal
communication differ profoundly from those of kinesics and
paralanguage. But the point has been made that there is a great deal
of resemblance between the codes of kinesics and paralanguage and
the codes of nonhuman mammals.

We may, I think, state categorically that man’s verbal system is
not derived in any simple way from these preponderantly iconic
codes. There is a general popular belief that in the evolution of man,
language replaced the cruder systems of the other animals. I believe
this to be totally wrong and would argue as follows:

In any complex functional system capable of adaptive
evolutionary change, when the performance of a given function is
taken over by some new and more efficient method, the old method
falls into disuse and decay. The technique of making weapons by the
knapping of flint deteriorated when metals came into use.

This decay of organs and skills under evolutionary replacement
is a necessary and inevitable systemic phenomenon. If, therefore,
verbal language were in any sense an evolutionary replacement of
communication by means of kinesics and paralanguage, we would
expect the old, preponderantly iconic systems to have undergone
conspicuous decay. Clearly they have not. Rather, the kinesics of
men have become richer and more complex, and paralanguage has
blossomed side by side with the evolution of verbal language. Both
kinesics and paralanguage have been elaborated into complex forms
of art, music, ballet, poetry, and the like, and, even in everyday life,
the intricacies of human kinesic communication, facial expression,
and vocal intonation far exceed anything that any other animal is
known to produce. The logician’s dream that men should

* This essay appeared as Chapter 22 in Animal Communication: Techniques of
Study and Results of Research, edited by Thomas A. Sebeok. Copyright 1968 by
Indiana University Press. Reprinted by permission of the publisher.

communicate only by unambiguous digital signals has not come true
and is not likely to.

I suggest that this separate burgeoning evolution of kinesics and
paralanguage alongside the evolution of verbal language indicates
that our iconic communication serves functions totally different
from those of language and, in-deed, performs functions which
verbal language is unsuited to perform.

When boy says to girl, “I love you,” he is using words to convey
that which is more convincingly conveyed by his tone of voice and
his movements; and the girl, if she has any sense, will pay more
attention to those accompanying signs than to the words. There are
people—professional actors, confidence tricksters, and others—who
are able to use kinesics and paralinguistic communication with a
degree of voluntary control comparable to that voluntary control
which we all think we have over the use of words. For these people
who can lie with kinesics, the special usefulness of non-verbal
communication is reduced. It is a little more difficult for them to be
sincere and still more difficult for them to be believed to be sincere.
They are caught in a process of diminishing returns such that, when
distrusted, they try to improve their skill in simulating paralinguistic
and kinesic sincerity. But this is the very skill which led others to
distrust them.

It seems that the discourse of nonverbal communication is
precisely concerned with matters of relationship—love, hate,
respect, fear, dependency, etc.—between self and vis-à-vis or
between self and environment and that the nature of human society
is such that falsification of this discourse rapidly becomes
pathogenic. From an adaptive point of view, it is therefore important
that this discourse be carried on by techniques which are relatively
unconscious and only imperfectly subject to voluntary control. In
the language of neurophysiology, the controls of this discourse must
be placed in the brain caudad of the controls of true language.

If this general view of the matter be correct, it must follow that to
translate kinesics or paralinguistic messages into words is likely to
introduce gross falsification due not merely to the human propensity
for trying to falsify statements about “feelings” and relationship and
to the distortions which arise whenever the products of one system
of coding are dissected onto the premises of another, but especially


to the fact that all such translation must give to the more or less
unconscious and involuntary iconic message the appearance of
conscious intent.

As scientists, we are concerned to build a simulacrum of the
phenomenal universe in words. That is, our product is to be a verbal
transform of the phenomena. It is necessary, therefore, to examine
rather carefully the rules of this trans-formation and the differences
in coding between natural phenomena, message phenomena, and
words. I know that it is unusual to presume a “coding” of nonliving
phenomena and, to justify this phrase, I must expand somewhat on
the concept of “redundancy” as this word is used by the communications

The engineers and mathematicians have concentrated their
attention rigorously upon the internal structure of message material.
Typically, this material consists of a sequence or collection of events
or objects (commonly members of finite sets—phonemes and the
like). This sequence is differentiated from irrelevant events or
objects occurring in the same region of time-space by the
signal/noise ratio and by other characteristics. The message material
is said to contain “redundancy” if, when the sequence is received
with some items missing, the receiver is able to guess at the missing
items with better than random success. It has been pointed out that,
in fact, the term “redundancy” so used becomes a synonym for
“patterning.”1 It is important to note that this patterning of message
material always helps the receiver to differentiate between signal
and noise. In fact, the regularity called signal/noise ratio is really
only a special case of redundancy. Camouflage (the opposite of
communication) is achieved (1) by reducing the signal/noise ratio,

(2) by breaking up the patterns and regularities in the signal, or (3)
by introducing similar patterns into the noise.
By confining their attention to the internal structure of the
message material, the engineers believe that they can avoid the
complexities and difficulties introduced into communication theory
by the concept of “meaning.” I would argue, however, that the
concept “redundancy” is at least a partial synonym of “meaning.” As

1 F. Attneave, Applications of Information Theory to Psychology, New York,
Henry Holt and Co., 1959.

I see it, if the receiver can guess at missing parts of the message,
then those parts which are received must, in fact, carry a meaning
which refers to the missing parts and is information about those

If now we turn away from the narrow universe of message
structure and consider the outer world of natural phenomena, we
observe at once that this outer world is similarly characterized by
redundancy, i.e., that when an observer perceives only certain parts
of a sequence or configuration of phenomena, he is in many cases
able to guess, with better than random success, at the parts which he
cannot immediately perceive. It is, indeed, a principal goal of the
scientist to elucidate these redundancies or patternings of the phenomenal

If we now consider that larger universe of which these two
subuniverses are parts, i.e., the system: message plus external
phenomena, we find that this larger system contains redundancy of a
very special sort. The observer’s ability to predict external
phenomena is very much increased by his receipt of message
material. If I tell you that “it is raining” and you look out the
window, you will get less information from the perception of
raindrops than you would have got had you never received my
message. From my message you could have guessed that you would
see rain.

In sum, “redundancy” and “meaning” become synonymous
whenever both words are applied to the same universe of
discourse. “Redundancy” within the restricted universe of the
message sequence is not, of course, synonymous with “meaning” in
the wider universe that includes both message and external referent.

It will be noted that this way of thinking about communication
groups all methods of coding under the single rubric of part-forwhole.
The verbal message “It is raining” is to be seen as a part of a
larger universe within which that message creates redundancy or
predictability. The “digital,” the “analogic,” the “iconic,” the
“metaphoric,” and all other methods of coding are subsumed under
this single heading. (What the grammarians call “synecdoche” is the
metaphoric use of the name of a part in place of the name of the
whole, as in the phrase “five head of cattle.”)


This approach to the matter has certain advantages: the analyst is
forced at all times to define the universe of discourse within which
“redundancy” or “meaning” is supposed to occur. He is forced to
examine the “logical typing” of all message material. We shall see
that this broad view of the matter makes it easy to identify major
steps in the evolution of communication. Let us consider the
scientist who is observing two animals in a physical environment.
The following components then must be considered:

(1)The physical environment contains internal patterning or
redundancy, i.e., the perception of certain events or objects
makes other events or objects predictable for the animals and/or
for the observer.
(2)Sounds or other signals from one animal may con-tribute
redundancy to the system, environment plus signal; i.e., the signals
may be “about” the environment.
(3)The sequence of signals will certainly contain redundancy—
one signal from an animal making another signal from the same
animal more predictable.
(4)The signals may contribute redundancy to the universe; A’s
signals plus B’s signals, i.e., the signals may be about the
interaction of which they are component parts.
(5)If all rules or codes of animal communication and
understanding were genotypically fixed, the list would end at this
point. But some animals are capable of learning, e.g., the repetition
of sequences may lead to their becoming effective as patterns. In
logic, “every proposition proposes its own truth,” but in natural
history we deal always with a converse of this generalization. The
perceivable events which accompany a given percept propose that
that percept shall “mean” these events. By some such steps an
organism may learn to use the information contained in patterned
sequences of external events. I can therefore predict with bet-ter
than random success that in the universe, organism plus
environment, events will occur to complete patterns or
configurations of learned adaptation between organism and

(6)The behavioral “learning” which is usually studied in
psychological laboratories is of a different order. The redundancy
of that universe, which consists of the animal’s actions plus

external events, is increased, from the animal’s point of view,
when the animal regularly responds to certain events with certain
actions. Similarly, this universe gains redundancy when the
animal succeeds in producing those actions which function as
regular precursors (or causes) of specific external events.

(7)For every organism there are limitations and regularities
which define what will be learned and under what circumstances
this learning will occur. These regularities and patterns become
basic premises for the individual adaptation and social
organization of any species.
(8)Last but not least, there is the matter of phylogenetic
learning and phylogeny in general. There is redundancy in the
system, organism-plus-environment, such that from the
morphology and behavior of the organism a human observer can
guess with better than random success at the nature of the
environment. This “information” about the environment has
become lodged in the organism through a long phylogenetic
process, and its coding is of a very special kind. The observer who
would learn about the aquatic environment from the shape of a
shark must deduce the hydrodynamics from the adaptation which
copes with the water. The information contained in the phenotypic
shark is implicit in forms which are complementary to characteristics
of other parts of the universe, phenotype plus environment
whose redundancy is increased by the phenotype.

This very brief and incomplete survey of some of the sorts of
redundancy in biological systems and the universes of their
relevance indicates that under the general rubric “part-for-whole” a
number of different sorts of relationship between part and whole are
included. A listing of some of the characteristics of these formal
relations is in order. We consider some of the iconic cases:

(1)The events or objects which we here call the “part” or
“signal” may be real components of an existing sequence or
whole. A standing trunk of a tree indicates the probable presence
of invisible roots. A cloud may indicate the coming storm of
which it is a part. The bared fang of a dog may be part of a real
(2)The “part” may have only a conditional relationship to its
whole: the cloud may indicate that we shall get wet if we don’t go

indoors; the bared fang may be the beginning of an attack which
will be completed unless certain conditions are met.

(3)The “part” may be completely split from the whole which is
its referent. The bared fang at the given instant may mention an
attack which, if and when it occurs, will include a new baring of
the fangs. The “part” has now became a true iconic signal.
(4)Once a true iconic signal has evolved—not necessarily
through steps 1, 2, or 3, above—a variety of other pathways of
evolution become possible:
(a)The “part” may become more or less digitalized, so that
magnitudes within it no longer refer to magnitudes within the
whole which is its referent but, for example, contribute to an
improvement of the signal/noise ratio.

(b)The “part” may take on special ritual or metaphoric
meanings in contexts where the original whole to which it once
referred is no longer relevant. The game of mutual mouthing
between mother dog and puppy which once followed her weaning
of the pup may become a ritual aggregation. The actions of
feeding a baby bird may become a ritual of courtship, etc.

Throughout this series, whose branches and varieties are here
only briefly indicated, it is notable that animal communication is
confined to signals which are derived from actions of the animals
themselves, i.e., those which are parts of such actions. The external
universe is, as already noted, redundant in the sense that it is replete
with part-for-whole messages, and—perhaps for that reason—this
basic style of coding is characteristic of primitive animal communication.
But in so far as animals can signal at all about the external
universe, they do so by means of actions which are parts of their
response to that universe. The jackdaws indicate to each other that
Lorenz is a “jackdaw-eater” not by simulating some part of the act
of eating jackdaws but by simulating part of their aggression vis-avis
such a creature. Occasionally actual pieces of the external
environment—scraps of potential nest-building material,
“trophies,” and the like—are used for communication, and in these
cases again the messages usually contribute redundancy to the
universe message plus the relationship between the organisms rather
than to the universe message plus external environment.

In terms of evolutionary theory, it is not simple to ex-plain why
over and over again genotypic controls have been evolved to
determine such iconic signaling. From the point of view of the
human observer such iconic signals are rather easy to interpret, and
we might expect iconic coding to be comparatively easy for animals
to decode—in so far as the animals must learn to do so. But the
genome is presumed not capable of learning in this sense, and we
might therefore expect genotypically determined signals to be
aniconic or arbitrary rather than iconic.

Three possible explanations of the iconic nature of genotypic
signals can be offered:

(1)Even genotypically determined signals do not occur as
separate and isolated elements in the life of the phenotype but are
necessarily components in a complex matrix of behavior some, at
least, of which is learned. It is possible that the iconic coding of
genotypically determined signals renders these easy to assimilate
into this matrix. There may be an experiential “schoolmarm”
which acts selectively to favor those genotypic changes which
will give rise to iconic rather than arbitrary signaling.
(2)A signal of aggression which places the signaler in a
position of readiness to attack probably has more survival value
than would a more arbitrary signal.
(3)When the genotypically determined signal affects the
behavior of another species—e.g., eye marks or postures which
have a warning effect, movements which facilitate camouflage or
aposematic mimicry—clearly the signal must be iconic to the
perceptive system of that other species. However, an interesting
phenomenon arises in many instances where what is achieved is a
secondary statistical iconicism. Labroides dimidiatus, a small
Indo-Pacific wrasse, which lives on the ectoparasites of other
fishes, is strikingly colored and moves or “dances” in a way which
is easily recognized. No doubt these characteristics attract other
fish and are part of a signaling system which leads the other fish to
permit the approaches of the cleaner. But there is a mimic of this
species of Labroides, a saber-toothed blenny (Aspidontus

taeniatus), whose similar coloring and movement permit the

mimic to approach—and bite off pieces of the fins of other fishes.2

Clearly the coloring and movements of the mimic are iconic and
“represent” the cleaner. But what of the coloring and movements of
the latter? All that is primarily required is that the cleaner be
conspicuous or distinctive. It is not required that it represent
something else. But when we consider the statistical aspects of the
system, it becomes clear that if the blennies become too numerous,
the distinctive features of the wrasses will become iconic warnings
and their hosts will avoid them. What is necessary is that the signals
of the wrasse shall clearly and indubitably represent wrasse, i.e., the
signals, though perhaps aniconic in the first instance, must achieve
and maintain by multiple impact a sort of autoiconicism. “When I
say it three times, it is true.” But this necessity for autoiconicism
may also arise within the species. Genotypic control of signaling
ensures the necessary repetitiveness (which might be only fortuitous
if the signals had to be learned) .

(4) There is a case for asserting that the genotypic determination
of adaptive characteristics is, in a special sense, more economical
than the achievement of similar characteristic by somatic change or
phenotypic learning. This matter has been argued elsewhere.3
Briefly it is asserted that the somatic adaptive flexibility and/or
learning capacity of any organism is limited and that the demands
placed upon these capacities will be reduced by genotypic change in
any appropriate direction. Such changes would therefore have
survival value because they set free precious adaptive or learning
capacity for other uses. This amounts to an argument for Baldwin
effects. An extention of this argument would suggest that the
iconic character of genotypically controlled signaling
characteristics may, in some cases, be explained by supposing that
these characteristics were once learned. (This hypothesis does not,
of course, imply any sort of Lamarckian inheritance. It is obvious
(1) that to fix the value of any variable in a homeostatic circuit by
2 J. E. Randall and H. S. Randall, “Examples of Mimicry and Protective
Resemblance in Tropical Marine Fishes,” Bulletin of Marine Science of the Gulf
and Caribbean, 1960, 10: 444-80.

3 G. Bateson, “The Role of Somatic Change in Evolution,” Evolution, 1963,

17: 529-39.

such inheritance would soon gum up the homeostatic system of the
body, and (2) that no amount of modification of the dependent
variables in a homeostatic circuit will change the bias of the

(5) Last, it is unclear at what level genotypic determination of
behavior might act. It was suggested above that iconic codes are
easier for an organism to learn than more arbitrary codes. It is
possible that the genotypic contribution to such an organism might
take the form, not of fixing the given behavior, but rather of
making this behavior easier to learn—a change in specific learning
capacity rather than a change in genotypically determined
behavior. Such a contribution from the genotype would have
obvious advantages in that it would work along with ontogenetic
change instead of working possibly at cross-purposes with it.
To sum up the argument so far:

(1)But the evolution of aniconic verbal coding remains unexplained.
It is understandable that an early (in an evolutionary
sense) method of creating redundancy would be the use of iconic
part-for-whole coding. The external nonbiological universe
contains redundancy of this kind, and in evolving a code of
communication it is expectable that organisms would fall into the
same trick. We have noted that the “part” can be split from the
whole, so that a showing of the fangs can denote a possible but as
yet nonexistent fight. All this provides an explanatory background
for communication by means of “intention movements” and the
(2)It it partly understandable that such tricks of coding by
iconic parts might become genotypically fixed.

(3)It has been suggested that the survival of such primitive (and
therefore involuntary) signalling in human communication about
personal relationship is explained by a need for honesty in such

We know from studies of aphasia, from Hockett’s enumeration at
this meeting of the characteristics of language and from elementary
common sense that the component processes of creating and
understanding verbal communication are many and that language
fails when any one of those component processes is interrupted. It is
possible that each of these processes should be the focus of a


separate study. Here, however, I shall consider only one aspect of
the matter: the evolution of simple indicative assertion.

An interesting intermediate between the iconic coding of animals
and the verbal coding of human speech can be recognized in human
dreaming and human myth. In psychoanalytic theory, the
productions of dream process are said to be characterized by
“primary-process” thinking.4 Dreams, whether verbal or not, are to
be considered as metaphoric statements, i.e., the referents of dream
are relationships which the dreamer, consciously or unconsciously,
perceives in his waking world. As in all metaphor, the relata remain
unmentioned and in their places appear other items such that the
relationships between these substitute items shall be the same as
those between the relata in the waking world.

To identify the relata in the waking world to which the dream
refers would convert the metaphor into a simile, and, in general,
dreams contain no message material which overtly performs this
function. There is no signal in the dream which tells the dreamer that
this is metaphor or what the referent of the metaphor may be.
Similarly, dream contains no tenses. Time is telescoped, and
representations of past events in real or distorted forms may have
the present as their referent—or vice versa. The patterns of dream
are timeless.

In a theater, the audience is informed by the curtain and the
framing of the stage that the action on the stage is “only” a play.
From within that frame the producers and actors may attempt to
involve the audience in an illusion of reality as seemingly direct as
the experience of dream. And, as in dream, the play has metaphoric
reference to the out-side world. But in dream, unless the sleeper be
partly conscious of the fact of sleep, there is no curtain and no
framing of the action. The partial negative—”This is only
metaphor”—is absent.

I suggest that this absence of metacommunicative frames and the
persistence in dream of pattern recognition are archaic
characteristics in an evolutionary sense. If this be correct, then an
understanding of dream should throw light both on how iconic

4 4 O. Fenichel, Psychoanalytic Theory of Neu York, Norton, 1945.

communication operates among animals and on the mysterious
evolutionary step from the iconic to the verbal.

Under the limitation imposed by the lack of a metacommunicative
frame, it is clearly impossible for dream to make an
indicative statement, either positive or negative. As there can be no
frame which labels the content as “metaphoric,” so there can be no
frame to label the content as “literal.” Dream can imagine rain or
drought, but it can never assert “It is raining” or “It is not raining.”
Therefore, as we have seen, the usefulness in imagining “rain” or
“drought” is limited to their metaphoric aspects.

Dream can propose the applicability of pattern. It can never
assert or deny this applicability. Still less can it make an indicative
statement about any identified referent, since no referent is

The pattern is the thing.

These characteristics of dream may be archaic, but it is important
to remember that they are not obsolete: that, as kinesic and
paralinguistic communication has been elaborated into dance,
music, and poetry, so also the logic of dream has been elaborated
into theater and art. Still more astonishing is that world of rigorous
fantasy which we call mathematics, a world forever isolated by its
axioms and definitions from the possibility of making an indicative
statement about the “real” world. Only if a straight line is the
shortest distance between two points is the theorem of Pythagoras

The banker manipulates numerals according to rules sup-plied by
the mathematician. These numerals are the names of numbers, and
the numbers are somehow embodied in (real or fictitious) dollars. To
remember what he is doing, the banker marks his numerals with
labels, such as the dollar sign, but these are nonmathematical and
no computer needs them. In the strictly mathematical procedure, as
in the process of dream, the pattern of relationships controls all
operations, but the relata are unidentified.

We return now to the contrast between the iconic method of
creating redundancy in the universe, organism plus other organism,
by the emission of parts of interactive pat-terns and the linguistic
device of naming the relata. We noted above that the human
communication which creates redundancy in the relationships


between persons is still preponderantly iconic and is achieved by
means of kinesics, paralinguistics, intention movements, actions,
and the like. It is in dealing with the universe, message plus
environment, that the evolution of verbal language has made the
greatest strides.

In animal discourse, redundancy is introduced into this universe
by signals which are iconic parts of the signaler’s probable response.
The environmental items may serve an ostensive function but
cannot, in general, be mentioned. Similarly, in iconic
communication about relationship, the relata—the organisms
themselves—do not have to be identified because the subject of any
predicate in this iconic discourse is the emitter of the signal, who is
always ostensively present.

It appears then that at least two steps were necessary to get from
the iconic use of parts of patterns of own behavior to the naming of
entities in the external environment: there was both a change in
coding and a change in the centering of the subject-predicate frame.

To attempt to reconstruct these steps can only be speculative,
but some remarks may be offered:

(1)Imitation of environmental phenomena makes it possible. to
shift the subject-predicate frame from the self to some
environmental entity while still retaining the iconic code.

(2)A similar shifting of the subject-predicate frame from self to
other is latent in those interactions between animals in which A
proposes a pattern of interaction and B negates this with an iconic
or ostensive “don’t.” The subject of B’s message here verbalized as
“don’t” is A.
(3)It is possible that the paradigms of interaction which are
basic to iconic signaling about relationship could serve as
evolutionary models for the paradigms of verbal grammar. We
should not, I suggest, think of the earliest rudiments of verbal
communication as resembling what a man does with only a few
words of a foreign language and no knowledge of its grammar and
syntax. Surely, at all stages of the evolution of language, the
communication of our ancestors was structured and formed-complete
in itself, not made of broken pieces. The antecedents of
grammar must surely be as old or older than the antecedents of

(4)For actions of the self, iconic abbreviations are readily
available, and these control the vis-à-vis by implicit reference to
interactional paradigms. But all such communication is necessarily
positive. To show the fangs is to mention combat, and to mention
combat is to propose it. There can be no simple iconic
representation of a negative: no simple way for an animal to say “I
will not bite you.” It is easy, however, to imagine ways of
communicating negative commands if (and only if) the other
organism will first propose the pat-tern of action which is to be
forbidden. By threat, by inappropriate response and so on, “don’t”
can be communicated. A pattern of interaction, offered by one
organism, is negated by the other, who disrupts the proposed
But “don’t” is very different from “not.” Commonly, the
important message “I will not bite you” is generated as an
agreement between two organisms following real or ritual combat.
That is, the opposite of the final message is worked through to reach
a reductio ad absurdum which can then be the basis of mutual
peace, hierarchic precedence, or sexual relations. Many of the
curious interactions of animals, called “play,” which resemble (but
are not) combat are probably the testing and reaffirmation of such
negative agreement.

But these are cumbersome and awkward methods of achieving
the negative.

(5)It was suggested above that the paradigms of verbal grammar
might somehow be derived from the paradigms of interaction. We,
therefore, look for the evolutionary roots of the simple negative
among the paradigms of interaction. The matter, however, is not
simple. What is known to occur at the animal level is the
simultaneous presentation of contradictory signals—postures which
mention both aggression and flight, and the like. These ambiguities
are, however, quite different from the phenomenon familiar among
humans where the friendliness of a man’s words may be contradicted
by the tension or aggressiveness of his voice or posture. The man is
engaging in a sort of deceit, an altogether more complex
achievement, while the ambivalent animal is offering positive
alternatives. From neither of these patterns is it easy to derive a
simple “not.”


(6)From these considerations it appears likely that the
evolution of the simple negative arose by introjection or imitation
of the vis-à-vis, so that “not” was somehow derived from “don’t.”
(7)This still leaves unexplained the shift from communication
about interaction patterns to communication about things and
other components of the external world. This is the shift which
determines that language would never make obsolete the iconic
communication about the contingency patterns of personal
Further than that we cannot at present go. It is even possible
that the evolution of verbal naming preceded the evolution of the
simple negative. It is, however, important to note that evolution of
a simple negative would be a decisive step toward language as we
know it. This step would immediately endow the signals—be they
verbal or iconic–with a degree of separateness from their
referents, which would justify us in referring to the signals as
“names.” The same step would make possible the use of negative
aspects of classification: those items which are not members of an
identified class would become identifiable as nonmembers. And,
lastly, simple affirmative indicative statements would become

Conscious Purpose versus Nature*

Our civilization, which is on the block here for investigation and
evaluation, has its roots in three main ancient civilizations: the
Roman, the Hebrew and the Greek; and it would seem that many of
our problems are related to the fact that we have an imperialist
civilization leavened or yeasted by a downtrodden, exploited colony
in Palestine. In this conference, we are again going to be fighting out
the conflict between the Romans and the Palestinians.

You will remember that St. Paul boasted, “I was born free.” What
he meant was that he was born Roman, and that this had certain
legal advantages.

We can engage in that old battle either by backing the
downtrodden or by backing the imperialists. If you are going to fight
that battle, you have to take sides in it. It’s that simple.

On the other hand, of course, St. Paul’s ambition, and the
ambition of the downtrodden, is always to get on the side of the
imperialists—to become middle-class imperialists themselves—and
it is doubtful whether creating more members of the civilization
which we are here criticizing is a solution to the problem.

There is, therefore, another more abstract problem. We need to
understand the pathologies and peculiarities of the whole Romano-
Palestinian system. It is this that I am interested in talking about. I
do not care, here, about defending the Romans or defending the
Palestinians—the upper dogs or the underdogs. I want to consider
the dynamics of the whole traditional pathology in which we are
caught, and in which we shall remain as long as we continue to
struggle within that old conflict. We just go round and round in
terms of the old premises.

Fortunately our civilization has a third root—in Greece. Of
course Greece got caught up in a rather similar mess, but still there
was a lot of clean, cool thinking of a quite surprising kind which
was different.

* This lecture was given in August, 1968, to the London Conference on the
Dialectics of Liberation, and is here reprinted from Dialectics of Liberation by
permission of the publisher, Penguin Books Inc.

Let me approach the bigger problem historically. From St.
Thomas Aquinas to the eighteenth century in Catholic countries, and
to the Reformation among Protestants (be-cause we threw out a lot
of Greek sophistication with the Reformation), the structure of our
religion was Greek. In mid-eighteenth century the biological world
looked like this: there was a supreme mind at the top of the ladder,
which was the basic explanation of everything downwards from that
—the supreme mind being, in Christian