The Organization

of Behavior
A NEUROPSYCHOLOGICAL THEORY
This page intentionally left blank
The Organization
of Behavior
A NEUROPSYCHOLOGICAL THEORY
D. O. HEBB
McGill University
|Q\ LAWRENCE ERLBAUM ASSOCIATES, PUBLISHERS
^" Mahwah, New Jersey London
Copyright 1949 by John Wiley & Sons Inc.
Copyright 2002 by Mary Ellen Hebb
This book or any part thereof must not be reproduced in
any form without the written permission of the pub-
lisher or copied in the United States in excess or fair use
without a license from the Copyright Clearance Center
(978-750-8400) or in Canada in excess of fair dealing
without a license from CANCOPY (1-800-893-5777).
Lawrence Erlbaum Associates, Inc., Publishers
10 Industrial Avenue
Mahwah, NJ 07430
Cover design by Kathryn Houghtaling Lacey
Library of Congress Cataloging-in-Publication Data
Hebb, D. O. (Donald Olding)
The organization of behavior: a neuropsychological theory /
D. O. Hebb.
p. cm.
Originally published: New York : Wiley, 1949. (A Wiley
book in clinical psychology). With new foreword.
Includes bibliographical references and index.
ISBN 0-8058-4300-0 (paper : alk. paper)
1. Neuropsychology. I. Title.
BF181 .H4 2002
150.19'8dc21 2002018867
CIP
Books published by Lawrence Erlbaum Associates
are printed on acid-free paper, and their bindings are
chosen for strength and durability.
Printed in the United States of America
1 0 9 8 7 6 5 4 3 2
To
G. C. H.
y
C. Q H.
and
A. M. H.
DONALD OLDING HEBB
Photograph by Richard E. Brown, April 1979.
Foreword
Richard E. Brown
Dalhousie University
Peter M. Milner
McGill University
Since the publication of The Organization of Behavior in 1949, it
has become one of the most influential books in Psychology and
Neuroscience (over 4100 SCI citations since 1989). According to
Adams (1998, p. 419): "Two of the most influential books in the
history of biology are Darwin's On the Origin of Species (1859) and
Hebb's Organization of Behavior (1949). The books are famous be-
cause of the simple but powerful ideas that they contain." While
Darwin's book can be purchased in any bookshop, Hebb's book
has been out of print since 1966, ensuring that Hebb's comment
that a classic normally means "cited but not read" (Fentress,
1987, p. 105) would be true in his case. One of us (REB) found this
situation intolerable and decided to organize a reprinting of this
book. This new edition of The Organization of Behavior, therefore,
rectifies a long-standing problem for behavioral neurosci-
entiststhe inability to obtain one of the most cited publications
in the field. A complete list of Hebb's publications from 1925 to
1987 is also included.
The Organization of Behavior played a significant part in the de-
velopment of physiological psychology; the investigation of the
neural foundations of behavior (Fentress, 1999; Milner 1999). It
introduced the idea of the "Hebb synapse" and the "Hebbian cell
assembly." The more frequently cited of these two ideas is the
Hebb synapse, but, as pointed out by Seung (2000, p. 1166):
"Hebb's synapse is not his only legacy. Even after half a century,
his theory as a whole is still inspiring, because it is a general
Fl
F2 Foreword
framework for relating behavior to synaptic organization
through the dynamics of neural networks." Hebb was amused
that as a result of his neurophysiological postulate, learning syn-
apses had acquired his name"because this postulate is one of
the few aspects of the theory he did not consider completely origi-
nal. Something like it had been proposed by many psychologists
before him, including Freud in his early years as a neuro-
biologist." (Milner, 1993, p.127).
1. A Short Biography of Donald O. Hebb
Donald Olding Hebb was born on July 22nd, 1904 in the village of
Chester, Nova Scotia, Canada. Both of his parents were physi-
cians and his mother was the third woman to obtain an M.D. from
the medical school of Dalhousie University in Halifax, Nova Sco-
tia. His earliest ambition was to write a successful novel so he
studied English and Philosophy at Dalhousie University, gradu-
ating in 1925. Although his degree was in Arts, he also took
classes in chemistry, physics and mathematics, obtaining a sec-
ond-class distinction in mathematics and a first class distinction
in physics. He spent the next year teaching at the high-school he
had attended in Chester. During this time he discovered Freud
and James and began to study psychology. He then moved to
Montreal and in 1929 became a part-time graduate student at
McGill University, taking a graduate seminar in systematic psy-
chology (Dalhousie University had no Department of Psychol-
ogy at the time). To support himself he took a job as a teacher and
after a year he was promoted to principal of an elementary school
in Montreal. During the next few years he became an enthusiastic
reformer of teaching methods [1; numbers refer to Hebb's publi-
cation list], but his reforms were not always appreciated and he
became a full-time graduate student in psychology at McGill in
1931. He completed an MA thesis, entitled "Conditioned and
Unconditioned Reflexes and Inhibition" in April 1932. Al-
though Hebb [ 112, p. 283], in the light of subsequent work, dis-
missed this MA thesis as nonsense, it does contain original ideas
and was passed "cum laude" by two examiners. As the thesis was
theoretical, Hebb was encouraged to do experimental work,
which he started under the supervision of one of his examiners,
Dr. Boris Babkin, a former student of Pavlov.
Foreword F3
During the writing of his MA thesis, Hebb was bedridden with
a tubercular hip, and his experimental work ended when his wife
was killed in a car accident. He became depressed and disen-
chanted with the direction of his research and the McGill psy-
chology department. Encouraged by Professor Babkin, Hebb
wrote to Karl Lashley at the University of Chicago and was ac-
cepted there as a PhD student in July 1934. Lashley was an emi-
nent and respected physiological psychologist. In fact, during
the 1920's his laboratory was almost the only one in America
where genuine research on the neural bases of behavior was being
done. He was at the peak of his career when Hebb became his stu-
dent. Hebb learned a great deal during the year he spent in Chi-
cago, with lectures from such giants as Thurstone, Kohler,
Herrick, and Lashley.
The next year Lashley was offered a professorship at Harvard
and Hebb obtained permission to accompany him, so his PhD was
awarded by Harvard University. His PhD thesis (Hebb, 1936) ex-
amined the innate organization of visual perception in the rat and
the results of this research were published in a series of three pa-
pers [2,3,4], In these papers, Hebb concluded that some aspects of
rat vision are innate, though by the time he wrote this book he had
reverted to a more empiricist position. Hebb remained at Harvard
for another year as a teaching assistant and to complete research on
place learning in rats that he had started in Chicago.
His first appointment after leaving Harvard (1937) was a
two-year fellowship with the neurosurgeon Wilder Penfield, di-
rector of the newly established Montreal Neurological Institute.
His responsibility was to investigate intellectual deficits in pa-
tients with cortical lesions, mostly those resulting from opera-
tions performed for the relief of epilepsy. He was particularly
intrigued by the finding that patients with large frontal-lobe le-
sions had normal intelligence-test scores, some even improved
after operation. He also detected visual impairment in a patient
after right temporal lobectomy, a cortical region not previously
associated with vision. Hebb's work with Penfleld at the Mon-
treal Neurological Institute developed his interest in human
brain function and the behavioral effects of neurological damage
[9], especially with respect to intelligence [7,8,13] and initiated a
long-term collaboration between the two men. Indeed, Hebb's
student, Brenda Milner, began her work with Penfield and went
F4 Foreword
on to study H. M., surely one of the most famous patients in the
history of neuropsychology.
At the end of his Fellowship Hebb accepted a position at
Queen's University in Kingston, Ontario and studied the effect
of cortical lesions on rat learning. He wrote a paper in which he
made the unheard of proposal that early experience could perma-
nently influence intelligence [18], a theme that runs through
much of Hebb's later writing, including this book. He also devel-
oped the Hebb-Williams maze during this time, although the pa-
per was not published until much later [34]. This work arose from
Hebb's interest in intelligence testing and he was attempting to
develop a Stanford-Binet test for rats.
In the meantime, Lashley had been appointed Director of the
Yerkes Laboratories of Primate Biology in Orange Park, Florida.
In 1942 he persuaded Hebb to join him in an investigation of the
effects of brain lesions on intelligence and emotion in chimpan-
zees. During the five years he spent in Florida, Hebb devised
tests of chimpanzee emotion as well as making some pioneer ob-
servations of dolphin behavior [38]. His work with chimpanzees
developed his interest in emotionality [28] and led him to com-
pare emotional problems in chimpanzees and humans [35]. Con-
cerning his time working with chimpanzees, Hebb said that he
"learned more about human beings during that time than in any
other five-year period of my life, except the first." [112, p. 293]. It
was during this period with Lashley and his colleagues that Hebb
wrote most of The Organization of Behavior.
In 1947 Hebb returned to McGill [at a salary of $5500 per year,
plus $500 moving expenses] as one of four new professors re-
cruited to revive a psychology department that had all but ceased
to exist during the war. The Chairman, Robert MacLeod, moved
to Cornell the next year and Hebb occupied the Chair for the next
ten years. When The Organization of Behavior was published in
1949, it was, to Hebb's astonishment, an immediate success, but
he was in no position to sit back and enjoy the acclaim. After the
rather sedate pace of life in Orange Park he was now teaching
three undergraduate courses, including the large introductory
course, plus an evening graduate seminar, setting up animal labs
and arranging funding for research on the effects of rearing on
perception and intelligence in rats and dogs. He was also prepar-
ing an introductory textbook for his course [61]. This left him
Foreword F5
with little time, or incentive, to tinker with his neural theory or
to revise it in the light of the revolutionary developments in
neuroanatomy, neurophysiology and information theory that
were taking place.
Under Hebb, McGill was at the forefront of the physiological
approach to psychology. McGill graduates of that era were in de-
mand in all parts of the world. Hebb's textbook, published in
1958, was distinctly physiological in scope and further contrib-
uted to the dissemination of his neural theories of mental pro-
cesses. For the next 25 years he taught his notoriously rigorous
graduate seminar (mandatory for every graduate student) and In-
troductory Psychology to classes of more than a thousand stu-
dents. In 1970 he was appointed to a four-year term as Chancellor
of McGill. He was successively elected President of the Canadian
Psychological Association and the American Psychological As-
sociations; he was also elected a Fellow of both the Royal Society
of London and the American Academy of Arts and Sciences. He
received many other honors, including innumerable honorary
degrees. In 1976 he retired to his family home in Nova Scotia and
was appointed honorary Professor at Dalhousie University in
1977. He died in 1985 following a hip replacement operation.
2. The Organization of Behavior
When did Hebb develop the ideas that went into The Organization
of Behavior'? The idea for the Hebb synapse first appeared in 1932
in his MA thesis, where he discussed the functioning of the syn-
apse in conditioned reflexes. Hebb opens his introduction by
stating that: "The purpose of this paper is to present a theory of
the functioning of the synapse based on the experimental work of
Sherrington and Pavlov, on reflexes and inhibitions. The impli-
cations of these things for psychological theory, in some aspects,
has been far from clear. In looking for a firm basis for psychology
in physiology, there are some peculiarities about the results of
both investigators which demand serious consideration and sug-
gest another interpretation of their work" (Hebb, 1932, p. 2).
In Figure 1 of his thesis, Hebb makes his first attempt to illus-
trate the synaptic changes associated with conditioning. The
figure shows branches of a stimulus input going to two uncondi-
tioned reflex arcs. Pavlov's experiments show that if one of the
F6 Foreword
FIG. I
w-xl
reflex area.
T-Zj
A la another receptor.
b, c , d, e , connecting neurons.
(Arrowhead gives direction In which lapulsea
are transmitted.)
FIGURE 1.
reflex arcs is active when a stimulus is applied to the input, the
branch going to that reflex becomes more potent; branches go-
ing to inactive effectors do not. Hebb went further, assuming
that the potency of other branches declines. He summarized the
first section of his thesis as follows: "An excited neuron tends to
decrease its discharge to inactive neurons, and increase this dis-
charge to any active neuron, and therefore to form a route to it,
whether there are intervening neurons between the two or not.
With repetition this tendency is prepotent in the formation of
neural routes" (Hebb, 1932, p. 13).
This may be seen as the emergence of the Hebb synapse idea.
Hebb underlined it (the typewriter code for italics) as he later
italicized the corresponding postulate in his book.
On November 24th 1934, Hebb submitted a revised version of
his MA thesis as a term paper for Anatomy 316 (Elementary Neu-
Foreword F7
rology), which he took as a PhD student at the University of Chi-
cago. This paper, entitled "The interpretation of experimental
data on neural action" was noted as "A very thoughtful paper" by
C. J. Herrick. In this paper, Hebb shows two diagrams of possible
synaptic activity during conditioning and states that: "There is
as yet no understanding theoretically of the process of condition-
ing, and when this is understood, it may throw light on the reflex
activity of the [spinal] cord, and perhaps account for the exis-
tence of some of the apparently rigidly inherited patterns. The
conception of block and facilitation at the synapse postulates a
process which can strengthen and perpetuate a route once
formed, but none whatever to account for the establishment of
the route in the first placea most important weakness in the
whole theory" (Hebb, 1934, p. 15).
Hebb's first use of the term "Organization of Behavior" was in
the title of two papers published in 1938 [5,6] on the field orienta-
tion of the rat. In these papers Hebb examined the effects of corti-
cal lesions on learning visual discriminations. Although these
papers have the mark of Lashley on them, Hebb's focus was to de-
velop a better behavioral test for the study of the effects of cere-
bral destruction. He states that: "It is clear that the success of the
physiological analysis depends on the adequacy of the behavioral
tests used. Much of the difficulty found in the evaluation of the
effects of cerebral destruction, both clinically and with animal
experimentation, is due to the fundamental difficulty of the anal-
ysis of behavior and to the unsatisfactory available accounts of
it." [5, p. 333]. In light of recent research on context condition-
ing, it is interesting that these studies discussed the importance
of room cues on the discrimination of objects in an open field.
During the years between 1934 and 1946, Hebb continued to
work on his ideas about the organization of behavior. While still
in Chicago in 1934, he wrote five chapters (90 pages) of a first
draft of a book entitled "Scientific methods in psychology: A
theory of epistemology based on objective methods in psychol-
ogy." Parts of this book were to be included as a final chapter in
The Organization of Behavior but were deleted during the edito-
rial process.
In Feburary, 1944 Hebb discovered that Lorente de No had
recently shown that closed circuits were to be found through-
out the brain, and "that one neuron by itself may not be able to
F8 Foreword
excite a second neuron at the synapse, but can do so if sup-
ported by simultaneous action from another neuron" [112, p.
295]. It was these ideas (of neural loops) of Lorente de No that
Hebb realized "were just what he needed to develop a more re-
alistic theory of the mind" (Milner, 1993, p. 127). On 28 April,
1944, Hebb wrote to Lorente de No to ask if he could spend a
month at the Rockefeller Institute to "become familiar with
your work." When Lorente de No replied on 1 May 1944 that
"At present my work is concerned with the relationship be-
tween the production of the nerve impulse and the metabo-
lism of the nerve, a problem that is of little immediate interest
to a psychologist," Hebb decided to delay his study trip to
Rockefeller and wrote to Lorente de No (4 May 1944) that "I
am profoundly convinced that psychological progress will de-
pend on research such as yours ... that psychological theory
can go no f ar t her wi t hout a more det ai l ed basi s in
neurophysiological fact."
Hebb first outlined his ideas for The Organization of Behavior
in five pages of typed notes in June 1944. These notes contain
his early thoughts on the term "organization." In this defini-
tion, he considered both behavioral and physiological organiza-
tion and considered the "concept of set" which he used to
denote the pre-existing state of the nervous system encountered
by an incoming stimulus. The second draft of The Organization
of Behavior was developed in March-July 1945 and written un-
der the title "Precis: The structure of a set of neuropsycho-
logical speculations." A complete draft of The Organization of
Behavior was finished in 1946 and sent to Lashley, who wrote
two pages of comments on the manuscript in February 1947. In
June 1947, the book was submitted to Charles Thomas publish-
ers. Henry Nissen received the manuscript from Thomas on 21
June 1947. It was 150 pages long and insured for $300.00.
Nissen edited this manuscript from June 1947 to September
1948, sending Hebb numerous comments and suggestions for
changes. On 22 September, 1948, Hebb sent the final draft of the
manuscript to the publishers. However, in January, 1949,
Thomas decided not to publish the book, citing "circumstances
beyond their control" and returned it to Hebb. After consulting
with Beach, who suggested Hoeber (the publishers of his 1948
book, Hormones and Behavior), and considering Appleton-Cen-
Foreword F9
tury as publishers, Hebb sent the book to John Wiley and Sons,
who accepted it with great praise on 18 March 1949 and pub-
lished it in the fall of 1949.
On April 26, 1949, Hebb wrote to Lorente de No that "I be-
lieve that my book will be able to show that modern ideas in
neurophysiology, and particularly some of those that you have
developed, have a revolutionary significance for psychological
theory."
3. The Cell Assembly: Conception.
Ever since his observations of patients at the Montreal Neurolog-
ical Institute, Hebb had puzzled over the nature of thought and
the way ideas might be represented in the brain. The relevance to
his problem of Lorente de No's recurrent loops struck home to
him in 1944, as mentioned earlier, and provided the key to his
theory. Lashley (1942), in a paper written before Hebb arrived in
Orange Park, raised the possibility that reverberations of
Lorente de No's loops might play a role in vision. His use of the
loops as oscillators for his wave-interference theory was more
limited and very different from Hebb's grand scheme to repre-
sent concepts with them, but Hebb never refers to it in his book,
which maybe one reason for Lashley's less than enthusiastic re-
ception of it.
In the absence of evidence to the contrary, Hebb assumed that
before they became cell assemblies, Lorente de No's loops were
randomly connected to sensory input. When a stimulus was im-
pressed on the system some loops would receive more input than
others, depending on the pattern of the connections, and they
would therefore reverberate more intensely and persistently. Af-
ter a number of such episodes the connections within the most
vigorously reverberating group of cells would be strengthened to
such an extent that partial input, or even remote associations,
would subsequently activate the whole assembly.
Hebb postulated that when an assembly was firing as a result
of sensory input, its activity represented the perception of the
stimulus; when it fired in the absence of the corresponding sen-
sory input the activity represented the concept of the stimulus.
When a sequence of stimuli occurred frequently, Hebb assumed
that the aroused assemblies would become associated with each
FIO Foreword
other via strengthened cross-linkages, to establish what he
called phase sequences.
Having established the physical reality of ideas and their
mode of association, Hebb was able to discuss expectancies, at-
tention, attitudes, intelligence, even mental illness, in the rest
of the book with a freedom not permitted to a serious psycholo-
gist in America for almost half a century. It is probable that this
contributed to the popularity of the book among psychologists
who had been deprived of such rich fare for too long. The ear-
lier, more speculative chapters, on the other hand, attracted
the attention of computer scientists, engineers, philosophers
and biologists of all persuasions. Before its publication, Hebb
used a manuscript of the book in his graduate seminar at
McGill and it fell into the hands of one of us (PMM, who had
spent the war working on radar and atomic energy). After read-
ing a few chapters he applied to do graduate work with Hebb
and was accepted.
Hebb's neurophysiological postulate (p. 62) was of particular
interest to physiologists. Psychologists have always assumed
that association requires contiguity of two or more events, but
many physiologists thought that learning occurred when a con-
nection was used repeatedly. As Hebb pointed out in his MA
thesis, if such a synapse is initially ineffective it can never in-
crease in strength, however often it fires. The implication of
Hebb's postulate is that synapses increase in strength only if the
post-synaptic neuron is fired by some other input at about the
same time as a pre-synaptic impulse occurs. When synapses that
behaved in this way were found, neurophysiologists called them
"Hebb synapses" and the name has stuck.
Learning synapses in computer models of the brain are also
often referred to as Hebb synapses, though usually they do
not adhere strictly to Hebb's definition. In order to prevent
all synapses from increasing indefinitely in strength, it is ad-
ditionally postulated that synaptic effectiveness is reduced
when pre- and post-synaptic elements fire at separate times.
It is interesting that in his 1932 MA thesis, the synaptic learn-
ing process described by Hebb (see page xi above) included a
process by which synaptic strength was reduced when the
post-synaptic membrane did not fire at the same time as the
incoming impulse.
Foreword Fll
4. The Cell Assembly: Haifa century later.
Hebb's notes indicate that the chapters in which the cell assembly
and phase sequence are developed cost Hebb much time and effort,
and he was aware that too often speculation had to be substituted
for vital neurological and anatomical information that was not
available. On page xl of the Introduction he gives the reader per-
mission to skip those chapters. He argues strongly in defense of his
neural speculation, but on page 80 he tells us that he expects the
model to be amended in the future, as better information about the
nervous system becomes available. The future is now with us, and
it is possible to see that Hebb was not fortunate in his timing rela-
tive to the state of neurophysiological knowledge.
As the book was being published, information was coming to
light concerning the importance of the reticular system of the
brainstem for arousal. During the next few years the relevance to
behavior of subcortical structures such as the amygdala, septal
area and hippocampus began to be evident. Furthermore, as he
worked on the book, Hebb accepted the widespread belief of most
physiologists and avoided incorporating synaptic inhibition
(though a theory to account for it was advanced in his MA thesis),
only to have J. C. Eccles (Eccles et al., 1954), previously the most
vociferous opponent of the idea, conclusively demonstrate its ex-
istence a year or two later.
Somewhat more recently, new genetic, histological and brain
imaging techniques have fueled the rise of a modern phrenology,
quite at variance with the anti-localization sentiment prevailing
when this book was being written. Even if the present trend to-
ward nativism is sometimes exaggerated there seems little doubt
that Hebb's assumption that cortical connections are initially
random was overcautious. His argument that vision developed
very slowly in human babies was based on two very dubious ob-
servations. The impaired vision of chimpanzees reared in the
dark turned out to be caused by irreversible damage to the retina,
and von Senden's reports of the difficulties of people given sight
as adults after being born blind could be plausibly explained as
damage to innate connections due to lack of normal stimulation
during development.
Even the simple purity of Hebb's neurophysiological learning
postulate has proved illusory. We now know that synaptic activity
F12 Foreword
gives rise to a complex sequence of molecular changes, some
brief, others lasting for hours, days, and sometimes permanently,
that change the characteristics of both sides of the synaptic junc-
tion (e.g. Greengard, 2001). The inelegant reverberating loops of
the cell assembly, almost as impervious to research as the mental
activity they were meant to replace, are no longer needed to ex-
plain the persistence of thoughts. The stimulated neurons may
continue to fire because of a persisting intrinsic change rather
than by recurrent restimulation.
Without Hebb's speculation many of these discoveries might
not have been made, or only after a longer delay. The biological
speculations in chapters 4 and 5 of this book are interesting from
a historical point of view, though as neuroscience they have now
been largely superceded. Many of the later chapters, being based
mainly on more reliable behavioral observations, continue to be
valid and to the new reader they frequently bring the tingle of en-
countering a brilliant, and often witty, insight into a problem.
When he embarked upon the book, Hebb's original purpose
was to explain in neural terms the processes whereby early expe-
riences build the concepts required for subsequent intelligent be-
havior. He made some progress towards that end and pioneered
the way for the vast armies of neuroscientists presently engaged
in the task. What he succeeded in doing was to engineer a change
in the philosophical outlook of behavioral scientists. Tradi-
tionally psychologists were dualists, believing that mental pro-
cesses take place in an incorporeal mind; a "black box" whose
workings could only be inferred by introspection or by observa-
tions of the behavior it produced. Hebb offered a more direct win-
dow to the mind, investigation of its neural machinery. His
monograph held the promise of a genuinely monist psychology.
5. Why Hebb remains important today
It is difficult to read a paper in the neurobiology of learning and
memory today that does not mention its debt to Hebb. To cite but
one example, Dobrossy and Dunnett (2001, p. 871) state that "The
mechanisms by which environment and experience can influence
brain structure and function first received direct experimental at-
tention after Hebb's observations and theoretical proposals that
learning is represented in the brain by synaptic changes in re-
Foreword F13
sponse to repeated concurrent activation." The reviews of Hebb's
ideas by Fentress (1999) and Milner (1999) show how Hebb has in-
fluenced research in many areas of psychology and neuroscience
and how our findings have now gone beyond Hebb's wildest specu-
lations. Hebb's cell assembly theory is becoming more important
for guiding research (Sakurai, 1999) while the Hebb synapse re-
mains a central component of molecular biological (Tsien, 2000)
and computational (Klemm et al, 2000) models of learning. As
stated by Ray Klein (1999, p. 3), "In the scientific literature, refer-
ences to Hebb, the Hebbian cell assembly, the Hebb synapse and
the Hebb rule increase each year. These forceful ideas of 1949 are
now applied in engineering, robotics, and computer science as well
as neurophysiology, neuroscience and psychologya tribute to
Hebb's scientific acumen, foresight and courage to put forth a
foundational neuropsychological theory of the organization of be-
haviour."
Acknowledgements
We would like to thank the McGill University Archives and, in
particular, Gordon Burr, Senior Archivist, for his help in obtain-
ing copies of archival material.
References
Adams, E (1998). Hebb and Darwin. Journal of Theoretical Biology, 195:
419-438.
Dobrossy, M. D., & Dunnett, S. B. (2001). The influence of environment
and experience on neural grafts. Nature Reviews Neuroscience, 2:
871-879.
Eccles, J. C., Fatt, E, & Landgren, S. (1954). The 'direct' inhibitory pathway
in the spinal cord. Australian Journal of Science, 16: 130134.
Fentress, J. C. (1987). D. O. Hebb and the developmental organization of
behavior. Developmental Psychobiology, 20: 103-109.
Fentress, J. C. (1999). The organization of behaviour revisited. Canadian
Journal of Experimental Psychology, 53: 8-19.
Greengard, E (2001). The neurobiology of slow synaptic transmission. Sci-
ence 294: 1024-1030.
Hebb, D. O. (1932). Conditioned and unconditioned reflexes and inhibi-
tion. Unpublished MA Thesis, McGill University, Montreal, Quebec,
April 1932.
Hebb, D. O. (1934). The interpretation of experimental data on neural ac-
tion. Unpublished manuscript, University of Chicago, Nov 24,1934.
F14 Foreword
Hebb, D. O. (1936). The innate organization of visual perception in the rat.
Unpublished PhD Thesis, Harvard University, March, 1936.
Klein, R. M. (1999). The Hebb legacy. Canadian Journal of Experimental
Psychology, 53: 1-3.
Klemra, K., Bornholdt, S., & Schuster, H. G. (2000). Beyond Hebb: Exclu-
sive-OR and biological learning. Physical Review Letters, 84:3013-3016.
Milner, P M. (1993). The mind and Donald O. Hebb. Scientific American,
268(1), 124-129.
Milner, P M. (1999). The Autonomous Brain. Lawrence Erlbaum Associates,
Mahwah, NJ.
Sakurai, Y. (1999). How do cell assemblies encode information in the
brain? Neuroscience and Biobehavioral Reviews, 23: 785-796.
Seung, H. S. (2000). Half a century of Hebb. Nature Neuroscience, supple-
ment, 3,1166.
Tsien, J. Z. (2000). Linking Hebb's coincidence-detection to memory for-
mation. Current Opinion in Neurobiology, 10: 266-273.
Donald Olding Hebb: Publications in Chronological Order:
1925-1987.
Numbering system based on: Milner, P M. & Milner, B. (1996). Donald
Olding Hebb: 22 July 190420 August 1985. Biographical memoirs of the
Royal Society of London 42: 193-204.
_. Hebb, D. O. (1925). 25 class history. Dalhousie Gazette, 25-29.
1. Hebb, D. O. (1930). Elementary school methods. Teacher's magazine
12(51): 23-26.
2. Hebb, D. O. (1937). The innate organization of visual activity: I. Percep-
tion of figures by rats reared in total darkness.Journal of Genetic Psychol-
ogy 51: 101-126.
3. Hebb, D. O. (1937). The innate organization of visual activity. II. Trans-
fer of response in the discrimination of brightness and size by rats
reared in total darkness.Journal of Comparative Psychology 24:277-299.
4. Hebb, D. O. (1938). The innate organization of visual activity. HI. Dis-
crimination of brightness after removal of the striate cortex in the rat.
Journal of Comparative Psychology 25: 427-437.
5. Hebb, D. O. (1938). Studies of the organization of behavior. I. Behavior of
the rat in a field orientation. Journal of Comparative Psychology 25:
333-353.
6. Hebb, D. O. (1938). Studies of the organization of behavior. II. Changes
in the field orientation of the rat after cortical destruction. Journal of
Comparative Psychology 26: 427-441.
7. Hebb, D. O. (1939). Intelligence in man after large removals of cerebral
tissue: Defects following right temporal lobectomy. Journal of General
Psychology 21: 437-446.
Foreword F15
8. Hebb, D. O. (1939). Intelligence in man after large removals of cerebral
tissue: Report of four left frontal lobe casts.Journal of General Psychology
21:73-87.
9. Hebb, D. O., & Penfield, W. (1940). Human behavior after extensive bi-
lateral remov,al from the frontal lobes.Archives of Neurology & Psychiatry
(Chicago) 43: 421-438.
10. Hebb, D. O. (1940). Clinical tests of adult intelligence. Psychological
Bulletin 37: 513-514.
11. Hebb, D. O. (1941). Clinical evidence concerning the nature of normal
adult test performance. Psychological Bulletin 38: 593 (Abstract).
12. Hebb, D. O. (1941, April). Higher level difficulty in verbal test material.
Bulletin of the CanadianPsychological Association 1: 29 (Abstract).
13. Hebb, D. O. (1941). Human intelligence after removal of cerebral tissue
from the right frontal lobe. Journal of General Psychology 25: 257-265.
14. Hebb, D. O. (1941). The McGill Picture Anomaly Series. Kingston, On-
tario, Author.
15. Hebb, D. O. (1941). The McGill Picture Anomaly Series: Data on 100
unsophisticated adults. Bulletin of the Canadian Psychological Association
1:47-49.
16. Hebb, D. O. & Williams, K. (1941, February). Experimental control of
cues determining the rat's orientation. Bulletin of the Canadian Psycho-
logical Association 1: 22-23 (Abstract)
17. Hebb, D. O. (1942). Observations on cerebral dysfunction. Psychological
Bulletin 39: 491- 492 (Abstract).
18. Hebb, D. O. (1942). The effect of early and late brain injury upon test
scores, and the nature of normal adult intelligence. Proceedings of the
American Philosophical Society 85: 275-292.
19. Hebb, D. O. (1942). The McGill Picture Anomaly Series, M and N. Orange
Park, Fla.: Yerkes Laboratories. 34 pp.
20. Hebb, D. O., & Morton, N. W. (1942). The McGill verbal situation se-
ries. Bulletin of the Canadian Psychological Association 2: 26 (Abstract).
21. Hebb, D. O. (1942). Verbal test material independent of special vocabu-
lary difficulty. Journal of Educational Psychology 33: 691-696.
22. Hebb, D. O., & Morton, N. W. (1942). The McGill Verbal Situation:Series
A andB. Montreal: McGill University. 19 pp. [Book]
23. Hebb, D. O. (1942). Directionsfor the use of the McGill PictureAnomaly Se-
ries, M andN. Montreal: McGill University.
24. Hebb, D. O., & Morton, N. W. (1943). The McGill Adult Comprehen-
sion Examination: Verbal Situation and Picture Anomaly Series.Jour-
nal of Educational Psychology 34: 16-25.
25. Hebb, D. O., & Riesen, A. H. (1943). The genesis of irrational fears. Bul-
letin of the CanadianPsychological Association 3: 49-50.
26. Hebb, D. O., & Morton, N. W. (1944). Note on the measurement of adult
intelligence. Journal of General Psychology 30: 217-223.
27. Hebb, D. O. (1945). Man's frontal lobes:A critical view. Archives of Neu-
rology & Psychiatry (Chicago) 54: 10-24.
F16 Foreword
28. Hebb, D. O. (1945). The forms and conditions of chimpanzee anger.
Bulletin of the CanadianPsychological Association 5: 32-35.
29. Hebb, D. O., & Foord, E. N. (1945). Errors of visual recognition and the
nature of the trace. Journal of Experimental Psychology 35: 335-348.
30. Hebb, D. O. (1946). Behavioral differences between male and female
chimpanzees. Bulletin of the Canadian Psychological Association 6:56-58.
31. Hebb, D. O. (1946). Emotion in man and animal: an analysis of the intu-
itive processes of recognition. Psychological Review 53: 88-106.
32. Hebb, D. O. (1946). On the nature of fear. Psychological Review 53:
259-276.
33. Hebb, D. O. (1946). The objective description of temperament.Ameri-
can Psychologist 1: 275-276 (Abstract).
34. Hebb, D. O., & Williams, K. (1946). A method of rating animal intelli-
gence. Journal of General Psychology 34: 59-65.
35. Hebb, D. O. (1947). Spontaneous neurosis in chimpanzees; theoretical
relations with clinical and experimental phenomena. Psychosomatic
Medicine 9: 3-19.
36. Hebb, D. O. (1947). The effects of early experience on problem solving
at maturity. American Psychologist 2: 306-307 (Abstract).
37. Hebb, D. O. (1948). Research planning in the Canadian Psychological
Association. I. Report on experimental, physiological, and comparative
psychology. Canadian Journal of Psychology 2: 13-14.
38. McBride, A. E, & Hebb, D. O. (1948). Behavior of the captive bot-
tle-nose dolphin, Tursiops truncatus. Journal of Comparative &Physiologi-
cal Psychology 41:111-123.
39. Hebb, D. O. (1949). Temperament in chimpanzees: I. Method of analy-
sis. Journal of Comparative & Physiological Psychology 42: 192-206.
40. Hebb, D. O. (1949). The organization of behavior: a neuropsychological the-
ory. NY: Wiley, xix, 335 pp.
41. Hebb, D. O. (1950). Animal and physiological psychology. Annual Re-
view of Psychology 1: 173-188.
42. Clarke, R. S., Heron, W, Fetherstonhaugh, M. L., Forgays, D. G., &
Hebb, D. O. (1951). Individual differences in dogs: Preliminary report
on the effects of early experience. Canadian Journal of Psychology 5:
150-156.
43. Hebb, D. 0. (1951). The role of neurological ideas in psychology.Jour-
nal of Personality 20: 39-55.
44. Hoyt, R., Elliott, H., & Hebb, D.O. (1951). The intelligence of schizo-
phrenic patients following lobotomy. Treatment Service Bulletin: Depart-
ment of Veteran's Affairs, Canada 6: 553-557.
45. Hebb, D. O., & Bindra, D. (1952). Scientific writing and the general
problem of communication. American Psychologist 7: 569-573.
46. Hebb, D. O., Heron, W, & Bexton, W. H. (1952). The effect of isolation
upon attitude, motivation, and thought. In Fourth Symposium, Military
medicine 1, in cooperation with McGill University. Ottawa: Defense Re-
search Board.
Foreword F17
47. Hebb, D. O. (1953). Heredity and environment in mammalian behav-
iour. British Journal of Animal Behaviour 1: 43-47.
48. Hebb, D. O. (1953). On human thought. Canadian Journal of Psychology
7:99-110.
49. Hebb, D. O. (1953). On motivation and thought. Contributions a Etude
des Sciences de I'Homme 2: 41-47.
50. Hebb, D. O., Heron, W., & Bexton, W. H. (1953). Cognitive effects of a
decreased variation to the sensory environment. American Psychologist
8: 366.
51. Hebb, D. O. (1954). The problem of consciousness and introspection.
In: J. F. Delafresnaye (Ed.) Brain mechanisms of consciousness, pp.
402-421. Oxford: Blackwell.
52. Hebb, D. O., Heath, E. S., & Stuart, E. A. (1954). Experimental deafness.
Canadian Journal of Psychology 8: 152-156.
53. Hebb, D. O., & Thompson, W. R. (1954). The social significance of ani-
mal studies. In: G. Lindzey (Ed.). Handbook of social psychology, Vol. 1,
pp.532-561. Cambridge Mass: Addison-Wesley.
54. Hebb, D. O. (1955). Drives and the C. N. S. (conceptual nervous system).
Psychological Review 62: 243-254.
55. Hebb, D. O. (1955). The mammal and his environment. American Jour-
nal of Psychiatry 111: 826-831.
56. Hebb, D. O., & Heron, W. (1955). Effects of radical isolation upon intel-
lectual function and the manipulation of attitudes. In Terminal report on
conditions of attitude change in individuals. Ottawa: Defense Research
Board.
57. Hebb, D. O., & Mahut, H. (1955). Motivation et recherche du
changement perceptif chez le rat et chez l'homme [Motivation and
search for perceptual change in rat and man.] Journal de Psychologic
Normak et Pathologique 52: 209-221.
58. Hebb, D. O., Murphy, C. W., Kurlents, E., & Cleghorn, R. A. (1955). Ab-
sence of increased corticoid excretion with the stress of perceptual depri-
vation. Canadian Journal of Biochemistry and Physiology 33: 1062-1063.
59. Hebb, D. O. (1956). The distinction between "classical" and "instru-
mental. "Canadian Journal of Psychology 10: 165-166.
60. Bloch, V, & Hebb, D. O. (1956). Etude des phenomenes d'enrayement et
d'activation du comportement par stimulation thalamique et reticulaire
chez le rat non anesthesie. [Phenomena of cessation and initiation of be-
havior by thalamic and reticular stimulation in the non-anesthetized
rat.] Psychologie Fran$aise 1: 8-9.
61. Hebb, D. O. (1958). A textbook of psychology. Philadelphia: Saunders. x,
276 pp.
62. Hebb, D. O. (1958). Alice in Wonderland or Psychology among the bio-
logical sciences. In: H. F. Harlow & C. N. Woolsey (Eds.) Biological and
biochemical bases of behavior, pp. 451-467. Madison: University of Wis-
consin Press [Symposium on interdisciplinary research, University of
Wisconsin, 1955]
F18 Foreword
63. Hebb, D. O. (1958). The motivating effects of exteroceptive stimula-
tion. AmericanPsychologist 13: 109-113.
64. Hebb, D. O. (1959). A neuropsychological theory. In: S. Koch (EdJPsy-
chology: A study of a science, Vol. l,pp. 622-643. New York: McGraw Hill.
65. Hebb, D. O. (1959). Intelligence, brain function and the theory of mind.
Brain, 82,260-275. [23rd Hughlings Jackson Memorial Lecture, Mon-
treal Neurological Institute, 1958]
66. Hebb, D. O. (1959). Karl Spencer Lashley: \89Q-l958.AmericanJournal
of Psychology 72:142-150.
67. Hebb, D. O. (1959, Apr.). Motivation and thought. Bulletin of the Mari-
time Psychological Association 8: 4-9.
68. Beach, F. A., Hebb, D. O., Morgan, C. T., & Nissen, H. W. (Eds.) (1960).
The Neuropsychology of Lashley. New York: McGraw-Hill, xx, 564 pp.
69. Hebb, D. O. (1960). The American revolution. American Psychologist 15:
735-745. [APA Presidential Address]
70. Pritchard, R. M., Heron, W., & Hebb, D. O. (1960). Visual perception
approached by the method of stabilized images. Canadian Journal of Psy-
chology 14: 67-77.
71. Hebb, D. O. (1961). Call for Dr. Finagle, [review of Plans and the structure
of behavior] Contemporary Psychology 5: 209-211.
72. Hebb, D. 0. (1961). Distinctive features of learning in the higher ani-
mal. In: J. F. Delafresnay (Ed.). Brain mechanisms and learning, pp. 42-53.
Oxford: Blackwell.
73. Hebb, D. O. (1961). On the meaning of objective psychology. Transac-
tions of the Royal Society of Canada 55: 81-86.
74. Hebb, D. O. (1961). Sensory deprivation: Facts in search of a theory.
Journal of Nervous & Mental Disease 132: 40-43. [Discussion of sympo-
sium presentation]
75. Hebb, D. O. (1961). The role of experience. In: S. M.Farber and R. H. L.
Wison (Eds.). Man and civilization: Control of the mind, pp. 37-51. New
York: McGraw Hill.
76. Hebb, D. O. (1962). Auditory-oculomotor reflexes at birth. Science 135:
998-999.
77. Hebb. D. O. (1962) John Davidson Ketchum. Transactions of the Royal
Society of Canada, 56,197-198.
78. Hebb, D. O. (1962). The mind of man. McGillNews 44:34-35. [Text of
symposium presentation]
79. Hebb, D.O. (1963). Introduction to Dover reprint. K.S. Lashley, Brain
mechanisms and intelligence, pp. v-xiii. New York: Dover.
80. Hebb, D. 0. (1963). The semiautonomous process: Its nature and nur-
ture. A merican Psychologist 18: 16-27.
81. Hebb, D. O. & Milner, P (1963). Aktivitatsformen der Gehirns und
Verhaltensorganisation. Naturwissenschaftliche Rundschau 16: 258-262.
82. Hebb, D.O. (1965). The evolution of mind. Proceedings of the Royal Soci-
ety, B. 161,376-383.
Foreword F19
83. Hebb, D. O. (1966).^ textbook of psychology. (2nd ed.). Philadelphia: W.
B. Saunders. xvi, 353 pp.
84. Hebbj D. O. (1967). Cerebral organization and consciousness. Research
Publication of the Association for Research in Nervous and Mental Diseases
45: 1-7.
85. Hebb, D. O. (1966, January). Education for research. Canadian Federa-
tion News 8: 53-57.
86. Hebb, D. O. (1968). Concerning imagery. Psychological Review 75:
466-477.
87. Hebb, D. O., & Thompson, W. R. (1968). The social significance of ani-
mal studies. In: G. Lindzey & E. Aronson (Eds.). Handbook ofsocialpsy-
chology. Vol. 2 (2nd edition), pp.729-774. New York: Addison-Wesley.
88. Pribram, K. H., Hebb, D., & Macdonald, G. (1968, Sept.). The ghost in
the machine. Psychological Scene 2: 28-43.
89. Hebb, D. O. (1969, May) The mind's eye.Psychology Today 2(12): 54-57,
67-68.
. Hall, E. (1969, November). Hebb on hocus-pocus: A conversation with
Elizabeth Hall. Psychology Today 3 (6): 21-28.
90. Hebb, D. O., & Favreau O. (1969). The mechanism of perception. Radi-
ology Clinics of North America 7: 393-401.
91. Hebb, D. O. (1970). A return to Jensen and his social science critics.
American Psychologist 25: 568.
92. Hebb, D. O., & Krebs, D. (1971). Comment on altruism: The compara-
tive evidence: Infrahuman altruism. Psychological Bulletin 76:409-414.
93. Hebb, D. O. (1971). Concerning Hebb's criticism of Jensen and the he-
redityenvironment argument: Response to Gordon by Hebb. Ameri-
can Psychologist 26: 665.
94. Hebb, D. O. (1971). The nature of a university education. McGill Jour-
nal of Education 6: 514.
95. Hebb, D. O. (1971). Whose confusion? American Psychologist 26: 736.
96. Hebb, D. O., Lambert, W. E., & Tucker, G. R. (1971). Language, thought,
and experience. Modern Language Journal 55: 212-222.
97. Hebb, D.O. (1972M textbook of psychology. (3rd ed.). Philadelphia: W. B.
Saunders.
98. Hebb, D. O. (1972). Possible test for Hebb's hypothesis concerning im-
ageryReply. Psychology Review 79: 368.
99. Hebb, D. O., Lambert, W. E., & Tucker, R. (1973, April). A DMZ in the
language war. Psychology Today 5, 55-58,60,62.
100. Hebb, D. O. (1974). What psychology is about. American Psychologist
29: 71-79.
101. Hebb, D. O. (1975). Psychological aspects of imagery. In The Brain
mechanisms. A collection of papers dedicated to the 90th birthday of Ivan
Beritashvili, pp. 64-68. Georgian Academy of Sciences, Tiflis.
102. Hebb, D. O. (1975). Science and the world of imagination. Canadian
Psychological Review 16: 4-11.
F20 Foreword
103. Hebb, D. O. (1976). Physiological learning theory.Journal of Abnormal
Child Psychology 4: 309-314.
104. Hebb, D. O. (1977). The frontal lobe. In W. Feindel (Ed.) Wilder Pen-
field: His legacy to neurology,. Canadian Medical Association Journal
116: 1373-1374.
105. Hebb, D.O. (1977). To know your own mind. In: John M. Nicholas
(Ed.) Images, perception and knowledge, pp. 213-219. D. Reidel Pub-
lishing Co., Dordrecht-Holland.
106. Hebb, D. O. (1977). What he gives with one hand.. Contemporary Psy-
chology 22: 849. [Comment on Jerison's overview of Bindra, Contempo-
rary Psychology 22: 417-419].
107. Hebb, D. O. (1978). A problem of localization. Behavioral and Brain
Sciences 1: 357.
108. Hebb, D. O. (1978). Behavioral evidence of thought and conscious-
ness. Behavioral and Brain Sciences 1: 577.
109. Hebb, D. O. (1978, November). On watching myself get old. Psychology
Today 12(6): 15-23.
110. Hebb, D. O. (1978). Open letter: To a friend who thinks IQ is a social
evil. A merican Psychologist 33: 1143-1144.
111. Hebb, D. O. (1978). Review of Divided consciousness: Multiple controls in
human thought and action. American Journal of Psychology 91: 545547.
112. Hebb, D. O. (1980). D. O. Hebb. In: G. Lindzey (Ed.). A history of psy-
chology in autobiography. Vol.VII. pp. 273-309. San Francisco: W.H.
Freeman.
113. Hebb, D. O. (1980). Essay on mind. Hillsdale NJ: Erlbaum.
114. Hebb, D. O. (1980). The structure of thought. In R W. Juscyzk & R. M.
Klein (Eds.) The nature of thought: Essays in honour of D. O. Hebb, pp.
19-35. Hillsdale, N.J.: Lawrence Erlbaum Associates.
115. Hebb, D. O. (1980). The view from without. Philosophy of Social Science
10: 309-315. [Review of The self and its brain].
116. Hebb, D. O. (1980, November). An inside look at ageing. American Psy-
chological Association Monitor, 4-5.
117. Hebb, D. O. (1981). Consider mind as a biological problem. Neurosci-
ence 6: 2419-2422.
118. Hebb, D. O. (1981). Open letter in response to D. O. Hebb: Reply irrel-
evant? American Psychologist 36: 423-424.
119. Hebb, D. O. (1982). Comment on a commentaryIntrospicere ergo
esse?Reply. Neuroscience 7: 2300.
120. Hebb, D. O. (1982, May). Hilgard's discovery brings hypnosis closer to
everyday experience. Psychology Today. 16(5): 52-54.
121. Hebb, D. O. (1983) Neuropsychology: Retrospect and prospect. Cana-
dian Journal of Psychology 37: 47.
122. Hebb, D. O. (1984). Clinical psychology training in Canada: Its devel-
opment, current status, and the prospects for accreditation: Response to
Con way. Canadian Psychology 25: 192.
Foreword F2l
. Hebb, D. O., & Donderi, D. C.(1987). Textbook of psychology (4th ed.).
Hillsdale, NJ, USA: Lawrence Erlbaum Associates, xii, 384 pp.
Note: Buchtel, H. A., Editor. (1982). The Conceptual Nervous System. Ox-
ford: Pergamon Press. [An edited volume containing 21 of Hebb's pub-
lications from 1951 to 1977. These are publications 43,47,49,51,54,62,
65,69,72,73,75,80,82,84,86,90,96,100,102,103,105. But they are not
in their original published format]
Donald Olding Hebb: Biographies/Obituaries
Fentress, J. C. (1987). D. O. Hebb and the developmental organization of
behavior. Developmental Psychobiology 20: 103-109.
Fentress, J. C., & Klein, R. M. editors. (1999). Special issue: The Hebb leg-
acy. Canadian Journal of Experimental Psychology 53:1-131. [contains 10
papers]
Harnad, S. (1985). D. O. Hebb: father of cognitive psychobiology: 1904-1985.
Behavioral and Brain Sciences 8: opp. p. 5 29.
Klein, R. M. (1981). D. O. Hebb: An appreciation. In: E W. Jusczyk & R. M.
Klein (Eds.). The nature of thought: Essays in honour of D.O.Hebb, pp.
1-18. Hillsdale N.J.: Lawrence Erlbaum Associates.
Milner, R (1985). Donald Olding Hebb: 1904-1985.Proceedings of 'the Royal
Society of Canada, Series IV 23: 93-96.
Milner, P (1986). Donald Olding Hebb (1904-1985). Obituary. Trends in
Neuroscience 9: 347-351.
Milner, P M. (1993). The mind and Donald O. Hebb. Scientific American
268(1): 124-129.
Milner, P M. and Milner, B. (1996). Donald Olding Hebb: 22 July 1904-20
August 1985. Biographical memoirs of the Royal Society of London 42:
193-204.
Mogenson, G. J. (1988). Obituaries. Donald Olding Hebb: 1904-1985. Ca-
nadian Psychology 29: 315-316.
This page intentionally left blank
Preface
In this book I have tried to bring together a number of differ-
ent lines of research, in a general theory of behavior that at-
tempts to bridge the gap between neurophysiology and psy-
chology, as well as that between laboratory psychology and the
problems of the clinic.
The book is written, consequently, with the hope that it will
be intelligible to the clinician and the physiologist as well as to
psychologists. The development of psychological theory can be
a cooperative affair, and has often been so in the past: witness
Helmholtz, Jackson, Pavlov, Freud. The clinician and the physi-
ologist frequently have direct access to data of first importance
for psychology, sometimes without recognizing the fact. Though
I have done my best, it may be chimerical to hope that my discus-
sion is extensive and clear enough to stand on its own feet, for
the nonpsychological reader. The reader who needs it will find
more of the details of psychological theory in Morgan (1943)
on physiplogical psychology, Hilgard and Marquis (1940) on the
theory of learning, Woodworth (1938) on "experimental" (nor-
mal human adult) psychology, and Moss (1942) or Maier and
Schneirla (1935) on animal psychology. Of these, Morgan is
most directly relevant, and in several places I have assumed a
knowledge of fact to the extent provided by his text.
It is a pleasure to record my indebtedness to the colleagues
who have read and improved the contents of this book. I owe
much to students in a seminar at Harvard University in the
summer of 1947, and in another at McGill University in the
following winter. Part or all of the manuscript was read by Pro-
fessor Harlow W. Ades, Professor Frank A. Beach, Dr. J. G.
Beebe-Center, Professor R. B. MacLeod, Dr. Francis McNaugh-
ton, Dr. G. A. Miller, Dr. Karl Pribram, Professor H. E. Rosvold,
and Professor R. W. Sperry, and their help is gratefully acknowl-
vii
viii Preface
edged. I am much indebted to Professor Edwin G. Boring,
Professor George Clark, Professor K. S. Lashley, and Dr. J. C. R.
Licklider for their painstaking and detailed criticism of a large
part of the manuscript, improving both matter and style. My
greatest debt, perhaps, is to the weekly colloquium and the per-
sistent theoretical debate at the Yerkes Laboratories of Primate
Biology between 1942 and 1947; and to a small group taking
part therein who have also read the entire manuscript and have
contributed greatly to itProfessor Henry W. Nissen, Mr. and
Mrs. Robt. Blum, and Dr. Austin Riesen. This board of critics
should also include my wife; and to it, for the positive contribu-
tions made as well as for some of the changes and deletions
forced on me, both the reader and I owe much.
Finally I should like to express my gratitude to those who have
helped with clerical work in the preparation of this book, in its
various stages: particularly, Alice Sellers, Therese Mahoney, and
Margaret Halliday. The McGill University Research Fund sup-
ported some of the experimental work that is briefly reported
here, and also made it possible to secure additional clerical as-
sistance.
D. O. HEBB
Montreal
April, 1949
Contents
FOREWORD TO THE REPUBLICATION Fl
Richard E. Brown and Peter M. Milner
PREFACE mi
INTRODUCTION xi
1. THE PROBLEM AND THE LINE OF ATTACK . . 1
2 . SUMMATION AN D LEARNING I N PERCEPTION . . . 1 7
3 . FIELD THEORY AN D EQUIPOTENTIALITY . . . . 3 8
4. THE FIRST STAGE OF PERCEPTION: GROWTH OF THE
ASSEMBLY 60
5. PERCEPTION OF A COMPLEX: THE PHASE SEQUENCE . 79
6. DEVELOPMENT OF THE LEARNING CAPACITY . . 107
7. HIGHER AND LOWER PROCESSES RELATED TO LEARN-
ING 140
8. PROBLEMS OF MOTIVATION: PAIN AND HUNGER . . 171
9. THE PROBLEM OF MOTIVATIONAL DRIFT . . . . 2 0 7
10. EMOTIONAL DISTURBANCES 235
11. THE GROWTH AND DECLINE OF INTELLIGENCE . . 275
BIBLIOGRAPHY 305
INDEX 321
This page intentionally left blank
Introduction
It might be argued that the task of the psychologist, the task
of understanding behavior and reducing the vagaries of human
thought to a mechanical process of cause and effect, is a more
difficult one than that of any other scientist. Certainly the
problem is enormously complex; and though it could also be
argued that the progress made by psychology in the century fol-
lowing the death of James Mill, with his crude theory of asso-
ciation, is an achievement scarcely less than that of the physical
sciences in the same period, it is nevertheless true that psycho-
logical theory is still in its infancy. There is a long way to go
before we can speak of understanding the principles of behavior
to the degree that we understand the principles of chemical
reaction.
In an undertaking of such difficulty, the psychologist pre-
sumably must seek help wherever he can find it. There have
been an increasing number of attempts to develop new mathe-
matical methods of analysis. With these, in general, I do not
attempt to deal. The method of factor analysis developed by
Spearman (1927) and greatly elaborated by Thurstone (1935) is
well established as a powerful tool for handling certain kinds of
data, though the range of its use has been limited by depend-
ence on tests that can be conveniently given to large groups of
subjects. Another method is the application of mathematics
more directly to the interaction of populations of neurons, by
Rashevsky, Pitts, Householder, Landahl, McCulloch, and others.*
Bishop (1946) has discussed the work from the point of view of
neurophysiology, and his remarks are fully concurred with here.
The preliminary studies made with this method so far have been
obliged to simplify the psychological problem almost out of
* Two papers by Culbertson (Bull. Math. Biophys., 1948, 10, 31-40 and
97-102), and Bishop's review article, list some of the more important of
the actual titles in this field.
xi
xii Introduction
existence. This is not a criticism, since the attempt is to develop
methods that can later be extended to deal with more complex
data; but as matters stand at present one must wait for further
results before being sure that the attempt will succeed. Un-
doubtedly there is great potential value in such work, and if
the right set of initial assumptions can be found it will pre-
sumably become, like factor analysis, a powerful ally of other
methods of study.
However, psychology has an intimate relation with the other
biological sciences, and may also look for help there. There is
a considerable overlap between the problems of psychology and
those of neurophysiology, hence the possibility (or necessity) of
reciprocal assistance. The first object of this book is to present
a theory of behavior for the consideration of psychologists; but
another is to seek a common ground with the anatomist, physi-
ologist, and neurologist, to show them how psychological theory
relates to their problems and at the same time to make it more
possible for them to contribute to that theory.
Psychology is no more static than any other science. Physi-
ologists and clinicians who wish to get a theoretical orientation
cannot depend only on the writings of Pavlov or Freud. These
were great men, and they have contributed greatly to psycho-
logical thought. But their contribution was rather in formulat-
ing and developing problems than in providing final answers.
Pavlov himself seems to have thought of his theory of condi-
tioned reflexes as something in continual need of revision, and
experimental results have continued to make revisions necessary:
the theory, that is, is still developing. Again, if one were to re-
gard Freud's theory as needing change only in its details, the
main value of his work would be stultified. Theorizing at this
stage is like skating on thin icekeep moving, or drown. Ego,
Id, and Superego are conceptions that help one to see and state
important facts of behavior, but they are also dangerously easy
to treat as ghostly realities: as anthropomorphic agents that want
this or disapprove of that, overcoming one another by force or
guile, and punishing or being punished. Freud has left us the
task of developing these provisional formulations of his to the
point where such a danger no longer exists. When theory be-
comes static it is apt to become dogma; and psychological
xiii Introduction
theory has the further danger, as long as so many of its prob-
lems are unresolved, of inviting a relapse into the vitalism and
indeterminism of traditional thought.
It is only too easy, no matter what formal theory of behavior
one espouses, to entertain a concealed mysticism in one's think-
ing about that large segment of behavior which theory does not
handle adequately. To deal with behavior at present, one must
oversimplify. The risk, on the one hand, is of forgetting that
one has oversimplified the problem; one may forget or even deny
those inconvenient facts that one's theory does not subsume.
On the other hand is the risk of accepting the weak-kneed dis-
couragement of the vitalist, of being content to show that exist-
ing theories are imperfect without seeking to improve them.
We can take for granted that any theory of behavior at present
must be inadequate and incomplete. But it is never enough to
say, because we have not yet found out how to reduce behavior
to the control of the brain, that no one in the future will be able
to do so.
Modern psychology takes completely for granted that behavior
and neural function are perfectly correlated, that one is com-
pletely caused by the other. There is no separate soul or life-
force to stick a finger into the brain now and then and make
neural cells do what they would not otherwise. Actually, of
course, this is a working assumption onlyas long as there are
unexplained aspects of behavior. It is quite conceivable that
some day the assumption will have to be rejected. But it is
important also to see that we have not reached that day yet:
the working assumption is a necessary one, and there is no real
evidence opposed to it. Our failure to solve a problem so far
does not make it insoluble. One cannot logically be a deter-
minist in physics and chemistry and biology, and a mystic in
psychology.
All one can know about another's feelings and awarenesses is
an inference from what he doesfrom his muscular contractions
and glandular secretions. These observable events are deter-
mined by electrical and chemical events in nerve cells. If one
is to be consistent, there is no room here for a mysterious agent
that is defined as not physical and yet has physical effects
(especially since many of the entities of physics are known only
xiv Introduction
through their effects). "Mind" can only be regarded, for scien-
tific purposes, as the activity of the brain, and this should be
mystery enough for anyone: besides the appalling number of
cells (some nine billion, according to Herrick) and even more
appalling number of possible connections between them, the
matter out of which cells are made is being itself reduced by
the physicist to something quite unlike the inert stick or stone
with which mind is traditionally contrasted. After all, it is that
contrast that is at the bottom of the vitalist's objection to a
mechanistic biology, and the contrast has lost its force (Herrick,
1929). The mystic might well concentrate on the electron and
let behavior alone. A philosophical parallelism or idealism,
whatever one may think of such conceptions on other grounds,
is quite consistent with the scientific method, but interactionism
seems not to be.
Psychologist and neurophysiologist thus chart the same bay-
working perhaps from opposite shores, sometimes overlapping
and duplicating one another, but using some of the same fixed
points and continually with the opportunity of contributing to
each other's results. The problem of understanding behavior is
the problem of understanding the total action of the nervous
system, and vice versa. This has not always been a welcome
proposition, either to psychologist or to physiologist.
A vigorous movement has appeared both in psychology and
psychiatry to be rid of "physiologizing," that is, to stop using
physiological hypotheses. This point of view has been clearly
and effectively put by Skinner (1938), and it does not by any
means represent a relapse into vitalism. The argument is re-
lated to modern positivism, emphasizes a method of correlating
observable stimuli with observable response, and, recognizing
that "explanation" is ultimately a statement of relationships be-
tween observed phenomena, proposes to go to the heart of the
matter and have psychology confine itself to such statements
now. This point of view has been criticized by Pratt (1939)
and Kohler (1940). The present book is written in profound dis-
agreement with such a program for psychology. Disagreement
is on the grounds that this arises from a misconception of the
scientific method as it operates in the earlier stages. Those ap-
parently naive features of older scientific thought may have had
xv Introduction
more to do with hitting on fertile assumptions and hypotheses
than seems necessary in retrospect. The anti-physiological po-
sition, thus, in urging that psychology proceed now as it may be
able to proceed when it is more highly developed, seems to be
in short a counsel of perfection, disregarding the limitations of
the human intellect. However, it is logically defensible and
may yet show by its fertility of results that it is indeed the
proper approach to achieving prediction and control of behavior.
If some psychologists jib at the physiologist for a bedfellow,
many physiologists agree with them heartily. One must sympa-
thize with those who want nothing of the psychologist's hair-
splitting or the indefiniteness of psychological theory. There is
much more certainty in the study of the electrical activity of a
well-defined tract in the brain. The only question is whether
a physiology of the human brain as a whole can be achieved
by such studies alone. One can discover the properties of its
various parts more or less in isolation; but it is a truism by now
that the part may have properties that are not evident in isola-
tion, and these are to be discovered only by study of the whole
intact brain. The method then calls for learning as much as
one can about what the parts of the brain do (primarily the
physiologist's field), and relating behavior as far as possible to
this knowledge (primarily for the psychologist); then seeing
what further information is to be had about how the total brain
works, from the discrepancy between (1) actual behavior and
(2) the behavior that would be predicted from adding up what
is known about the action of the various parts.
This does not make the psychologist a physiologist, for pre-
cisely the same reason that the physiologist need not become a
cytologist or biochemist, though he is intimately concerned with
the information that cytology and biochemistry provide. The
difficulties of finding order in behavior are great enough to re-
quire all one's attention, and the psychologist is interested in
physiology to the extent that it contributes to his own task.
The great argument of the positivists who object to "physiolo-
gizing" is that physiology has not helped psychological theory.
But, even if this is true (there is some basis for denying it), one
has to add the words so far. There has been a great access of
knowledge in neurophysiology since the twenties. The work of
xvi Introduction
Berger, Dusser de Barenne, and Lorente de No (as examples)
has a profound effect on the physiological conceptions utilized
by psychology, and psychology has not yet assimilated these re-
sults fully.
The central problem with which we must find a way to deal
can be put in two different ways. Psychologically, it is the
problem of thought: some sort of process that is not fully con-
trolled by environmental stimulation and yet cooperates closely
with that stimulation. From another point of view, physio-
logically, the problem is that of the transmission of excitation
from sensory to motor cortex. This statement may not be as
much oversimplified as it seems, especially when one recognizes
that the "transmission" may be a very complex process indeed,
with a considerable time lag between sensory stimulation and
the final motor response. The failure of psychology to handle
thought adequately (or the failure of neurophysiology to tell us
how to conceive of cortical transmission) has been the essential
weakness of modern psychological theory and the reason for
persistent difficulties in dealing with a wide range of experi-
mental and clinical data, as the following chapters will try to
show, from the data of perception and learning to those of
hunger, sleep, and neurosis.
In mammals even as low as the rat it has turned out to be
impossible to describe behavior as an interaction directly be-
tween sensory and motor processes. Something like thinking,
that is, intervenes. "Thought" undoubtedly has the connotation
of a human degree of complexity in cerebral function and may
mean too much to be applied to lower animals. But even in
the rat there is evidence that behavior is not completely con-
trolled by immediate sensory events: there are central processes
operating also.
What is the nature of such relatively autonomous activities in
the cerebrum? Not even a tentative answer is available. We
know a good deal about the afferent pathways to the cortex,
about the efferent pathways from it, and about many structures
linking the two. But the links are complex, and we know prac-
tically nothing about what goes on between the arrival of an
excitation at a sensory projection area and its later departure
from the motor area of the cortex. Psychology has had to find,
Introduction
xvii
in hypothesis, a way of bridging this gap in its physiological
foundation. In general the bridge can be described as some
comparatively simple formula of cortical transmission.* The
particular formula chosen mainly determines the nature of the
psychological theory that results, and the need of choosing is
the major source of theoretical schism.
Two kinds of formula have been used, leading at two extremes
to (1) switchboard theory, and sensori-motor connections; and
(2) field theory. (Either of these terms may be regarded as
opprobrium; they are not so used here.) (1) In the first type
of theory, at one extreme, cells in the sensory system acquire
connections with cells in the motor system; the function of the
cortex is that of a telephone exchange. Connections rigidly de-
termine what animal or human being does, and their acquisition
constitutes learning. Current forms of the theory tend to be
vaguer than formerly, because of effective criticism of the theory
in its earlier and simpler forms, but the fundamental idea is still
maintained. (2) Theory at the opposite extreme denies that
learning depends on connections at all, and attempts to utilize
instead the field conception that physics has found so useful.
The cortex is regarded as made up of so many cells that it can
be treated as a statistically homogeneous medium. The sensory
control of motor centers depends, accordingly, on the distribu-
tion of the sensory excitation and on ratios of excitation, not on
locus or the action of any specific cells.
Despite their differences, however, both theoretical approaches
seem to imply a prompt transmission of sensory excitation to the
motor side, if only by failing to specify that this is not so. No
one, at any rate, has made any serious attempt to elaborate
ideas of a central neural mechanism to account for the delay,
between stimulation and response, that seems so characteristic
of thought. There have indeed been neural theories of "motor"
thought, but they amount essentially to a continual interplay of
proprioception and minimal muscular action, and do not provide
for any prolonged sequence of intracerebral events as such.
But the recalcitrant data of animal behavior have been draw-
* The simplicity possibly accounts for the opinion expressed by an anato-
mist who claimed that psychologists think of the brain as having all the
finer structure of a bowlful of porridge.
xviii Introduction
ing attention more and more insistently to the need of some
better account of central processes. This is what Morgan (1943)
has recognized in saying that "mental" variables, repeatedly
thrown out because there was no place for them in a stimulus-
response psychology, repeatedly find their way back in again in
one form or another. The image has been a forbidden notion
for twenty years, particularly in animal psychology; but the
fiend was hardly exorcised before "expectancy" had appeared
instead. What is the neural basis of expectancy, or of attention,
or interest? Older theory could use these words freely, for it
made no serious attempt to avoid an interactionist philosophy.
In modern psychology such terms are an embarrassment; they
cannot be escaped if one is to give a full account of behavior,
but they still have the smell of animism: and must have, until
a theory of thought is developed to show how "expectancy" or
the like can be a physiologically intelligible process.
In the chapters that follow this introduction I have tried to
lay a foundation for such a theory. It is, on the one hand and
from the physiologist's point of view, quite speculative. On the
other hand, it achieves some synthesis of psychological knowl-
edge, and it attempts to hold as strictly as possible to the psy-
chological evidence in those long stretches where the guidance
of anatomy and physiology is lacking. The desideratum is a
conceptual tool for dealing with expectancy, attention, and so
on, and with a temporally organized intracerebral process. But
this would have little value if it did not also comprise the main
facts of perception, and of learning. To achieve something of
the kind, the limitations of a schema are accepted with the pur-
pose of developing certain conceptions of neural action. This
is attempted in Chapters 4 and 5; Chapters 1 to 3 try to clear
the ground for this undertaking. From Chapter 6 onward the
conceptions derived from schematizing are applied to the prob-
lems of learning, volition, emotion, hunger, and so on. (In gen-
eral, the reader may regard Chapters 1 to 5 as mainly prepara-
tory, unless he is particularly interested in the neurological de-
tails, or in the treatment of perception; to get the gist of the
theory that is presented here one should read the two following
paragraphs, and turn directly to Chapter 6.) In outline, the
conceptual structure is as follows:
xix Introduction
Any frequently repeated, particular stimulation will lead to
the slow development of a "cell-assembly," a diffuse structure
comprising cells in the cortex and diencephalon (and also, per-
haps, in the basal ganglia of the cerebrum), capable of acting
briefly as a closed system, delivering facilitation to other such
systems and usually having a specific motor facilitation. A series
of such events constitutes a "phase sequence"the thought
process. Each assembly action may be aroused by a preceding
assembly, by a sensory event, ornormallyby both. The cen-
tral facilitation from one of these activities on the next is the
prototype of "attention." The theory proposes that in this cen-
tral facilitation, and its varied relationship to sensory processes,
lies the answer to an issue that is made inescapable by Hum-
phrey's (1940) penetrating review of the problem of the direc-
tion of thought.
The kind of cortical organization discussed in the preceding
paragraph is what is regarded as essential to adult waking be-
havior. It is proposed also that there is an alternate, "intrinsic"
organization, occurring in sleep and in infancy, which consists
of hypersynchrony in the firing of cortical cells. But besides
these two forms of cortical organization there may be disorgani-
zation. It is assumed that the assembly depends completely on
a very delicate timing which might be disturbed by metabolic
changes as well as by sensory events that do not accord with the
pre-existent central process. When this is transient, it is called
emotional disturbance; when chronic, neurosis or psychosis.
The theory is evidently a form of connectionism, one of the
switchboard variety, though it does not deal in direct connec-
tions between afferent and efferent pathways: not an "S-R" psy-
chology, if R means a muscular response. The connections serve
rather to establish autonomous central activities, which then
are the basis of further learning. In accordance with modern
physiological ideas, the theory also utilizes local field processes
and gradients, following the lead particularly of Marshall and
Talbot (1942). It does not, further, make any single nerve cell
or pathway essential to any habit or perception. Modern physi-
ology has presented psychology with new opportunities for the
synthesis of divergent theories and previously unrelated data,
and it is my intent to take such advantage of these opportuni-
ties as I can.
This page intentionally left blank
/. The Problem
and the Line of Attack
This book presents a theory of behavior that is based as far
as possible on the physiology of the nervous system, and makes
a sedulous attempt to find some community of neurological and
psychological conceptions.
The purpose of this first chapter is to define the major diffi-
culties in the way of such an undertaking and to show how
these difficulties determine the line it must take. The great
need, as I have tried to make clear in the Introduction, is for a
physiological theory of thought (or some equivalent term that
can be applied to animals as well as to man).
Let us look first at the specific problem from which the present
speculations began, to see why the discussion is oriented as it is.
The problem lay in certain puzzling effects of operation on
the human brain. The effect of a clearcut removal of cortex
outside the speech area is often astonishingly small; at times no
effect whatever can be found (Hebb, 1942a, 1945&). It is pos-
sible that there is always a loss of intelligence in aphasia, when
the "speech area" is seriously damaged, but this does not, of
course, explain why damage elsewhere should have no effect.
It would be unreasonable to suppose that most of the cortex has
nothing to do with intelligence, and there are in fact definite
indications that this is not true. Intelligence must be affected
by any large brain injuryyet sometimes it seems not to be.
A final touch is added to the puzzle when we find that it is
tests of the Binet type that least show the effect of injuries out-
side the speech area. The Binet is the measure of intelligence
that is most sensitive and accurate with normal subjects. How
can it be possible for a man to have an IQ of 160 or higher,
1
2 The Problem and the Line of Attack
after a prefrental lobe has been removed (Hebb, 1939), or for
a woman (described by Rowe, 1937) to have an IQ of 115, a
better score than two-thirds of the normal population could
make, after losing the entire right half of the cortex?
Those two are perhaps the most striking cases, but high scores
after brain operation have been reported by nearly everyone
who has used standard tests as a method of study, and they
have to be explained.. Most investigators have preferred to
forget them, and to search only for intellectual defects. The
defects certainly must exist, and it is important to find them;
but it is just as important to learn why Binet-type tests, the most
valid and sensitive indices of normal ability, should often show
no effect of injury to the brain.
The only explanation that has been proposed, and perhaps the
only feasible one, has to do with perceptual learning and with
concepts as distinct from conditioned responses or motor learn-
ing (Hebb, 1942a). The explanation, roughly, is this. The level
of intelligence-test performance is a function of the concepts a
patient has already developed. Once developed, a concept is
retained, despite brain damage that, if it had occurred earlier,
would have prevented the development. The patient with brain
injury at maturity may continue to think and solve problems
normally (in familiar fields), although his intelligence would
have been far from normal if a similar injury had happened at
birth. The explanation meets the clinical facts and, moreover,
is supported by the way in which some intellectual capacities are
retained in old age when others are disappearing.
But now we come to the crux of the matter. As an explana-
tion, this is only a good start. What is a concept, physiologi-
cally and a propos of the loss of neural cells? Though it has
been tried before now, an explanation cannot be half neural
anatomy and half consciousness. They are different levels in a
logical hierarchy. A problem recognized at one level can be
solved by recourse to a more fundamental set of conceptions,
but an hypothesis cannot well comprise two levels, take in two
universes of discourse at the same time. We want to explain
certain clinical facts. To really do so, we must find an ana-
tomical and physiological understanding of what is known psy-
3 Rejecting Sensory Control
chologically as a concept; and we must be able to deal with its
relation to perception and to learning.
And with that, we land right in the middle of the generalized
problem of explaining mammalian behavior. What is a con-
cept, if it is not a conditioned response? What is perceptual
learning? And so on. Before such questions can be answered,
psychological theory must have a new base of operations. As
an illustration, a particular problem has been cited; but the diffi-
culty is really general, as the following section will try to show,
and there is no phase of psychological theory in which the same
central weakness does not appear.
It has already been suggested that the essential need is to find
out how to handle thought, and related processes, more ade-
quately. The difficulty in doing so goes back to fundamental
assumptions. If this discussion is to rest on solid ground, it
must start with certain ideas with which every psychologist is
concerned but which have been confused, vague, or ill defined.
This first chapter, accordingly, will dot some i's and cross
some t's, so that we can get down to business in what follows.
There are two fundamental assumptions to be dealt with. One
of them must be rejected, one accepted. The rest of this chapter
will show how they determine a point of attack for the revision
of theory.
REJECTING THE ASSUMPTION OF A COMPLETE
SENSORY CONTROL
The first to be discussed is what can be called the assumption
of a sensory dominance of behavior. It is the idea that be-
havior is a series of reactions (instead of actions), each of which
is determined by the immediately preceding events in the sen-
sory systems. The idea is not altogether consistent with recog-
nizing the existence of set, attitude, or attention; and an implicit
inconsistency of this sort is at the root of the current confusion
in psychological theory.
It may be noted in passing that the assumption of a sensory
dominance of behavior is not the property of any particular
theory. Theories differ as to how a sensory event has its effect,
but not as to its all-important role. Gestalt psychology avoids
4 The Problem and the Line of Attack
words like "stimulus" and "sensation" but as a system is not less
preoccupied with stimulus configurations than other systems are
with their stimuli (p. 23).
Now for the source of confusion.
In the simplest terms, "attention" refers to a selectivity of
response. Man or animal is continuously responding to some
events in the environment, and not to others that could be re-
sponded to (or "noticed") just as well. When an experimental
result makes it necessary to refer to "set" or "attention," the ref-
erence means, precisely, that the activity that controls the form,
speed, strength, or duration of response is not the immediately
preceding excitation of receptor cells alone. The fact that a
response is not so controlled may be hard to explain, theo-
retically; but it is not mystical, and "attention" is not necessarily
anthropomorphic, or animistic, or undefinable.
Now the tradition in psychology has long been a search for
the property of the stimulus which by itself determines the
ensuing response, at any given stage of learning. This approach
seems partly a consequence of psychology's persistent fight
against animism * and deserves respect for that reason; but it
is no longer satisfactory as theory. Almost without exception
psychologists have recognized the existence of the selective cen-
tral factor that reinforces now one response, now another. The
problem is to carry out to its logical conclusion an incomplete
line of thought that starts out preoccupied with stimulus or
stimulus configuration as the source and control of action, even-
tually runs into the facts of attention and so on, and then simply
* And partly, as we shall see, the product of antiquated physiological con-
ceptions. Fighting animism meant that psychological phenomena had to be
reduced to a pattern of cause and effect. The only means at hand was a
physiology of the nervous system in which a knowledge of sense organs and
peripheral nerve was the main content. As a result, it is still easy to feel
that psychology becomes part of a larger demonology with any retreat from
the stimulus-response formula. As for "insight," "purpose," "attention"
any one of these may still be an invocation of the devil, to the occasional
psychologist. However, this attitude should not be too much made fun of.
It cannot be emphasized too strongly that there is continual danger of slip-
ping momentarily into animistic thinking; and consistent use of the S-R for-
mula does at least avoid that danger effectively. It behooves those of us
who wish to use other terms to see that they are clearly defined.
5 Rejecting Sensory Control
agrees that attention is an important fact, without recognizing
that this is inconsistent with one's earlier assumptions. To com-
plete this process, we must go back and make a change in the
basis of the theory.
There are three points here: one is that psychologists have
generally recognized the existence of attention or the like; an-
other that they have done so reluctantly and sparingly, and
have never recognized the fact in setting up theories. The third
point is obvious enough, that we need to find some way of deal-
ing with the facts consistently. Since everyone knows that at-
tention and set exist, we had better get the skeleton out of the
closet and see what can be done with it.
The first two of these points have been pretty clearly estab-
lished by Gibson (1941). His review needs some clarification
in one respect, since he declined to attempt any definition of
"set" or any other of the long list of terms with a similar mean-
ing that he gathered together, although he evidently recognized,
in classifying them so, that they have something in common.
This common meaning has already been defined. When one
considers the problem in the light of the implicit assumption of
a sensory dominance of behavior it becomes clear at once that
the notions of set, attention, attitude, expectancy, hypothesis, in-
tention, vector, need, perseveration, and preoccupation (Gibson,
1941, pp. 781-782) have a common element, and one only.
That element is the recognition that responses are determined
by something else besides the immediately preceding sensory
stimulation. It does not deny the importance of the immediate
stimulus; it does deny that sensory stimulation is everything in
behavior.
All such terms, then, are a reference to the "central process
which seems relatively independent of afferent stimuli," defined
by Hilgard and Marquis (1940), which I shall call here the
autonomous central process. Gibson's review shows in effect
that the process is ubiquitous, that it crops up in every sort of
psychological investigationand that almost everyone has recog-
nized its existence, in one form or another. To Gibson's list
can be added Pavlov's (1928) and Hull's (1943) stimulus trace
a lasting cerebral state, set up it is true by a specific stimulus
but not transmitted and dissipated at once; Beach's (1942) cen-
6 The Problem and the Line of Attack
tral excitatory mechanism; Morgan's (1943) central motive state;
and Kleitman's (1939) "interest"a factor in wakefulness. All
these things have the same property of an activity that has a
selective effect on behavior without being part of the present
afferent excitation.
Everyone has had such ideas about the control of behavior;
and yet, as Gibson noted, "The meaning [of the term "set"] is
felt to be unsatisfactory, and the concept is employed reluctantly
and only because the facts make it absolutely unavoidable."
The reluctance is partly no doubt because of a feeling that the
concept is animistic, in some obscure way. But why animism,
if the facts of behavior make it unavoidable? The trouble really
seems to have been in finding how to make an essential idea
intelligible.
Hilgard and Marquis' central process, and Beach's central
excitatory mechanism, are hypothetical entities, but they cer-
tainly have no flavor of animism about them. "Attention" and
"set" are now seen to fall in the same class: it may well be that
their connotations are misleading and that we shall have to look
for new terms, but the idea itself is respectable, and such lan-
guage need no longer risk starting a witchhunt.
THE NEUROLOGICAL PROBLEM OF ATTENTION
There is a further hazard on the course. This is the apparent
lack of a theoretical rationale for the autonomous central process.
Actually, modern neurophysiology has already removed this
difficulty.
Here again the situation can only be understood historically.
A main function of the neural cell is of course to transmit exci-
tations, and earlier ideas of anatomy and physiology made the
central nervous system appear, in principle, a collection of
routes, some longer, some shorter, leading without reversal from
receptors to effectorsa mass of conductors that lies inactive
until a sense organ is excited, and then conducts the excitation
promptly to some muscle or gland. We know now that this is
not so, but the older idea still has a profound effect on psycho-
logical thoughtdemonstrated, for example, in the assumption
of sensory dominance, discussed above.
7 Neurological Problem of Attention
The lack of a rationale for nonsensory influences on behavior
that seemed to exist in 1920 certainly exists no more. Psycholo-
gists have long had to recognize (since the days of the Wiirzburg
school at least) the existence of a determining tendency, whether
physiology made it comprehensible or not. Modern electro-
physiology has more than caught up with psychology and now
provides abundant evidence to support the same idea. When
the detailed evidence of neurophysiology and histology are con-
sidered, the conclusion becomes inevitable that the nonsensory
factor in cerebral action must be more consistently present and
of more dominating importance than reluctant psychological
theory has ever recognized. Instead of a joker to occasionally
confuse the student of behavior, nonsensory activities appear
in every fall of the cards and must make up a large share of
the deck. Neurophysiologically, it may even become a problem
to account for any consistent effect of a specific stimulus (pp.
121, 126).
Electrophysiology of the central nervous system indicates in
brief that the brain is continuously active, in all its parts, and
an afferent excitation must be superimposed on an already
existent excitation. It is therefore impossible that the conse-
quence of a sensory event should often be uninfluenced by the
pre-existent activity. If we recognize in that activity the psy-
chologically known factor of set and the like (as Denny-Brown,
1932, suggested some time ago), the problem for psychology is
no longer to account for the existence of set but to find out how
it acts and above all to learn how it has the property of a con-
sistent, selective action instead of producing the random-error
distribution postulated by Hull (1943) in his "oscillation prin-
ciple."
So there really is a rational basis for postulating a central
neural factor that modifies the action of a stimulus. The theo-
retical problem now is to discover the rules by which it oper-
ates. At first glance this is a problem for the neurophysiologist
only. But look closer; much of the evidence, from which these
rules must be worked out, is psychological, or behavioral. The
problem is after all the problem of attention, and seen best in
the activity of the whole animal. It is in the highest degree
unlikely that it can be solved either from the physiological evi-
8 The Problem and the Line of Attack
dence alone or from the behavioral evidence alone. What we
need, evidently, is some synthesis of both kinds of datum. The
psychological data have been reviewed briefly (for a fuller ac-
count, see Hilgard and Marquis, 1940; Gibson, 1941; Morgan,
1943under such headings as attitude, expectancy, and set).
Let us turn now, again briefly, to the anatomical and physio-
logical evidence that has made a drastic change in the theo-
retical problems of behavior.
For our purposes, the physiological evidence can be treated
under two heads, as bearing on (1) the existence and properties
of a continuous cerebral activity, and (2) the nature of synaptic
transmission in the central nervous system.
1. The studies reviewed by Jasper (1937), beginning with
Berger's work and its confirmation by Adrian, have shown with
practical certainty that the central nervous system is continu-
ously active, in all its parts, whether exposed to afferent stimula-
tion or not. It appears that the EEG, or electroencephalogram,
must be at least in part a summation of action potentials, an
index of actual cellular firing. There are changes of cell poten-
tial without active transmission of impulses (Gibbs, 1945), and it
is quite possible that these are the source of much of the EEG.
However, there is a considerable body of evidence to show that
neural tissue is persistently active; and presumably the EEG
includes a record of that activity. It is taken here as a working
assumption that the EEG is correlated with neural firingthat
large potentials indicate a local synchrony of firing, even though
other factors contribute to the size of the potential. The psy-
chological usefulness of this assumption will perhaps become
evident in the following chapters.
The evidence concerning persistent activity consists first of a
number of studies showing directly that the neural cell fires
spontaneously, i.e., under no stimulation except that of the nu-
trient fluids bathing it (Fessard, cited by Jasper, 1937; Prosser,
1934; Lehmann, 1937a, 1937&; Bronk, 1939; Dubner and Gerard,
1939; Libet and Gerard, 1939). Secondly, a very important
paper for the theory of behavior, by Weiss (1941a), reports
spontaneous, almost incessant motor activity in a transplanted
amphibian limb innervated from a pool without afferent fibers.
The activity of the limb was not normally coordinated, which
9 Neurological Problem of Attention
agrees with the general idea that sensory activity is essential to
the regulation of central neural firing but not essential to initiat-
ing it. Finally, the work of Adrian and his collaborators: Adrian
and Matthews (1934) rejected the earlier opinion of Adrian
(1931) and Adrian and Buytendijk (1931), and concluded defi-
nitely that the EEC is the summation of single sharp potentials,
the result of actual cellular firing. They concluded further that
the activity is not necessarily maintained by sensory activity.
Elsewhere Adrian (1934, p. 1126) has stated his general con-
clusions: "There are cell mechanisms in the brain which are set
so that a periodic discharge is bound to take place. The mo-
ment at which it occurs can be greatly altered by afferent in-
fluences, but it cannot be postponed indefinitely."
Several comments may be made on these conclusions. Spon-
taneity of firing by central neural cells is not philosophic inde-
terminacy, as some writers have thought; the "spontaneity"
means only that synaptic stimulation is not the sole cause of
firing. As Adrian pointed out, also, the cell that is capable of
firing spontaneously is also open to synaptic control. If it is left
undisturbed the processes of metabolism will lead in the end to
firing, but obviously an afferent stimulation could always catch
the cell in the period between the absolute refractory state and
the time of spontaneous discharge. With a frequent arrival of
impulses at the synapse, therefore, the cell can never reach the
point of spontaneous action but must remain under afferent
control.
A very interesting relationship exists between sensory activity,
EEC, and coordinated, adaptive behavior (or "conscious state").
There is, in the first place, Adrian and Matthews' observation
that sensory processes, instead of supporting synchronous,
rhythmic firing and large potentials in the EEG, have the op-
posite effect. They introduce irregularity and flattening of the
electrical record. In the second place, large potentials, or
"hypersynchrony," negate or may negate normal function (Jas-
per, 1941). That is, sensory activity breaks up hypersynchrony
and makes for normal, coordinated, adaptive activity. Bartley
and Bishop (1933) and Adrian and Matthews have even pro-
posed that anesthetics produce "unconsciousness" by suppressing
sensory activity rather than acting directly on the cerebrum
10 The Problem and the Line of Attack
itself. It does seem clear from the facts discussed, from the
large potentials observed in sleep, and from the hypnotic effect
of minimizing the normal variation of sensory activity, that the
sensory input to the brain has a constantly necessary function,
for adaptive behavior. We shall return to this point later
(p. 145). For the present, it has been advisable to point out
that recognizing the existence of spontaneous neural activity is
neither indeterminacy nor a denial of the importance of sensory
processes in normal behavior.
2. The nature of synaptic transmission in the central nervous
system is also of fundamental importance for the theory of be-
havior. There are two radical modifications of earlier ideas:
transmission is not simply linear but apparently always involves
some closed or recurrent circuits; and a single impulse cannot
ordinarily cross a synapsetwo or more must act simultane-
ously, and two or more afferent fibers must therefore be active
in order to excite a third to which they lead.
The concepts of neural action chiefly developed by Lorente
de No (19380, 1938&, 1939, 1943) are well enough known by
now to need no elaborate review. It is necessary, however, to
point out that they have revolutionary implications for psycho-
logical theory.
In particular, the psychological criticism advanced by Koffka
(1924), Lashley (1929&, 1930), and Kohler (1929), against the
theory of neural connections as the basis of learning, applies
only to the older theory of linear, sensori-motor connections, in
which a single cell was supposed to be always capable of excit-
ing a second cell with which it synapsed. The criticism is effec-
tive against the idea that synaptic resistances are all that deter-
mines the direction of transmission at the synapse. A radical
change in the whole problem, however, has been made by con-
siderations of timing in transmission (Gasser, 1937) and by the
conception of "optional transmission" (Lorente de N6, 1939).
In a single system, and with a constant set of connections be-
tween neurons in the system, the direction in which an entering
excitation will be conducted may be completely dependent on
the timing of other excitations. Connections are necessary but
may not be decisive in themselves; in a complex system, espe-
11 Neurological Problem of Attention
cially, time factors must always influence the direction of con-
duction. The older ideas of neural transmission gave synaptic
connections too much rigidity, as a determinant of behavior.
The rigidity has now disappeared; the idea of connections can
again be useful in psychological theory, and the question of
"synaptic resistances" is completely reopened.
Let me now summarize what has been said, in this and the
preceding section, about attention, set, attitude, and so on. (1)
All psychologists have recognized some such factor in be-
havior. It undoubtedly exists. (2) Recognizing it is really a
denial that behavior is only a series of responses to environ-
mental stimulation. One important meaning of "attention" or
the like is the reference to a partly autonomous, or nonsensory,
cerebral activity: the "autonomous central process." (3) The
problem for psychology then is to find conceptions for dealing
with such complexities of central neural action: conceptions
that will be valid physiologically and at the same time "molar"
enough to be useful in the analysis of behavior. (4) Psychology
is still profoundly influenced by the very "molecular" conception
of linear transmission through a sequence of single cells. The
conception is no longer valid physiologically, just as it has long
been without psychological usefulness. The attack on neural
connections as an explanation of behavior was really an attack
on this particular conception of the way connections operate;
modern neuroanatomy and electrophysiology have changed the
question completely, and the significance of synaptic connec-
tions must be examined all over again.
Our problem, then, is to find valid "molar" conceptions of
neural action (conceptions, i.e., that can be applied to large-
scale cortical organizations). Bishop (1946, p. 370) has made
the point, in another context, that this is an essential problem
for neurophysiology also. But psychologists can hardly sit
around with hands folded, waiting for the physiologist to solve
it. In its essence the problem is psychological and requires a
knowledge of the psychological as well as the physiological evi-
dence for its solution.
12 The Problem and the Line of Attack
PERCEPTUAL GENERALIZATION AND THE ASSUMP-
TION OF A STRUCTURAL MEMORY TRACE
Now for a second fundamental assumption of psychological
theory: this time, one that must, it seems, be accepted; but in
accepting it we must also recognize the difficulties it entails
and provide for them. These difficulties in fact determine the
main features of the theory presented in this monograph.
The assumption we must accept is that the memory trace, the
basis of learning, is in some way structural and static; and the
difficulties in the way of making the assumption are mainly in
the facts of perceptual generalization that have been emphasized
by Gestalt psychologists (Koffka, 1935; Kohler, 1929, 1940) and
Lashley (1938&, 1942a). The problem raised by these writers
is crucial and must be disposed of before we touch anything
else.
Lashley has concluded that a learned discrimination is not
based on the excitation of any particular neural cells. It is sup-
posed to be determined solely by the pattern, or shape, of the
sensory excitation. Kohler, also stressing the apparent fact that
the pattern and not the locus of stimulation is the important
thing, has developed a theory of electrical fields in the brain
which control cerebral action. Like Lashley, he explicitly de-
nies that the same cells need be excited to arouse the same
perception.
This suggests that the mnemonic trace, the neural change
that is induced by experience and constitutes "memory," is not
a change of structure. Other facts, at the same time, are an
even stronger argument that it must be structural. A structural
trace, as we shall see in a moment, must be assumed; but when
we do so we have to find some way of fitting in the facts of
perception.
If it is really unimportant in what tissues a sensory excitation
takes place, one finds it hard to understand how repeated sen-
sations can reinforce one another, with the lasting effect we call
learning or memory. It might be supposed that the mnemonic
trace is a lasting pattern of reverberatory activity without fixed
locus, like some cloud formations or an eddy in a millpond. But
if so it seems that the multitudinous traces in the small confines
13 Generalization and Trace
of the cerebral cortex would interfere with one another, produc-
ing a much greater distortion of early memories by later ones
than actually occurs.
Moreover, violent cortical storms can occur (as in grand mal
epilepsy or cerebral concussion) without a detectable effect on
earlier memories. That the trace should be purely "dynamic"
a pattern of activity not dependent on structural changes for its
permanencethus seems in the highest degree unlikely. No one
has explicitly made such an assumption; yet how otherwise are
the known properties of a learned discrimination to be accounted
for, with its inevitable tendency to be generalized beyond what
has already been experienced by the animalits apparent inde-
pendence of excitation in specific cells?
In addition to the facts of perceptual generalization, two
other forms of evidence might make it difficult to postulate a
structural trace as the basis of memory. One is from Lashley's
(1929a) extirpation experiments, showing that the removal of
blocks of the rat's cerebral cortex does not affect habits selec-
tively. If one habit is affected, others are also. From this,
Lashley has concluded that memory traces are not localized in
the cerebral cortex, but himself has pointed out (Lashley, 1929fo)
another possible interpretation. His evidence is consistent with
the idea that the trace is structural but diffuse, involving, that
is, a large number of cells widely spaced in the cortex, physio-
logically but not anatomically unified. This is not, conse-
quently, crucial evidence for or against the notion of structural
traces in the cortex.
The other evidence that seemed once to prevent postulating
a structural trace is found in the work of Wulf (cited by Koffka,
1935) and later investigators who have interpreted their studies
of human memory for patterns to mean that the trace is spon-
taneously active, and does not lie dormant or merely deteriorate
with the passage of time. Hanawalt (1937), however, effec-
tively criticized the earlier evidence for this idea; and Hebb
and Foord (1945), having obtained data inconsistent with
Wulf's hypothesis, re-examined the later work that managed to
avoid Hanawalt's criticism. They have shown that there is no
evidence to even faintly support the idea of slow, spontaneous
changes in the trace. This conception must be abandoned.
14 The Problem and the Line of Attack
Thus the only barrier to assuming that a structural change in
specific neural cells is the basis of memory lies in the generaliza-
tion of the perception of patterns. Man sees a square as a
square, whatever its size, and in almost any setting. A rat
trained to look for food behind a horizontal rectangle will there-
FIGURE 1. Illustrating generalization in perception. A rat is first trained
with the two diagrams at the top, in the left half of the figure, until he
always avoids the vertical bar and chooses the horizontal. He is then tested
with the next pair of diagrams, the horizontally and vertically arranged
squares; and with the next pair, circles side by side versus circles one above
the other. Transfer of response occurs, the rat in each case preferring the
horizontal diagram, showing that perception of horizontal and vertical is
generalized beyond any particular pattern. Similarly, after training with
the large plain triangles, top right, the animal is tested with smaller triangles
and with circumscribed triangles. If the erect triangle, the one to the left,
is positive in training, the rat will choose the left-hand diagram when pre-
sented with either of the two lower pairs, again demonstrating perceptual
generalization.
after choose almost any horizontal figure, such as an interrupted
line or a pair of circles side by side (figure 1). Trained to
choose a solid upright triangle and to avoid an inverted triangle,
he will discriminate consistently between outlines of triangles;
triangles with confusing figures added (such as circumscribing
circles); and triangles of different size, which cannot thus excite
the same retinal cells simultaneously (Lashley, 19386). Rats
reared in darkness, then trained in the same way, show the
same perceptual generalizations (Hebb, 1937a).
15 Generalization and Trace
These are concrete, undisputed facts of behavior. They have
been interpreted as meaning that perception is independent of
the locus of excitation; and this interpretation has been tacitly
accepted as inescapable. The result is an awkward dilemma for
theory, since, as we have seen, it is hard to reconcile an un-
localized afferent process with a structural (and hence localized)
mnemonic trace.
Lashley's (1942a) hypothesis of interference patterns is the
one explicit attempt to solve this difficulty and to deal ade-
quately with both perception and learning. As such it deserves
special mention here, although we shall see that in other re-
spects it faces great difficulties.
Other writers have had to choose one horn of the dilemma.
Kohler (1940), for example, starts out with the facts of per-
ceptual generalization, in his theory of cerebral fields of force,
and then cannot deal with learning. He has no apparent way
of avoiding a fatal difficulty about the nature of the trace, its
locus and structure. This is another aspect of the difficulty for
Gestalt theory raised by Boring (1933), who pointed out that
at some point the perceptual process must act on specific cells
in order to determine a specific response.
The theory elaborated by Hull (1943), on the other hand, is
to be regarded as providing first of all for the stability of learn-
ing. It then has persistent difficulty with perception. The prin-
ciple of "afferent neural interaction" appears to be a concession
extorted by the facts of perceptual generalization. With this,
there is some danger that the entire system may lose its mean-
ing. The great value of Hull's theory is in showing how one
may conceive of variable behavior as determined by constant
causal relationships between stimulus, intervening variables, and
response. This is brilliantly achieved, for an important segment
of behavior. But then the postulate of afferent neural interac-
tion adds that anything may happen when two sensory events
occur at the same timewhich of course they are always doing.
Evidently no prediction is possible until the limits, and the de-
terminants, of afferent neural interaction can be given in detail.
This it seems demands that the neurological reference, already
present in the theory, be made explicit, and detailed. For our
present purposes, at any rate, Hull must be regarded as not
16 The Problem and the Line of Attack
yet having solved the problem of dealing with the perceptual
process in a theory of learning, although it remains possible that
his program (Hull, 1945) will do so in the future.
THE MODE OF ATTACK
How are we to provide for perceptual generalization and the
stability of memory, in terms of what the neuron does and what
happens at the synapse? We must suppose that the mnemonic
trace is a structural change; the difficulty, in supposing it, is a
conflict with the idea that only the pattern, and not the locus
of sensory stimulation, is important in perception; so let us
begin by asking whether that idea is, after all, securely estab-
lished.
That paragraph puts the theoretical approach of this mono-
graph. I propose to go over the evidence on perception again
and show that it is not what it seems. We do not know that
pattern is everything, locus nothing. Next, an alternative set
of ideas about perception will be developed. According to
these ideas, perception does depend on exciting specific parts
of the receptor surface; mnemonic trace and perceptual gener-
alization need no longer be at odds with one another; and a
physiological meaning of attention (or set, or expectancy)
shows up.
Now let us see, in the following two chapters, what revisions
can be made in current ideas about perception so that they can
be extended into a more general theory. The reader will ob-
serve that the discussion deals almost entirely with visual per-
ception. This is not because vision has any unique significance,
but because it is in visual perception, with few exceptions, that
the problem of patterning and form has been studied experi-
mentally. It can be assumed that perception in other sense
modes does not depend on fundamentally different principles.
2
Summation and Learning
in Perception
This chapter begins a revision of perceptual theory.* The
immediate objective is to show that "simple" perceptions are in
fact complex: that they are additive, that they depend partly on
motor activity, and that their apparent simplicity is only the end
result of a long learning process.
The preceding pages have tried to show that the crucial
problem of perception must be dealt with explicitly at the very
beginning of a psychological theory; and that one has the choice
of two approaches to this problem. One must decide whether
perception is to depend (1) on the excitation of specific cells or
(2) on a pattern of excitation whose locus is unimportant. Cur-
rent opinion seems tacitly to have accepted the Gestalt argu-
ment (and Lashley's argument) that the only tenable assump-
tion is the second of these possibilities.
The theory to be presented here is diametrically opposed to
this aspect of Gestalt theory, and is based on assumption 1, that
a particular perception depends on the excitation of particular
cells at some point in the central nervous system. Now the
Gestalt argument depends, I believe, on another assumption:
that when one perceives a simple figure (such as square or
circle) one perceives it directly as a distinctive whole, without
need of any learning process and not through a prior recogni-
* This and the following chapter may be disregarded by the reader who
is not particularly interested in the theory of perception. These two chap-
ters are in a sense preventive. The theory of behavior presented later
would appear impossible to the reader who is familiar with the current lit-
erature on perception if he had not been shown that the alternative (and
at first sight preferable) approach of "field theory" or "equipotentiality" has
very serious difficulties indeedmore serious than has been recognized.
17
18 Summation and Learning in Perception
tion of the several parts of the figure. If one makes this assump-
tionif the perception of a square is as simple and immediate
as it seems to us as adultsI believe that the Gestalt argu-
ment is unanswerable. But if on the other hand the perception
is additive, a serial reconstruction (though very rapid and
"unconscious" for the normal adult), the theoretical problem
would be very much changed.
In this chapter, accordingly, an attempt is made to show that
quite simple diagrams are not perceived directly as distinctive
wholesthat, though the stimulus has a unitary action in the
figure-ground relationship, the perception of identity depends
on a series of excitations from the parts of the stimulating dia-
gram. If this can be established, it will remove the necessity of
accepting field theory; and the following chapter will then at-
tempt to show that field theory actually is not consistent with
some of the facts of perception.
The work of Senden (1932) and of Riesen (1947) is funda-
mental to my argument here. Senden's monograph is a compi-
lation of all published reports on the vision of the congenitally
blind given sight by a surgical operation after the patient was
old enough to talk to the examiner and describe what he saw.
In some respects the data are incomplete, but the report is re-
peatedly referred to here because it contains the only existent
evidence concerning the course taken by the early development
of human perception. At first sight, some of the reported facts
are literally almost incredible, since they differ so much from
what would be predicted by current theory (either of perception
or of learning). There is, however, a considerable unanimity
among the writers reviewed by Senden, some of whom evidently
were not aware of the reports made by others; and the work of
Riesen (1947), who reared chimpanzees in darkness to an age
when the normal chimpanzee makes an effective use of vision,
fully confirms Senden's clinical evidence.
The two reports, by Senden and Riesen, are complementary;
and, though many details are not clear, the human and chim-
panzee data taken together seem to require radical changes in
the theory of perception and of learning.
19 Primitive Unity
DISTINCTION OF "PRIMITIVE UNITY" FROM OTHER
PROPERTIES OF THE PERCEIVED FIGURE
As a preliminary, certain terms must be defined. I want to
show that simple figures do not always act as wholes, innately.
But it is undoubtedly true that they sometimes do so in one re-
spectin the figure-ground relationship: so this property of a
perceived figure is to be distinguished from others, in which
summation and learning are important. Accordingly, the fol-
lowing are distinguished: the conceptions of (1) a primitive,
sensorily determined unity, (2) a nonsensory unity, affected by
experience, and (3) the identity (also affected by experience),
of a perceived figure.
The primitive unity of a figure is defined here as referring to
that unity and segregation from the background which seems to
be a direct product of the pattern of sensory excitation and the
inherited characteristics of the nervous system on which it acts.
Rubin (1921) elaborated the conception of the figure-ground
relationship, in a study of visual perception of patterns with
clearly marked boundaries. In order to elucidate the relation of
figure to ground (the relation of the perceived object or surface
to other objects and surfaces which make up the background)
he put particular stress on "ambiguous" figures (figure 2). This
is the special case in which either of two parts of a diagram may
be seen as figure, each alternating as part of the ground when
the other is figure. The principles he established, however, are
most fully operative in the unambiguous figure, made up of a
homogeneous zone of color surrounded by another color and
having a sharply defined boundaryan abrupt transition in
brightness from one zone to the other at all points. Such a
figure may or may not be regular. It is as well illustrated by
a splash of ink or by a silhouette of an animal as by a white
circle on a black stimulus card.
An area thus sensorily delimited is seen as one, unified and
distinct from its surroundings, by any normal person, by the
congenitally blind on the first occurrence of vision following
operation for cataract (Senden, 1932), by the normal rat
(Lashley, 1938Z?), and apparently also at first vision by the rat
20 Summation and Learning in Perception
that has been reared in darkness (Hebb, 1937a). The unity
and distinctiveness of such figures from their background, then,
is independent of experience, or "primitive."
It is not possible to specify exactly the stimulating conditions
which determine the primitive figure-ground organization. I do
FIGURE 2. Ambiguous figure. In this diagram it is possible to see either
of two profiles, but only rarely can the two be seen simultaneously; they
alternate instead. When a figure appears on the left, the profile of a rather
chubby man, the white area on the left appears as a more solid surface and
one that is nearer to the observer. This relationship is reversed when a
figure appears on the rightthe profile of one suffering from gigantic tumors
of the neck and of the frontal bone. The theoretical importance of such re-
versals of figure-ground relationships has been stressed by Gestalt psycholo-
gists, though there is considerable difficulty in subsuming the instability
under perceptual theory.
not exclude as nonprimitive the perception of groupings; nor
the segregation of a patch of color which has ill-defined bound-
aries. There are suggestions in Senden's (1932) monograph
and in rat behavior that the perception of such units in the
visual field may also be independent of experience. Senden's
monograph is of the first importance for understanding the per-
ceptual process, but in several respects the evidence is far from
clear. The earlier writers whose reports are summarized did
not recognize some of the psychological problems to which their
21 The "Nonsensory" Figure
observations are relevant. What Senden does show, in the fact
that patients always responded to certain objects as wholes and
could on occasion detect differences between objects even in
spite of nystagmus, is that there is a primitive or innate figure-
ground mechanism. He does not make it possible to state its
limiting conditions.
THE "NONSENSORT FIGURE
The nonsensory figure-ground organization is defined as one
in which the boundaries of the figure are not fixed by gradients
of luminosity in the visual field. It is affected by experience and
other nonsensory factors, and is not inevitable in any perception.
In contrasting the primitive and the nonsensory figure, however,
one need not imply that a perceived figure must be either of
one or the other kind. They are rather two extremes, and in
most perceptions both sensory and nonsensory factors affect
figure-ground organization.
The one of these extremes, in which exactly the same figural
boundaries are recognized by anyone, has already been illus-
trated. The other is quite common in ordinary experience,
although I have not found any discussion that makes its mean-
ing explicit. It is implied, for example, by Gibson and Crooks
(1938) in their discussion of the perceptual field of the automo-
bile driver, and the shrinking or expanding of a perceived zone
of safety from potential collision. In general terms, the non-
sensory figure occurs in perception whenever the subject re-
sponds selectively to a limited part of a homogeneous area in
the visual field. One may look at the "middle part" of a rope as
distinct from the rest, in knot-tying, or perceive the "foreground"
of an unbroken landscape. It is a commonplace that in human
perception many entities exist, like the place in a continuous
expanse of lawn where a flowerbed is to be put in, such that
no sensory delimitation of figure from ground can even be sug-
gested. The "corner" of a room certainly does not always refer
to a geometrical point; it is extended, and yet does not comprise
all of the two (or three) rectangular surfaces which determine
it. Its locus is sensorily fixed, but not its boundaries, and Ian-
22 Summation and Learning in Perception
guage is full of terms with a similar implication for the theory
of perception.
Even commoner in everyday perception is the perceived entity
in which both sensory and nonsensory factors cooperate. Here
the figural boundary may follow one of a number of possible
luminosity gradients in the field, the particular one that is effec-
tive being usually determined by experience in one form or
another. This is illustrated by Leeper's (1935) experiment,
which demonstrated a lasting control, by earlier experience, of
the particular perception that is made with Boring's ambiguous
figure, "My wife and my mother-in-law" (Boring, 1930). An-
other illustration is the process of slow learning to see a con-
figuration in a particular way, as in the gradual decrease of the
difficulty a novice has in following the boundaries of a thalamic
nucleus that are obvious to the expert.
There is reason to believe that the rat, as well as man, finds
some figure-ground relations obvious and inescapable, and de-
tects others only after prolonged experience. That is, the dis-
tinction of primitive and nonsensory figures applies also to the
rodent's perception. Lashley (1938&, pp. 156, 185) points out
that success in discrimination and generalization evidently de-
pends on the rat's finding a differentiating characteristic between
two figures, or one that is common to a pattern already known
and the new one which is presented in the testing situation. It
is equally clear that the distinctive part of the test pattern is
not obvious to the rat at once; this part, whatever it is, is what
determines the response and becomes the true figure (as con-
trasted with what the experimenter meant to be the figure), the
rest merging into ground.
In general, Gestalt writers on the organization of the figure
have been concerned to show that it cannot be reduced to
experience and learning, and have thus selected cases for dis-
cussion in which sensory dynamics alone is enough to produce
an effective figure-ground organization. They have, that is,
concentrated attention on the primitive figure, and they give the
impression that the spontaneity of its organization is a property
of any figure. True, one can find many passages in the litera-
ture on Gestalten that refer to figures not sensorily delimited,
23 The "Nonsensory" Figure
but no stress on the fact that this implies some role of learning
in the delimitation.
Kohler (1929), for example, has written: "Since 'real form'
presupposes a segregated whole, the existence of 'form' depends
upon factors of stimulation similar to those upon which the
segregation and organization of wholes depend. Again, definite
relations in the total constellation of retinal stimuli are found to
be decisive for the existence of real form" (p. 202; my italics
draw attention to the stress put on sensory factors). Again,
after referring to Gottschaldt's experiments, Kohler says: "After
these results, whoever defends the automatic influence of past
experience upon our seeing definite forms, will have incumbent
upon him the task of supporting his theory by other experi-
ments. If such an influence exists, it must be restricted to
rather special cases" (p. 208; italics mine).
Attention is drawn to this aspect of Gestalt theory because it
helps one to define the point at which one can diverge from the
theory without failing to recognize the great contribution it has
made to modern psychology, which has been shaped to a great
extent by the impact of Gestalt ideas on behaviorism. There
are few psychologists who would not own a debt of this kind
to Gestalttheorie, and few who do not also feel the need of
qualifying the theory in some way, though it is not easy to see
just how this is to be done without losing the values that have
been obtained with its help.
In the pages from which the passage above is taken, Professor
Kohler, I believe, provides the necessary clue. His argument
appears to depend on a complete antithesis of experience and
innate sensory dynamics, and it is this antithesis that we may
avoid. The question that is asked by Kohler is whether sensory
organization is wholly innate, or wholly experiential.* If these
are the only two alternatives, the argument is unanswerable.
Kohler and Koffka and Lashley have unquestionably slain the
*In the monograph of Kohler and Wallach (1944, pp. 316, 323) there
are also passages that seem not intelligible unless one is thinking in terms
of such an extreme dichotomy between completely organized sensory proc-
esses (innately) and completely unorganized ones. This does not do jus-
tice, either to current learning theory, or to the effectiveness of the earlier
Gestalt arguments in favor of an innate sensory organization.
24 Summation and Learning in Perception
dragon of pure learning theory, in the field of perception, and
no one today would argue that perceptual organization is wholly
acquired; there is some innate organization. But this of course
does not show that the organization is entirely innate. There is
always a possibility that perception has a partly innate, partly
learned organization; and that besides the figure that has a
"primitive unity" there are "nonsensory figures" in which expe-
rience has an important role.
I propose, then, that Gestalt theory made an essential contri-
bution when it showed that there are innate factors in percep-
tion, but that it has tended to carry the argument too far in deny-
ing that learning and experience have any important role in the
perception of simple configurations (apart, of course, from learn-
ing the meaning of the configurations). This, as we have seen,
has led to emphasis on a dominant role of sensory dynamics;
important as sensory processes are, however, they do not com-
pletely dominate either behavior or perception.
It is also important to see that the argument against an effect
of experience on perception often requires the assumption that
any perceived figure is perceived as a whole, in all respects.
Thus the upper diagram of figure 3 has been regarded as quite
unfamiliar to the observer (Kb'hler, 1929). Subjects shown this
diagram for the first time failed to find another which is con-
cealed in it, and which they had been shown before. The con-
clusion was drawn that an unfamiliar configuration obscured a
familiar, smaller onethat sensory dynamics dominates experi-
ence and the effects of learning. But the experimental diagram,
actually, was unfamiliar only as a whole: its parts, two parallelo-
grams and a set of parallel lines forming a Z, were certainly not
unfamiliar to the experimental subjects. Consequently, the con-
clusion is valid only if the total figure is an unanalyzable whole,
which it surely is not.
In the lower diagram of figure 3, a "4" is concealed. That it
may not be recognized is evidence that sensory dynamics on
occasion may override experience, in the delimitation of a figure.
This is a valid point. But when one looks for it, the 4 can be
seen, and this is not explained by Gestalt theory. A "special
vector" (Kohler, 1929) is operating, but this special vector seems
to be the factor of attention and experience that is involved
25 The "Nonsensory" Figure
in the nonsensory figure. Our problem is to find out how to
modify Gestalt theory so that it can comprise this factor of at-
tention and not have to deny the importance of learning in
perception.
Furthermore, Gestalt emphasis on the primitive figure, which
has a marked stability, obscures the fact that in ordinary per-
FIGURE 3. Diagrams from Kohler, Gestalt psychology, 1929 (figures 10
and 12). Courtesy of Liveright Publishing Corporation and of the author.
ception any figure-ground relationship is a highly unstable one,
with a practically constant fluctuation from one organization to
another. Even when one perceives the compact, clearly de-
limited figure which is such that any observer at once sees it,
one finds that the stability of the figure-ground relation is not
great; or, better, that its stability consists of a continual recur-
rence, instead of a constancy, of the figure. It is notorious that
attention wanders, and this is another way of saying that in per-
ception any figure is unstable; one looks at this part of the con-
figuration and that, and notices its corners or smooth contour, in
the intervals between seeing the figure as a whole. In ordinary
perception, moreover, the instability is far greater (Boring, 1933;
Pillsbury, 1913).
26 Summation and Learning in Perception
An adequate account of perception cannot be given in terms
of a figure-ground organization that has any great duration in
time, even when gaze is fixated at a single point in a configura-
tion. The fluctuations of attention which occur point directly to
a fundamental importance, in any perception, of nonsensory
factors. This of course has already been demonstrated for com-
plex indeterminate figures by Carmichael, Hogan, and Walter
(1932) and by Zangwill (1937); I propose that the same factors
must be taken into account in the perception of square or circle
as well.
"IDENTITY" IN PERCEPTION
Identity is defined here as referring to the properties of asso-
ciation inherent in a perception. The reference has two aspects:
first, a figure is perceived as having identity when it is seen im-
mediately as similar to some figures and dissimilar to others
that is, when it falls at once into certain categories and not into
others. This similarity can be summed up as spontaneous asso-
ciation, since it may occur on the first exposure to the stimulus
object. Secondly, the object that is perceived as having iden-
tity is capable of being associated readily with other objects or
with some action, whereas the one that does not have identity
is recalled with great difficulty or not at all, and is not recog-
nized or named easily. Identity of course is a matter of degree
and, as I shall try to show, depends on a considerable degree of
experience; it is not innately given.
Thorndike (1931, p. 87) has presented an approach to this
conception, using the term "identifiability," and has proposed
that identifiability promotes the formation of associations. If
one carries the analysis a little further, it appears that the propo-
sition is circular. Identifiability is not merely a perceptible dif-
ference of one figure from another when the two are side by
side, but implies a rememberable differenceidentifiability is
clearly, in the instances Thorndike gives, recognizability; and
recognition is one form of association. Thus Thorndike's propo-
sition is that associability affects the occurrence of associations.
The real point at which he is driving seems to be that there
are genuine differences of associability in different patterns.
Also, more is involved in these differences than the number of
27 "Identity" in Perception
trials necessary to establish recognition; there are also the spon-
taneous associations referred to in speaking of similarity. Rec-
ognizability goes with selective similarity, or generalization: the
figure that is readily remembered is also perceived as belonging
to a particular class of figures, and remembered so.
An irregular mass of color or a pattern of intersecting lines
drawn at random has some coherence and unity, but one such
figure is not readily recognized and distinguished from others
when it is seen a second time, and generalization (or similarity)
is not selective among a number of such stimuli. There is not
a total lack of distinctiveness and of generalization, however.
Two of Thorndike's figures which lack identity are not indistin-
guishable when seen together; and mistaking one figure for
another can be called generalization. Lashley and Wade (1946)
distinguish between the "so-called generalization" which means
only a failure to observe differences and the generalization
which involves perception of both similarities and differences.
The amorphous figure, lacking in identity, is generalized in the
first sense only.
A further illustration of these points is found in the develop-
ment of identity in the perception of chimpanzee faces by one
who has seen no chimpanzees before. Two animals seen side
by side are obviously different in details, but the inexperienced
observer is not able easily to remember one selectively. Also,
all chimpanzees at this stage look alike; the "so-called generali-
zation" occurs. With experience the perception of identity in-
creases. Similarity is still perceived between animals, and con-
fusion between some animals is still possible; but there is a
marked change in the perception, as shown in a much more
selective similarity of one animal to others, and in the radical
increase of the observer's capacity to associate even a new chim-
panzee with a specific name. Thus identity is a matter of de-
gree: readiness of recognition, and the extent to which gener-
alization is selective.
This discussion has been meant to establish the conception
of "identity" as an important property of perception which
should be kept carefully distinct from the "unity" of the per-
ceived figure (as well as from its "meaning"). Unity may be
innately determined, an immediate property of sensory dy-
28 Summation and Learning in Perception
namics, whereas identity is dependent on a prolonged experi-
ence. Because these two things have not been separated in the
past, it has appeared that perceptual organization is innate.
Some aspects of the organization are present, apparently, at the
first experience; but others are not.
INDEPENDENCE OF UNITY AND IDENTITY IN
SIMPLE FIGURES
The examples of the preceding section dealt with complex
perceptions. I wish next to review the evidence that shows that
unity and identity have separate determinants even in quite
simple perceptions. Seeing a circle as a single coherent object
is not the same as seeing it as a distinctive object, selectively
recognizable.
First, the evidence for man. It has already been said that the
figure-ground segregation is good at the initial occurrence of
human vision, as shown by Senden's (1932) reports, at the very
time when perception of identity appears to be practically nil.
Unity then can be perceived without identity. Investigators (of
vision following operation for congenital cataract) are unani-
mous in reporting that the perception of a square, circle, or
triangle, or of sphere or cube, is very poor. To see one of these
as a whole object, with distinctive characteristics immediately
evident, is not possible for a long period. The most intelligent
and best-motivated patient has to seek corners painstakingly
even to distinguish a triangle from a circle. The newly seeing
patient can frequently find a difference between two such figures
shown together, just as the normal adult can readily detect dif-
ferences between two of Thorndike's figures lacking "identifia-
bility," but the differences are not remembered. There is for
weeks a practically zero capacity to learn names for such figures,
even when tactual recognition is prompt and complete.
Another facet of the same lack of capacity is the failure to
generalize as normal persons do. When the patient first gets to
the point of being able to name a simple object promptly, rec-
ognition is completely destroyed if the object is slightly changed
or put into a new setting. The patient who had learned to
name a ring showed no recognition of a slightly different ring;
29 Independence of Unity and Identity
having learned to name a square, made of white cardboard,
could not name it when its color was changed to yellow by turn-
ing the cardboard over; and so on. These reports consistently
indicate that the perceived whole at first vision is simultane-
ously unified and amorphous. There is not a single instance
given in which the congenitally blind after operation had trouble
in learning color names; but a great number in which the per-
ception of identity in a simple figure was poor indeed (Senden,
1932, pp. 135-141).
A second evidence comes from species differences in the per-
ception of identity. For coherent patterns and simple groupings
the figure-ground relationship appears to be the same from rat
to man, so the significant differences of perceptual generaliza-
tion (one aspect of identity) argue strongly that figural unity
and identity have separate bases, (a) The discrimination of
simple geometrical figures by man and chimpanzee is unaffected
by reversal of brightness relations between figure and ground
(Gellerman, 1933); a white triangle on a black ground is gen-
eralized by these anthropoids to include a black triangle on a
white ground. For the rat, such a reversal completely disrupts
discrimination, and no amount of training with the white tri-
angle alone will produce "recognition" of a black one. There is
some uncertainty about the phylogenetic level at which the
capacity for this generalization appears; see Lashley (1938&,
p. 144), Smith (1936), and Neet (1933); but between the rat,
and chimpanzee or man, the difference in this respect seems
complete (although for simple figures only; human perception
of more complex figures in reversed brightness is definitely de-
fective, as in recognition of photographic negatives), ( b ) The
perception of a triangle or square by the rat is not generalized,
without specific training, to include a similar rotated figure
(Fields, 1932) but is generalized so by chimpanzee and by two-
year-old children, although a corresponding head rotation also
occurs (Gellerman, 1933). ( c ) Perception of a triangle is not
generalized either by rat or chimpanzee to include a triangle
made up of small circles, but the generalization is made by two-
year-old children (Gellerman, 1933; Lashley, 1938&).
Thus the perception of identity is different in different mam-
mals; the perception of primitive unity is practically the same.
SO Summation and Learning in Perception
Further evidence of the independence of unity and identity is
found in the peculiar "equivalences" or generalizations often ob-
served in rat behavior, and in the anomalies to be found in the
relative difficulty with which the rat learns to discriminate be-
tween certain patterns. Two discriminations that are equally
easy, from the human point of view, present very unequal diffi-
culty for the rat. For normal man a square and a circle are no
less obviously distinct than erect and inverted triangles; yet the
rat readily learns to discriminate between the triangles and con-
sistently has trouble with circle versus square (Lashley, 1938&,
pp. 155, 156). Some animals that have learned other patterns
(thus showing that their acuity is sufficient for the task) show
no sign of discriminating circle and square at all. A five-pointed
star and an H are for man clearly, inevitably distinct, while the
rat distinguishes them with difficulty and no faster learning
occurs than with quite irregular figures (Lashley, 1938b, p. 157).
Such facts mean a great difference of rat from man in perceiv-
ing the identity of simple, regular figures.
The same is evident in the patterns which may be "equivalent"
for individual rats, and in some failures of equivalence. One
animal may transfer his response from an erect triangle (versus
an inverted one) to a single horizontal line low on the stimulus
card (versus another higher on the card), while another animal
fails to discriminate the complete triangles with which he was
trained, when a slight change is made of their positions on the
respective stimulus cards. It is very doubtful in such cases that
the rat has perceived the pattern as a distinctive whole, and it
seems that a response is frequently determined by only a part
of a figure as simple as a triangle.
Lashley (1938&, p. 182) has recognized this, but interpreted
it by comparing the rat to a ski-jumper who does not pay atten-
tion in his jump to nonessentials such as the shape of a spec-
tator's hat. The general interpretation was made that there is
little significant difference in the perception of simple figures
by rat and man, but this conclusion is not supported by the
evidence.
When identity is distinguished from unity, we find great spe-
cies differences in the perception of identity and not in what is
seen as a primitive unity. The analogy between the rat's per-
31 Learning to Perceive Simple Figures
ceiving all the detail of a triangle in the jumping apparatus, and
the ski-jumper's perception of a hat, is validexcept that it sets
the human subject a much more difficult task. It is inconceiv-
able that a ski-jumper would not perceive a large triangle as
such, fully structured, if it lay close to and marked the point of
safe landing in an otherwise plain field, as it does for the rat in
the jumping apparatus. The rat, however, may respond selec-
tively to only the part of the figure which lies nearest the point
to which he jumps, although his field of vision is wide and
although (as Lashley has demonstrated) he perceives much
more than he has to in making a discrimination.
These considerations are the more convincing because the evi-
dence of the rat's frequent failure to see the triangle as a dis-
tinctive whole figure coincides with Senden's description of the
congenitally blind after acquiring visionthe normally reared
rat, and a man seeing for the first time, both have precisely the
same kind of difficulty.
It is reasonable to conclude that the perception of identity
(as defined) follows the same principles in rat and man but is
much more highly developed in normal man. Since there is no
evidence of any clear difference of the primitive figure-ground
organization perceived by these two species, but strong sugges-
tions that it is the same for both, the evidence is consistent with
the idea that identity and figural unity have separate physio-
logical bases. They are, that is, psychologically independent.
This makes it possible to suppose that additive processes may
occur in the development of identity without denying that the
figure has a primitive unity.
LEARNING TO PERCEIVE SIMPLE FIGURES
The facts already adduced in the last section have indicated
a role of learning in the perception of triangle or square as dis-
tinctive wholes. The idea that one has to learn to see a triangle
must sound extremely improbable, and so I shall now present
the evidence to this effect more systematically.
We have seen that the perceptions of the congenitally blind
after operation are almost completely lacking in identity. Sen-
den (1932, pp. 155-157) reports cases in which there was an
32 Summation and Learning in Perception
immediate perception of differences in two figures seen together,
but also a number of instances in which even this was not pos-
sible. Thus a patient sometimes saw differences between a
sphere and cube, sometimes not (p. 91). Color has been
found to dominate form persistently in the first vision of these
patients. Eleven months after operation the color names learned
by a patient in hospital were retained, but the little that had
been learned of form was forgotten (p. 135). An egg, potato,
and cube of sugar were seen by a patient repeatedly, until
naming was prompt, but then were not recognized when put
into colored light; the cube of sugar was well named when it
was seen on the table or in the investigator's hand but not rec-
ognized when suspended by a thread with a change of back-
ground (p. 138).
Such patients, when learning has proceeded far enough, mani-
fest the characteristic generalizations of the normal person, so
the initial difficulties are not to be put down to structural de-
fects of the sensory apparatus (Senden, pp. 173-175).
Riesen (1947) has fully confirmed the conclusion that ordi-
nary visual perception in higher mammals presupposes a long
learning period. His observations concerning the almost com-
plete visual incapacity of chimpanzees reared in darkness, and
the slowness of learning, are of the greatest importance. They
show that Senden's similar results with man are not due to some
inadequacy of the clinical tests, nor peculiarly human.
The course of perceptual learning in man is gradual, proceed-
ing from a dominance of color, through a period of separate
attention to each part of a figure, to a gradually arrived at iden-
tification of the whole as a whole: an apparently simultaneous
instead of a serial apprehension. A patient was trained to dis-
criminate square from triangle over a period of 13 days, and
had learned so little in this time "that he could not report their
form without counting corners one after another. And yet it
seems that the recognition process was beginning already to be
automatic, so that some day the judgment 'square' would be
given with simple vision, which would then easily lead to the
belief that form was always simultaneously given" (Senden,
1932, p. 160). The shortest time in which a patient approxi-
mated to normal perception, even when learning was confined
33 Learning to Perceive Simple Figures
to a small number of objects, seems to have been about a
month.
It is possible then that the normal human infant goes through
the same process, and that we are able to see a square as such
in a single glance only as the result of complex learning. The
notion seems unlikely, because of the utter simplicity of such a
perception to the normal adult. But no such argument can be
valid, since Lashley (1937) has shown that subjective simplicity
and immediacy may be very deceptive as an index of physio-
logical simplicity. There are moreover residual traces of learn-
ing in normal perception, and hints of its complexity.
Gellerman (1933) reports that chimpanzees and two-year-old
children recognized a triangle that had been rotated through
120 from the training position, but (in the one protocol that
is given) responded selectively only after a head rotation; and
persistent head rotation continued in the later discriminations.
Older human subjects do not need to make the same receptor
adjustment to recognize the figure in two positions, and so this
generalization may be a learned capacity, simple as it seems
to us.
Consider also the following evidence, which is suggestive
though perhaps not conclusive. When a simple figure such as
square, circle, or triangle, subtending a retinal angle of from
2 to 10, is fixated at one point, it tends in a second or so to
become almost amorphous except near the point of fixation. The
effect is not due to fatigue alone, nor to poor acuity outside the
macular zone: since (1) a single straight line does not suffer
nearly as much, (2) shifting fixation back and forth between
two corners of a triangle does not restore the clarity with which
the third is seen, and (3) an 8 circle drawn with a line 2 mm.
wide, and a 4 circle drawn with a line 1 mm. wide, seem to
give approximately the same effect.
The factors involved are evidently complicated; it will be
found, for example, that with a large figure merely imagining eye-
movements (of following the contours) will restore definition of
the figure. Also, these "imaginary" eyemovements, or subliminal
activations of the motor system, occur more frequently and are
less easy to control in looking at a smaller than at a larger
figure, and it is hard to be sure that the size of the figure is un-
34 Summation and Learning in Perception
important. But this at least seems definite, that a stable, clear,
and effective perception of circle or square is more possible with
eyemovement than without. Once the question is asked, any-
one can verify for himself the falsity of the implicit assumption
(usually made in the study of perception) that the figure acts
always as one, without a reinforcing perception of its parts as
distinct from the whole.
My point is not that eyemovements are essential to perception
by a sophisticated observer (nor, in the following paragraph,
that they are completely necessary for an image); but that the
perception is definitely clearer, more effective, with them than
without. This is really an evident fact. It is to be interpreted
in the light of all evidence, cited above, showing that the per-
ception of square or circle is slowly learned and depends origi-
nally on multiple visual fixations.
Directly in line with such phenomenological observations are
the following introspections. I find it very difficult to have a
clear image of a triangle, square, or circle without imagining or
actually making a series of eyemovements. Several others, asked
to make observation on this point, have reported the same thing.
It is hard or impossible, that is, to have a clear image of a tri-
angle as fixated at one point. Eyemovements definitely improve
the "image." They do not take the form, necessarily, of follow-
ing the figure's contours, but are apt to jump from point to
point, perhaps three to four points in all. Thus the distinctive-
ness of the image is not merely in the eyemovement pattern,
for approximately the same series of eyemovements may con-
tribute to a good image either of circle or square. Activation of
the motor system, overt or implicit (even possibly within the
cerebrum alone, with no activity of the final common paths),
contributes essentially to the development of visual integration
without being sufficient to it. As I have said, such evidence is
hard to evaluate, but it points to the same conclusion as Senden's
evidence, already cited, and is supported by it.
Clark and Lashley (1947) have confirmed the observations of
Kennard and Ectors (1938) and Kennard (1939), and have pro-
vided what appears to be an independent confirmation of the
argument above. Kennard found a one-sided loss of vision by
monkeys on extirpation of the opposite frontal eyefield, a cortical
35 Conclusions
motor area for head-and-eye movement. Clark and Lashley
have demonstrated this phenomenon convincingly, with an ade-
quate method of testing. The most significant and striking ob-
servation was startle by the monkey when an object was passed
from the blind side into the seeing side, at the moment of pass-
ing the midline. One might have argued that the animal could
"see" an object in his apparently hemianopic field but was not
able to move his eyes toward it. The observation referred to
rules that interpretation out, and other observations showed that
the hemianopia is a genuine failure to see (though it is tran-
sient, disappearing in a week or two).
Now the question is what the motor cortex can have to do
with visual perceptionunless perception intimately involves a
motor activity, liminal or subliminal. There is no reason to
think that the frontal cortex has anything to do with the recep-
tion of visual sensation, and the alternative seems to be that it
must have something to do with the elaboration of sensation
into visual perceptions. In Chapter 5 will be found a treatment
of perception which supposes that perception of even a simple
object involves a "phase sequence." This is a chain of central
cortical events with motor links. Although the motor activa-
tions may be subliminal and do not always produce overt re-
sponse, their role is essential in any perception. This would
account for the observations of Kennard and Ectors, and Clark
and Lashley, which therefore can be considered to be a con-
firmation of the argument of Chapter 5 as well as the argument
of the preceding paragraphs.
CONCLUSIONS
Animal experiments and the human clinical data alike indi-
cate that the perception of simple diagrams as distinctive wholes
is not immediately given but slowly acquired through learning.
Introspective observations which would not carry much weight
in themselves appear to agree fully with other evidence, show-
ing vestiges of a summative process involved in perceiving the
identity of circle or triangle; although such a figure is seen by
the adult clearly and is effectively discriminated at a single
glance, there are still traces left of complexities such as the
36 Summation and Learning in Perception
learning process described by Senden would produce, which for
normal persons must have occurred in early infancy and which
makes the unified perception possible.
The subjective experience of an irreducible simplicity in the
perception of square or circle may then be fully analogous to
the illusion of the image of the letter-square (Woodworth, 1938,
p. 42), where the subject thinks he has an actual image of the
square but can be shown not to have.* Although the percep-
tion of identity is good at a glance, it is improved by several
glances at different parts of the figure. This process of "succes-
sive part reinforcement," as an aid to perception, exists at the
same time as an essential unity of the whole; and theory must
provide for the additive process, with its motor elements, as well
as for the primitive unity.
This argument is not in any way a return to the old idea that
sensory integration occurs solely through motor activity, or that
the distinctiveness of a perception is solely or mainly due to
distinctive eyemovement. We know that this is not true. But
there are three theoretical possibilities, not two: (1) perceptual
integration is wholly the result of motor activity; (2) it is wholly
independent of motor activity; and (3) the motor activity is
important but not all-importantthe position that is taken here.
Grant fully that visual integration cannot be reduced to a
synthesis of unrelated elements through effector activity, and
the question remains, how much significance the motor factor
* The experiment is as follows. The subject is shown a diagram such as
x e a q
r 1 i s
o f z g
d y u p
and studies it until he has, apparently, an image of the whole square and
can "look at" it and read the letters off, one by one. If he really has such
an image, it will not matter in what direction he is asked to "read." Ac-
tually, it is found that the subject cannot reproduce the letters as fast from
right to left as from left to right, or promptly give the four letters, p, z, I, x,
that make up the diagonal from lower right to upper left. So what seems
a simple, immediately given image of the whole is actually a serial recon-
struction of parts of the figure. An "image" of triangle or square is simpler,
longer practiced, but may be fundamentally the same. The perception of
such figures, also, may involve a temporal sequence.
37 Conclusions
may still have. Receptor adjustment (head-and-eye movement)
is the most prominent feature of visual perception whether in
rat, chimpanzee, or manexcept in long-practiced habits. The
assumption has been tacitly made that the adjustments are un-
important for theory (once it was shown that they were not the
whole answer). The fact of eyemovement has been treated
only as a further (negative) evidence that the locus of excita-
tion in the retina is unimportant, since the changing retinal pro-
jection does not disturb perception. But obviously another
point of view is possible. The thesis of this chapter is that eye-
movements in perception are not adventitious. They contribute,
constantly and essentially, to perceptual integration, even though
they are not the whole origin of it.
^j. Field Theory
and Equipotentiality
The last chapter examined the proposition that perception is
a unitary process, in all its aspects. We can turn now to an in-
vestigation of the two theories, Kohler's and Lashley's, that have
resulted from that idea. In effect, this will be a critique of the
theory of perception, since Kohler and Lashley are the only
writers who both recognize the real problem of the neural
mechanisms of perceptual integration and attempt an adequate
solution. Others have avoided the crucial difficulty of specify-
ing the way in which perceptual generalization occurs. Either
like Pavlov (1928, 1932) they have not seen the difficulty at all,
or like Hull (1943) have put it to one side for solution later.
Hull has discussed the problem, but only to show that it is not
necessarily insoluble for his theory of learning (Hull, 1945).
For the present, Kohler and Lashley are the only ones who have
attempted to say where and how perceptual generalization takes
place.
For reasons already mentioned, however, the line of thought
that they have chosen may be a blind alley, and it must be
shown that there are other possible solutions. Until someone
challenges the theory of innate stimulus equivalence and equi-
potentiality, and of neural fields and gradients as the only
mechanism of integration, any theory with neural connections in
its ancestry is damned before it is born. Being gravid, and so
disapproving of infant damnation, let me look for another
gospel.
38
39 Clarification of Terms and Issues
CLARIFICATION OF TERMS AND ISSUES
As elsewhere, the issues here have sometimes been lost sight
of in the dust of a prolonged battle. There is one main ques-
tion: whether recognition, or a selective discriminatory re-
sponse, requires the excitation of specific neural cells or not.
The doctrine that it does not, we can refer to as equipoten-
tiality: Kohler does not use the term, but his position is the
same as Lashley's on this point, and a single term will do.
Explanations of the supposed equipotentiality differ greatly in
some respects but agree in their stress on patterning in the cen-
tral nervous system as against locus of excitation: so we can
again use a single term, configuration theory, to designate
Kohler's (1940; Kohler and Wallach, 1944) explanation by elec-
trical fields in the cortex and Lashley's (1942a) alternative idea
of irradiating waves of excitation and their interference patterns.
Next, sensory equipotentiality can be coined for Lashley's
"equivalence of stimuli," which is ambiguous. One may want
to refer to a limited degree of equipotentiality, in afferent struc-
tures, without accepting the whole theory of equipotentiality in
central action, and this term can be used for the purpose.
"Equivalence of stimuli" has a double reference. It may mean
only (1) that different stimuli can arouse the same response.
This is an observed fact of behavior, whatever one's interpreta-
tion of the fact. But Lashley has also used the term to mean
(2) that it does not matter what sensory cells are excited in
order to get a certain response, and this is interpretation. The
fact that two different patterns have the same effect may be
due to eyemovement, or to multiple learning processes (the po-
sition adopted here); or it may be because similar patterns of
excitation have the same ultimate motor effect, innately and
necessarily, whether the receptors excited are the same or not.
Let us then separate fact from interpretation; and let us refer
(1) to the behavioral evidence in speaking of the equivalence
of stimuli, and (2) to a particular interpretation in speaking of
sensory equipotentiality.
Finally, the idea of equipotentiality itself should be made
more explicit. "Potentiality" literally implies only that different
40 Field Theory and Equipotentiality
cells are able to acquire the same function in behavior. Anyone
must agree that this can happen, with separate learning proc-
esses. It is the basic idea of conditioning. But Lashley has
given the word another meaning, that all cells in a system neces-
sarily do acquire the same function, when excited in a given
pattern, so that separate learning processes are not necessary for
habits involving different stimuli or different parts of a cortical
field.
Suppose for example that a habit is set up in this way: the
stimulus is a circle, let us say, that falls always on the same
retinal cells in the lower right quadrant; the response is a move-
ment of the left hand. The stimulation then is confined to one
half of the visual area in the right hemisphere, and the response
is determined by motor systems that are also in the right hemi-
sphere. Equipotentiality implies (1) that any other retinal cells,
excited in a circular pattern, will elicit the same responsewith
either left or right hand; (2) that the right hemisphere may be
extirpated, and the left will be found then to have "learned"
whatever the right did; and (3) that this transfer of learning
from one set of cells, primarily excited, to other sets does not
depend on an earlier experience that set up connections between
them. The idea is that the transfer is an innate property of the
central nervous system, so that it is not possible for a particular
habit to depend on any particular neural cell within an ana-
tomical system, in any circumstances. If a whole system is
extirpated, habits will be lost (as when both visual areas are
removed); but it is not possible for a habit to be dependent on
one half of an anatomical system only (for the habit to depend
on one visual area alone, or on any part of it).
This is obviously a rather extreme position. An alternative
hypothesis to be proposed later recognizes a limited field action
in the sensory projection areas, and something like a limited
equipotentiality between cells that are in functional parallel in
physiological systems (this will be clear later on); but in the
main it derives transfer of response from earlier-established
habits and specific connections.
The issue is thus defined. We must recognize that Kohler
and Lashley have been completely successful in their original
attack on stimulus-response theory such as Watson's. No one
41 Brightness, Size, Pitch
now believes that retinal cells acquire direct connections with
muscle cells, as in a child's learning to say "cat" when he sees a
cat. The elimination of such oversimplified forms of switch-
board theory was a fundamental contribution. One need only
glance at the Journal of Neurophysiology to see that there are
gradients of neural activity; there is some mutual influence of
neighboring parts as an excitation is conducted from the retina
to area 17. Electrophysiology has fully confirmed the value of
such ideas, and they must enter into psychological theory.
But in fact, they have been generally accepted already, so
this is no longer an issue. Pavlov's irradiation and concentra-
tion is a field action. Hull and Spence both make use of the
notion of gradients. The question now is whether gradients
and fields are the only mechanism of a selective neural action
or whether they are combined with an equally important mech-
anism of connections and specialized conduction paths.
In short, configuration theory represents two things. It has
been, first, a necessary corrective to earlier ideas, such as
Titchener's and Watson's. This positive contribution must be
recognized as an important step forward. But there is also an
over-reaction from punctate sensory elements and sensori-
motor connections that is about as extreme as these older ideas.
As a theory, therefore, configurationism is just as vulnerable as
structuralism and the early behaviorism were.
Now let us see what the evidence is by which we can accept
or reject field theory and equipotentiality. Some of the evi-
dence that has been used to support these ideas actually is not
relevant to the question, and such evidence can be discussed
first, leaving the more relevant evidence for later consideration.
THE PERCEPTION OF BRIGHTNESS, SIZE, PITCH
In the literature, the perception of relative intensity has been
used almost entirely by writers on configuration theory for the
embarrassment of learning theory. Actually, it is no easier for
configuration theory to explain, and so is not really relevant to
the debate between the two. It has been thought to be crucial,
however, and so must be considered here. Man or animal tends
to perceive relative rather than absolute intensity, extent, or fre-
42 Field Theory and Equipotentiality
quency. One can readily train an animal to choose the larger
of two surfaces; it is extraordinarily hard to get him to choose
a particular size, except when differences are very great. These
facts are psychologically important, and they provided a deci-
sive answer to learning theory in its older forms; but we know
too little of the physiology of the relative perception to use it
as evidence for or against any currently maintained psycho-
logical theory.
The actual experimental evidence is as follows, taking the dis-
crimination of brightness as an example.
A rat is trained to go toward a light of intensity 1 and to keep
away from intensity 2 (i.e., twice as strong as intensity 1). He
is then presented with intensities 2 and 4, and chooses (if he
chooses at all) intensity 2which he was taught to avoid. (In
the test, discrimination may break down, choices being made at
random. This happens perhaps less than half the time; but a
consistent choice of an absolute stimulus value never occurs in
the rat in these circumstances, and the relative value is what
the animal discriminates, when he does discriminate.)
Or, if the rat is taught in the first place to avoid an intensity 1
and to choose intensity 2, in the test with 2 and 4 he will reject
the stimulus he was trained to respond to in favor of a new
one, intensity 4. This happens even when the animal has been
reared in total darkness and has never in his life seen that
stimulus before. This is, so far as one can see, still a conclusive
answer to any theory of behavior that deals in connections from
specific rods and cones to specific muscle fibers (Hebb, 1937&).*
But the perception of brightness is commonly cited in a
broader context, as supporting configuration theory in general
against "learning" theory in general. By implication, it is treated
as support for the idea that discrimination is not dependent on
the excitation of any particular cells, at any neural level. This
does not follow. The truth is that no one has produced any
good explanation of the relative perception, so it does not sup-
port one camp against the other; and the truth is, also, that
* This author, however, seems to have supposed that such experiments
rule out all possibility that connections are the basis of learning. The
hypothesis presented in the following section is enough to show that the
conclusion was unwarranted.
43 Hypothesis of Brightness
we know little or nothing about the locus of the cells whose
activity mediates the perception "brighter," or "smaller," or
"higher" (in pitch), so the facts neither deny nor support the
hypothesis of equipotentiality.
We know, with reasonable certainty, that the training process
does not form a connection all the way from receptors to ef-
fectors, but it is still possible to suppose that connections may
be formed at some intermediate levels. I shall outline here a
possible hypothesis, mainly to show that we must know more of
the physiology of sensation before brightness perception (for
example) can be made an argument for or against any existent
theory.
AN HYPOTHESIS OF BRIGHTNESS PERCEPTION
Marshall and Talbot's (1942) treatment of visual acuity makes
it possible to assume that the perception "brighter" depends on
the activity of particular cortical cells, but not of particular
receptor cells. We take for granted that intensity 2 excites
a larger population of first-order (retinal) neurons than inten-
sity 1 and with a higher rate of firing. Intensity 4 excites a
still larger population. The fact that 2 and 4 may elicit the
same selective response is an argument for a sensory equipoten-
tiality, but the equipotentiality may disappear farther on in the
system.
Marshall and Talbot point out that the whole visual system,
from receptors to the several layers of the cerebral cortex, must
act to damp strong stimulations, amplify weak ones. See what
this means for the behavioral problem. At each synaptic stage
in transmission the damping, or amplification, increases. A level
X may be reached in the system where the degree of activity is
constant for an adapted retinal area, with a brightness in the
middle range. The amount of activity at X, then, will be the
same whether the eye is adapted to intensity of 1, 2, or 4.
This of course is true only in adaptation. The system must
remain very sensitive to changes of stimulus intensity. There-
fore, when the eye after being adapted to 1 is focused on 2, the
level of activity at X goes up, say, from n impulses per second
to 1.5n. The same would happen when the eye adapted to 2
44 Field Theory and Equipotentiality
is focused on 4. In one part of the system, the same central
neurons might thus be active in the perception "brighter," re-
gardless of absolute stimulus values. (In other parts, the cells
that are aroused would still correspond partly to the absolute
intensity, and this helps to account for the fact that absolute
values can be detected, though the tendency is toward the rela-
tive discrimination.)
My point is not that this has explained the relative percep-
tion. I do not suppose that the problem is as simple as this has
made it sound. The point is that the known facts are not in-
consistent with the idea that the perception "brighter" involves
an activity in specific neural cells at some point in the system,
and thus that a perception of relative values is not incompatible
with some specific neural connections. Something else, a "sen-
sory equipotentiality" akin to neural gradients and field action, is
certainly involved; but the establishment of new synaptic con-
nections is not ruled out as the basis of an animal's learning to
respond to the brighter of two surfaces.
In the next two chapters some neurophysiological ideas are
proposed about visual perception. They could be elaborated
speculatively to deal with the perception of brightness and size,
though I have not done so. The perception of size, brightness,
and pitch should be written down for the present as not yet
accounted for, by any theory. Before they are, we shall have
to know much more about the physiology of sensation.
THE PERCEPTION OF PATTERNS
In the perception of patterns we reach surer ground, having
more of the relevant physiological information available and
being better able to see what bearing the facts have on the
theory of equipotentiality.
An animal that has learned to look for food behind a small
square and not behind a small circle will, without further train-
ing, choose a large square and reject a large circle. Small and
large squares are "equivalent stimuli." If each is perceived with
a single fixation, it is impossible for the total pattern in each
case to excite the same set of retinal cells. The current treat-
ment of perception disregards eyemovement; the inference has
45 The Perception of Patterns
been drawn, therefore, that the locus of the cells stimulated by
the square is unimportant. Only the pattern of stimulation
matters.
But we have seen, in the last chapter, that eyemovement can-
not be disregarded. A rat trained to jump toward a square does
not respond only to the total figure, but at times clearly responds
to parts of it, as separate entities. Eyemovement has never
been controlled in animal studies of perception. If the rat fix-
ates parts of the figure in succession the result is to superimpose
its parts, one after another, on the same retinal cells that were
involved in the original training. The locus of excitation may
then be essential. Evidently the equivalence of similar patterns
different in size does not in itself provide support for the theory
of equipotentiality.
For the evidence to become crucial, two conditions must be
fulfilled. The equivalent stimuli must be shown to be con-
ducted to different parts of the central nervous system; and
there must have been no previous experience that could have
set up connections between those parts before the experiment
began.
One of these conditions has been met by Levine's (1945fl,
1945&) experiment with pigeons. The optic chiasm in birds is
completely crossed, so one eye conducts to one optic lobe, the
other to the other. Levine achieved the truly surprising result
of finding transfer for patterns in the lower half of the visual
field and not in the upper half. For configuration theory, the
best that can be said with these results is that there is a limited
equipotentiality, between parts of the optic lobes (from the op-
posite point of view, these results definitely refute the theory
of a general equipotentiality in vision).
Levine's experiment did not, however, control earlier experi-
ence in his birds, and it remains possible that even the limited
equipotentiality suggested by his results is illusory. It is pos-
sible that patterns in the lower half of the bird's visual field-
as when the bird is foraging on the ground, or alighting on a
perch that extends into the field of vision of both eyeshad
established, by perceptual learning before the experiment began,
interconnections between the two optic lobes which might later
be a basis of transfer. This would not require that the bird
46 Field Theory and Equipotentiality
have learned beforehand the actual patterns used in the experi-
ment; only that the parts of those patterns (or "perceptual ele-
ments": see Chapter 5) have been learned.* That this possi-
bility is not far-fetched will be seen as we consider next the
instantaneous recognition of patterns by man, where it is evi-
dent that the perception of one or two parts of a figure may be
the clue to recognizing the whole.
Human beings readily recognize certain patterns in the tachis-
toscope. The time of exposure of the pattern is of the order of
one-fifth of a second. The subject cannot fixate the separate
parts of the pattern one after another, and cannot adjust his
eyes in advance so as to make it fall on any particular retinal
cells. This might be interpreted as equipotentiality, but there
are facts that make that interpretation practically impossible.
I propose that the human capacity for recognizing patterns
without eyemovement is possible only as the result of an inten-
sive and prolonged visual training that goes on from the mo-
ment of birth, during every moment that the eyes are open, with
an increase in skill evident over a period of 12 to 16 years at
least. The evidence supporting this proposition is found (1) in
the errors made in tachistoscopic recognition, (2) in the way
in which reading skill develops, and (3) in the rate and course
of learning by the congenitally blind after operation.
1. The recognition of patterns momentarily perceived, as in
the tachistoscope, is extremely defective except with simple
and conventional patterns, or very simple combinations of
them. The process is exactly that named by Woodworth (1938)
"schema with correction." This means that the pattern is per-
ceived, first, as a familiar one, and then with something missing
or something added. The something, also, is familiar; so the
total perception is a melange of the habitual.
* In particular, the slope of lines as they extend into the fields of both
eyes simultaneously. This explanation by a learning process is not, how-
ever, the only alternative to equipotentiality. Another possibility is that
part of each optic lobe, receiving impulses from the upper hemiretina only,
has a point-to-point projection to the other lobe. Transfer would thus be
innately provided for. The explanation by learning, however, should be
examined first, by repeating Levine's experiment with pigeons reared in
darkness and kept in darkness except when they are actually being trained
or tested.
47 The Perception of Patterns
The subject's reports are such as "a triangle with the top cut
off" or "a square with a crooked bottom." It is thus clear that
the subject is not only responding to the diagram as a whole;
he perceives its parts as separate entities, even though presen-
tation is so brief. Errors are prominent, and such as to show
that all the subject really perceivesand then only with rough
accuracyis the slope of a few lines and their direction and dis-
tance from one another. A drawing or a report of what is seen
tachistoscopically is not unlike a paleontologist's reconstruction
of early man from a tooth and a rib. There is a clear effect of
earlier experience, filling in gaps in the actual perception, so
that the end result is either something familiar or a combination
of familiar thingsa reconstruction on the basis of experience.
All that is needed to account for the recognition of pattern
without special adjustment of the eyes, to the extent that such
recognition really occurs, is ability to recognize roughly (1) the
slope of line, (2) degree of separation of points, in any retinal
projection. These in combination cover intersection of lines,
sharp inflections of line (or corners), and curvature of line (by
the variation of slope at different points, and the absence of
sharp inflection). What I am proposing, then, is that, during
the continuous, intensive, and prolonged visual training of in-
fancy and childhood, we learn to recognize the direction of line
and the distance between points, separately for each grossly
separate part of the visual field.*
* Here, as elsewhere, I have chosen to be more definite than is actually
required by the theory to be developed; this is done with the conviction
(1) that vagueness is not desirable as such, in psychological theory, and
(2) that the more definite assumption may turn out to be correct, and at
any rate should be explored.
If it is not correct, however, the later theorizing is unaffected. An
alternative possibility to be considered is that some form of Lashley's
interference-pattern theory may hold in modified form. Lashley has used
the theory to explain perception of a square or triangle as a single unified
process, and it will be seen in the later part of this chapter that the theory
in this form cannot be maintained; but it might be feasible to apply it to
single lines instead. Then the stimulation from a line of a certain slope in
one part of the visual field would be innately equivalent to the stimulation
of a similar line in another part of the field. I know of no evidence that
opposes this, but no evidence for it either.
Psychologically, such an explanation might be more plausible; what it
48 Field Theory and Equipotentiality
I have already admitted that such ideas sound unlikely. The
perception of simple objects seems so simple and direct, so ob-
viously immediate, that it cannot be complex and the result of
a long-drawn-out learning process. But it was shown in Chap-
ter 2 that the learning process is necessary, that the end result
is complex, whatever it may seem.
It should be noted that these ideas do not mean that each
separate row of single cells in the retina must be separately con-
ditioned for the perception of horizontal line (e.g.) in different
retinal projections. Marshall and Talbot's discussion of visual
acuity implies a significant overlap of excitation, at the level of
the primary visual cortex, from lines that have distinct retinal
projections. It is not clear how great this spread is, in terms of
retinal angle, but it may be of the order of 2 in central vision
and considerably greater at the periphery. Accordingly, the
number of separate visual habits necessary to ensure the recog-
nition of horizontal line in any retinal projection might be, per-
haps, not greater than 10 or 15. Lines of slightly different slope,
likewise, would not have to be learned separately (remember
that accuracy in discriminating slope tachistoscopically is not
great). The number of separate perceptual habits needed to
account for tachistoscopic vision, consequently, is by no means
infinite.
Now consider how intensive and prolonged the human visual
training is. As we shall see later, every single movement of the
eyes, when exposed (as they always are) to an unevenly lighted
field, produces a number of excitations corresponding exactly to
the excitation from a number of lines all parallel to the direc-
tion of eyemovement. The eyes are constantly moving when the
subject is awake. Every waking moment then provides the
necessary conditions for reinforcing the perception of lines, in
every part of the retina.
would postulate is that parallel "ridges" of activity in area 17 would tend
to excite the same, diffusely arranged, cells in 18. My neural schematizing,
in the following two chapters, would then go on as before, but without the
necessity of supposing that different parts of the retina must be separately
conditioned to produce recognition of the slope of line in any part of the
field.
The question is, I believe, open to experimental test, and will be inves-
tigated.
49 The Perception of Patterns
We have already seen that it takes months for the first direct
apprehension of a figure such as a plain, well-marked triangle to
be established. The normal human infant, apparently, reaches
this stage quite early in life; but his further training continues
every moment that his eyes are open, and must extend his ca-
pacity for prompt recognition of patterns falling outside the
macula.
2. Speed of reading increases up to the age of 12 to 16 years,
or perhaps even later. This speed, depending on the instan-
taneous recognition of larger and larger blocks of letters, may be
regarded as a function of the training of peripheral vision; it
implies that learning to recognize a word immediately (as a
single Gestalt) in direct central vision does not mean recogniz-
ing it immediately at any angle of regard, though this is what
the theory of equipotentiality requires.
It might be argued that the theory does not require this, be-
cause of differences of acuity between central and peripheral
vision. Peripheral recognition of a word may depend, for ex-
ample, more on its general outline and less on particular letters.
It might be argued, that is, that an increased speed of reading
depends on learning to use different cues from those used in cen-
tral vision. This may well be true, but such an argument aban-
dons the fundamental thesis of equipotentiality, namely, that the
same properties of form are perceived whatever the retinal pro-
jection, provided acuity is sufficient. Above all, such an argu-
ment would imply that peripheral recognition and central recog-
nition depend on separate visual habits, which is the argument
of this chapter.
More direct evidence on the point is available, however, than
when the above considerations were first formulated. These
considerations suggested the need of a re-examination of some
features of tachistoscopic perception, and this has been begun
by M. Mishkin and D. Forgays. Their experiments show di*
rectly that reading does not train all parts of the retina in the
same way, even when acuity does not enter the picture. The
work will be reported in detail elsewhere; in summary, the fol-
lowing results have been obtained. Fluent readers of English,
used to reading from left to right, are able to recognize words
to the right of the fixation point between two and three times
50 Field Theory and Equipotentiality
as well as words to the left of fixation. Readers of Jewish (in
which the words run from right to left) recognize more words
to the left of fixation. These results can be obtained when the
subject does not know whether an English or a Jewish word is
coming next, or whether it will fall to the right or the left of
fixationthat is, with a random order of left-right, English-Jewish
presentations. The subjects that have been studied were not
very fluent in Jewish, apparently, and the left-right difference,
with Jewish words, was not statistically significant (P = 0.15);
but it is certain that whatever factors make the English word
better recognized to the right of fixation does not affect recog-
nition of Jewish words. Either there is no difference, or the
Jewish word is recognized better to the left.
This shows directly that there is not an equipotentiality be-
tween the left and the right hemiretina, or left and right oc-
cipital cortex, in man. A learned response can be more readily
elicited from one part of the retina than another, which provides
a good deal of support for the assumption, made in these chap-
ters, that a separate learning process is necessary for the elicita-
tion of the same response by stimulation of a separate region in
a receptor surface.
3. Finally, there are Senden's explicit reports of the way pat-
tern vision develops after the congenitally blind are operated on
(see Chapter 2). The patient, despite weeks of practice, must
still count corners to distinguish a square from a triangle: here
there is no unimportance of the locus of stimulation, no evidence
of equipotentiality. When the patient first gets to the point of
naming an object promptly, the recognition is destroyed by put-
ting the object into a new setting or by changing its color. The
significant fact is that the characteristic normal generalization
shows up later, after a prolonged and arduous training process.
The evidence is clearly that the apparent equipotentiality is the
sequel and product of learning; the perception of pattern is spe-
cific and limited at first, generalized only with further practice;
and in these cases there is no support whatever for the idea that
when a pattern is recognized at one angle of regard it will be
at once recognized in any other.
51 Clinical and Anatomical Evidence
CLINICAL AND ANATOMICAL EVIDENCE
We can turn next to the clinical and anatomical evidence that
is related to the question of equipotentiality in perception.
The conclusions reached already are supported by the per-
ceptual completions in hemianopia, described by Fuchs (1920)
and Lashley (1941). When one occipital pole of the brain has
been destroyed by injury, or temporarily loses its function in a
migraine attack, the patient becomes blind in half the visual
field. If a simple symmetrical object such as a solid white
square or a billiard ball is fixated in the midline, the patient
nevertheless reports that he sees all of it. That he does not, but
instead completes in his perception what he really sees only half
of, is shown by another fact. If half the object is presented
instead, in such a way that the missing half would have fallen in
the blind side, the patient still "sees" the whole object.
What is the basis of this completion? Koffka (1935) treats it
as a further evidence of Gestalt principlesan instance of the
operation of field forces. This, however, is precisely what it
cannot be.
In two of Fuchs' cases at least (Fuchs, 1920, pp. 424, 436) the
gunshot wound that produced hemianopia must have destroyed
one area 17. The bullet entered the skull in the midline at the
occipital pole, and could hardly have destroyed the optic radia-
tions on one side without also destroying the visual cortex on
that side. But area 17 is the cortical tissue in which a dynamic
completion, due to field forces, would have to occur; there is no
isomorphism in the excitation conducted beyond 17conduction
is diffuse, and the topological relationship to the visual stimulus
that is retained from the retina to area 17 is lost beyond it, so
the field-force idea is applicable only at this level. The com-
pletion, then, is not the result of a field process.
In the case of migraine scotoma described by Lashley, the
symptoms indicate that one visual area was nonfunctional dur-
ing an attack. But this again is the tissue in which interference
patterns would have to originate, to produce the perception of
a complete object fixated in the center. According to the theory
52 Field Theory and Equipotentiality
of interference patterns (Lashley, 1942a) the perception of a
semicircle lying immediately to the right of the fixation point
depends on the particular interference pattern that is set up by
that diagram in the left visual cortex. A whole circle, fixated in
the middle, sets up two such interference patterns, one from
each visual area. In the hemianopic, only one can occur; why,
then, should the patient see a whole circle? The conclusion is
FIGURE 4. Diagramming roughly the changes in cortical projection of a
square when the fixation point only is changed: based on the data of Polyak,
1941, and his figure 100, for a square subtending a visual angle of 18 20'
(the size of the "central area" of the retina). 1, fixation on the upper right
corner of the square, which thus falls in the lower left visual field and pro-
duces an excitation in the upper right cortex only; 2, fixation on the lower
right corner; 3, bilateral projection with fixation on the center of the square;
4, bilateral, fixation on the midpoint of the top line of the square; 5, fixation
on midpoint of bottom line. F, projection of fixation point; VM, vertical
meridian.
inevitable that the "completion" occurs in some other, undam-
aged, area of the cortexand that it cannot, consequently, be a
field process.
Next, consider some of the anatomical properties of the visual
cortex in which the field process must occur.
There is a point-to-point correspondence between the retina
and the visual cortex, or area 17 (Lashley, 1934; Polyak, 1941).
Thus there is an isomorphism of the visually stimulating surface
with the resulting excitation in area 17, and both the theories
that we are examining assume that this fact accounts for the per-
ception of a square, for example, as a distinctive whole. But
there are definite limits to what isomorphism can account for in
perception.
Figure 4 represents roughly the patterns of cortical excitation
53 Clinical and Anatomical Evidence
that are aroused by looking at different parts of the same pat-
tern, a square. These show how hard it is to suppose that the
perception of a square, as distinct from other quadrilaterals, is
determined by the shape of the isomorphic excitation in area 17,
since that shape varies so much.
(There are undoubtedly major errors in the figure, which is
based on a diagram by Polyak [1941, figure 100]. Polyak made
no attempt to represent corresponding retinal and cortical points
in detail; the necessary evidence for doing so is actually not in
existence. Polyak's diagram also deviates in some respects from
his text, which is presumably the more accurate; his figure 100,
for example, shows the "central fovea" half as wide as the "cen-
tral area," whereas his text, pp. 230-231, suggests that the differ-
ence in size is greater. Figure 4 follows his diagram, not the
text. Finally, the human visual area cannot be represented ac-
curately as a flat sheet.)
Figure 4 is only roughly accurate; it seems certain, however,
that the distortion, in the pattern of cortical excitation, that oc-
curs when fixation is changed from one to another part of the
figure is at least as great as that diagrammed. If the data of
Marshall and Talbot (1942, p. 134) had been used instead, the
discrepancies would have been greater. Furthermore, a differ-
ence in size (in terms of retinal angle) between two squares
would produce even greater changes in the shape of the cortical
excitation, although squares of different size are also supposed
to derive their common identity from the distribution of excita-
tion in area 17.
The reproduction of retinal patterns in area 17 is topological,
not topographical; and if identity were completely determined
by the shape of the cortical excitation, regardless of what the
fixation point may be, the perceived properties of a square with
shifting gaze or at different distances should change markedly-
remaining, it is true, a quadrilateral (topological reproduction
provides for this), but not retaining the same proportions. That
is, errors should occur, in the perception of quadrilaterals, which
do not occur; or in the perception of triangles; and so on. A
circle in one angle of regard should be confused with an oval
in another angle. One might postulate a differential cell den-
sity to correct the distortion of topographical relations which I
54 Field Theory and Equipotentiality
have diagrammed, but there is no support from histological
study of area 17 to support this ad hoc assumption, and I can
find no way of doing it for one pattern which would not mean
an even greater distortion for another.
However, another way of dealing with this obstacle for field
theory has been proposed by Kohler and Wallach (1944), who
suggest that a differential "satiation" or fatigue would correct the
anatomical distortion of the cortical retina. This proposal will
be considered with other aspects of their theory of figural after-
effects.
THE THEORY OF FIGURAL AFTER-EFFECTS
Kohler and Wallach (1944) have proposed a new treatment
of perception which is in some respects very attractive and deals
with a certain set of facts very efficiently. On the other hand,
I believe that the theory is inconsistent with some of the facts
reported by Kohler and Wallach themselves, and that it cannot
be considered to have disposed of the anatomical difficulty that
has just been discussed.
Following some original experiments by Gibson, Kohler and
Wallach have studied certain peculiar fatigue effects resulting
from prolonged inspection of visual objects. The method of
experiment is this: The subject keeps his eyes fixed, for some
two to three minutes, on a single "fixation point" in a large sur-
face on which a diagram is also presented. The fixation point
is determined by the experimenter so that the "inspection figure"
stimulates a particular part of the retina. Then, after the in-
spection period, the subject looks at a second surface, again with
a fixation point determined by the experimenter, which is placed
so that a "test figure" on this second surface also stimulates a
particular part of the retina. Thus in figure 5 are represented
both an inspection and a test figure, superimposed so that one
can see the relationship between the retinal projection of each.
The fixation point is shown by a small x; the inspection figure
(at which the subject stared first) is the large rectangle marked
I, and the test figure is the set of four small squares each
marked T. These four squares are so placed that two fall in a
region of greater fatigue or "satiation," near the boundary of the
55 he Theory of Figural After-Effects
rectangle, and two fall farther away, allowing one to discover
the changes that occur in the neighborhood of the inspection
figure, where satiation has occurred.
When a test figure is projected on the retina near the place at
which the inspection figure was projected, several effects can be
observed in the test figure. These
effects include a color effect (fad-
ing), a depth effect (apparent in-
crease of distance from the observer),
and displacement effects (apparent
change in direction of line, in posi-
tion, or in size). It is the displace-
ment effects with which we are con-
cerned here. Kohler and Wallach
propose, first, that the basis of per-
ception is the occurrence of electric
currents aroused by the visual ex-
citation in the cortex of area 17; sec- FIGURE 5. Diagram from
ondly, that the flow of these currents,
Kohler and Wallach, Proc.
in the neighborhood of the excitation,
Amer. Phil Soc., 1944, 88,
increases the resistance of the tissue
p. 280. Courtesy of Ameri-
can Philosophical Society
through which they flow. This in-
and of the authors. The
crease of resistance is satiation, and large square, /, is the "in-
the changes it causes in current flow
spection figure" to which the
account for displacement effects.
subject is first exposed; the
large square is then removed
This is on the assumption that per-
and the subject looks at th
ceived distance between two objects four smaller "test" squares
"varies with the degree to which
each marked T.
corresponding cortical objects [i.e.,
the excitations in area 17] are interrelated by their figure cur-
rents" (Kohler and Wallach, 1944, p. 334). Just what this inter-
relation is and how it operates is not specified, but its degree
decreases when the tissue between two areas of excitation has
an increased resistance. The two corresponding objects in the
visual field then look farther apart. It is also important to ob-
serve that Kohler and Wallach state explicitly (p. 337) that ap-
parent displacements are not dependent on actual changes in the
locus of the excitations in area 17: only on an increased resistance
to current flow in the tissue between them.
These assumptions account economically for some of the phe-
56 Field Theory and Equipotentiality
nomena described by Kohler and Wallach. Other phenomena,
however, seem inconsistent with the theory. Consider first the
perceived size of a figure within an area of satiation, including
the special case where the test figure coincides with the inspec-
tion figure. According to the theory, the perceived size of a
square or circle should increase with increased satiation, since
the greater resistance to current flow between opposite sides
should produce an overestimation of the distance between them.
In fact, the opposite is frequently reported by Kohler and Wal-
lach: when a test figure, such as square or circle, coincides
exactly with the locus of the previously seen inspection figure,
its apparent size diminishes. Again, in figure 5, the excitations
produced by the two small test squares on the right, near the
boundary of the inspection rectangle, are separated by tissue
that is more satiated than that separating excitations from the
two small squares on the leftyet the two on the right appear
closer together than those on the left. This difficulty for the
theory is found in the data obtained by Kohler and Wallach
with their figures 1, 2, 3, 4, 5, 7, 10, and so on.
The problem has been recognized by the authors and is
briefly discussed at the end of their monograph (pp. 351, 356).
They suggest two possible solutions: one is that apparent size
may be affected by the presence of a large area of satiation sur-
rounding the figure, as well as by the degree of satiation within
it (p. 351); another is the possibility that an immediate self-
satiation, by the test figure itself, may have a greater effect than
immediate self-satiation plus a pre-existent satiation (p. 356). If
I understand these explanations, they may undermine all the
preceding argument, much as "afferent neural interaction" under-
mines Hull's theory by removing the definiteness of inference
from his other postulates (p. 15). Certainly these two influ-
ences, of immediate self-satiation and of a surrounding area of
satiation, would have to be reckoned with in the production of
other phenomena, else their existence would become an ad hoc
postulate to be used only where difficulty appears. This of
course may mean only that the theory needs to be worked out
in more detail, as Kohler and Wallach themselves have pointed
out.
57 The Theory of Figural After-Effects
Next, a difficulty of a somewhat different kind, related to the
anatomical distortions of the cortical retina (the difficulty for
configuration theory that was discussed in the preceding sec-
tion).
To deal with the anatomical fact that objects equally spaced
in the visual field are not equally spaced for the cortical retina,
the key assumption is made by Kohler and Wallach that a per-
manent satiation in the peripheral cortical retina would be
greater than in the cortical fovea; and, accordingly, that ana-
tomical distortion such as has been diagrammed in figure 4
would be corrected functionally. The greater peripheral satia-
tion is thought to occur as follows. Contours equally spaced on
the retina would not be equally spaced in area 17, because the
peripheral retina has a smaller cortical representation. So an
equal distribution of contours in the visual field, and conse-
quently on the retina, would result in a bunching of contours in
the peripheral part of the cortical retina, and a wide spacing of
contours in the cortical macula. On this argument, the closer
spacing of contours would produce a higher level of satiation in
peripheral vision, which would compensate for the fact that a
given retinal angle here is represented by a much shorter cor-
tical distance than the same angle in central vision (Kohler and
Wallach, 1944, p. 345). Although the cortical distance is shorter,
resistance is higher too: so perceived distance remains the same.
But all this has overlooked an essential fact. Acuity in cen-
tral vision is very much higher than in peripheral vision, and so
there is no basis for assuming that the central cortical retina
would be the site of a smaller number of excitations per centi-
meter. Kohler and Wallach's argument assumes, essentially, that
all contours in the visual field are supraliminal for the periphery;
given this assumption, their argument stands; but we know ac-
tually that the assumption cannot be made, that many contours
are supraliminal for central vision and subliminal for peripheral
vision. These contours, subliminal at the periphery, would fill
in the gaps between the more widely spaced excitations in cen-
tral vision that originate from contours which are supraliminal
for the whole visual system, peripheral as well as central. There
is no evidence permitting one to be certain that the frequency
of contours that are subliminal for the peripheral field is such
58 Field Theory and Equipotentiality
as to compensate exactly for the greater spacing of other con-
tours in the macular area of the visual cortex; but what evidence
we have implies this, since differences of acuity appear to be
closely related to the ratio of centimeters of cortex per degree
of visual angle (Marshall and Talbot, 1942).
The theory of figural after-effects, therefore, has not avoided
the difficulties for configuration theory that are to be found in
the anatomy of the visual cortex.
CONCLUSIONS
The conclusions that have been arrived at concerning field
theory and equipotentiality, in this and the preceding chapter,
may be summarized as follows. The examination of these theo-
ries of perception was undertaken in the first place because no
way could be found of dealing with attention and attitude, in
terms of the configurational hypothesis. This still stands as an
essential weakness, which by itself would justify a search for
some other approach to perception. On closer inspection, how-
ever, one also finds that configuration theory has flaws even in
the treatment of perception itself; and some of the flaws are
serious. It is not possible to reject the theory finally, for an
answer may be found for each of the objections that have been
raised, and the alternative treatment that is proposed in the two
following chapters has its own difficulties and shortcomingsit
is in fact because my theorizing about perception is in some re-
spects implausible that I am obliged to emphasize weaknesses in
other theories, just as later (in Chapter 8) a rather vague treat-
ment of motivation is bolstered up by trying to show that Hull's
alternative is not as precise as it seems.
The fundamental difficulty with configuration theory, broadly
speaking, is that it leaves too little room for the factor of expe-
rience. It makes difficulty in seeing how learning can occur
(particularly, perceptual learning), and in seeing how autono-
mous central processes can exist and influence behavior. Asso-
ciation (as distinct from conditioned reflex) theory, on the other
hand, may make it possible to adopt a halfway position in which
one can take advantage of some of the obvious values both of
configurationist and of connectionist theories.
59 Conclusions
It is important as psychology comes of age to avoid, if pos-
sible, the extreme positions that have often been adopted in the
discussions of the past. It is consequently relevant here to point
out that a large part of the original thesis of the configuration-
ists (the Gestalt group, Lashley, Tolman, et al.) has already
been accepted and is fully made use of in current theorizing;
and also that the theory to be presented in the following chap-
ters is explicitly designed to deal from the first with the prob-
lems of form perception and attention or setproblems whose
existence has been most insisted on in recent years by the con-
figurationistsas well as those problems of learning and memory
emphasized by learning theorists. Thus the present argument
is based at least as much on Gestalt as on learning theory.
The current debate between these two theories cannot be
understood on the assumption that they are utterly opposed to
each other. It is rather to be seen as like the running battle
between the Left and the Right in governmental policy: the
Left continually insisting on the inadequacy of present theory,
working to broaden and elaborate the conceptual armament of
the experimenter; the Right insisting always on the dangers of
being too free with changes that may open the doors to animism,
on the importance of demonstrable evidence and intelligible
definitions, and on the necessity of fully exploring the ideas
we have now before rejecting them as insufficient. But in this
conflict, as in the political one, it will be found over any
great period of time that the Right (in current psychology,
learning theory) has adopted again and again the ideas origi-
nally advocated by the Left (currently configurationism), once
they are properly aged and found not to have spoiled. Thus
modern learning theory is "sophisticated"; it deals freely in
ideas that were once anathema to itideas of gradients, gener-
alizations, interaction among afferent processes as well as the
classical afferent-efferent action, and so on. Obviously both
Left and Right are needed for the development of psychology,
though it may be easy for the configurationist to forget the
value of the systematizing and search for rigor made by his
opponent, or for the learning theorist to forget where some of
his ideas originally came from.
4. The First Stage of Perception:
Growth of the Assembly
This chapter and the next develop a schema of neural action
to show how a rapprochement can be made between (1) per-
ceptual generalization, (2) the permanence of learning, and (3)
attention, determining tendency, or the like. It is proposed first
that a repeated stimulation of specific receptors will lead slowly
to the formation of an "assembly" of association-area cells which
can act briefly as a closed system after stimulation has ceased;
this prolongs the time during which the structural changes of
learning can occur and constitutes the simplest instance of a
representative process (image or idea). The way in which this
cell-assembly might be established, and its characteristics, are
the subject matter of the present chapter. In the following
chapter the interrelationships between cell-assemblies are dealt
with; these are the basis of temporal organization in central
processes (attention, attitude, thought, and so on). The two
chapters (4 and 5) construct the conceptual tools with which,
in the following chapters, the problems of behavior are to be
attacked.
The first step in this neural schematizing is a bald assumption
about the structural changes that make lasting memory possible.
The assumption has repeatedly been made before, in one way
or another, and repeatedly found unsatisfactory by the critics
of learning theory. I believe it is still necessary. As a result, I
must show that in another context, of added anatomical and
physiological knowledge, it becomes more defensible and more
fertile than in the past.
The assumption, in brief, is that a growth process accompany-
ing synaptic activity makes the synapse more readily traversed.
60
61 Dual Trace Mechanism
This hypothesis of synaptic resistances, however, is different
from earlier ones in the following respects: (1) structural con-
nections are postulated between single cells, but single cells are
not effective units of transmission and such connections would
be only one factor determining the direction of transmission;
(2) no direct sensori-motor connections are supposed to be
established in this way, in the adult animal; and (3) an inti-
mate relationship is postulated between reverberatory action
and structural changes at the synapse, implying a dual trace
mechanism.
THE POSSIBILITY OF A DUAL TRACE MECHANISM
Hilgard and Marquis (1940) have shown how a reverbera-
tory, transient trace mechanism might be proposed on the basis
of Lorente de No's conclusions, that a cell is fired only by the
simultaneous activity of two or more afferent fibers, and that
internuncial fibers are arranged in closed (potentially self-excit-
ing) circuits. Their diagram is arranged to show how a rever-
beratory circuit might establish a sensori-motor connection be-
tween receptor cells and the effectors which carry out a condi-
tioned response. There is of course a good deal of psycho-
logical evidence which is opposed to such an oversimplified
hypothesis, and Hilgard and Marquis do not put weight on it.
At the same time, it is important to see that something of the
kind is not merely a possible but a necessary inference from
certain neurological ideas. To the extent that anatomical and
physiological observations establish the possibility of reverbera-
tory after-effects of a sensory event, it is established that such
a process would be the physiological basis of a transient
"memory" of the stimulus. There may, then, be a memory trace
that is wholly a function of a pattern of neural activity, inde-
pendent of any structural change.
Hilgard and Marquis go on to point out that such a trace
would be quite unstable. A reverberatory activity would be
subject to the development of refractory states in the cells of
the circuit in which it occurs, and external events could readily
interrupt it. We have already seen (in Chapter 1) that an
"activity" trace can hardly account for the permanence of early
62 Growth of the Assembly
learning, but at the same time one may regard reverberatory
activity as the explanation of other phenomena.
There are memories which are instantaneously established,
and as evanescent as they are immediate. In the repetition of
digits, for example, an interval of a few seconds is enough to
prevent any interference from one series on the next. Also,
some memories are both instantaneously established and perma-
nent. To account for the permanence, some structural change
seems necessary, but a structural growth presumably would re-
quire an appreciable time. If some way can be found of sup-
posing that a reverberatory trace might cooperate with the
structural change, and carry the memory until the growth change
is made, we should be able to recognize the theoretical value
of the trace which is an activity only, without having to ascribe
all memory to it. The conception of a transient, unstable rever-
beratory trace is therefore useful, if it is possible to suppose
also that some more permanent structural change reinforces it.
There is no reason to think that a choice must be made between
the two conceptions; there may be traces of both kinds, and
memories which are dependent on both.
A NEUROPHYSIOLOGICAL POSTULATE
Let us assume then that the persistence or repetition of a re-
verberatory activity (or "trace") tends to induce lasting cellular
changes that add to its stability. The assumption * can be pre-
cisely stated as follows: When an axon of cell A is near enough
to excite a cell B and repeatedly or persistently takes part in
firing it, some growth process or metabolic change takes place
in one or both cells such that A's efficiency, as one of the cells
firing B, is increased.
The most obvious and I believe much the most probable
suggestion concerning the way in which one cell could become
more capable of firing another is that synaptic knobs develop
and increase the area of contact between the afferent axon and
efferent soma. ("Soma" refers to dendrites and body, or all of
* See p. 229 for a further discussion of this point and an elaboration of
the assumption made concerning the nature of memory.
63 A Neurophysiological Postulate
the cell except its axon.) There is certainly no direct evidence
that this is so, and the postulated change if it exists may be
metabolic, affecting cellular rhythmicity and limen; or there
might be both metabolic and structural changes, including a
limited neurobiotaxis. There are several considerations, how-
ever, that make the growth of synaptic knobs a plausible con-
ception. The assumption stated above can be put more defi-
nitely, as follows:
When one cell repeatedly assists in firing another, the axon
of the first cell develops synaptic knobs (or enlarges them if
they already exist) in contact with the soma of the second cell.
This seems to me the most likely mechanism of a lasting effect
of reverberatory action, but I wish to make it clear that the
subsequent discussion depends only on the more generally
stated proposition italicized above.
It is wise to be explicit on another point also. The proposi-
tion does not require action at any great distance, and certainly
is not the same as Kappers' (Kappers, Huber, and Crosby,
1936) conception of the way in which neurobiotaxis controls
axonal and dendritic outgrowth. But my assumption is evi-
dently related to Kappers' ideas, and not inconsistent with
them. The theory of neurobiotaxis has been severely criticized,
and clearly it does not do all it was once thought to do. On
the other hand, neurobiotaxis may still be one factor determin-
ing the connections made by neural cells. If so, it would
cooperate very neatly with the knob formation postulated above.
Criticism has been directed at the idea that neurobiotaxis di-
rects axonal growth throughout its whole course, and that the
process sufficiently accounts for all neural connections. The
idea is not tenable, particularly in view of such work as that
of Weiss (1941&) and Sperry (1943).
But none of this has shown that neurobiotaxis has no influ-
ence in neural growth; its operation, within ranges of a centi-
meter or so, is still plausible. Thus, in figure 6 (Lorente de
No, 1938a), the multiple synaptic knobs of fiber 2 on cell C
might be outgrowths from a fiber passing the cell at a distance,
and determined by the fact of repeated simultaneous excita-
tions in the two. Again, the course followed by fiber 7 in the
64 Growth of the Assembly
FIGURE 6. Relationships between synaptic knobs and the cell body. From
Lorente de N6, 1938a. Courtesy of Charles C. Thomas and of the author.
65 A Neurophysiological Postulate
neighborhood of cell D may include deflections from the
original course of the fiber, determined in the same way.
The details of these histological speculations are not impor-
tant except to show what some of the possibilities of change at
the synapse might be and to show that the mechanism of learn-
ing discussed in this chapter is not wholly out of touch with
what is known about the neural cell. The changed facilitation
that constitutes learning might occur in other ways without
affecting the rest of the theory. To make it more specific, I
have chosen to assume that the growth of synaptic knobs, with
or without neurobiotaxis, is the basis of the change of facilita-
tion from one cell on another, and this is not altogether im-
plausible. It has been demonstrated by Arvanitaki (1942) that
a contiguity alone will permit the excitation aroused in one cell
to be transmitted to another. There are also earlier experi-
ments, reviewed by Arvanitaki, with the same implication.
Even more important, perhaps, is Erlanger's (1939) demonstra-
tion of impulse transmission across an artificial "synapse," a
blocked segment of nerve more than a millimeter in extent.
Consequently, in the intact nervous system, an axon that passes
close to the dendrites or body of a second cell would be capable
of helping to fire it, when the second cell is also exposed to
other stimulation at the same point. The probability that such
closely timed coincidental excitations would occur is not con-
sidered for the moment but will be returned to. When the
coincidence does occur, and the active fiber, which is merely
close to the soma of another cell, adds to a local excitation in
it, I assume that the joint action tends to produce a thickening
of the fiberforming a synaptic knobor adds to a thickening
already present.
Lorente de N6 (1938a) has shown that the synaptic knob is
usually not a terminal structure (thus the term "end foot" or
"end button" is misleading), nor always separated by a stalk
from the axon or axon collateral. If it were, of course, some
action at a distance would be inevitably suggested, if such con-
nections are formed in learning. The knob instead is often a
rather irregular thickening in the unmyelinated part of an axon
near its ending, where it is threading its way through a thicket
of dendrites and cell bodies. The point in the axon where the
66 Growth of the Assembly
thickening occurs does not appear to be determined by the
structure of the cell of which it is a part but by something ex-
ternal to the cell and related to the presence of a second cell.
The number and size of the knobs formed by one cell in con-
tact with a second cell vary also. In the light of these facts it
is not implausible to suppose that the extent of the contact
established is a function of joint cellular activity, given propin-
quity of the two cells.
Also, if a synapse is crossed only by the action of two or more
afferent cells, the implication is that the greater the area of con-
tact the greater the likelihood that action in one cell will be
decisive in firing another.* Thus three afferent fibers with
extensive knob contact could fire a cell that otherwise might be
fired only by four or more fibers; or fired sooner with knobs
than without.
In short, it is feasible to assume that synaptic knobs develop
with neural activity and represent a lowered synaptic resist-
ance. It is implied that the knobs appear in the course of
learning, but this does not give us a means of testing the as-
sumption. There is apparently no good evidence concerning
the relative frequency of knobs in infant and adult brains, and
the assumption does not imply that there should be none in the
newborn infant. The learning referred to is learning in a very
general sense, which must certainly have begun long before
birth (see e.g., the footnote on pp. 121-2).
* One point should perhaps be made explicit. Following Lorente de No,
two afferent cells are considered to be effective at the synapse, when one
is not, only because their contacts with the efferent cell are close together
so their action summates. When both are active, they create a larger
region of local disturbance in the efferent soma. The larger the knobs in
a given cluster, therefore, the smaller the number that might activate the
cell on which they are located. On occasion, a single afferent cell must be
effective in transmission. It is worth pointing this out, also, because it
might appear to the reader otherwise that there is something mysterious
about emphasis on the necessity of activity in two or more cells to activate
the synapse. All that has really been shown is that in some circumstances
two or more afferent cells are necessary. However, this inevitably implies
that an increase in the number of afferent cells simultaneously active must
increase the reliability with which the synapse is traversed.
67 Conduction from Area 17
CONDUCTION FROM AREA 17
In order to apply this idea (of a structural reinforcement of
synaptic transmission) to visual perception, it is necessary first
to examine the known properties of conduction from the visual
cortex, area 17, to areas 18, 19, and 20. (In view of the criti-
cisms of architectonic theory by Lashley and Clark [1946], it
may be said that Brodmann's areas are referred to here as a
convenient designation of relative cortical position, without sup-
posing that the areas are necessarily functional entities or always
histologically distinctive.)
It has already been seen that there is a topological reproduc-
tion of retinal activities in area 17, but that conduction from 17
to 18 is diffuse. Von Bonin, Carol, and McCulloch (1942) have
found that a localized excitation in 17 is conducted to a large
part of 18, a band lying along the margins of 17. There is no
point-to-point correspondence of 17 and 18. Excitation from 18
is conducted back to the nearest border region of 17; to all parts
of area 18 itself; and to all parts of the contralateral 18, of area
19 (lying anterior to 18), and of area 20 (in the lower part of
the temporal lobe).
The diffusity of conduction from area 17 is illustrated by the
diagram of figure 7. Cells lying in the same part of 17 may
conduct to different points in 18. The cells in 18, thus stimu-
lated, also lead to points in 18 itself which are widely separated;
to any part of the ipsilateral areas 19 and 20; and, through one
synapse, to any part of the contralateral 19 and 20. Conversely,
cells lying in different parts of 17 or 18 may have connections
with the same point in 18 or 20.
Thus there is convergence as well as spread of excitation. The
second point illustrated by figure 7 is a selective action in 18,
depending on the convergence of fibers from 17. In the figure,
F and G are two cells in area 18 connecting the same macro-
scopic areas. F, however, is one that happens to be exposed to
excitations from both A and B (two different regions in area
17). When an area-17 excitation includes both A and B, F is
much more likely to be fired than G. The figure does not show
the short, closed, multiple chains which are found in all parts
68 Growth of the Assembly
of the cortex and whose facilitating activity would often make
it possible for a single fiber from B to fire G. But the same
sort of local bombardment would also aid in firing F; and the
cell which receives excitations from two area-17 fibers simul-
taneously would be more likely to fire than that which receives
excitation from only one.
FIGURE 7. Illustrating convergence of cells in Brodmann's area 17 upon
cells in area 18, these cells in turn leading to other areas. A, B, C, three
grossly distinct regions in area 17; D, JE, F, G, H, cells in area 18. See text.
On the other hand, when B and C (instead of A and B) are
excited simultaneously, G would be more likely to fire than F.
Any specific region of activity in area 17 would tend to excite
specific cells in area 18 which would tend not to be fired by the
excitation of another region in 17. These specific cells in 18
would be diffusely arranged, as far as we know at random.
They would be usually at some distance from one another and
would always be intermingled with others which are not fired
by the same afferent stimulation, but because of their lasting
structural connections would tend always to be selectively ex-
. cited, in the same combination, whenever the same excitation
recurs in area 17. This of course would apply also in areas 19
and 20. Since a single point in 18 fires to many points through-
69 The Cell-Assembly
out 19 and 20, excitation of any large number of area-18 cells
means that convergence in 19 and 20 must be expected. How
often it would happen is a statistical question, which will be
deferred to a later section.
The tissues made active beyond area 17, by two different
visual stimuli, would thus be (1) grossly the same, (2) histo-
logically distinct. A difference of stimulating pattern would
not mean any gross difference in the part of the brain which
mediates perception (except in the afferent structures up to and
including area 17, the visual cortex). Even a completely uni-
lateral activity, it should be noted, would have diffuse effects
throughout areas 18, 19, and 20 not only on one side of the brain
but on both. At the same time, a difference of locus or pattern
of stimulation would mean a difference in the particular cells in
these areas that are consistently or maximally fired.
MODE OF PERCEPTUAL INTEGRATION: THE
CELL-ASSEMBLY
In the last chapter it was shown that there are important
properties of perception which cannot be ascribed to events in
area 17, and that these are properties which seem particularly
dependent on learning. That "identity" is not due to what hap-
pens in 17 is strongly implied by the distortions that occur in
the projection of a retinal excitation to the cortex. When the
facts of hemianopic completion are also considered, the conclu-
sion appears inescapable. Perception must depend on other
structures besides area 17.
But we now find, at the level of area 18 and beyond, that all
topographical organization in the visual process seems to have
disappeared. All that is left is activity in an irregular arrange-
ment of cells, which are intertangled with others that have noth-
ing to do with the perception of the moment. We know of
course that perception of simple objects is unified and determi-
nate, a well-organized process. What basis can be found for an
integration of action, in cells that are anatomically so disor-
ganized?
An answer to this question is provided by the structural
change at the synapse which has been assumed to take place in
70 Growth of the Assembly
learning. The answer is not simple; perceptual integration
would not be accomplished directly, but only as a slow devel-
opment, and, for the purposes of exposition, at least, would in-
volve several distinct stages, with the first of which we shall
now be concerned.
The general idea is an old one, that any two cells or systems
of cells that are repeatedly active at the same time will tend to
become "associated," so that activity in one facilitates activity in
the other. The details of speculation that follow are intended
to show how this old idea might be put to work again, with the
equally old idea of a lowered synaptic "resistance," under the
eye of a different neurophysiology from that which engendered
them. (It is perhaps worth while to note that the two ideas
have most often been combined only in the special case in
which one cell is associated with another, of a higher level or
order in transmission, which it fires; what I am proposing is a
possible basis of association of two afferent fibers of the same
orderin principle, a sensori-sensory association,* in addition to
the linear association of conditioning theory.)
The proposal is most simply illustrated by cells A, B, and C
in figure 8. A and B, visual-area cells, are simultaneously ac-
tive. The cell A synapses, of course, with a large number of
cells in 18, and C is supposed to be one that happens to lead
back into 17. Cells such as C would be those that produce the
local wedge-shaped area of firing in 17 when a point in 18 is
strychninized (von Bonin, Carol, and McCulloch, 1942). The
cells in the region of 17 to which C leads are being fired by the
same massive sensory excitation that fires A, and C would almost
necessarily make contact with some cell B that also fires into 18,
or communicate with B at one step removed, through a short-
axon circuit. With repetition of the same massive excitation in
17 the same firing relations would recur and, according to the
assumption made, growth changes would take place at synapses
AC and CB. This means that A and B, both afferent neurons
* It should be observed, however, that some theorists have continued to
maintain that "S-S" (sensori-sensory) associations are formed in the learn-
ing process, and have provided experimental evidence that seems to estab-
lish the fact. See, e.g., Brogden, J. Exp. Psychol, 1947, 37, 527-539, and
earlier papers cited therein.
71 The Cell-Assembly
of the same order, would no longer act independently of each
other.
At the same time, in the conditions of stimulation that are
diagrammed in figure 8, A would also be likely to synapse (di-
rectly, or via a short closed link) with a cell D which leads
back into an unexcited part of 17, and there synapses with still
FIGURE 8. Cells A and B lie in a region of area 17 (shown by hatching)
which is massively excited by an afferent stimulation. C is a cell in area
18 which leads back into 17. E is in area 17 but lies outside the region of
activity. See text.
FIGURE 9. A, B, and C are cells in area 18 which are excited by converging
fibers (not shown) leading from a specific pattern of activity in area 17.
D, E, and X are, among the many cells with which A, B, and C have con-
nections, ones which would contribute to an integration of their activity.
See text.
another cell E of the same order as A and B. The synapse DE,
however, would be unlikely to be traversed, since it is not like
CB exposed to concentrated afferent bombardment. Upon fre-
quent repetition of the particular excitation in area 17, a func-
tional relationship of activity in A and B would increase much
more than a relationship of A to E.
The same considerations can be applied to the activity of the
enormous number of individual cells in 18, 19, and 20 that are
simultaneously aroused by an extensive activity in 17. Here, it
should be observed, the evidence of neuronography implies that
there are anatomical connections of every point with every other
point, within a few millimeters, and that there is no orderly ar-
rangement of the cells concerned.
72 Growth of the Assembly
Figure 9 diagrams three cells, A, B, and C, that are effectively
fired in 18 by a particular visual stimulation, frequently repeated
(by fixation, for example, on some point in a constant distant
environment). D, E, and X represent possible connections
which might be found between such cells, directly or with in-
tervening links. Supposing that time relations in the firing of
these cells make it possible, activity in A would contribute to
the firing of E, and that in E to firing C and D. Growth changes
at the synapses AE, BC, BD, and so on, would be a beginning
of integration and would increase the probability of coordinated
activity in each pair of neurons.
The fundamental meaning of the assumption of growth at the
synapse is in the effect this would have on the timing of action
by the efferent cell. The increased area of contact means that
firing by the efferent cell is more likely to follow the lead of the
afferent cell. A fiber of order n thus gains increased control
over a fiber n + 1, making the firing of n -f- 1 more predictable
or determinate. The control cannot be absolute, but "optional"
(Lorente de No, 1939), and depends also on other events in the
system. In the present case, however, the massive excitation in
17 would tend to establish constant conditions throughout the
system during the brief period of a single visual fixation; and
the postulated synaptic changes would also increase the degree
of this constancy. A would acquire an increasing control of
E, and E, with each repetition of the visual stimulus, would fire
more consistently at the same time that B is firing (B, it will be re-
called, is directly controlled by the area-17 action). Synaptic
changes EB would therefore result. Similarly, B acquires an
increasing control of D; and whenever a cell such as D happens
to be one that connects again with B, through X, a closed cycle
(BDXB) is setup.
It is, however, misleading to put emphasis on the coincidences
necessary for the occurrence of such a simple closed circuit.
Instead of a ring or hoop, the best analogy to the sort of struc-
ture which would be set up or "assembled" is a closed solid
cage-work, or three-dimensional lattice, with no regular struc-
ture, and with connections possible from any one intersection to
any other. Let me say explicitly, again, that the specificity of
such an assembly of cells in 18 or 20, to a particular excitation
73 The Cell-Assembly
in 17, depends on convergences. Whenever two cells, directly
or indirectly controlled by that excitation, converge on another
cell (as and X converge on B in figure 9) the essential con-
dition of the present schematizing is fulfilled; the two converg-
ing cells need not have any simple anatomical or physiological
relation to one another, and physi-
ological integration would not be
supposed to consist of independent
closed chains.
This has an important conse-
quence. Lorente de No (1938&)
has put stress on the fact that ac-
tivity in a short closed circuit must
be rapidly extinguished, and could
hardly persist as long as a hun-
dredth of a second. It is hard, on
the other hand, to see how a long,
many-linked chain, capable of
FIGURE 10. Arrows represent a
longer reverberation, would get
simple "assembly" of neural
established as a functional unit.
pathways or open multiple
chains firing according to the
But look now at figure 10, which
numbers on each (the pathway
diagrams a different sort of possi-
"1, 4" fires first and fourth, and
bility. Arrows represent not neu-
so on), illustrating the possibil-
rons, but multiple pathways, of
ity of an "alternating" rever-
beration which would not ex-
whatever complexity is necessary
tinguish as readily as that in a
so that each arrow stands for a
simple closed circuit.
functional unit. These units fire in
the order 1, 2, 3, 15. The pathway labeled (1, 4) is the first
to fire, and also the fourth; (2, 14) fires second and fourteenth;
and so on. The activity 1-2-3-4 is in a relatively simple closed
circuit. At this point the next unit (2, 14) may be refractory,
which would effectively extinguish reverberation in that simple
circuit. But at this point, also, another pathway (5, 9) may be
excitable and permit activity in the larger system to continue in
some way as that suggested by the numbers in the figure. The
sort of irregular three-dimensional net which might be the ana-
tomical basis of perceptual integration in the association areas
would be infinitely more complex than anything one could show
with a diagram and would provide a large number of the multi-
74 Growth of the Assembly
pie parallel (or alternate) units which are suggested by figure
10. If so, an indefinite reverberation in the structure might be
possible, so long as the background activity in other cells in the
same gross region remained the same. It would not of course
remain the same for long, especially with changes of visual fixa-
tion; but such considerations make it possible to conceive of
"alternating" reverberation which might frequently last for
periods of time as great as half a second or a second.
(What I have in mind, in emphasizing half a second or so as
the duration of a reverberatory activity, is the observed dura-
tion of a single content in perception [Pillsbury, 1913; Boring,
1933]. Attention wanders, and the best estimate one can make
of the duration of a single "conscious content" is of this time-
order. )
This then is the cell-assembly. Some of its characteristics
have been defined only by implication, and these are to be de-
veloped elsewhere, particularly in the remainder of this chapter,
in the following chapter, and in Chapter 8 (see pp. 195-7). The
assembly is thought of as a system inherently involving some
equipotentiality, in the presence of alternate pathways each
having the same function, so that brain damage might remove
some pathways without preventing the system from functioning,
particularly if the system has been long established, with well-
developed synaptic knobs which decrease the number of fibers
that must be active at once to traverse a synapse.
STATIST/GAL CONSIDERATIONS
It must have appeared to the reader who examined figures
8 and 9 carefully that there was something unlikely about its
being arranged at the Creation to have such neat connections
exactly where they were most needed for my hypothesis of per-
ceptual integration. The answer of course is statistical: the
neurons diagrammed were those which happen to have such
connections, and, given a large enough population of connecting
fibers distributed at random, the improbable connection must
become quite frequent, in absolute numbers. The next task is
to assess the statistical element in these calculations, and show
that probability is not stretched too far.
Statistical Considerations 75
The diagrams and discussion of the preceding section require
the frequent existence of two kinds of coincidence: (1) syn-
chronization of firing in two or more converging axons, and (2)
the anatomical fact of convergence in fibers which are, so far
as we know, arranged at random. The necessity of these coin-
cidences sets a limit to postulating functional connections ad lib.
as the basis of integration. But this is not really a difficulty,
since the psychological evidence (as we shall see) also implies
that there are limits to perceptual integration.
Consider first the enormous frequency and complexity of the
actual neural connections that have been demonstrated histo-
logically and physiologically. One is apt to think of the neural
cell as having perhaps two or three or half a dozen connections
with other cells, and as leading from one minute point in the
central nervous system to one other minute point. This impres-
sion is far from the truth and no doubt is due to the difficulty
of representing the true state of affairs in a printed drawing.
Forbes (1939) mentions for example an estimate of 1300
synaptic knobs on a single anterior horn cell. Lorente de No's
drawings (1943, figures 71-73, 75) show a complexity, in the
ramification of axon and dendrite, that simply has no relation
whatever to diagrams (such as mine) showing a cell with one
or two connections. The gross extent of the volume of cortex
infiltrated by the collaterals of the axon of a single neuron is
measured in millimeters, not in microns; it certainly is not a
single point, microscopic in size. In area 18, the strychnine
method demonstrates that each tiny area of cortex has connec-
tions with the whole region. (These areas are about as small
as 1 sq. mm., according to McCulloch, 1944&.) It puts no great
strain on probabilities to suppose that there would be, in area
18, some anatomical connection of any one cell, excited by a
particular visual stimulation, with a number of others excited in
the same way.
There is, therefore, the anatomical basis of a great number of
convergences among the multitude of cortical cells directly or
indirectly excited by any massive retinal activity. This is to be
kept in mind as one approaches the physiological question of
synchronization in the converging fibers. In the tridimensional,
lattice-like assembly of cells that I have supposed to be the basis
76 Growth of the Assembly
of perceptual integration, those interconnecting neurons which
synapse with the same cell would be functionally in parallel.
Figure 10 illustrates this. The pathways labeled (1, 4), (8),
and (13), converging on one synapse, must have the same func-
tion in the system; or the two-link pathway (5, 9)-(6, 10) the
same function as the single link (2, 14). When impulses in one
such path are not effective, those in another, arriving at a dif-
ferent time, could be.
Once more, the oversimplification of such diagrams is highly
misleading. At each synapse there must be a considerable dis-
persion in the time of arrival of impulses, and in each individual
fiber a constant variation of responsiveness; and one could never
predicate a determinate pattern of action in any small segment
of the system. In the larger system, however, a statistical con-
stancy might be quite predictable.
It is not necessary, and not possible, to define the cell-
assembly underlying a perception as being made up of neurons
all of which are active when the proper visual stimulation oc-
curs. One can suppose that there would always be activity in
some of the group of elements which are in functional parallel
(they are not of course geometrically parallel). When for
example excitation can be conducted to a particular point in the
system from five different directions, the activity characteristic
of the system as a whole might be maintained by excitation in
any three of the five pathways, and no one fiber would have to
be synchronized with any other one fiber.
There would still be some necessity of synchronization, and
this has another aspect. In the integration which has been
hypothesized, depending on the development of synaptic knobs
and an increasing probability of control by afferent over efferent
fibers, there would necessarily be a gradual change of the fre-
quency characteristics of the system. The consequence would
be a sort of fractionation and recruitment, and some change in
the neurons making up the system. That is, some units, capable
at first of synchronizing with others in the system, would no
longer be able to do so and would drop out: "fractionation."
Others, at first incompatible, would be recruited. With per-
ceptual development there would thus be a slow growth in the
assembly, understanding by "growth" not necessarily an increase
77 Statistical Considerations
in the number of constituent cells, but a change. How great
the change would be there is no way of telling, but it is a
change that may have importance for psychological problems
when some of the phenomena of association are considered.
This then is the statistical approach to the problem. It is
directly implied that an "association" of two cells in the same
region, or of two systems of cells, would vary, in the probability
of its occurrence, over a wide range. If one chose such pairs
at random one would find some between which no association
was possible, some in which association was promptly and easily
established when the two were simultaneously active, and a
large proportion making up a gradation from one of these ex-
tremes to the other. The larger the system with a determinate
general pattern of action, the more readily an association could
be formed with another system. On a statistical basis, the more
points at which a chance anatomical convergence could occur,
the greater the frequency of effective interfacilitation between
the two assemblies.
Psychologically, these ideas mean (1) that there is a pro-
longed period of integration of the individual perception, apart
from associating the perception with anything else; (2) that an
association between two perceptions is likely to be possible only
after each one has independently been organized, or integrated;
(3) that, even between two integrated perceptions, there may
be a considerable variation in the ease with which association
can occur. Finally, (4) the apparent necessity of supposing
that there would be a "growth," or fractionation and recruit-
ment, in the cell-assembly underlying perception means that
there might be significant differences in the properties of per-
ception at different stages of integration. One cannot guess
how great the changes of growth would be; but it is conceiv-
able, even probable, that if one knew where to look for the evi-
dence one would find marked differences of identity in the per-
ceptions of child and adult.
The psychological implications of my schematizing, as far as
it has gone, have been made explicit in order to show briefly
that they are not contrary to fact. We are not used to thinking
of a simple perception as slowly and painfully learned, as the
present chapter would suggest; but it has already been seen, in
78 Growth of the Assembly
the discussion of the vision of the congenitally blind after opera-
tion, that it actually is. The slowness of learning, and the fre-
quent instances of total failure to learn at all in periods as
great as a year following operation (Senden, 1932), are extraor-
dinary and incredible (if it were not for the full confirmation by
Riesen, 1947). The principles of learning to be found in psy-
chological textbooks are derived from the behavior of the half-
grown or adult animal. Our ideas as to the readiness with
which association is set up apply to the behavior of the devel-
oped organism, as Boring (1946) has noted; there is no evidence
whatever to show that a similarly prompt association of separate
perceptions can occur at birththat it is independent of a slow
process in which the perceptions to be associated must first be
integrated.
As to the wide range in difficulty of associating two ideas or
perceptions, even for the adult, this is psychologically a matter
of common experience. Who has not had trouble remembering,
in spite of repeated efforts, the spelling or pronunciation of some
word, or the name of some acquaintance? The fact of the un-
equal difficulty of associations is not stressed in the literature,
probably because it does not fit into conditioned-reflex theory;
but it is a fact. My speculations concerning the nature of the
trace and the aboriginal development of perception thus are
not obviously opposed to the psychological evidence. Further
evaluation can be postponed until the speculations have been
fully developed.
O. Perception of a Complex:
The Phase Sequence
The reader will remember that what we are aiming at here is
the solution of a psychological problem. To get psychological
theory out of a difficult impasse, one must find a way of recon-
ciling three things without recourse to animism: perceptual
generalization, the stability of memory, and the instabilities of
attention. As neurophysiology, this and the preceding chapter
go beyond the bounds of useful speculation. They make too
many steps of inference without experimental check. As psy-
chology, they are part of a preparation for experiment, a search
for order in a body of phenomena about which our ideas are
confused and contradictory; and the psychological evidence does
provide some check on the inferences made here.
Although this discussion is not anchored with sufficient neu-
rophysiological evidence, it still has a compass. It is guided
throughout by the necessity of conforming to the concrete evi-
dence of behavior, and any virtue it may have derives from
coordinating the two sources of information. The details of
these two chapters, speculative as they are, have to be given
for two reasons.
First, competent psychologists have asserted that field or con-
figuration theory is the only conceivable solution of the prob-
lems of behavior. The argument is stronger when a physiolo-
gist such as Sherrington (1941), without benefit of field theory,
can find no possible way of reducing "mind"which must be the
control of behaviorto neural action. The argument is even
strengthened by the vagueness of Hull's treatment of percep-
tion: "afferent neural interaction" and generalization gradients
have the smell of brimstone about them and, when they are
79
80 Perception of a Complex: The Phase Sequence
made specific, in neurological terms, may turn out to be only a
new form of field theory (just as Pavlov's waves of inhibition
and excitation are field theory). Improbable as Kohler's theory
of electrical fields may be, hard as it may be to reconcile Lash-
ley's theory of interference patterns with the anatomical and
clinical evidence, these writers still have a powerful argument
for holding psychology to what I believe is a blind alley, so long
as it remains true that an alternative explanation cannot be
elaborated in enough detail to be conceivably an account of real
processes.
So one must speculate. But there is a second reason for the
form of the discussion in these two chapters. In speculating,
one can at least be specific enough so that, when further ana-
tomical and physiological information is made available to the
psychologist, the errors of earlier speculation such as this will
be apparent at once, and the necessary changes clearly indi-
cated.
In this chapter, accordingly, the new foundation for associa-
tion theory is to be completed. The preceding chapter dealt
only with the effects of a single visual stimulus, with a constant
retinal projection. We can next ask how the several effects of
a variable stimulation may be combined, or associated.
LINES AND ANGLES IN PERCEPTION
It has already appeared that the corners of a rectilinear figure
are of special importance in first vision. The first crude per-
ception of an object in the visual field as distinctive, as having
identity, seems to be related to the sharp inflection of contours
and the direction of line. Senden describes the congenitally
blind patient, after operation, as frequently not being able to
distinguish (for example) square from circle; but when the dis-
tinction is made, it is through a search for corners, apparently
as foci in an otherwise amorphous mass of light. There is also
evidence of an immediate distinction between narrow strips of
light, when one is vertical and one horizontal in the visual field.
Here again the perception of identity in such objects is defec-
tive, since the patient is extremely slow to learn names for the
81 Lines and Angles in Perception
two visual patterns, although he can name horizontal and ver-
tical at once by touch.
From Lashley's experiments, it is also evident that line and
angle dominate the rat's perception of patterns. We may there-
fore consider that these things are among the elements from
which more complex perceptions develop. "Perceptual element"
is meant to contrast sharply with the punctate sensory elements
of classical association theory, on the one hand; and, on the
other, the term does not imply that the perception of line and
angle is fully innate. There is a prolonged learning period be-
fore these "elements" are promptly and clearly perceived, either
by rat or man.
The evidence on this point concerning man is to be found in
Senden's (1932) compilation and has been cited earlier; some
evidence follows which indicates that the rat also has to learn
to see lines distinctively.
Eighteen rats reared in complete darkness (Hebb, 1937a)
were trained, following Lashley's procedure exactly, to discrimi-
nate horizontal from vertical lines. The only comparable data
for normal animals are from Lashley (1938fo), who gives a mean
of 21 trials for normal learning of this discrimination. The mean
number of trials for the rats reared in darkness was 129, six
times as great, with a range from 40 to 190. These animals had
been trained first to jump to an open window, and then to dis-
criminate black and white cards (as were Lashley's animals)
before being trained with the horizontal and vertical striations.
In subsequent tests they behaved like normal animals, showing
that the slow original learning with striations was not due to
structural defects.
It is to be observed, in this experiment, that the range of
learning scores overlapped the normal range. Some animals
reared in darkness learned as quickly as some normally reared,
Also, in preliminary experiments, two animals showed a prompt
but not completely consistent discrimination of horizontal and
vertical line. These facts fit with Senden's, for human subjects,
very closely; they indicate that both rat and man have a crude
immediate perception of a difference between the horizontal and
vertical, but that the identity of these perceptions is radically
improved with experience.
82 Perception of a Complex: The Phase Sequence
It may not be necessary to seek far for a physiological under-
standing of the primitive significance of line and the sharp in-
flection of line. In their stimulating statistical approach to the
problem of visual acuity, Marshall and Talbot (1942) have dis-
cussed a mechanism of summation at the borders of the projec-
tion of a visual excitation in area 17. In the absence of other
influence, the heightened activity corresponding to the margin
of a figure might dominate the total process. Walker and
Weaver (1940) have shown a direct control of eyemovement by
peripheral stimulation of the visual cortex. The movement is in
such a direction that the light source which produced such a
peripheral activity would be fixated. In view of Walker and
Weaver's work, Marshall and Talbot's discussion suggests that
there would be a tendency to fixate, successively, various parts
of the contour of an object. Also, with the intersection of lines
there may be a further summation of the border effects, so that
the foci of greatest activity aroused by a pattern in the visual
field would correspond to its corners.
It is important to observe that the figure made up of straight
lines, instead of irregular curves, has special physiological prop-
erties. When one point of a straight line is fixated, every point
to one side of the fixation tends to arouse exactly the same di-
rection of eyemovement, and every point in the line on the other
side exactly the opposite direction. At times the two vectors
may balance, but often they will not. It follows that there is
a strong tendency for the eye to make a sweep along the line,
in one direction or the other; and, as the sweep is made, at the
moment when a corner is reached the stimulation of the inter-
secting line is at the maximum, since at this moment every point
in the second line has the same vector, for a new direction of
eyemovement. It appears from these considerations that the eye
would tend to seek out the contours of a figure and follow them
irregularly, and with reversals, it is true, and subject to disturb-
ance by other events, but it seems that such a tendency must
exist.
It is also important to note that every movement of the eye
in a single direction, for whatever reason, stimulates a "ridge"
of activity in area 17 corresponding to the projection of a straight
line, wherever the visual field contains a point of light. Except
83 Lines and Angles in Perception
in the special case of a field consisting of uniform parallel stria-
tions, every change of fixation from one point to another means
that rows of cells in area 17 are excited together or left unexcited
together. From the assumptions of the preceding chapter, the
first visual learning would to a great extent be the integration
of the effects, in the peristriate region, of the action of such rows
of cells in the striate region. This is a further emphasis on the
importance of straight lines in perception, and, as I have argued
(in Chapter 3), makes it possible to assume, from such a con-
stant and extended training, that the characteristics of adult
tachistoscopic vision may be accounted for by a learning process.
Lines and angles, then, can be treated as perceptual elements,
not fully innate in perception, but partly so, and likely to be
learned before more complex patterns are. A very incomplete
analogy will serve here as a mnemonic device, to help keep in
mind the different processes that enter into perceptual learning.
If line and angle are the bricks from which form perceptions
are built, the primitive unity of the figure might be regarded as
mortar, and eyemovement as the hand of the builder. The
analogy is poor because the bricks contain mortar, they grow
while they are being used, and the house may change beyond
recognition as it is being built. Nevertheless, two things must
be kept in mind. First, the primitive figure-ground relationship
stressed by the Gestalt psychologists remains of fundamental
importance even though one recognizes that it is not the be-all
and end-all of perception. Secondly, one may agree with Lash-
ley and the Gestalt psychologists that motor activity in itself
cannot possibly explain the organization of perception and never-
theless recognize that it has an essential role. Bricks alone are
not enough for building, nor yet is mortar.
A triangle then is a complex entity in perception, not primi-
tive. As a whole, it becomes distinctive and recognizable only
after a prolonged learning period in which there is a good deal
of receptor adjustmenthead-and-eye movementas the psycho-
logical evidence of Chapter 2 showed. From the point of view
we have now arrived at, the difficulty of explaining the experi-
mental facts of pattern equivalence has been greatly reduced.
Similar triangles of unequal size, a solid and an outline triangle,
even a solid plain triangle and one circumscribed with a circle
84 Perception of a Complex: The Phase Sequence
(Lashley, 1938&) contain a number of identical elementslines
and angles with the same orientation. (For some readers, "iden-
tical elements" will be an invitation to battle: let me say again
that these are not the sensory elements of the classical dispute.)
It is of course still necessary to show how these elements, the
parts of the figure, are integrated in perception. This I shall try
to do next, taking an example that is as simple as possible. The
present schema of perception and perceptual learning deals only
with the case in which changes of visual fixation and some loco-
motion occur freely. The problem is to show how the variable
stimulation which results from such movements can have a
single effect, the perception of a single, determinate pattern.
MOVEMENTS DURING PERCEPTUAL INTEGRATION
In the perception of a triangle, there are three nodal points in
the pattern on which fixation would be made repeatedly. A
single fixation point was assumed in discussing perceptual inte-
gration in the last chapter. Let us next try to elaborate the
schema to provide explicitly for movement of the eyes, and for
those movements of the whole animal that produce variations in
the size of the retinal projection of the stimulating pattern.
The account of neural integration now needs to be elaborated
in three respects: (1) The activity in the peristriate, temporal,
and other association areas of the cortex, that is aroused by
fixation on any one of the three corners of the triangle ABC
(figure 11), is always accompanied by motor processes which
control the changes of fixation to the other corners. (2) Because
of these changes, the integration of activity in area 18 and be-
yond occurs without constant conditions of excitation in area 17:
when for example B is fixated, which cells will be excited in 18
is determined in part by the preceding activity in 18, and this
varies according to whether the preceding fixation was on A or
on C. (3) Finally, each of the three retinal patterns produced
by fixation on the corners of the triangle is itself variable, with
different distances of the animal from the stimulus object.
To show as simply as possible what these three points involve,
suppose now that an animal sees a triangle repeatedly, having
had no previous visual experience and having no other stimula-
85 Movements During Integration
tion at the same time. Disregard any changes of background
activity in the brain, apart from those induced by the changes
of fixation mentioned; and, as a final simplification, treat the
three angles A, B, and C of figure 11 as the only perceptual ele-
ments involved: that is, disregard fixation on the sides of the
triangle, and the perception of the lines as distinctive entities.
The three changes to be made
in the schema now can be dealt
with one by one.
1. First, the concomitant mo-
tor excitation: with fixation on
A, as represented in figure 11,
B and C falling outside the mac-
ular field, the peripheral stimu-
lation of B and C excites two
motor responses simultaneously.
The excitation may of course be
subliminal, particularly at the
FIGURE 11. The triangle ABC is
seen with fixation on the point A.
first moment of fixation on A,
The macular field is represented by
since a fixation often endures for
the hatched circle, so that the
an appreciable time. The rela- points B and C fall in peripheral
vision. Arrows represent the direc-
tive strengths of the motor com-
tion and strength of eyemovement
ponents of the excitation aroused
tendencies aroused by the stimula-
by B and C can be supposed to
tion from B and C.
be the same, statistically, as
shown by the vectors (the length of the arrows) in the figure.
At any one moment, variations of excitability would make one
dominant over the other, but one can suppose that on the aver-
age they are equal. Looking at A would be followed by looking
at B about as often as C.
In the normal adult animal, at least, the control of the direc-
tion and extent of eyemovement is accurate within quite narrow
limits and does not work through a series of gross approxima-
tions. This means that the neural activity leading to the motor
centers, before an eyemovement begins, is determinate, and
specific to the locus of the peripheral stimulus. Fixation on A,
then, has two effects: one "central" (i.e., the arousal of specific
cells in the peristriate and temporal regions, without a direct
86 Perception of a Complex: The Phase Sequence
motor significance); and one primarily motor (i.e., the arousal
of specific fibers in a pathway that is not known but which
originates in the visual cortex and which presumably involves
the frontal eyefields [Clark and Lashley, 1947]).
These two effects, central and motor, correspond roughly to
a dichotomy of the total excitation in area 17. The central
effects are mainly determined by the part of the excitation fall-
ing in the macular zone, where acuity is greatest and the motor
components least; the motor effects, by the peripheral part of the
excitation, where acuity is poor, the ratio of cortical cells to de-
grees of visual angle least, and the motor limen lowest (Walker
and Weaver, 1940).
If the cells in area 18 and beyond which are aroused by fixa-
tion on A are designated by the letter a, the composition of a
is determined by the whole triangle, but the angle A has a dis-
proportionately great influence because it falls in the field of
central vision. Millimeter for millimeter, the part of the triangle
which falls on the macula excites more cortical cells. Thus the
main determinant of the central effect of fixating A is the angle
A itself, rather than B or C. We shall also find that this influ-
ence of A in organizing the neural structure a must be further
increased if the triangle is seen at a variable distance; but it is
important to remember that a is defined originally as made up
of the cells in the association areas of the cortex that are acti-
vated, not by A alone, but by the whole triangle when fixation is
on A.
In the same notation, b and c are aroused by fixation on B
and C. The distinction of a from b is a distinction of the exci-
tations aroused with two retinal projections of the same figure.
In brief, the angle A is the main determinant of the structure a,
but not the only one; while B and C, when A is fixated, are the
main determinants of the subsequent motor response.
As the integration of a, b, and c proceeds, repeated fixation on
the three corners of the triangle slowly establishes three diffuse,
irregular cell-assemblies (Chapter 4), each of which is capable
of acting momentarily as a closed system. As this goes on, the
activity of each structure would at all stages of its development
be accompanied by two specific motor excitations. Activity in a
occurs in the constant presence of a liminal or subliminal exci-
87 Movements During Integration
tation tending to move the eyes to the right (fixating B, figure
11) or upwards and to the right (fixating C). From the assump-
tions of the preceding chapter, this means some mutual facilita-
tion between a and the motor activity in question. Psycho-
logically, in the particular limited conditions of the schema, per-
ception of the triangle with fixation at one corner would involve
a set to look at one of the other two corners (facilitation by a
of the motor responses, strengthening the motor facilitation from
B and C); and a sensorily aroused change of fixation from A
to B would produce an expectancy of seeing the triangle with
fixation at B (facilitation from the motor activity, arousing or
reinforcing activity in b).*
2. The second point (on which the schema as it appeared in
* That is, when the animal looks at B after looking at A, the assembly b
is excited in two ways: centrally, by the facilitation from a and from the
motor activity; and sensorily, by B itself. In these circumstances, the two
facilitations would just about coincide in time. "Expectancy" implies that
the central facilitation definitely precedes the sensory, so it would be better
to speak here of attention (central reinforcement of a sensory process).
But the same process is called expectancy when the sensory reinforcement
is delayed.
The later chapters will consider the interaction of sensory and central
facilitations, and the disturbing effect on behavior when they conflict or
when an expectancy fails to get sensory reinforcement. In a thoroughly
familiar environment, the two facilitations would be synchronized, mainly
because of the motor links in the central "phase sequence" discussed in the
following pages. The facilitation of a is equally on b and on c; so a might
arouse either one. But the cortical motor activity that results in looking
from A to B is specific to b; and the animal will not "expect" to see the
point C.
None of these relationships can be so precise outside of this schema, but
the principles can hold. Central and sensory processes will stay more or
less in synchrony in a familiar environment, both because of the motor links
in the central sequence, and because the facilitation of a perception X will
not be specific to another particular perception Y unless X was invariably
followed by Y in past experiencethe sight of an unsupported object, and
the sound of a crash. But a sequence of visual sensations which is not
organized by one's own eyemovements can be very disturbing to behavior,
as the occasional case of seasickness at the movies shows. Again, the sight
of a large body of calm water is not disturbing in a strange landscape,
where one's expectancies are not specific, but would be very much so if it
followed a movement of the head and eyes toward a place where one had
always seen dry land. These points are developed further in Chapter 7.
88 Perception of a Complex: The Phase Sequence
the last chapter must be modified) has to do with an interfer-
ence between the suprasensory effects of fixation on different
points in the stimulating pattern.
It was proposed that the cells which are excited in area 18, by
the first visual stimulation, are determined by chance anatomical
convergences. Those which are active when the angle A is
looked at must, on a statistical basis, include some which are
active when B or C is looked at. The number of convergences
in common would also be increased by any overlap in the retinal
projections of a figure. Consequently, when A and B, or B and
C, are looked at in close succession, there must be an influence
of the one activity, in the peristriate region, on the next. When
B is looked at, some cells in 18 will be active if A was looked
at just before, but not if C was; others inhibited, or refractory,
following a look at A. In the early stages of the visual develop-
ment, there must be an interaction between a, b, and c, which
in this schema amounts to an interference. In the ordinary
course of perceptual learning, evidently, where the eyes are first
opened on the complications of an ordinary room without the
artificial limits of the schema, the amount of interference be-
tween the first perceptual activities must be much greater.*
The interference has to be fully recognized here if the schema
is to have any theoretical value later.
Such an interference, it seems, can occur only in the first
stages of perceptual development and would disappear later.
It is important to recognize it, however, since it affects the
course that integration takes. In terms of the notation used
above, the organization of the structure a (excited by fixation on
A) must proceed mainly in cell groups (effective units of trans-
mission) which are reliably excited whenever fixation is on A,
whether B or C, or neither, had been fixated just before. I have
* And the resultant generalization of perception will be greater also. In
this schematizing, the conditions of visual learning have been simplified to
an absurd degree, but the ideas that emerge from it can be applied to be-
havior in more natural circumstances. The initial interference between the
three primary visual habits epitomizes a relationship between a much larger
number of habits in real perceptual development. The larger number of
habits means that each one will be more an abstraction, less a perception
of a particular angle (or line) in a particular setting.
89 Movements During Integration
emphasized that there must be fractionation and recruitment, or
growth, as the individual closed cell-assembly becomes physio-
logically organized. When the growth occurs in a variable back-
ground of visual stimulation, those cells which are affected by
the variability would not be consistently active at each repeti-
tion of the specific stimulus which excites the assembly, and
therefore would not be part of it. In other words, as growth
proceeds in a, b, and c, fractionation would eliminate elements
that are active only when a certain fixation has preceded, and
those which are inactive following any particular fixation. Ac-
tivity in a can therefore occur independently of b or c, once the
integration of these assemblies is complete.
In general terms, what has just been discussed is the case in
which sensory events recur in random order in a short time
interval, as one must suppose visual fixations to do. The alterna-
tive is a fixed sequence of sensory events, which of course also
happens in normal perceptual development. The difference be-
tween the two cases is instructive.
Using the same notation, let us consider for a moment what
would be the result of looking at A, B, and C always in the same
order. (A is now a "signal" for B, and B for C, but this is not
an analogue of the conditioned reflex: in the hypothetical con-
ditions of the schema, B occurs only following A; in the classical
conditioning procedure, B is an event such as feeding which
follows A, the conditioned stimulus, but which also occurs at
other times.) With a fixed order of stimulation, activity in the
assembly b occurs always in the same setting, the after-effects
of activity in a. The structure b therefore would include
transmission units whose activity is especially facilitated by a,
whereas, when stimulation occurs at random, b must be made
up of units that are independent of the preceding activity.
Also, when activity in a, b, and c occurs always in that order,
no reverberation in a can be expected to coexist and be inde-
pendent of activity in b, but must fuse with it. Any reverbera-
tory after-effects which are not extinguished by the changed
sensory influx are constantly present during the physiological
integration of b and could not be distinct from it.
When fixation is at random among A, B, and C, however, we
have seen that fractionation would tend from the first to keep
90 Perception of a Complex: The Phase Sequence
the structures a, b, and c distinct, and make activity in b pos-
sible without preceding facilitation from a. This leaves the pos-
sibility that the three assemblies, once integrated, could still
develop the interfacilitation that is discussed below, while the
three are independent in the sense that one can be active apart
from another, or that two may be active at the same time.
3. The third point, on which the schema as it was developed
in the last chapter is to be elaborated, concerns the effect of a
variable distance of the animal from the stimulus object. The
result must be an increased importance of the part of the pat-
tern which falls in the field of central vision.
When an animal looks at any part of a rectilinear pattern
while moving toward it, the excitation aroused by the part
looked at, falling in central vision, remains constant. The rest
of the excitation, in the peripheral field, varies steadily. If the
point of fixation is, say, the midpoint of one of the sides of the
figure, the locus of the retinal projection of that line does not
change, except to be extended at each end. If an angle of a
triangle is fixated, as in figure 11, the projection of two lines
and their intersection in the macula is unchanged as the animal
moves closer to the figure; the third line, and the two remaining
intersections, move farther and farther toward the periphery of
the visual field.
The rat in the Lashley jumping stand, learning to discriminate
a square from a triangle, does not see the figures at a constant
distance. The retinal angular extent might vary by 50 per cent.
The variability is very much greater in a Yerkes box, where the
animal runs toward, or may back away from, a door on which
the stimulus figure is placed. Now, using the same notation, in
the composition of the assembly a (the organized structure in
the association areas of the brain which is aroused by looking
at the point A in figure 11), the influence of the angle A, rela-
tive to B and C, becomes even greater. The two lines and their
intersection, fixated at a variable distance, have a constant effect;
the other two intersections have a variable effect and do not
repeatedly excite the same cells in area 18. In the growth of
a, b, and c, fractionation would eliminate most of the cells
whose excitation is due to the extramacular part of the figure.
This excitation would of course always coexist with the activity
91 The Spatial Hypothesis
of the organized assemblies, but it would remain unorganized.
The organized activity would have to be determined by the
macular excitation, which, with appropriate fixation, is constant
despite a variable angular size of the stimulus object. This con-
clusion is clearly significant in understanding stimulus equiva-
lence, as demonstrated by a rat's choice of a large triangle after
learning to choose a small one.
THE SPATIAL HYPOTHESIS AND PERCEPTION OF THE
REMOTE ENVIRONMENT
My discussion is already getting pretty far from the actual
known facts of neural function. If it is not to become fantasy,
it must at least stick close, at every point, to the facts of be-
havior. Here I digress for a moment to consider a group of
facts that are well known but have not been explained. They
serve as a check on the present theorizing.
The preceding discussion implies that a variable stimulation
from the same stimulus object would retard the rate at which
the subject could learn to perceive it de novo. Some of the dis-
cussion has suggested that differences of size, in terms of retinal
angle, might not greatly affect learning; but differences of pat-
tern, as an irregular object is seen from different directions,
would be more important. Each grossly different pattern of
stimulation, as the object is seen from one side or another, re-
quires the establishment of a separate set of cell-assemblies.
I can anticipate the discussion of the following section, and add
that when this happens the various sets of assemblies would
gradually acquire an interfacilitationif sight of the object from
one angle is often followed by sight of it from another. Arous-
ing one set would then mean arousing the others, and essentially
the same total activity would be aroused in each case.
Each perception would thus involve a conceptual activity (an
activity, i.e., not directly controlled by sensory processes).
There is plenty of evidence in children's drawings, and in adult
errors in perspective drawing, to show that a person looking at
an object thinks he sees more of it than he does. What he
knows about the object appears in his drawing, as well as what
is visible at the moment; and the significant fact is that neither
92 Perception of a Complex: The Phase Sequence
child nor adult can usually say where his drawing departs from
what is actually presented to the sense organ.
An object seen from various aspects, then, requires a more
complex learning process for its recognition than one seen always
from the same direction and the same distance. Such learning
should take longer; and, if we assume that a smaller number of
cortical cells means a greater limitation to the number of assem-
blies that can be set up as separately functioning systems, learn-
ing to recognize a variably stimulating object should be harder
for lower animals than recognizing a constant object.
Now it happens that the visual activity of lower species is
dominated by the perception of place. This turns out experi-
mentally to mean a dominance of cues from remote objects in-
stead of near ones; and remote objects provide the most stable
and constant stimulation of the animal's environment.
In discrimination training, the "position habit," which Kre-
chevsky (1932) has called a "spatial hypothesis," is a constant
nuisance in animal experiments. The method of training is to
offer a choice of two alleys, or doors or windows, one of which
leads the animal to food. He is also given a sign as to which
one contains the food: the correct door has a black card on it,
or a circle, say; the wrong one has a white card or a square.
Food, and the sign of food, are sometimes on the right, some-
times on the left. But what the rat, dog, or chimpanzee persist-
ently tries to find out is something different. He wants to know
whether food is always on the right, or always on the left; or
perhaps, in desperation, whether it alternates, once right, once
left, and so on. He wants nothing of the rarefied intellectual
problem of the signs the experimenter has put on the doors;
it is only after repeated discouragement of the position habit
the attempt to find the food in some one placethat one can get
him to learn anything else.
It has sometimes been supposed that this is a visual lack, or
that the vision of lower animals is dominated by other senses.
But this interpretation is not right, for two reasons. First, the
position habit is just as annoying when one is trying to get blind
animals to discriminate tactual cues (Smith, 1939) or auditory
ones (Pennington, 1938). Secondly, the perception of place
itself is visual, as much as kinesthetic or tactual. When visual
93 The Spatial Hypothesis
cues from the distant environment are available, they dominate
behavior. So the difference between lower and higher species is
that the lower species is more dominated by the least variable
of the environmental stimulation, not that vision in general has
less influence.
This can be demonstrated strikingly in the rat. The rat is first
accustomed to feeding on a bare table. He is then taught to
run across the table, from a fixed starting point, to a food dish
that is always in the same place on the opposite side of the
table. The table is then rotated through 90, but nothing else
is changed: the rat is dropped at the same place on the table,
with the food still opposite him. He runs at least once to the
side of the table where the food used to be, with respect to the
room, although neither food nor food dish is there now.
If one changes the method slightly, the dominance of room
cues as against table cues becomes clearer. Put four small
shields on the table, one in the middle of each edge. Mark the
one that contains food by painting it white and making it larger
than the others. Train the rat first to run to that one only, by
putting no food in the others. Then rotate the table through
90, and put food in all four shields. The rat will choose, not
the plainly marked shield he was taught to go to, but the un-
marked shield which, after rotation, now occupies the place in
the room that the food used to occupy (Hebb, 1938a, 1938&).
A still more striking demonstration has been mentioned by
Lashley (1938&) and further investigated by several students of
Queen's University (Hebb and Williams, 1941, and further un-
published experiments). Teach a rat to jump from a small
platform to another one near by. The second platform is just
large enough for him to land on safely, and holds food. After
he has made ten jumps, move the second platform through 90.
The rat hesitates, shows disturbance, but finally jumpsinto
space, in the former direction of food.
A final demonstration that visual cues are controlling the re-
sponse can be made in a small cabinet, 6 by 6 feet square and
6 feet high, in a building quiet enough to provide no auditory
cues to direction. Each wall is identical with the others, and
each contains a door. The ceiling is homogeneous. One door
is opened only, and a thin curtain is hung over it to admit light
94 Perception of a Complex: The Phase Sequence
without allowing the animal to see outside. Now the rat is
again taught to jump from one platform to another, in this
cabinet. When he has learned to jump, the door that was
opened is closed, and one in the next wall, at a 90 angle, is
opened, and screened as before. Nothing is changed but the
more distant visual cues of the door; the nearby platform re-
mains as it was; but the rat now jumps out into space, at a 90
angle from his goal.
Why? And why is the position habit so persistent? The an-
swer seems to be that the animal mainly perceives, and responds
to, the least-variable objects in his environment, which are the
ones at the greatest distance.
The stimulation received from any object varies with the ani-
mal's movements, but there are important differences that de-
pend on the distance of that object. When the animal turns
round, excitation is changed equally from near and far objects;
but not when he moves from one place to another. Changes of
position affect retinal locus, extent, and intensity only slightly
when the stimulus is remote, very greatly when it is near. Even
with body rotation, the order in which nearby objects are seen
has no constancy, unless the animal always turns at precisely
the same point. But the order in which distant objects are seen
is the same, no matter where the animal is in the experimental
apparatus.
It seems therefore to be true that, the more constant the
stimulation from an object is, the more readily it will be iden-
tified and responded to, as the schema of these chapters requires.
This is relevant also to the fact that as one goes up the phylo-
genetic scale one finds an increasing tendency toward visual dis-
crimination of objects, without special training. The rat shows
no sign of distinguishing between persons, or between inert
objects visually presented; the dog often distinguishes a number
of objects by vision, and probably a few persons; the chim-
panzee is very selective in his behavior toward a great many
objects, and is clearly able to distinguish a large number of
persons (whom he sees often) from strangers.
The fact therefore is that perception of the intrinsic visual
properties of a near or movable object is a less primitive and
more difficult feat than perception of its place. This too is rele-
95 Superordinate Perceptions
vant to the fact that spatial delayed response is obtained easily
in monkey and chimpanzee, nonspatial only with the greatest
difficulty (Riesen and Nissen, 1942).
My account in these pages evidently has a long way to go
before the generalized human concept of a triangle, or any other
figure or object, is provided for; but it is essential to remember
that the rat's perception is also a long way removed from
man's. He does not recognize the triangle he was trained to
recognize if it is rotated by 60, nor a black triangle on a white
ground after learning a white one on a black ground. He has,
in fact, some considerable difficulty in seeing the difference be-
tween a triangle and a square when their base-lines are iden-
tical. Normal man can recognize simple regular figures at a
single glance, but it is very unlikely that a rat does. When
some change in the training figure is made, such as making it
larger, the rat's behavior is clearly disturbed, while man might
be quite unaware of the change; and the rat often discriminates
only part of as simple a figure as a triangle.
None of these differences between rat and man is accounted
for in current theories of perception. The critic of the account
of perception that is offered here is likely to find it distasteful
because it implies that with limited visual experience perception
would be little generalized. Clinical and experimental evidence,
however, does not support the objection but shows that the
characteristic human perception of a triangle is the product of
a long experience. Thus the schema agrees with fact on this
point and in addition has the advantage of suggesting, at least,
why the perception of place and direction is so prominent in
animal behavior.
THE DEVELOPMENT OF SUPERORDINATE
PERCEPTIONS
We can now turn to the question of an integration of the sev-
eral parts of a figure into a distinctive whole, as contrasted with
the amorphous whole that is perceived in first vision.
The most direct way of accounting for the superordinate inte-
gration is as follows, as long as this is still recognized as frankly
schematic. Activity in the assembly a, aroused by fixation on
96 Perception of a Complex: The Phase Sequence
an angle A of a triangle, can occur independently of b or c.
When A, B, and C are looked at successively, in any order, but
in a short period of time, activity may continue by reverberation
in two of the structures while the third is sensorily aroused.
Just such a series of fixations would be the result of the behavior
described by Senden, as the congenitally blind patient after
operation learns to count the corners of a square or triangle and
becomes quicker and quicker at doing so before he learns to
recognize the figure at one glance. In these circumstances,
conceivably, there is a frequent occurrence of activity in the
three assemblies a, b, and c at the same time. These lie inter-
laced with each other in what is grossly the same tissue of the
cerebrum, and according to the assumptions of the last chapter
the simultaneous activity would result in an integration of the
three systems.
It is perhaps necessary to remind the reader that the three
systems do lie in the same tissue, although two of the sensory
excitations concerned are unilateral, one in one striate area ex-
clusively, one in the other. All such suprasensory systems must
develop in parallel, in both hemispheres. The purely sensory
activity, up to and including area 17, is unilateral when the
stimulating diagram falls wholly to one side of the fixation point.
But when area 18 is strychninized, it fires into the contralateral
18 as well as into the ipsilateral 19 and 20 (von Bonin, Carol,
and McCulloch, 1942). The suprasensory integration initiated
by a unilateral sensory event must be bilateral and consist of
two halves, in the two hemispheres. Each half has the same
functional significancethat is, it mediates the same perception
and facilitates the same responseseven apart from a coordinat-
ing action of the corpus callosum and other cerebral commis-
sures (though the development of the contralateral half depends
on the commissures). This is evidently relevant to the clinical
and experimental reports of slight effects from damage to the
callosum (Bridgman and Smith, 1945), or failure to find an effect
of unilateral extirpation of association areas; but it is important
for the present as showing that the assemblies a (excited by a
figure which falls wholly in the right homonymous visual field)
and b (arising from the left visual field) are structures which
must develop in the same gross tissues of the brain.
97 Superordinate Perceptions
According to the assumptions made earlier, simultaneous ac-
tivity in a, b, and c would establish facilitation between them,
through their chance anatomical interconnections and the en-
largement of synaptic knobs. An effective facilitation from one
system on another means a change of frequency characteristics
in the system receiving the facilitation. It therefore means some
fractionation and recruitment in the constituent units (p. 76).
With three extensive systems involved, each facilitating action
in the other two, these growth changes must be considerable.
The resulting superordinate system must be essentially a new
one, by no means a sum or hooking together of a, b, and c. In-
stead of abc, which might suggest such an idea, a better nota-
tion for the new structure is t: the assembly of cells whose ac-
tivity, in the schema, is perception of the triangle as a distinc-
tive whole. As Gestalt writers would say, this is something other
than the sum of its parts; but, unlike Gestalt theory, the schema
derives the distinctiveness of the whole from perception of the
parts.
Now for the defects in such a formulation. Supposing that
the general idea is right, it is still unlikely that the synthesis of
t from a, b, and c would be made as a single step, however
gradual. A single step would depend on simultaneous activity
in three systems and so requires a frequent fixation of the three
corners of the triangle in quick succession. Looking at two of
the corners only would not contribute to the integration.
A much more plausible idea is that one or more intermediate
stages would occur, such as an integration of a with b before
that of "ab" with c (ab is used for brevity, though it is mislead-
ing in the same way as abc in the preceding paragraph). This
would call for simultaneous activity in only two systems at a
time. I mention ab as the first stage, rather than ac, because
the horizontal line (AB, in figure 11) seems of fundamental
importance in human perception, and certainly is so for the rat
in the usual conditions of testing. The perception of lines as
distinctive entities has been disregarded in this schema, in order
to avoid a cumbrous and unwieldy discussion; but as we have
seen the perception of lines is primitive, as the perception of
angles or corners is, and a triangle has six instead of the three
perceptual elements dealt with by the schema. This would
95 Perception of a Complex: The Phase Sequence
make probable the occurrence of several intermediate stages
between the perception of "elements" only, and perception of a
distinctive total figure.
The argument up to this point can be summarized in general
terms. Reasons have been given for believing (1) that fixation
on each of the several parts of a figure would have an increas-
ingly determinate effect, as arousing one specific structure; (2)
that these structures, each corresponding to a frequently made
fixation, are anatomically diffuse and interlaced with one another
in the same gross cerebral tissue; and (3) that the several ac-
tivities may coexist, and be aroused in any order. It is a reason-
able inference (4) that two of these determinate actions simul-
taneously would have a determinate effect, tending to excite spe-
cific transmission units, and that the action of these units would
tend to organize in the same way that the earlier established
systems were organized. Activity in a superordinate structure
(in this case, t] is then best defined as being whatever deter-
minate, organized activity results from repeated activity in the
earlier-developed or subordinate structures giving rise to it (in
this case, a, b, and c; or "ab" and c, assuming two steps in de-
velopment, and that a and b are first integrated).
THE PHASE SEQUENCE IN PERCEPTION
Next, let us consider the temporal relationship of activity in
these various structures. During the development of the assem-
blies a, b, and c, arousal of a as we have seen is accompanied
by two motor activities. Of these, one always becomes liminal
(producing a change of fixation) before b or c is sensorily
aroused. The sequence of events can be schematized as
a-b-c-b-a-c-a-b-a-
and so on. Each of these events is associated with two specific
motor excitations. One of them at least is subliminal, and one
becomes liminal as an event intervening between a and b, for
example, or between c and a.
This "ideational" series with its motor elements I propose to
call a "phase sequence."
When the assembly t has become organized, the psychological
The Phase Sequence in Perception 99
evidence indicates that its activity intervenes between the ac-
tivities of the subordinate assemblies a, b, and c and does not
supersede them. The sequence now becomes something like
this:
a-b-t-a-c-t-c-t-b-
Such complication of a simple perception has important con-
sequences for theorizing. The reader is briefly reminded of the
reasons for thinking that perception of a simple pattern is not a
single lasting state, terminated by an external event, but a se-
quence of states or processes. The congenitally blind patient
after operation at first sees any figure as an amorphous mass, but
may be able with effort to count its corners; the perception is
then alternately of the whole and of its parts. As the figure
becomes a distinctive whole, there is still the same fluctuation
of the figure-ground relationshipattention directed now to the
whole, now to its parts. This is a phenomenon which as a matter
of common observation is always present in perception (and in
a "concept," as one thinks about an object), as Chapter 2
showed, although the fact is not recognized in current discus-
sions of the figure-ground relationship. Exactly the same sort
of thing is implied by Lashley's inference that the rat succes-
sively isolates (i.e., sees as figure) various parts of a unified pat-
tern before making a response.
In terms of the schema, the alternate perception of whole and
parts is an alternation of activity among a, b, c, and t, with cor-
responding directions of fixation (except for the entity t, which
is accompanied by no determinate eyemovement; since the
average values of the six eyemovement vectors associated with
the three part perceptions of the triangle add up to zero, but
also fluctuate from moment to moment, their resultant would
fluctuate in direction and amount, and would produce neither
a fixation of gaze nor any predictable change of fixation).
It follows that the integration of t, the basis of perceiving a
distinctive total figure, essentially involves a sequence of cor-
tical events with motor components. Activity in a facilitates the
arousal of both b and c, with the appropriate intervening eye-
movement, and activity in b or c facilitates the arousal of a in
the same way. Whether b or c is aroused following a would
100 Perception of a Complex: The Phase Sequence
depend on the momentary conditions of excitability. Activity
in a would also facilitate that in t. In the early stages of per-
ceptual development, t might be excited only after repeated ac-
tivations of a, b, and c, but later (with the extensive develop-
ment of synaptic knobs in the system and the consequent in-
crease in the strength of facilitation) might be aroused follow-
ing sensory activation of a alone, so that the triangle would be
recognized with a single glance,at A (figure 11). But the ac-
tivity so aroused must be transient, as we have seen; perception
of the whole as such is momentary, and alternates with percep-
tion of the various parts. Instead of an indefinitely prolonged
reverberation, interrupted only by some event outside the system,
excitation in one of the assemblies a, b, c, and t is an unstable
equilibrium which moves readily into another phase.
The schema that has been developed requires only that re-
verberation continue in one of these structures long enough so
that temporal overlap can occur. The psychological evidence
reinforces the idea, based on the physiological evidence, that
reverberation is short-lived: if the duration of an idea, or a per-
ception, is the duration of reverberatory activity in a closed
system, one can say that the pattern of activity rarely lasts
without change for as long as a second. The stability of a per-
ception is not in a single persistent pattern of cerebral activity
but in the tendency of the phases of an irregular cycle to recur
at short intervals.
It will be proposed in the following chapter that the train of
thought is also a "phase sequence" of the same kind, but more
extended, consisting of a series of phase cycles. The present dis-
cussion, besides dealing with perception, is also meant to lay a
groundwork for dealing with the temporally organized processes
of thinking.
THEORETICAL PROPERTIES OF THE SCHEMA
With this we are done schematizing. It remains to ask what
theoretical significance the schema has, and how its ideas are to
be applied to the development of behavior in normal circum-
stances.
Actually, all the rest of this monograph is devoted to answer-
Theoretical Properties of the Schema 101
ing the questions, but it will be worth while first to strike a
trial balance and see what has been accomplished already.
The main conceptions used in the following chapters have now
been developed. The relationship of the schema to the follow-
ing chapters will be more easily kept in mind by making its
psychological reference more explicit.
Within limits imposed by the needs of exposition, a concep-
tual system has been elaborated which relates the individual
nerve cell to psychological phenomena. A bridge has been
thrown across the great gap between the details of neuro-
physiology and the molar conceptions of psychology. The
bridge is definitely shaky in the middle, but it is well buttressed
at each end; and we have a psychological bridgehead which
can be widened and which already includes some strategic
points. In other words, the schema has some theoretical value
already. It shows, more or less explicitly, how it is possible:
(1) to conceive of a conjoined action of the primitive figure-
ground mechanism, eyemovement, and learning (specifically de-
fined synaptic changes), in the development of simple percep-
tions; (2) to provide for an action of set, attention, or expect-
ancy, also defined physiologically, in the perceptual process; and
(3) to provide at the same time for Gestalt completion; simi-
larity; generalization; and abstractionthese, with attention also,
being essentially different aspects of the same process and closely
related to association itself.
1. The interrelationships of eyemovement, figure-ground seg-
regation, and learning are explicit in the schema. Perception
depends on learning first to see the parts of an object clearly, a
process involving a series of visual fixations, and proceeds from
seeing, at first, an amorphous mass containing several foci (the
corners), to seeing a distinctive figure at a glance. Even at this
final stage we know that perception of the whole is dependent
on eyemovements for maximal clarity (Chapter 2). According
to the schema, the perception is constituted by a temporal se-
quence of activity in suprasensory (or association-area) struc-
tures, which owe their organization to changes at the synapse:
it is an irregular cycle of recurring events which can continue
momentarily without the corresponding sensory stimulations, but
102 Perception of a Complex: The Phase Sequence
which is reinforced by them and by the appropriate eye-
movements.
2. References to set, attention, and expectancy were made in
developing the schema, which can be more precise here. The
term "attention" is ambiguous in the literature, and has several
meanings: it may refer (a) to the state or end result of attend-
ingthe subjective clarity of what is attended to, or the neces-
sary receptor adjustment; (b) to the selectivity of the process;
(c) to the hypothetical agency or process which produces the
selectivity; and (d} to various properties of "mind" which ap-
parently cannot be defined or understood. It is in the third
sense (c) that the term is used here, and in the schema atten-
tion may be defined as a central facilitation of a perceptual ac-
tivity. So used, "attention" has exactly the same meaning as a
"perceptual set," a process which makes one thing seen more
readily than another. When this facilitation is effective in ad-
vance of the corresponding sensory process, expectancy is said
to occur.
In the notation of this chapter, activity in the cell-assembly
a facilitates an arousal of the assemblies b and c, as well as
facilitating two motor responses, to fixate B or C (points of the
triangle ABC, figure 11). In one way or another, this facilita-
tion has inherent in it the notions of association (the whole
schema makes this explicit), attention, and expectancy. Activity
in b may be aroused in two ways: sensorily, by looking at B, or
centrally, by the "association of ideas." When B is looked at
just after A, activity in b is aroused in both ways, and the cen-
tral facilitation, from a, is an instance of attention: a central
reinforcement of a particular sensory event.
This illustration of the physiological meaning of attention is
complete logically but is not particularly effective, since it does
not make the selectivity of the process explicit. Our hypo-
thetical animal has not been allowed to see a variety of patterns
as yet, among which selectivity could operate; but this will be
clearer in a moment, in discussing abstraction. Similarly, an
eyemovement from A to B may be aroused in two ways: by the
sensory stimulation from B, or by the facilitation from a on the
motor system. Here the selectivity of the central facilitation is
more evident. Eyemovements may be made from A to B or
Theoretical Properties of the Schema 103
from A to C, and which of these is made evidently depends
often on central events. We have seen that the relative strength
of the two motor facilitations from a fluctuates from moment to
moment; when one becomes liminalthat is, becomes decisive in
determining which eyemovement is madethe selectivity of the
"motor set" is illustrated.
3. Suppose now that our hypothetical animal, after thorough
habituation to his limited environment ("limited" puts it mildly,
since he was permitted no experience except to see a single tri-
angle, at varying distances but always with the same angle of
regard)suppose that the animal is now tested with the two
patterns represented in figures 12 and 13. Let him first see
figure 12, a triangle such as he is used to, but with apex missing.
What neural processes will result, according to the schema?
Fixation near the top of the figure will have no organized
effect (remember that the schema disregarded perception of
lines as distinctive entities, discussing the intersection of lines
only). If the animal looks at A and B, however, the assemblies
a and b are excited, with facilitation on each other, on c, and
on t. A momentary excitation in t, a perception of the whole
triangle, is then possible. Here is an instance of Gestalt com-
pletion, but derived as an associative process, with no field forces
operating. According to the schema, it could happen only with
a simple and thoroughly familiar figure (or thoroughly familiar
part of a complex and unfamiliar figure), which agrees with the
104 Perception of a Complex: The Phase Sequence
experimental facts. It is evident that such a completion be-
comes a simple corollary of association, once we can provide for
a perception of parts without denying the unity of the whole.
So too are similarity, abstraction, and generalization. Given
perception of the various parts or properties of a pattern, sepa-
rately; and the possibility of an association of these perceptions
with a perception of the whole: we have given at once the
possibility of a single response to two patterns which differ in
their total properties but which have a property or properties
in common. This of course is not new. What is new in the
schema is showing how, conceivably, perception of part and of
whole can be related to each other and to the nerve cell, and
to changes at the synapse. An immediate dividend is the pos-
sibility of clearly stating, in physiological terms, the meaning of
words like abstraction or similarity which are necessary to de-
scribe behavior but which have had, to say the least, a touch
of mystery about them.
When the animal whose perceptual processes have been
schematized looks at the point C in figure 13, at the apex of the
familiar triangle, the intersection of the two straight lines pro-
duces an organized neural activity, plus an unorganized one due
to the unfamiliar presence of the circumscribing circle. In the
notation of the schema, c is aroused, but also a new and random
activity of cells in area 18 and beyond. To interpret the effect
of the random firing in 18 we must fall back on the statistical
ideas of the last chapter. The number of convergences of this
new activity on the organized cell-structure c would vary from
animal to animal, since it is a chance matter; and in any one
animal, from moment to moment, with variations of excitability
in the individual unit of transmission. Accordingly, the organ-
ized activity in the assembly c would sometimes be disrupted,
sometimes not. So with activities of a and b, when the other
corners of the triangle are fixated.
When the organized activity does occur, the triangle is "rec-
ognized"the similarity of the figure to the familiar plain tri-
angle (figure 11) is perceived, the animal has abstracted from
the total complex, and generalization of perception has occurred.
Here too the selectivity of attention is more evident: once c is
aroused, a fixation immediately afterward on A will be more
Theoretical Properties of the Schema 105
likely to arouse the organized activity of athe sensory effect
of the triangle is selectively reinforced by the facilitation from
c, that of the circle is not.
The purpose of this schema has been to arrive at conceptions
that will be as near physiological fact as possible and that will
enable us to deal with the cerebral control of behavior and its
temporal integration. From this point onward, it is assumed
that some such structure as the cell-assembly is established as
the result of repeating any particular firing combination in an
afferent system, and that assemblies interact in some such way
as that proposed in the schema.
What follows is an attempt to generalize the ideas that have
been developed in these two chapters (4 and 5), rather than an
attempt at rigorous explanation by means of the assumptions
that have already been made. Two implications of those as-
sumptions, however, may be made explicit, since they broaden
the applicability of the schematizing to real problems:
1. The different properties of a single sensory event may have
separate central effects. This can be conceived of as follows:
The infant's hand making contact with two or three objects in
succession may receive from them stimulations which differ ex-
cept in one respect, such as the degree of pressure on the skin
as the hand closes. In the schematizing, there was only one
stimulating object in the environment; in the environment of an
actual infant, stimulation involves a number of objects and a
cumulative effect of stimulation would be established most
promptly by the property that several of these objects have in
common. Such a consideration makes abstraction fundamental
in the first learning; thus, in the context of one group of stimu-
lating objects, a metal bar on the crib might contribute only to
the development of the perception of hardness; but, while con-
tact is being made intermittently with another set of objects,
the same bar might contribute to the perception of (tactual)
roundness.
2. Another aspect of sensory processes can be taken account
of, in the temporal sequence of stimulating conditions. The
problem of the perception of black in the visual field, for
example, comes from the fact that this is not a static perception
determined only by an absence of light in part of the visual
106 Perception of a Complex: The Phase Sequence
field, but is transient and requires temporal as well as spatial
contrast for the full effect. A frequently repeated condition in
the optic system is a decline of activity up to and including the
lateral geniculate. During such a decline, and determined by
it, there may be a predictable and selective activity in areas 17
and 18, by cells which are not active in the same combination
during equilibration or during an increase of intensity. The re-
peated activity of such a particular group of cells would develop
an assembly whose activity would be the perception of black.
In general, then, whenever a change of stimulation produces
a transient but selective activity (that is, in certain cells only),
in the sensory projection areas or in the neighboring association
area, the necessary conditions are provided for the formation of
a cell-assembly. Such assemblies would be the basis for per-
ceiving contrasts, and the relative properties of stimulation: of
colder, larger, higher in pitch, and so on. By the very condi-
tions determining their occurrence, such perceptions must be
transient and brief but may nonetheless be definite and distinc-
tive for the period of their duration.
6.
Development
of the Learning Capacity
In the schema of perception, motor learning got a foot inside
the door through the emphasis on eyemovements. These move-
ments were treated as having a variable relationship to sensa-
tion. They may be elicited (1) directly, by stimulation of the
peripheral retina; (2) less directly, by a foveal stimulus that
arouses an assembly action, this in turn facilitating eyemove-
ment; or (3) still less directly, through a phase sequence, or suc-
cession of assembly actions. The stimulus may arouse assembly
a, a then arousing c, at which point the motor limen may be
low enough so that the activity of c results in overt eyemove-
ment.
This epitomizes the treatment to be made of motor learning.
The behavioral evidence shows a considerable variation in the
directness of sensori-motor control. Some responses have, at
maturity, all the properties of a reflex and yet are known to be
learned. Others remain quite unpredictable from a knowledge
of the stimulus alone; they are, that is, determined by an inter-
play of sensory stimulation and the autonomous activity of the
cerebrum (set).
We must remember both kinds of learning: the set-influenced
and the non-set-influenced. The reaction against early switch-
board theory and connections, and the current dogma that
learning occurs only with special conditions of motivation, have
both tended to draw attention away from a kind of learning
that, once established, is little affected by set and does not
seem to need reinforcement.
In the adult animal the eyeblink to a rapidly approaching
object, for example, is practically in the class of a spinal reflex.
107
108 Development of the Learning Capacity
Unlike the learned response studied in the usual experimental
approach, it is extremely resistant to extinction, and to distrac-
tion ("external inhibition"); and, in human subjects, very hard
to change by verbal preparation. Yet Riesen (1947) has shown
that it is in fact a learned response. It is absent at the twen-
tieth month in the chimpanzee reared in darkness, but then
appears slowly with visual experience; and it is found at about
three months in the normally reared animal. Such learning
must be thought of as depending essentially on sensori-motor
connections. The connections are presumably not from specific
receptors to specific effectors; but they must be rather direct
between afferent and efferent systems, since the response is with-
out important influence from the autonomous central process
(set, attention, or the like).*
The responses that have this property, however, appear to be
acquired early. They are by no means a paradigm of learning
in general.
At another extreme are those learned acts of the half-grown
or full-grown animal that we call both learned and "voluntary"
(a term that will be defined later). They are completely sub-
ject to extinction, and no matter how thoroughly learned they
remain completely a function of set and drive. There is no pos-
sibility of direct sensori-motor connections here; to understand
such behavior, and such learning, we shall first have to under-
stand the organization of the controlling cortical activity.
It is implied that the nature of the learning process changes
significantly with development; and the application of the ideas
of the schema of perception to the real behavior observed in an
experiment will best be seen in considering first the changes of
the learning capacity with growth.
* Which means that the learned eyeblink may be about as close as mam-
malian learning ever gets to setting up S-R (stimulus-response) connec-
tions. The experimental criterion of such a response is dual: its independ-
ence from set and attention, and its resistance to extinction in Pavlov's
sense. This comes down essentially to a single theoretical criterion: the
predictability of the response to a particular stimulation, in any circum-
stances, as long as no other physically incompatible response is aroused at
the same time. This is what sensori-motor connections imply. But when
a response is predictable only in the experimental apparatus, or following
verbal preparation, we are dealing with another class of behavior.
109 Early and Later Learning
It was possible to be specific about a hypothetical perceptual
learning as it is not possible about the motor learning of the
mature animal. Perception is firmly tacked down to known
stimulating conditions; its antecedents can be discovered. Motor
learning at maturity, however, is conditioned by all the un-
knowns of cortical action. But with the schema as a starting
point, and considering some properties of behavior that are often
overlooked, it will be possible to set up a tentative account of
learning at maturity.
THE RELATION OF EARLY TO LATER LEARNING
In this section I shall bring together the behavioral evidence
on the relationship between learning in infancy and that of the
normal adult animal, before returning in the next section to the
question of neural mechanisms.
It is proposed that the characteristics of learning undergo an
important change as the animal grows, particularly in the higher
mammals; that all learning tends to utilize and build on any
earlier learning, instead of replacing it (Mowrer, 1941), so that
much early learning tends to be permanent (Tinbergen, 1942;
Hunt, 1941); and, finally, that the learning of the mature animal
owes its efficiency to the slow and inefficient learning that has
gone before, but may also be limited and canalized by it.
THE GENERAL PROPOSITION. It is of course a truism that learn-
ing is often influenced by earlier learning. Innumerable experi-
ments have shown such a "transfer of training." Learning A
may be speeded up, hindered, or qualitatively changed by hav-
ing learned B before. The question for debate is how great the
effect may be in general behavioral development (as distinct
from the effect of some one specific habit on some other) and
what theoretical use is to be made of it.
McGeoch,* for example, has said:
After small amounts of learning early in the life of the individual,,
every instance of learning is a function of the already learned organi-
zation of the subject; that is, all learning is influenced by transfer.
* The psychology of human learning, copyright Longmans, Green & Co.>
Inc., 1942, pp. 445-446. By permission of the publishers.
110 Development of the Learning Capacity
The learning of complex, abstract, meaningful materials and the
solution of problems by means of ideas (reasoning) are to a great
extent functions of transfer. Where the subject "sees into" the funda-
mental relations of a problem or has insight, transfer seems to be a
major contributing condition. It is, likewise, a basic factor in origi-
nality, the original and creative person having, among other things,
unusual sensitivity to the applicability of the already known to new
problem situations. Perceiving, at whatever level, is probably never
free of its influence; and there is no complex psychological event
which is not a function of it.
Those are strong words, and I propose that they must be
taken literallyas presumably they were meant to be taken.
Unless we are to regard them as just a lip service to logic and
the known facts of behavior, they must influence general psy-
chological theory profoundly. If the learning we know and can
study, in the mature animal, is heavily loaded with transfer
effects, what are the properties of the original learning from
which those effects came? How can it be possible even to con-
sider making a theory of learning in general from the data of
maturity only? There must be a serious risk that what seems to
be learning is really half transfer. We cannot assume that we
know what learning transfers and what does not: for our knowl-
edge of the extent of transfer is also derived from behavior at
maturity, and the transfer from infant experiences may be much
greater and more generalized.
An example, in itself important for the theory of learning, will
also show the dangers of generalizing from adult to infant be-
havior with regard to transfer. A student once pointed out to
me that James' famous experiment on memorization begged the
question. James wanted to see if practice in memorization
would increase the ability to memorize. He found it did not;
and later writers have found the same thing. As a result, it has
been concluded that practice per se has little or no transfer
value, without instruction as to better methods of learning. But
all this is done with adults who have had long practice already,
and the student pointed out that the transfer effects must have
been complete before the experiments began, and could not be
demonstrated by the method used. James (1910) used highly
educated adult subjects, Woodrow (1927) college sophomores.
111 Early and Later Learning
What would have been the result with subjects who had done
no memorization before? We do not know, but it is certainly
quite illogical to conclude that undirected practice has no trans-
fer value because we find no evidence of it where there is no
reason for expecting it. Above all, it would be illogical to con-
clude from this sort of evidence that the incidental learning of
infancy has a negligible effect on later learning.
It has already been emphasized that perception is affected by
past experience (Gibson, 1929; Carmichael, Hogan, and Walter,
1932; Leeper, 1935; Zangwill, 1937; Krechevsky, 1938). What is
learned is in terms of what is perceived: what is not perceived
can hardly be remembered. Koffka (1935) has emphasized that
patterns may be seen and remembered by the arousal of "older
trace systems"; Woodworth (1938) says that all perceiving is
"schema with correction," that is, in terms of earlier perceptual
habits. How do these habits get established in the first place?
What are the properties of the learning that sets up the "older
trace systems," of learning in its first stages, before there are any
earlier habits to help it along? These questions cannot be com-
pletely answered at present, but even the skimpy evidence we
have is enough to reorient the whole problem of learning.
THE FIRST LEARNING OF PRIMATES IS EXTREMELY SLOW, AND
VERY DIFFERENT FROM THAT AT MATURITY. There are two kinds
of learning. One is that of the newborn infant, or the visual
learning of the adult reared in darkness or with congenital
cataract; the other that of the normal adult. I have repeatedly
cited the behavior of the patient born blind and given his sight
after motor (and speech) development was well along, to show
that the first learning is extremely inefficient as far as detectable
effects on behavior are concerned, despite the completion of
physical maturation (Senden, 1932). Here it is referred to
once more, partly to show that the early inefficiency is not due
to poor motivation and partly to make a comparison of man
with other species.
Senden reports a serious disturbance of motivation, apparently
in all cases, at one stage of learning. But this cannot be the
main cause of the slowness of learning, for two reasons.
1. Motivation is not disturbed in the first stage, immediately
upon beginning to use the eyes. At first there is a period of
112 Development of the Learning Capacity
delight, particularly in colors, and apparently a complete pre-
occupation with the new experience.
Before long, the patient finds out how hard it is to get an
effective use of pattern vision. The "crisis" of motivation then
ensues. Until that point there is interest and application; things
that are easy to learn are learned and not forgotten; color names
are readily remembered, but it takes a long apprenticeship be-
fore any useful or demonstrable learning occurs in pattern
vision.
Learning is evidently going on in this period, as long as the pa-
tient continues to keep his eyes open and makes any effort, but
it can hardly be demonstrated except in the later increase of
efficiency.
2. The second reason for denying that poor motivation ex-
plains the poor learning of man, in his first visual experience,
comes from the observations of Riesen (1947). His chimpan-
zees, reared in darkness, were certainly motivated: both by
hunger, and by their strong drive to find and cling tightly to an
attendant, when they were out of their living cages. Yet there
was no sign that either hunger or the desire to cling had taught
them, in 40 to 50 hours' visual experience, how to discriminate
the white-clad attendant from any other part of the environ-
ment. Astonishing as it was, the chimpanzees appeared to be
completely avisual at this stage of the experiment.
Moreover, in further tests, a strong electric shock failed in a
dozen trials to set up any avoidance whatever of a large, dis-
tinctive stimulus object. After a single trial, normal animals of
the same age and in similar circumstances showed violent avoid-
ance of the object with such painful properties. In the slow-
ness of their first visual learning, man and chimpanzee are in
the same class. The human slowness is not due to defects of
motivation, but points to some fundamental property of the
learning process in primates.
RELATIONSHIP OF LEARNING TO PHYLOGENESIS. The Conclusion
that the first learning differs radically from the later needs one
most important qualification: the difference depends on phylo-
genetic level.
The evidence so far has been for the higher primates only.
It was shown in an earlier chapter that training in pattern
113 Early and Later Learning
vision is slower for the rat reared in darkness than for normal
rats, but the difference is not nearly as great as for chimpanzee
or man. The rat reared in darkness is capable of a selective
visual discrimination, definitely learned, after a total visual
experience of less than 15 minutes (Hebb, 1937a, pp. 113-115).
He requires six times as many trials as the normal to learn a
discrimination of horizontal from vertical stripes, and twice as
many for erect versus inverted triangles, but within an hour or
so his behavior cannot be distinguished from that of normal
animals. As we have seen, the corresponding time for primates
is a matter of weeks or months.
There are no comparable data for other species, but some
insect behavior suggests strongly that the first learning of the
invertebrate is still quicker, and much prompter in reaching full
efficiency than the rat's. What learning ability there is seems
to appear full-blown, with little or no apprenticeship needed.
The bee for example on first emerging from a completely dark
hive flies off and is able to find the entrance to the hive again.
We know also that finding the hive depends on vision. The be-
havior indicates that the insect's learning starts out, from the
very first, at the mature level of efficiency. (Much that we
attribute to instinct, because no prolonged learning is evident,
might thus be due to learning that needs only a few seconds for
its completion. The associations that are formed may be only
certain ones to which the nervous system is especially adapted
[Tinbergen, 1942, p. 82]: heredity would still have an over-
mastering importance, but learning may nevertheless be essen-
tial to some apparently instinctive acts.)
As we go up the phylogenetic scale, then, we find in mature
animals an increasing ability to learn complex relationships, but
also, surprisingly, a slower and slower rate of learning in infancy.
This does not refer merely to the fact that higher animals
have a longer period of physical maturation. We have always
known that a rat grows up, and develops whatever capacities
the adult rat has, in three monthsor a dog in six months,
whereas a chimpanzee takes ten years, and a man twenty years.
We have thought, I suppose (if the question ever came up at
all), that this longer period of behavioral development meant
only that maturation takes longer in the primate, and that with
114 Development of the Learning Capacity
less instinct he has more to learn. But the clinical and experi-
mental evidence points to an additional factor. Given a really
new and unfamiliar set of sensations to be associated with motor
responses, selectively, the first definite and clearcut association
appears sooner in rat than in man, and apparently sooner in the
insect than in the rat.
We commonly regard quick learning as the main distinction
of higher species, and in certain conditions this is true. Normal
man can glance once at a face and remember it for years. The
chimpanzee Bimba was pricked once with a lancet and never
again would permit it to be brought near herbut with no
avoidance of other objects of the same size or roughly the same
proportions.* This is something completely outside the rat's
scope. We think of it as intelligent learning, and are prone to
regard it as an innate property of the primate brain. It cannot
be innate, however, as Riesen's evidence shows. So also with
the ability to remember faces: Miner's (1905) patient, described
as exceptionally intelligent despite her congenital cataract, two
years after operation had learned to recognize only four or five
faces and in daily conferences with two persons for a month did
not learn to recognize them by vision. The human baby takes
six months, the chimpanzee four months, before making a clear
distinction visually between friend and enemy. Evidently, this
is a period of learning as well as of maturation: not just a matter
of waiting until certain neural structures are fully grown, with
learning then at a typical adult rate.
There have been, in general, two schools of thought concern-
ing the rate of learning. The configurationists, stressing the
importance of insight, have been inclined to hold that learning
occurs as a single jump, an all-or-none affair proceeding by
discrete steps ("noncontinuity theory"); their opponents, that
* Dr. Glen Finch: diary of Bimba, Yerkes Laboratories of Primate Biol-
ogy, 1940. It is relevant here, in discussing the nature of learning in higher
species, to add a reference also to the remarkable learning capacity of
rhesus monkeys that Harlow (Psychol. Rev., 1949, 56, 51-65) has demon-
strated in a long series of experiments. Harlow's whole argument, showing
how the learning capacity may be changed out of all recognition by pro-
longed experience, is a powerful reinforcement of the position adopted in
these pages.
115 Early and Later Learning
learning is graded in amount, built up steadily by small incre-
ments, and typically independent of any special factor of insight
("continuity theory").
But it is impossible to avoid the conclusion that both types
of learning occur, and that one is characteristic of the mature
animal, the other mainly of the infant. There is insightful,
single-trial, all-or-none learningin the mature animal, but never
in the infant of a higher species. There is a slow-increment
learning in the infant, in which no trace of insight whatever can
be found, and in the mature animal also when he has been
reared in darkness and is learning to use vision. It is reasonable
to suppose in general that, the less familiar the situation or the
task to be performed, the more important slow-increment learn-
ing becomes. But it seems also that few situations can be set
up in which there is nothing familiar, so that it would be very
hard to find an instance of learning in the mature animal in
which there is not some effect of insight.
We are now in a position to define the relationship of the
learning capacity to phylogenetic level. There is no evidence
to support the idea that learning in general is faster in higher
specieseven at maturity. In the infant, the evidence is con-
clusive that the rate of the first learning is slowest in the highest
species, quite apart from slowness of maturation. The distinc-
tive characteristic of learning in higher species is the ability to
handle complex relationships, and handle them as quickly as
lower species can handle simpler ones. Man can learn to un-
fasten a latch quicker than the chimpanzee, the chimpanzee
quicker than a rat; but, if we take the learning at which each
species is most efficient, there is no good evidence that one is
faster than another.
Lashley (1929Z?) has made this point effectively. After dis-
cussing an experiment by Pechstein in which rats and human
subjects learned mazes of identical pattern, and in which the
rats showed to rather good advantagein one respect their
scores were better than those of the human subjectsLashley
goes on to point out that with simple enough habits lower spe-
cies and the feebleminded learn about as fast as normal man.
Such habits are not retarded, in rate of formation, by extensive
brain damage. There is also reason to think that immediate
116 Development of the Learning Capacity
incidental memory occurs in lower species, and Lashley con-
cludes: "The comparative study of learning in different animals
gives little evidence that evolution has brought any change in
the rate of formation of the simpler habits. On the other hand,
there is a fairly consistent rise in the limits of training and in
the formation of complex habits with ascent in the phylogenetic
scale."
In summary, then, the phylogenetic changes in the learning
capacity are as follows: (1) more complex relationships can be
learned by higher species at maturity; (2) simple relationships
are learned about as promptly by lower as by higher species;
and (3) the first learning is slower in higher than in lower
species.
CONCEPTUAL DEVELOPMENT AS THE BASIS OF LEABNING. Finally,
before turning to the question of neural mechanisms, I want to
bring together some of the behavioral evidence that throws light
on how the learning capacity changes with growth. In general,
it is a conceptual development, rather than the elaboration of
a number of specific motor responses. Perceptual organization
is also involved, but percept and concept are intimately related,
and the term "conceptual development" will do to cover both.
The best single illustration of how one set of experiences can
facilitate the formation of a new habit, involving a new stimulus
and a new response, is one that has already been given: in
learning to distinguish chimpanzee faces and remember them
(or of course the old example of the westerner's difficulty in
recognizing a particular Chinese face, before he has seen many
Chinese). Learning to name Pan, Jack, Frank, Don, and so on,
makes one later able to name a new chimpanzee, Bait, much
more quickly. The exposure to a number of individuals sets up
some sort of conceptual type, from which individual deviations
become very noticeable.
This sort of facilitation in learning seems quite general, par-
ticularly in what can be called intelligent (as distinct from
rote) learning, with meaningful material. That such a facilita-
tion affects intelligence-test scores was the only interpretation
that seemed possible of certain aspects of behavior following
brain injury, or in old age (Hebb, 19420). In a later chapter
I shall present evidence showing that the behavior of the blind
117 Early and Later Learning
rat is permanently changed by earlier visual experiences. Jack-
son (1942) and Birch (1945) have independently shown an
influence of experience on the insight of young chimpanzees.
Birch has made it clear that this does not reduce insight to rote
learning (thus Kohler's classification of behavior retains its
value), and yet insight depends on the earlier experience.
How then does the earlier experience work? In the very first
stages, to judge from Riesen's experiment and the material
brought together by Senden, it operates to establish the per-
ceptual elements discussed in Chapter 5. These are the entities
that make up more complex perceptions. Organizing such ele-
ments in the various sense modes would lay the foundation of
all later responses to the environment. Secondly, there is a
period of establishing simple associations, and with them con-
ceptual sequencesthe period in which meaning first begins to
appear. Finally, the learning characteristic of the mature animal
makes its appearance.
This later learning is essentially conceptual. Even in the rat,
maze learning requires the notion of the stimulus as acting to
arouse conceptual activities, which in turn control motor activity
(Lashley, 1944; Tryon, 1939). In man the conceptual activity
has an even greater and more obvious role. There have been
a number of attempts in psychology to treat language theoreti-
cally as a collection of conditioned reflexes, specific stimulations
directly controlling specific responses. But consider such peculi-
arities of human learning as the following.
Certain features of the development of language in small chil-
dren are very instructive. With opposites such as up-down,
back-front (of a house), in-out, or left-right, a very interesting
confusion can be observed in some children at about the age of
two to four years. (Some doubt about left and right, of course,
is often seen in adults; while other opposites, such as hot-cold,
or black-white, never seem to give any particular trouble.) The
confusion of up is only with its opposite and never with "in" or
"back." This means that the word has first acquired an asso-
ciation with a definite, limited set of conceptual coordinates.
The confusion, when it is observed, may last for months. In
this period the word up has its prompt, clearcut association
with the vertical dimension, but no association with a particular
118 Development of the Learning Capacity
motor act such as raising the hands or looking upward at sky
or ceiling. With such facts, it becomes nonsense to explain
man's conceptual development as exclusively consisting of
verbal associations.
The relationship of (a) stimulus, (b) central activity or con-
cept, and (c) motor response is clearest in the notorious diffi-
culty of choosing between left and right, to be observed by any-
one who tries to teach twelve-year-old children to "right turn"
promptly on command. Here we have a definite auditory
stimulus, the word right, and a definite motor act to be asso-
ciated with it. One might expect the association to be imme-
diate, or at the very least to be set up as a conditioned reflex
in a few trialsthe more so since the child has had some such
training since infancy. But we know that this does not happen;
the discrimination is certainly much harder to make than that
of up and down, and this in turn harder than black and white.
The child can very readily learn at the age of three that "right"
and "left" each refers to a side of the bodybut ah me, which
one?
Here, in the discrimination of right and left, is the real
paradigm of adult learning. What is set up first is a conceptual
organization. By the age of six the word "right" clearly and
immediately means sidedness to the child. A considerable con-
ceptual elaboration has already occurred, and the stimulus effec-
tively arouses that structure; but it arouses no prompt, specific
response. The response may come later, but for long it remains
vacillating and unpredictable.
Consider again the role of analogy in human thought, and
the figures of speech that betray it even in scientific work. The
pons and island and aqueduct of cerebral anatomy; the wave of
sound and cycle of sunspots in physics and meteorology; the
rise of the blood-sugar level, and the limen of stimulation in
physiologyall these are as enlightening, concerning the nature
of learning and intelligence, as the child's confusion of left and
right. Such figures of speech are at the very least an aid to
memory; even when it has a totally new reference, the familiar
term is more easily recalled than a neologism. Using it is
therefore more than an economy of language. The underlying
119 Early and Later Learning
analogy with something known already is an economy of thought
as well, an economy of effort in learning and understanding.
This fact is at the least surprising, since exactness of meaning
is essential to scientific thought, and since the laws of learning
as we know them would suggest that there must be a consider-
able interference between the old meaning and the new one
given to a familiar term. That is, there should be interference
unless terminology corresponds to thought and the new concept
is only a modification of the old.
The worker in the laboratory does not merely report and ex-
pound by the aid of analogy; that is how he thinks, also. The
atom was once a hard little round particle, or later one with
hooks on it. Recently it was a solar system. The classical dis-
pute of physics about the nature of light was really asking, Is
light like a shower of pebbles, or like ripples in a bathtub?
The ultimate answer, Both, was one that was hard to accept.
Why? Because it fitted into no pre-existing conceptions; waves
are waves, and pebbles are pebblesthere is nothing in common
experience that has the properties of both. We know well that
scientific thought travels by short steps; the individual thinker
never gets much in advance of his fellows, and his ideas are
born of existing ideas. A new conception, like Woodworth's
"schema with correction," must be mainly composed of earlier
ones. New learning is facilitated by old.
But, as the history of the theory of light suggests, it may be
limited or canalized as well. I have used the analogy of an
apprenticeship, in visual learning. Now an apprenticeship in
one trade may help in another, but not as much as the one di-
rectly studied. It is not just any learning that facilitates any
other learning at maturity.
Nissen, Machover, and Kinder (1935), for example, have
shown how the visual learning of childhood may be selective in
its later effects. West African natives made low scores on form
boards, in an intelligence test. In this sort of test the subject is
asked to fit a series of wooden blocks with simple geometrical
shapes into holes with the same shapes, as fast as he can. In
conversation, Dr. Nissen has made the point that the low scores
were not due to slowness of movement, but to a slowness in
identifying shapesa slowness of perception.
120 Development of the Learning Capacity
At the same time, he found himself just as inferior in seeing
things in the bush that seemed completely obvious to the na-
tive. Suppose now that a native and a city-dwelling scientist
were shown the trail of a new animal in the same habitat:
though it is strange to both, which would remember it better,
and be able to recognize it on a second occasion? What a na-
tive could learn from a text on geometry, or what one of us
could learn about following a trail, must be far more a function
of pre-existent learning than of the inherited properties of our
respective brains.
It is of course a commonplace that the experiences of child-
hood have a permanent effect on one's attitudes, interests, and
even abilities. Lorenz (cited by Tinbergen, 1942) has demon-
strated the effect experimentally in birds. With mammals,
everyone knows that taming is easiest in infancy, with a lasting
effect. The gun-shy dog, conversely, is an example of a lasting
failure of emotional adaptation: though the disturbance need
not be set up in infancy, it does demonstrate an intractable
effect of a first experience. Hunt (1941) has shown an effect
of early hunger on later hoarding by the rat. As to abilities or
intelligence, it is now generally accepted that races and peoples
cannot be compared in hereditary endowment, since low scores
may be due to cultural background. That the level of problem-
solving at maturity, then, may be permanently influenced by
childhood experience is an accepted psychological principle,
despite a certain inconsistency of the theorists who think that
the Negro's low IQ is to be explained so, but seem to have for-
gotten that the poor white's may be in the same class.
To this point we shall return, in discussing the nature of in-
telligence. Here I have attempted to define the general rela-
tionship of first learning to later learning, in terms of the be-
havioral evidence. Let us now turn to the physiological proc-
esses that must explain the relationship as well as the peculiari-
ties of primary or aboriginal learning.
THE STAGE OF PRIMARY LEARNING
Among others, the preceding discussion has raised the problem
of explaining the greater inefficiency of early learning in higher
121 The Stage of Primary Learning
species. To deal with it, we can return to the schema of per-
ception and repair a deliberate omission. The schema made
almost no mention of the spontaneous firing of the association
areas of the cerebral cortex. Taking this into account, a ques-
tion is raised that has really wide significance. For the moment,
it can be put in this form: In what circumstances can one sup-
pose that a stimulus will have the same central effects on two
different occasions, so that cumulative learning is possible?
It has already been said that an afferent excitation does not
arouse inactive tissue but feeds into an activity that is already
going on. A constant net result, when a stimulus is repeated,
requires some constancy in the spontaneous activity in the asso-
ciation areas upon which the afferent activity impinges. The
perceptual learning that has been schematized in the preceding
chapters depends on some consistent central action of a repeated
stimulus. We need not assume a constant pattern of the back-
ground activity, at all times, to satisfy the schema; but it does
demand at least a frequent recurrence of the same pattern. This
is necessary if the stimulus is to have a cumulative action and
build up the "assembly" of cells that constitutes the first learning.
In the answer to this problem lies an explanation of the slower
first learning in higher species and, indirectly, as we shall see
in later chapters, an understanding of some of the problems of
motivation and emotion in the half-grown or full-grown animal.
There seem to be two main factors that would make for some
consistency in the activity of the association areas at different
times. One can be referred to as an intrinsic organization, in
that activity; the other is the steadily increasing influence of the
infant's environment.
The intrinsic organization of cortical activity is so called be-
cause it is opposed to the organization imposed on the cortex
by sensory events. It appears at birth * in the large slow waves
* I.e., it is "innate"; but it may or may not be unlearned. There are two
possibilities: (1) that there are pacemakers which are inherently such as
to dominate other neural cells and produce the synchronous firing; and (2)
that the synchrony is "learned"established in utero as a result of the
neural activity itself. The comparatively slow change of the infant's EEG
toward the adult pattern, and the lack of any sharp discontinuity between
the two extremes, suggest the second interpretation. Supposing that the
122 Development of the Learning Capacity
of the infant's electroencephalogram, and in the sleep or coma
of older subjects. At birth, that is, the intrinsic organization is
completely dominant; in psychological development its domi-
nance decreases, but it continues to recur periodically, generally
in a diurnal rhythm. Its behavioral correlate is a heightened
limen for sensory stimulation and a marked absence of the per-
sistently directed activity that we call voluntary and purposeful.
This does not mean that there is any sharp line that divides
infant behavior, or sleep, from the behavior of the wakeful adult;
it does not mean, either, a hard-and-fast distinction between the
corresponding EEC's. But there appear to be two extremes in
the organization of cortical activity that correspond exactly with
the two extremes of learning ability discussed in the last sec-
tion. One is established early; interrupted rather than sup-
ported by a varied sensory activity, once the alternate organiza-
tion begins to appear; manifested by large potentials in the
EEC; and correlated with an inactivity of the musculature or an
aimless, undirected activity and a slow and inefficient learning.
firing of central cells is originally at random, the synaptic changes discussed
earlier would lead to an increased coordination of firing and the establish-
ment of massive self-exciting systems. These might be comparatively
simple closed circuits at first but would tend to steadily recruit other cells
and later other circuits. This is the "growth" referred to in the discussion
of perceptual development (p. 76). Without the constant sensory disturb-
ances resulting from the motor activity of the waking animal, the growth
would make for local integration, and hypersynchrony; with them, the in-
tegration is in anatomically diffuse systems and reduces local synchroniza-
tion.
This discussion makes a good deal of the electroencephalographic evi-
dence as the most direct index of physiological organization in the normal
living brain. It is wise to recognize explicitly, however, that the index is
a very rough one. I have treated the slow waves of the infant's EEG, and
those of adult sleep and coma, as having the same meaning. Apparently
they may have, in one respect: as indicating a local synchrony of firing in
cortical neurons (see the assumption made earlier, p. 8). But in other
respects there may be significant differences. Indeed, there must be; and
future developments in electrical recording may reveal them. For the
present, the important thing is to see that there is in one respect an iden-
tity of cortical activity in infancy, sleep, and pathological conditions which
is at the opposite extreme from that of the normal waking adult, as seen
both in the electrical record and in the presence or absence of sustained
purposeful behavior.
123 The Stage of Primary Learning
The other is slowly acquired; dependent on a constant change
of the sensory influx for its maintenance for any period of time;
produces a flattened, irregular EEC; and is the precondition of
the normal adult waking behavior and the adult level of learning.
Note that what is discussed here is a variation of organized
activity in the cortex. In the later discussion of fear and neu-
rosis the notion of a disorganization will appear, producing
neither quietude nor purposeful activity but incoordination and
autonomic activity. Intrinsic or primitive organization, and dis-
organization, are to be distinguished.
Against the background of the intrinsic organization, the in-
fant's first visual learning could proceed steadily, although
slowly, with no disruption. If the first sensory excitation feeds
into an organized activity, where cells are firing rhythmically in
large populations (which I assume is what the large waves of
the infant's EEG mean), the argument of the schema can still
stand, although visual learning must be slower than if the cells
of area 18 were always ready to be fired, and never refractory,
when an impulse arrives from 17. The only change in the
schema is to show still another reason for the slowness of first
learning in the higher primates.
At the same time, an explanation of the difference from lower
species appears. The process of perceptual learning must be
thought of as establishing a control of association-area activity
by sensory events. The larger the association areas, both abso-
lutely and relative to the size of the sensory projection areas,
the slower the establishment of such a control must be and the
less rigid and more complex its final form.
Let us suppose, as we safely may, that the principles of cen-
tral organization in other sense modes do not differ from those
of vision. Anatomical differences in the various sensory systems
may affect the complexity of discrimination, fineness or acuity
of discrimination, and so on. They may lead to a greater rela-
tive importance of temporal integration in one sense and of
spatial in another, as in hearing and sight. But these ana-
tomical differences do not imply different principles of central
neural integration in sight and hearing, or in touch and smell.
Our problem essentially is to see how a particular sensory
event can have the same central effects on different occasions
124 Development of the Learning Capacity
despite spontaneous central activity. If we conclude with
Adrian (1934) and Weiss (1941a) that the unstimulated neural
cell must eventually fire spontaneously, we must also accept
Adrian's conclusion that this cannot happen in the cell that is
exposed to persistent bombardment. As long as the receptor
surfaces are being stimulated, therefore, the sensory projection
areas of the cortex must remain completely and constantly under
environmental control.
Also under control are the fibers that lead from the sensory
areas into the association areas. According to the schema of
perception, this control is extended gradually, synapse by syn-
apse. Considering the association areas as made up of a popu-
lation of transmission units, two factors must affect the length
of time needed to bring all these units under control. One is
the number of controlling fibers leading from sensory areas into
association areas. The second is the number of transmission
units in the association areas themselves.
With cortex of a given size, these two factors may be con-
sidered to be roughly proportional to the size of the total sen-
sory cortex, and the total association cortex. It then follows that
the length of the primary learning period will be roughly pro-
portional to the ratio
total association cortex
total sensory cortex
which can be called the A/S ratio. The sensory projection
areas are directly under environmental control; and if they are
large, with respect to the association areas, and so project a
large number of fibers into the association cortex, their control
should be quickly established. If the sensory projection is small,
association cortex large, the control will take longer; the period
of "primary learning," that is, will be long.
But we may have to consider a further factor. With the same
A/S ratio, but with different absolute size of brain and a larger
absolute number of transmission units in the association areas,
it seems that the larger brain might also make for slower first
learning. A larger number of transmission units means a greater
variability in the spontaneous activity of the association areas,
as well as a larger number of synaptic junctions over which
125 The Stage of Primary Learning
sensory control must be extended. It will be recalled that the
synapses in question are not the minimum number through
which an afferent excitation might conceivably reach the motor
cortex. We are dealing with recurrent circuits and complex
closed systems. An absolutely greater number of transmission
units in the association areas would probably make for greater
variability in early spontaneous activity and a greater ultimate
complexity of organization. Both imply that a sensory control
might be established more slowly in a larger brain, even with
the A/S ratio constant.
The sensory projection areas are widely separated in the brain,
almost as if to maximize their influence. Only in the frontal
pole in primates (and to a less extent in the temporal pole) is
there cortical tissue that is far removed from a sensory projec-
tion area. We can then regard the stage of primary learning as
the period of establishing a first environmental control over the
association areas, and so, indirectly, over behavior. With this
proposition we have at once an explanation of the slowness of
the first learning in primates, in which brain and A/S ratio are
larger, and of the ultimate level of the primate learning capacity.
The numerical aspect of this fraction cannot be stressed, since
presumably there are other factors of finer neural anatomy, of
metabolism, and so on, to be taken into account. It may be
suggested, however, that the size of association cortex, and the
A/S ratio, when these can be expressed in terms of cell counts,
may provide the basis of a better morphological index of the
level of behavior than the brain-weight/body-weight ratio. This
of course is not testable at present. No cell counts are avail-
able, and psychology has so far found few satisfactory compari-
sons of intelligence as between different species. We have not
even any exact data on the gross size of the sensory cortex in
different species.
But, for widely differing phylogenetic levels, a hierarchy of
"intelligence" (or psychological complexity: McBride and Hebb,
1948) can be assumed, which corresponds to gross differences
(1) in size of the cerebrum, or (2) in the proportion of afferent
to internuncial neural tissues. In the lower vertebrates, the
cerebrum is small, and the afferent systems are massive in com-
parison with the internuncial. Within the mammalian series,
126 Development of the Learning Capacity
there are differences both of absolute and relative size of asso-
ciation cortex which may be assumed (in the total absence of
any exact data) to have a relevance to the greater speed with
which the "lower" species can learn to respond selectively to the
environment, and to the comparative simplicity of the behavior
when it is fully developed.
These anatomical considerations draw attention to another
point. The learning ability of higher species at maturity is not
merely the capacity for a greater number of associations or for
associations that involve finer sensory discriminations. The be-
havior also shows a less direct control by the stimulus of the
moment, from the immediate environment. In larger association
areas the central phase sequence can be more complex: it must
still be organized and ultimately controlled by the relatively
smaller sensory projection areas, but the phase sequence can
escape the direct control more frequently and for longer periods.
The possession of large association areas is an explanation both
of the astonishing inefficiency of man's first learning, as far as im-
mediate results are concerned, and his equally astonishing effi-
ciency at maturity.
MECHANISM OF ADULT LEARNING
The question has been raised as to how, if the association
areas of the brain have a continuous and variable activity, a re-
peated stimulus can have the same central effect on separate
occasions, to make cumulative learning possible. The question
was answered for the infant, and the answer helped to account
for the slow first learning of higher species. It can now be
answered for the adult.
Fortunately, it is no longer necessary to account for a cumu-
lative effect of stimulation that is superimposed on an inde-
pendent activity of the association areas (fortunately, because
this activity must be extremely variable in the waking adult).
The learning occurs when the events to be associated can
already command organized trains of cortical activity; in other
words, when the environment has a control of the association
areas that can be repeated, so that the central activity is not at
127 Mechanism of Adult Learning
random and the stimulation can impinge on the same central
pattern when the training situation is repeated.
I have made it an essential condition of learning that two
central events occur together. The reader is reminded, how-
ever, that this is not always a sufficient condition. In discussing
the question earlier (p. 77), I considered the probability that
two systems would acquire an interfacilitation by being active
at the same time. The systems referred to, at that point in the
development of the theory, were newly organized ones, and the
question was of the first association of one system with another
during primary learning. The conclusion was that the ease of
association must vary greatly, even when the two systems lie
intertangled in the same gross region of the brain. They might
promptly coalesce, at one extreme; or might, at the other, fail to
do so despite often being active at the same time, over a long
period. It was concluded also that the larger the system the
greater the probability of its establishing an effective inter-
facilitation with anotherprovided that the two are well organ-
ized, so that arousing part of one will arouse the whole.
But we are now interested in the associations formed at ma-
turity, between much more complex processes. Learning at
maturity concerns patterns and events whose parts at least are
familiar and which already have a number of other associations.
This changes the problem considerably. It means that the
learning is not an association between totally unrelated proc-
esses. It must concern a complex of cell-assemblies, and elabo-
rate phase cycles (in the jargon of my schema); and amounts
to a strengthening of facilitations, not a setting up of new con-
nections between wholly unrelated activities.
The characteristic adult learning (outside of psychological
laboratories) is learning that takes place in a few trials, or in
one only. It seems always to involve a recombination of fa-
miliar perceptions and familiar patterns of movement. Thus
one can typically remember a new name when given name and
surname are already familiar, or a new face when it is one of
a racial group to which one has long been exposed. Vary these
conditions, and repeated effort may not do the trick. Complicate
the problem even by requiring that the new name and new face
be learned simultaneously, and the difficulty is increased; it is
128 Development of the Learning Capacity
usually overcome only by repeating the name and looking at the
face a number of times in a short interval. So adult learning is
typically an interaction of two or perhaps three organized activi-
ties; being organized, they are capable of a continued existence
after cessation of the stimulation that set them off, which gives
time for the structural changes of permanent learning to take
place. For the theory being developed, a prompt learning is
possible when the stimulation sets off well-organized phase se-
quences, but not otherwise. This organized activity of the asso-
ciation areas is subject to environmental control. To the extent
that the control is effective, and re-establishes the same central
pattern of activity on successive trials, cumulative learning is
possible.
Adult learning is thus a changed relationship between the
central effects of separate stimulations, and does not directly
concern the precipitating stimulus or, primarily, the motor re-
sponse whose control is imbedded in the central activity.
The same conclusion is reached from another approach. Ex-
periments on' learning have shown again and again that the
nonsense syllables tob, del, rec, til, and so on, are harder to
memorize than the more complicated items tobacco, delights,
recommend, and heartily; and still harder than the sentence
"I recommend the delights of tobacco heartily." The event
with meaning is best remembered; moreover, it is the meaning
that is remembered rather than the specific stimulation that
aroused meaning (McGeoch, 1942). That is, the central effects
of sensation are what enter into an association, rather than the
comparatively simple sensory event itself. This seems especially
true of the most efficient learningthe kind that is established
most easily and persists longest.
Such learning may be diagrammed as in figure 14, where each
circle represents a conceptual activity. The concept of an ob-
ject or place is an irregular cycle, each phase of which is the
activity of a cerebral cell-assembly. If a large enough part of
this phase cycle is aroused, the whole becomes active. Thus in
figure 14 the concept A was originally organized by an inter-
action of hearing, touch, and vision. Once organized, it may be
aroused by hearing alone, or perhaps by hearing and touch; but
the essential association between A and B, resulting from simul-
129 Mechanism of Adult Learning
taneous activity, would be the same whether each was aroused
by vision, or whether one was aroused by hearing and the
other by touch.
In this way learning may be "free" of any particular sensory
cues (Tryon, 1939). The rat may learn a maze by the aid of
cues from vision, smell, touch, and kinesthesis; the maze once
FIGURE 14. A and B represent two conceptual activities. C, C, possible
connections between A and B. These are not simple closed neural circuits,
nor even the more complex "assemblies," but "phase cycles": systems of
assemblies whose several activities are temporally integrated and tend to
recur in an irregular cycle. A was originally organized by auditory, tactual,
and visual stimulation (that is, it involves assemblies in each of these
modes); B, by visual and tactual stimulation. When these cycles are well
organized, their activity may be initiated by part of the original stimula-
tionA for example by hearing, touch, or vision. When A and B are
simultaneously active, they may acquire an interfacilitation which is dia-
grammatically represented by C and C. The learning is independent of any
particular stimulation; the association might be set up by two visual stimu-
lations, but be manifest later when A is aroused by hearing, or B by touch.
having been learned, any of these cues may be disturbed or re-
moved, but the learning persists and utilizes those that remain.
We shall see in a later chapter that the behavior of the blind rat
is permanently affected by his having had vision during the
period of development. The rat blinded at maturity solves cer-
tain problems that rats blinded at birth do not. The properties
of the concept A (figure 14) depend originally on vision; but a
rat might be blinded, after A had been developed, and still re-
tain the concept as one that can be aroused by hearing and
touch alone.
Now as to the mode of association between two such con-
cepts. The two neural structures may have enough chance ana-
130 Development of the Learning Capacity
tomical convergences (Chapter 4) that a direct interfacilitation
is established. This is represented (still quite diagrammatically)
by C, C in figure 14. Then, even though the stimulation arous-
ing A and B is brief, they may continue active long enough for
synaptic knobs to develop. The result would be one-trial learn-
ing, involving two separate neural systems.
But this, it seems, is unlikely. When new connections are to
FIGURE 15. To illustrate the possibility that a subsystem, C, may act as a
link between two systems (conceptual complexes). One concept is repre-
sented by Aj, A
2
, and C; the second by B
lt
B
2
, and C. The two systems
have a subsystem, C, in common, to provide a basis of prompt association.
be set up, as in figure 14, it is probable that a number of trials
would be necessary, and the diagram is best considered as rep-
resenting rote learning. The facts already discussed have indi-
cated that one-trial learning occurs only as the association of
concepts with "meaning"having, that is, a large number of asso-
ciations with other concepts. Figure 15 diagrams another pos-
sibility, which can be given more weight. A perceived object
consists of a number of perceptual elements (p. 83). The same
elements recur in different perceptions, so that two concepts to
be associated may have phases (assembly actions) in common.
These would be ready-made links, needing only to be strength-
ened to establish the association.
But more: the perception of an actual object (that can be seen
from more than one aspect, and touched, heard, smelled, and
tasted) involves more than one phase cycle. It must be a hier-
131 Mechanism of Adult Learning
archy: of phases, phase cycles, and a cycle or series of cycles.
("Cycle" is of course temporal: referring not to a closed ana-
tomical pathway but to the tendency of a series of activities to
recur, irregularly.) The two ideas or concepts to be associated
FIGURE 16. Diagramming in another form the same principle as in figure 15.
The complexes A and B are sensorily aroused at the same time. Solid
circles represent systems that are reliably aroused; broken circles, ones that
are facilitated by the activity of adjacent systems, but not reliably aroused
by them. That is, the "fringe" systems X, Y, Z, C, and so on, receive a
facilitation from A which is often not sufficient to arouse them; which ones
are active will depend on preceding activity in the tissue and accidents of
concomitant stimulation. The complex C might or might not be aroused as
a sequel to A, or to B; but when both are active simultaneously, C is more
likely to be aroused. Thus the subject associates an object B with the ob-
ject A, because both are associated with something else, C: a familiar trick
in memorizing lists of words or nonsense syllables in learning experiments.
But the processes A, B, and C all occur within what is grossly the same
tissue, not spaced as in the diagram; and when by virtue of C's activity A
and B are persistently active together, they may slowly build up a direct in-
terfacilitationso that C can drop out eventually, leaving A linked to B
directly, by a short-circuiting. (Woodworth, 1938, p. 34; McGeoch, 1942,
p. 166.)
might have, not only phases, but one or more subsystems in
common. This is suggested by the diagram of figure 15. It
should provide an even more effective link. Such a mode of
association is possible only between complex systems, and it may
be recalled again that a complexity of meaning is more readily
remembered than a simpler perception without meaning.
The same general principle is involved in figure 16. This
132 Development of the Learning Capacity
represents the familiar fact that a subject required to learn
pairs of words with no obvious relationship will invent relation-
ships, and will himself supply extra terms in doing so. The
subject trying to associate conceal with above may do so by
thinking above = CEiLiNG-conceal; or with marble-punish, may
think of being punished FOR PLAYING marbles FOR KEEPS (Woo
worth, 1938). The subject is making his task more complex
when he chooses to remember three or four terms instead of
two, but he also makes it easier. Thus figure 16 shows how two
concepts A and B may be associated if a third one ( C ) is aroused
to which A and B are already related.
The prompt learning of maturity is not an establishing of new
connections but a selective reinforcement of connections already
capable of functioning. Observe, however, that this account
differs from traditional association theory in at least one respect,
which is of the greatest importance for applying the theory to
the results of experiment. Two concepts may acquire a latent
"association" without ever having occurred together in the sub-
ject's past experience. Although the "association by similarity"
of older theory recognized this fact, the explanation seemed to
depend on the idea of identical sensory elements. The present
theory suggests that the identical elements may be conceptual
rather than sensory; that is, two things may seem similar with-
out exciting the same receptors.
The concept A in figure 16 is capable of arousing, or facili-
tates, C; but the facilitation may be from one of the subsystems
of A on a subsystem of C, so that A and C need never have
been active together at any earlier time. Or, as in figure 15,
two systems may coalesce promptly because of having a sub-
system in common. It seems likely that something of this sort
is the explanation of the happy figure of speech, which links
two sets of ideas not previously associated: the more unlike
they are, superficially, the more effective the figure; and yet it
is effective only because something about the structure of the
two ideas is the same. (For examples of such pairs of ideas,
the reader may consult the discussion of scientific analogies
earlier in this chapter.)
My treatment of learning here is tailored carefully to the
experimental facts, but it also follows naturally from the original
133 Mechanism of Adult Learning
neural schema. At this point, in the emphasis on pre-existent
associations in learning, we have come to a classical problem
that we can examine a little more closely: the meaning of "mean-
ing."
The implication of the preceding paragraphs is that a concept
is not unitary. Its content may vary from one time to another,
except for a central core whose activity may dominate in arous-
ing the system as a whole. To this dominant core, in man, a
verbal tag can be attached; but the tag is not essential.* The
concept can function without it, and when there is a tag it may
be only part of the "fringe": sometimes aroused with the domi-
nant subsystem, sometimes not. The conceptual activity that
can be aroused with a limited stimulation must have its organ-
ized core, but it may also have a fringe content, or meaning,
that varies with the circumstances of arousal.
Thus in figure 16 the activity of A may be accompanied by
the activity of X on one occasion, or of Y and Z on another.
Which will happen is not indeterminate, but depends on the
excitability of each subsystem at the moment, and on the facili-
tation from other concurrent sensory and central activities.
Now consider the subject learning a list of words again. Fol-
lowing one train of thought, in which an activity occurred which
facilitates activity in C (figure 16), the occurrence of A and B
(also facilitating C) leads at once to the sequence A-C-B. On
another occasion, A and B at first might have excited no common
system. In such circumstances, the subject would look at the
words, "think" about them (that is, A and B are persistently
active, with a varying fringe activity), until the facilitation on C
* It seems likely that there is a great deal of conceptual activity that is
unreportable (and so "unconscious") in human thought: above all, in what
can be called intuitive judgments (Hebb, 1946o). Despite reliable and pre-
dictable verbal responses in the recognition of emotion, for example, a sub-
ject may be quite incapable of saying what considerations determined his
choice of terms. There may, consequently, be concepts in man that do not
have a verbal element, just as there are in animals. On the other hand, it
seems quite clear that many concepts are fundamentally and essentially
verbalthe "core" is a word or other symbol, without which the concept
could no longer be an element in thought (function as a neural system).
If we recognize, first, that there are nonverbal concepts, we must recognize
secondly that some concepts are verbal only.
134 Development of the Learning Capacity
from some one of the fringe activities suddenly becomes liminal,
and the relationship A-C-B is "perceived."
Evidently such changes are what has been called, in the psy-
chological literature, a "restructuring" of thought; and the sud-
den activation of an effective link between two concepts or per-
cepts, at first unrelated, is a simple case of "insight" (p. 158).
This is a topic that will be discussed more fully in the following
chapter. The mechanism of the sudden perception of new re-
lationships is, I propose, the one that has just been discussed, in
its simplest form.
This account does not, however, sufficiently recognize the or-
derly, directed, persistent feature of the phase sequence that
leads to insight: a problem that presents major difficulty (Hum-
phrey, 1940), and is considered in the following chapters in the
treatment of attitude, motivation, and pleasure. For the present
theory, insight is a "chance" combination of facilitations from
different phase sequences; but this is chance only in a limited
sense, and given phase sequences that persistently return to a
certain conceptual activity (motivation to solve a problem) the
ultimate occurrence of insight may be quite predictable. What
may not be predictable is just when the occurrence will take
place, and this of course agrees with the facts of behavior.
MAZE LEARNING
Having considered how a certain kind of human learning
might occur, in which the subject given one word must repro-
duce a second, let us next see how the theory of phase sequences
and "concepts" can be applied to the learning of a rat in a maze.
Suppose that we have a simple maze with a starting point A,
three choice points B, C, and D, and a goal E. As the rat runs
the maze correctly, the environmental or sensory sequence would
be A, B, C, D, E. The phase sequence determining such a run
cannot be so simple, because of the central facilitation that may
arouse a conceptual activity in advance of its sensory arousal
("expectancy"), and equally may arouse the concept of a point
that has already been passed ("recollection"). This can be dia-
grammed as
A-e-b-B-a-c-C >E
135 Maze Learning
where the small letters represent centrally aroused complexes
(each corresponding to the recognition of a point in the maze),
and capital letters represent sensorily aroused ones, or actual
perceptions.
The existence of the procurrent items in the sequence above
is both an inference from the neural schema and a reasonable
inference from the literature on expectancy. The recurrent
item (a following B) is mainly an inference from the schema;
but it is suggested also by the backtracking done by sophisti-
cated rats in a simple maze.
In a long and complex maze, with unsophisticated rats, re-
tracing is usually thought to be simply an error, part of the rat's
effort to reach food as quickly as possible. But in the method
of testing intelligence described by Hebb and Williams (1946),
the rat often acts in a way that suggests something different.
After reaching the food box the animal may turn and explore a
blind alley, or retrace the entire maze (sometimes without a
single error). His return to the food box is usually at a higher
speed than the first part of the run, and he eats enthusiastically
on the second time of reaching food. Behavior is determined
by the central phase sequence, and this means that the sequence
is recurrent as well as anticipatory. Similarly, "mental back-
tracking" is reported by the blindfold human subject learning a
stylus maze (Woodworth, 1938, p. 143).
Also, when one is finding one's way through an unknown city,
or by compass through the deep woods or at sea, one has a con-
tinual awareness of the supposed direction (1) of the goal and
(2) of the earlier route, though the apparent direction of either
may be quite wrong. There are indications in both animal and
human behavior that learning depends on setting up a unifica-
tion of the total situation even when it cannot be surveyed as
a whole. It is not necessary for the learning that an accurate
picture of the situation be achieved (Brown, 1932), only that
the goal-concept dominate the phase sequence, and that each
particular complex in the sequence corresponding to a correct
run reliably evoke the next complex, with the resultant recogni-
tion of the direction of the next choice point.
As an illustration: when driving to the neighboring city, 18
miles away by a winding route (on a modern highway, that is,
186 Development of the Learning Capacity
on which all the turns are very gradual), I persistently feel that
the part of the highway on which I am at the moment leads
nearly straight toward my objective. When a bend in the road
approaches, the apparent direction of the city changes; each
turn is remembered, inconsistently, as a turn toward the goal,
except when two turns are close together. I am unable to list
all the different segments of the route from memory, in the
right order; all I can do reliably is make the right turns at the
right time.
This proves nothing, perhaps, except that I am easily confused;
but it illustrates the point that a series of correct responses can
be organized into a smooth sequence under the influence of a
goal anticipation without requiring that the subject (whether
rat or man) have any accurate or detailed image of the total
situation. The phase sequence determining such behavior need
not be so well organized that it can run off complete without
sensory support. As I start driving, it consists of little more
than an awareness of the starting point, the goal in a certain
direction, and the next distinctive point in the route.
How is the sequence established? It will be clear, from the
facts reviewed in the early part of this chapter, that the half-
grown or adult animal comes to his task with an elaborate con-
ceptual organization ready made, and that the learning is a
modification of this to fit the particular properties of the new
situation. Before one can attempt anything like an explanation
of adult learning, therefore, it will be necessary to find out
much more about the ontogenetics of behavior. The experi-
ments reported in the following chapters hardly amount to a
beginning at this task; until they are carried much further a dis-
cussion of the details of maze learning must be speculation only.
The experiments referred to suggest, for example, that much of
a rat's learning to run a maze is done in his home cage, before
he ever enters the maze at all. What I propose to do here and
in the following chapters is outline the approach that is im-
plied by the theory and define the problems that it raises for
research.
One of these problems is the development and control of
hunger. Experimental evidence presented later (p. 193) indi-
cates that there is an important element of learning in hunger.
137 Maze Learning
It is not always enough to have an animal lack food in order to
get him to eat. At the same time, the analysis of the control of
eating made by Morgan (1943) shows that no simple learning
formula will provide for the known facts. This problem, then,
must be left to one side for the present. Assume here that the
foundations are laid: the rat put into the apparatus will eat food
when he finds it, will move about until he does find it, and has
FIGURE 17. A simple maze problem, the simplest of those used in the
method of testing rat intelligence, that is briefly described in Chapter 11.
S, starting box; F, food; A, B, C, D, points in the maze. The line C-D
represents the single barrier used in this test problem; in the preceding prob-
lem, four barriers, differently placed, presented the rat with a more difficult
problem. Crossing the broken line constitutes an "error."
become used in preliminary training to the general situation and
has organized his perception of room cues, the remote objects
that determine perceived position (p. 92).
With a simple problem such as that of figure 17, one of those
used in a rat "intelligence test" (Hebb and Williams, 1946),
learning takes from 1 to 5 trials. In this method the food box is
always in the same place, but there are movable barriers that
are changed for each test, so the rat is used to finding food by
a different route each day. With the problem shown, a bright
rat is likely to enter the error zone BC once or twice, and then
make no more mistakes, following the route A-D-F consistently.
At the point A, the rat's behavior on the first trial is deter-
mined by his experience with preceding problems. He has
sometimes found food by going toward B followed by a right
turn, sometimes toward D followed by a left turn, and always
138 Development of the Learning Capacity
in the general region of C and F. At A, accordingly, the phase
sequence is variable (unique at any one moment, but changing
from one moment to another as "meaning" variesp. 133). With
each phase sequence, there is a selective influence on the motor
system, determining head-and-eye movement toward B, C, or D.
Further, the phase sequence is dominated by two things: an
anticipatory concept or expectancy of reaching food, and per-
ception of the rat's present place in the maze. Each two con-
ceptual complexes are linked by activity of the cortical motor
system facilitating not only movements of the head but also loco-
motion. The latter facilitation is not always liminal, but when
it is liminal the rat runs, in the direction determined by the
earlier movement of head and eyes.
The sequence of events, as the rat reaches A, might then be
as follows: perception of Aexpectancy of food, with orienta-
tion toward Crunningperception of Cexpectancy of an alley
leading to F. No alley appears, and either of two things may
result.
1. The first possibility is that the controlling phase sequence is
not strongly established and readily gives way to another which
determines a different direction of movement. Expectancy was
not confirmed; the original phase sequence has not been rein-
forced by a corresponding sensory sequence, and the proba-
bility of its occurrence when the rat next arrives at A isat
mostnot increased. On a second trial, the run may be toward
D. Now the expectancy of an alley beyond D is reinforced sen-
sorily. The individual phase in the phase sequence is aroused
both by a sensory and a central facilitation, strengthening inter-
phase facilitations and so increasing the probability that on a
third trial the rat will go from A direct to D.
2. As the rat approaches C on his first run, the second possi-
bility is that the phase sequence determining this behavior is
strongly established, with a strong motor facilitation. The end
of the cul-de-sac not only fails to provide sensory reinforcement
for the phase sequence but disrupts it. There is a conflict be-
tween the original phase sequence and a new one aroused by
contact with the end of the cul-de-sac, which has been asso-
ciated in earlier experience with a right-about turn. This con-
flict may appear as an emotional disturbance (anticipating the
139 Maze Learning
discussion of emotion in the following chapter), and amounts to
breaking up a link in the chain of the recurrent organization of
the phase sequence. In succeeding trials when the rat reaches
the point A in the maze, the phase sequence may be, as before,
perception of Aexpectancy of food, with orientation toward
Canticipation of C. But the anticipatory concept of C is now
linked with a disruption of cortical organization. The cyclical
series A-C-F-A-C-F, and so on, is broken up, permitting
another cycle A-D-F (originally less strongly established) to
occur instead.
This, in outline, is the approach to adult learning that is sug-
gested by the phase-sequence hypothesis. As I have said, there
are large gaps in the evidence, and when these are filled in the
hypothesis may be found to need serious modifications. It has
not made use of motivation or insight as special processes inde-
pendent of the phase sequence, because these with emotion seem
already subsumed by the hypothesis. I shall try in the follow-
ing chapters to show how such processes may be dealt with in
more detail.
/. Higher and Lower Processes
Related to Learning
The course of learning is rarely a smooth accession of motor
skill, or a closer and closer relationship of response to some par-
ticular environmental event, as the learning process goes on.
Learning curves, for the individual animal, show great fluctua-
tions from day to daynow backsliding, now sudden improve-
ment. It is only by averaging the results from a number of ani-
mals, trial by trial, that one can manage to get a simple curve
showing steady improvement with practice.
It is usually considered that these fluctuations are caused by
a number of factors that are independent of learning and of
each other. Just what factors are invoked depends on the indi-
vidual writer and his theoretical point of view. In general,
they include (1) processes that guide and strengthen learning:
physiological needs and motivations (in one kind of theory), or
such things as attention, expectancy, and insight (in another);
(2) processes that weaken or disrupt learning, such as emotional
disturbances; and even (3) the innate processes of instinct that
are thought to take the place of learning.
The question of differences in the learning of different species
has already been broached. Amplifying this topic will also serve
to fill in some of the more important details in the account of
learning that has been outlined in preceding chapters. In the
present chapter, accordingly, the "higher" processes of con-
sciousness and insightful behavior will be discussed together
with the "lower" factor of instinct, and, more briefly, the rela-
tion of these to emotion and motivation, leaving a fuller account
of the latter processes to following chapters.
140
Attitude 141
ENDURING SELECTIVITY IN CENTRAL ACTION:
ATTITUDE
An important aspect of an animal's behavior is the occurrence
of periods of directed behavior, in which one kind of action
appears only, or in which responsiveness is to one kind of stimu-
lation. This selectivity appears to be like the selectivity of at-
tention but lasts longer and is less specific. It might be referred
to as attitude, interest, or intention (though these terms may be
objectionable), to distinguish it (1) from the more briefly last-
ing attention, and at the same time (2) from the stable selec-
tivity that may be either inborn (and referred to as instinct)
or established by slow learning, and remains unchanged from
one day to the next. The behavior discussed here is behavior
in which responses are now to one class of stimulus, now to
another, in the same environment.
At a purely descriptive level, this aspect of behavior must be
considered when one attempts to deal with species differences.
It seems first that the duration of the selectivitythe amount of
time likely to be spent continuously in one kind of activitymay
vary for higher and lower species. Periods of spontaneous play
are shorter in the rat than in the dog, and still shorter than in
man; the period of emotional disturbance following momentary
stimulation seems progressively shorter as one goes down the
phylogenetic scale; and so on. Secondly, the number and variety
of the distinctive forms of behavior involved is greater in higher
speciesthe number of recognizably different attitudes or inter-
ests, that is, is greater. And, finally, the variability of action
that still remains selective is also greater in higher species: an
aggressive attitude, for example, has a much more variable ex-
pression in the chimpanzee than in the dog.
At a theoretical level, it seems further that there can be no
explanation of learning and problem-solving in any mammal
without reference to the persisting central neural influence that
sustains activity in one particular direction. Even in the rat,
learning continually shows this selective responsiveness to one
aspect or part of the environment. In higher forms, where
expectancy has been most clearly demonstrated, it seems often
142 Processes Related to Learning
to be organized by an expectancy of a particular reward or goal.
This is clearly not sensory but a conceptual process (to which,
however, sensation must continually contribute).
To account for a series of responses to the environment in
which a lasting central influence is evident, it would be con-
venient to be able to think of reverberation in some thalamo-
cortical system, acting without interruption to influence other
cerebral systems in a particular way for extended periods of
time. But this is at the least unlikely. Even if such a system
were unexposed to external disturbance, simple reverberation
without afferent support cannot be expected to last more than a
fraction of a second (Lorente de No, 1939). The schema of
neural action on which the present theory depends has made it
appear that there is no constant, uninterrupted activity deter-
mining behavior, no one enduring pattern of neural firing, but
a continual fluctuation. If the theory up to this point is to be
accepted, a selectivity of behavior that may last for minutes or
longer must be ascribed to some other limitation on the varia-
bility of firing pattern in the cortex. How might this occur?
First, there may be classes of associated cortical activities.
Because of having occurred together frequently, each activity
in a class A, B, C, D, and so on, would facilitate other activities
in the same class more than those in another class L, M, N, O.
Arousal, of any one activity A would then arouse others
B-C-D-A' , but none of L, M, N, O. Each letter here stands
for a conceptual activity, or phase cycle, not a cell-assembly
action. The duration of each phase cycle might be a matter of
seconds; of the series of cycles (or conceptual series), corre-
spondingly longer. This would be one mechanism of a persist-
ing attitude, with a mainly intracerebral origin. Its maintenance
might also be contributed to by sensation; the behavior would
amount to a seeking out of certain stimulations rather than
others, and these might be such as to reinforce the selectivity
of further response.
But secondly, if the initiation of one of these protracted and
selective conceptual series involves some lasting bodily condi-
tion such as lack of food, which in the presence of a particular
environment sets up an expectancy of eating, the expectancy
might dominate the series and have a further organizing influ-
Attitude 143
ence. (The assumption is that hunger pangs, or a low level of
blood nutrients, would have a selective effect on the limen for
the conceptual activities determining eating. This assumption
will be discussed later, in relation to the problem of hunger.)
If A then stands for the expectancy of eating, and B, C, D for
conceptual activities associated with eating, the series might be
A-B-A-C-A-B-D-A, and so on, with A in a dominating role.
Thus a lasting bodily state might repeatedly induce a concep-
tual activity whose frequent recurrence would act to organize a
conceptual series and restrict its content. This second mech-
anism presumably would make possible a longer duration of
attitude than the first alone.
The duration of a self-maintained activity, as contrasted with
an activity that is frequently reinstated sensorily, depends on the
complexity of the system in which it occurs. A simple closed
circuit may reverberate for perhaps 0.001 to 0.05 second; the
cell-assembly, in which an alternating reverberation is possible
(by-passing refractory transmission units), has been assumed
to be capable of activity for periods as great as half a second
(p. 74); a phase cycle, or sequence of recurring assembly ac-
tions, might then endure for perhaps 1 to 5 or 10 seconds (the
apparent duration of conceptual processes in man). I have
assumed in the preceding paragraphs that a series of such cycles
or concepts might last for minutes or longer and constitute a
transient attitude. A greater complexity does not ensure a
longer persistence of self-maintained activity in the system, but
may permit it; and at the same time it permits or depends on
a greater variability of specific content from moment to moment.
In the conceptual class A, B, C, and so on, A facilitates B, C,
and D. A temporarily high limen in B at the time of A's activity
would not prevent continuance of selective behavior, since C
or D may be aroused by A if B is not. The idea in essence is
that a conceptual series A-B-C-D-A-B-C-D, and so on, might
have a greater duration, at its maximum, than A-B-C-A-B-C,
and so on.
In a higher brain, accordingly, in which more assemblies are
organized by learning and may occur in more varied combina-
tions, the upper time limit of a selective central influence is
greater and the selectivity at any one moment is less specific.
144 Processes Related to Learning
The anatomical factor here is both absolute size of the associa-
tion areas and the A/S ratio (p. 124). This provides some ra-
tionale for the observed relation between (1) phylogenetic level
and (2) the complexity and duration of attitude and interest.
The problem dealt with here has not of course been solved. It
will be returned to in Chapter 9.
Now the behavior that shows this sort of control is the be-
havior that is called conscious or voluntary. These two terms
are not in good standing in modern psychology, and yet they
refer to an important psychological distinction: not only for the
clinic, where it is fundamental, but even in the experimental
laboratory, where for example the differences in conditioning
voluntary and involuntary responses are well known. It is not
important whether one is to continue using the words conscious
and voluntary in psychological discussion; it is important to
have the distinction to which they refer made explicit, and to
be able to deal with it theoretically.
Let us start by recognizing that the distinction is not between
discrete, unrelated states but between the extremes of a con-
tinuum. The important thing is to define the continuum, not to
decide just where a line should be drawn to divide one end from
the other, or even to divide it at all.
The preceding discussion of attitude makes a theoretical defi-
nition of "voluntary" and "conscious" possible, and, just as im-
portant, draws attention to what seems to be the empirical mean-
ing of these terms, quite apart from theory. For the present
theory, a voluntary act is one that is determined by a phase
sequence or conceptual series with some duration in time, to
which both sensory and central facilitations contribute con-
stantly. These sequences are recurrent as well as anticipatory,
so the voluntary act is influenced by immediate memory and
set as well as by an expectancy of the futureparticularly, an
expectancy of the immediate consequences of the act. This
theoretical definition is considered also to cover the adjective
"purposeful"from the point of view of mechanism, there is no
apparent basis for distinguishing "voluntary" and "purposeful"
(the distinction in usage may actually be a distinction between
the circumstances in which a single kind of process occurs).
Empirically, a voluntary act appears during periods of selec-
Attitude 145
tive responsiveness, conditioned both by earlier stimulations (the
influence known as set) and by present stimulation. The influ-
ence of present stimulation is not always obvious, since the act
is not necessarily initiated by an environmental event. It be-
comes evident, however, whenever an environmental change oc-
curs during the act that would change its effect. There is at
once a variation of muscular pattern that is such as to tend
toward a constant end effect. This muscular variability with a
single result has been called "motor equivalence" by Lashley
(19420). It is an essential mark of the voluntary act (and is
essentially the only meaning of the adjective "purposeful").
The form of an involuntary act is characteristically not a func-
tion of set, or any preparation of the subject by instruction or
the like (Hilgard and Marquis, 1940). It is determined by the
stimulation of the moment plus the general physiological condi-
tion of the animal, and does not show motor equivalence. The
occurrence or strength of an involuntary response may be sub-
ject to the influence of set, as when ghost stories facilitate a
startle reaction, but the startle is involuntary as long as it main-
tains the characteristic rigidity of pattern. The distinction is of
course between two ends of a continuum, not attempting a sharp
separation of two totally different kinds of behavior. Some in-
voluntary acts may show "purposeful" motor variability, or vary
in pattern as a function of set, but the more they do so the less
certain the classification must be.
Consciousness then is to be identified theoretically with a cer-
tain degree of complexity of phase sequence in which both cen-
tral and sensory facilitations merge, the central acting to rein-
force now one class of sensory stimulations, now another. The
cortical organization consists of a diffuse firing in a succession of
assembly actions or phases that are organized first in phase
cycles (conceptual activities) and then in series of such cycles
that maintain a selective influence on behavior for appreciable
periods of time, but always with the possibility that the current
organization may be changed or disrupted by an unusual sensory
event. The limen for the unusual, that is, remains low. The
unresponsiveness to some stimuli that is characteristic of the
conscious animal is always for familiar features of the environ-
ment. Finally, this organization is opposed to the intrinsic or-
146 Processes Related to Learning
ganization of infancy and sleep (p. 121), and appears to be
maintained by the constant sensory flux to which motor activity
contributes. It is rarely possible for any prolonged period of
unchanging sensory pattern (when sleep or catalepsy is likely
to ensue).
Empirically, in behavior, the most important mark of con-
sciousness is the continually changing selective responsiveness
to different aspects of a familiar environment, the unchanging
responsiveness to unusual or unexpected events, together with
the continual presence of "purpose," "means-end-readiness" (Tol-
man, 1932), or motor equivalence. Another mark might be the
occurrence of "operants" (Skinner, 1938), or actions not initiated
by an environmental change. However, these occur also in
earliest infancy, in very primitive organisms, and in sleep, and
so do not very clearly distinguish a conscious from an uncon-
scious animal.
MODES OF SENSORY-CENTRAL INTERACTION
The chief need left by this discussion of attitude, or enduring
selectivity of central action, is a more detailed treatment of the
interplay of sensory and central facilitations. I have said that
the phase sequence persistently escapes from a direct sensory
control. But this by no means implies a total lack of sensory
influence, at any time. By "direct" control is meant that asso-
ciation-area activity is determined by the pattern of immediately
preceding sensory stimulation; since this activity determines
behavior, a direct control would mean that with any given sen-
sory stimulation an animal would respond in only one way.
This, as we know, does not occur in higher animals in a familiar
environment. At the same time, it is quite evident that there is
a continual influence, of some kind, from sensation.
Defining this influence in its various forms raises a number
of problems. The schema of neural action made the mode of
sensory-central interaction explicit as far as momentary excita-
tions are concerned, but new difficulties appear when one is
dealing with lasting central activities and a series of sensory
events. Some of these are handled reasonably well in the phase-
sequence hypothesis. Others remain as problems for experi-
147 Emotion and Learning
mental investigation. They can be defined here and at least re-
lated in a general way to the hypothesis.
At each point in a conceptual series, the ensuing activity is
determined by the total pattern of sensation at that moment and
by the residue of facilitation and negativity from the preceding
central activity. In general terms, this means that there are
three possible ways in which sensory and central facilitations
may interact. (1) They may conflict, producing phase sequences
that are mutually incompatible; (2) they may have unrelated
effects, tending to set up independent phase sequences in
parallel; and (3) they may reinforce one another's action, both
facilitating the same subsequent pattern of cortical action.
The second and third of these possibilities are the subject
matter of later sections, dealing respectively with the relation of
attention to learning, and the role of expectancy in a skilled
motor performance. The first possibility, a conflict of facilita-
tions, is related to the problem of emotional disturbance.
THE RELATION OF EMOTIONAL DISTURBANCE
TO LEARNING
The original idea of emotion was probably a reference to the
demons that now and then disturb ordinary behavior. Demons
or no, let it be observed that this was an inference from
behavior, demons (including the soul itself) not being easily
seen by most people.
Very much later a quite different idea appeared, and is still
current. Emotion is now generally thought of as an awareness,
a distinctive conscious process that is quite separate from intel-
lectual processes. This notion has led to a good deal of confu-
sion, for it has gradually become clear that no such distinct kind
of awareness exists (a surprising unanimity of psychological
opinion has been reviewed elsewhereHebb, 1946a). In this
sense, the term emotion has lost much of its meaning. Further,
it is meaningless as referring to any momentary muscular or
visceral response, for here also no distinctive pattern can be
found (Dashiell, 1928).
But it is still true that emotional disturbance exists, and can
be recognized as a long-term deviation from an animal's ordi-
148 Processes Related to Learning
nary behavior (Hebb, 19460). Something causes it. I propose
that we return to an earlier, behavioral, interpretation (but sub-
stituting neural events for the demon), and define emotion as
the neural process that is inferred from and causes emotional
behavior, with no reference to consciousness. However, for
reasons to be considered later (Chapter 10), we shall be con-
cerned in the present chapter with emotional disturbance rather
than emotion.
Now as to mechanism. The causes of rage (Hebb, 1945a)
and fear (Hebb, 1946&) are extremely diverse. Unless we are
to suppose not only that there are a number of separate emo-
tions, with independent neural mechanisms, but also a number
of angers and fears, each equally independent, we must ask what
these various causes can have in common.
One answer to this question has already been suggested.
Emotion may be a disruption of cortical organization, which
could occur in several ways: the occurrence of incompatible
phase sequences; the absence of a sensory facilitation that has
always contributed to the phase sequence; "pain" stimulation
that can be supposed to be innately disruptive of cortical ac-
tivity (see Pain, Chapter 8); and chemical change of the blood
content, altering the rate of firing of individual neurons and
so disrupting a cortical organization that is fundamentally a
matter of timing (see the discussion of hunger and addiction,
Chapter 8).
Some of the points involved must be left for later chapters.
For the present we are concerned with the possibility of a sen-
sory disturbance of the phase sequence, as an influence on learn-
ing. In the schema of neural action, it was concluded that a
neuron, or group of neurons acting as a transmission unit, must
be expected to enter into more than one assembly if the assem-
blies are organized at separate times. In two assemblies involv-
ing some of the same transmission units, a simultaneous activity
would not be possible. The activity of one would prevent ac-
tivity in the other.
In assemblies of cells that are frequently active at the same
time, fractionation eliminates from each the transmission units
whose action is interfered with or controlled by others (pp.
88-9). In a familiar environment, therefore, such conflict would
Emotion and Learning 149
not usually occur. The same process would also gradually elimi-
nate conflict occurring at the first presentation of a new stimulus
combination; as the combination is repeated, the make-up of one
or both conflicting assemblies would be changed by fractiona-
tion, and the disorganization would decrease. The strange com-
bination would now be familiar and no longer disturb behavior.
In the phase sequence, the central facilitation from one phase
on the next is usually not specific, since sensory events occur in
a variable order; and, because of this, the facilitation from one
phase cycle or conceptual activity is not completely specific but
tends to arouse one of a number of subsequent activities (pp.
133-4). This flexibility of the phase sequence permits a con-
siderable variation of the sequence of sensory events without a
disturbing effect. Such considerations do not permit one to say
how great an environmental change must be to produce an
overt disruption of learned behavior; but the assumption that
disruption of cortical organization may vary in extent and in
duration, and the assumption that the disruption when it has
more than a momentary duration means some incoordination of
effectors, make sense of the emotional disturbance that appears
when environmental change is great enough. What I postulate
is that, up to a certain point, lack of correspondence between
expectancy and perception may simply have a stimulating (or
"pleasurable") effect (Chapter 9); beyond this point, a disrup-
tive (or unpleasant) effect.
The changes that disturb behavior must also include the ab-
sence of a customary stimulation. The phase sequence is a con-
tinual interaction of central and sensory events, and the back-
ground facilitation from a constant sensory facilitation must
affect the composition of the cell-assembly, during its organiza-
tion, and be a factor in its later arousal. The absence of a cus-
tomary stimulation will be "strange," and exciting. But again,
if this condition is repeated, a reorganization of assemblies and
change of interassembly facilitations can re-establish the sta-
bility of the phase sequences concerned. With a varied experi-
ence, the animal will become less and less dependent on any
particular stimulation that is not a constant feature of his envi-
ronment.
This is an effect of learning on emotion, a mechanism of
150 Processes Related to Learning
adaptation to the strange. The converse is a directing influence
of emotional disturbance on learning, a problem raised in the
preceding chapter. It has been concluded that the prompt
learning of maturity depends on a phase sequence with a re-
current organization. Disruption of any link in this disrupts the
whole. Hypothetically, the disturbance varies in extent: when
it is widespread, it may cause deterioration of other, alternate,
phase sequencesan extreme instance, presumably, being the
paralysis of terror. With less marked disturbance, the effect
may be to prevent activity in the current phase sequence only,
and so allow another integrated sequence to take its place. In
this case, there would be little sign of emotion in the behavior,
although the disturbance would have its effect in a changed
direction of response.
This is supposed to be the way in which emotional disturb-
ance guides learning. An example will make the proposed idea
more definite. Let us suppose that a rat is being trained to
avoid a darkened door, by being given an electric shock, and
being allowed to find food when he goes through a lighted door.
As the rat makes his choice, suppose also that his perception is
of the relation of the right-hand door to the distant environ-
ment, the relative brightness of the door not arousing a con-
ceptual activity. He approaches the right-hand door, and re-
ceives a shock. The disruptive effect here is on the conceptual
series: perception of present placedoor Aexpectancy of food-
door A, and so on. Back at the choice point, an alternate series
involving perception of door B can displace the first, the rat
goes through B, and finds food. The effect of this one trial is
to teach the rat to go through the left-hand door. Unless there
was a conceptual activity aroused by the different lighting of
the two doors (p. 106), this trial would not set up a tendency
to avoid a darkened door as such.
Subsequent trials, with the dark door sometimes right and
sometimes left, would accumulate a disruptive effect on both
"spatial hypotheses." But on any trial when the conceptual con-
tent includes phases determined by the brightness or darkness
of a door, an association of darkness with shock can begin to
be established. With this, the phase sequence will alternate
between: perceiving present position at the choice pointper-
Intention and Attention as Factors in Learning 151
ceiving a brighter doorexpectancy of food; and perceiving
present positionperceiving darker doorexpectancy of food.
The first is reinforced by a corresponding series of sensory
events, as the rat runs under its influence; the second is unrein-
forced, and, further, disrupted. When this stage has been
reached, the final learning of the task should be prompt.
This, and the discussion of frustration in the preceding chapter
(pp. 138-9), represent in principle the role of emotional dis-
turbance as it might affect learning in an experimental study.
A special relationship between emotional disturbance and learn-
ing will be dealt with in Chapter 10, which tries to show how
learning might develop typical rage and fear, with their distinc-
tive patterns of organized response, out of what (in its origins)
is simply disorganization.
INTENTION AND ATTENTION AS FACTORS
IN LEARNING
We can now turn to the second way in which sensory events
may be related to the current central process. The first possi-
bility, just discussed, is that the two may conflict. The second
is that they may neither conflict nor reinforce one another, but
determine independent phase sequences. This possibility bears
directly on the commonly held opinion that no learning is pos-
sible without intention to learn, no memory of a sensory event
unless it was "attended to" at the time of its occurrence.
It is often supposed also that there is an extraordinary unity
about attentionthat one can attend to one thing, or to another,
but not to both at the same time unless the two are a single
unit, responded to as a whole. These conclusions need qualifi-
cation.
There is actually no evidence to justify the general statement
that learning never occurs without the help of attention (though
my treatment in the following paragraphs assumes that this is in
a sense true). It has been shown repeatedly that a subject may
not remember what he did not attend to, or that he remembers
more with attention than without. But other things are apt to
be remembered to some degree. It is common experience to
remember details of a situation to which one did not, appar-
152 Processes Related to Learning
ently, attendwhen the details later have become relevant to
one's thought. Similarly, the remark made by someone else,
when one is reading, may not be noticed or responded to in any
way, as far as anyone can tell; and yet 30 seconds or a minute
later one becomes aware of the remark and answers it. The
event was remembered. Are we to postulate an "unconscious"
attention to explain the fact? If so, attention is no longer uni-
taryeven if there were no other objection to such a treatment
of the problem.
This is a point, it seems, at which such theoretical concep-
tions as attention and intention are no longer precise enough.
I have already defined attention as the immediate facilitation
from one phase, or assembly action, on the ensuing one, with
no implication that two such processes might not occur at the
same time. In so far as the term is to be useful, it should be
restricted to this very brief central facilitation. "Intention"
seems a still more slippery term, related to "attitude" but with
the implication that motor components are prominent. If the
term has any intelligible reference to neural function, it must be
to the enduring selectivity of central action that was discussed
earlier.
There is a certain unity about this action, but not at all the
absolute unity with which attention, at least, has been endowed
in the older literature. There may be more than one phase se-
quence in the association areas at one time. The discussion of
meaning in the preceding chapter implied also that the phase
sequence may continually tend to branch out in more than one
direction, ameba-likesome, or all but one, of the branches even-
tually becoming abortive for lack of the general background
facilitation of "attitude" that determines the general trend of the
phase sequence. I have emphasized the idea that the motor
facilitation of phase or phase sequence may be subliminal; and,
to the extent that attention has the unity it is supposed to have,
the unity may consist of a control of the motor system by one
phase sequence only, when two or more run in parallel. Even
then it should be remembered that we very often carry on two
familiar activities at the same time. Arguing and driving a car
will do as an instance; neither seems possible without "attention."
It certainly seems that the unity of attention has been exag-
153 Motor Equivalence
gerated, and the phase-sequence hypothesis, at least, suggests
that it may often be multiple.
Memory for an incidental stimulation then may occur because
the stimulus has aroused a second phase sequence without in-
terrupting the prior dominant one. This can happen only with
familiar activities and perceptions, and only if a phase sequence
is aroused by the stimulus in question; and, in agreement with
the importance of meaning for learning, only if the incidental
phase sequence has a certain degree of complexity and persist-
ence. There will consequently be plenty of occasions on which
"unnoticed" events are not remembered, as the experimental lit-
erature has shown.
So also with the effect of set on perception (perceptual set is
another aspect of the phase sequence, closely related to atten-
tion). It has been shown that in certain circumstances one sees
what one expects, or is set, to see. But this is no general rule,
else one would be incapable of seeing the unexpected. When
a stimulation is sufficiently strong or distinctive from its back-
ground, and familiar or composed of familiar parts, it will be
capable of initiating its own phase sequence, regardless of what
other phase sequence may be going on at the moment. (If the
new phase sequence is incompatible with the pre-existing one,
startle or emotion may result.) The sensory event is capable of
taking control of central action, just as in other circumstances
the central action is capable of reinforcing one sensory event at
the expense of a second.
MOTOR EQUIVALENCE
Two modes of sensory-central interaction have been discussed.
The first, a conflict of sensory and central processes, was re-
garded as a source of emotion; the second, in which a sensory
event has central effects that neither support nor conflict with a
prior central process, was treated as accounting for certain kinds
of incidental or unintentional learning.
The third mode, an active support and direction of the phase
sequence by sensory processes, is at least as important. It is
best exemplified in the phenomenon of motor equivalence: a
variability of specific muscular responses, with circumstance, in
154 Processes Related to Learning
such a way as to produce a single result. The phenomenon and
its importance for the theory of learning have been emphasized
by Tolman (1932), Hull (1934), and Lashley (19420). At the
same time, theory has not handled the problem very success-
fully. It is usual, and in practice necessary, to define an ani-
mal's response in terms of its end effect: the rat reaches and
enters the food box, the chimpanzee's hand makes contact with
a stimulus plaque and moves it. It would be impractical to
record the actual series of muscular contractions that brings
such an event about. But one must not then assume that a
common end result means a single pattern of contractions.
It is necessary to use very special conditions of training indeed
to establish a "learned response" as a constant effector response.
A rat trained to depress a lever to get food may do so from any
of several positions, in each of which the muscular pattern is
different. He may climb on the lever; press it down with the
left forefoot, or the right; or use his teeth instead. Very often,
all that can be predicted after the response is learned is that
the lever will be moved downward. It is not necessary here to
multiply examples of such behavior; it will be evident to any-
one who cares to watch animal or man carry out a learned re-
sponse. With long overpractice it may seem at times that the
subject has settled down to the use of a specific muscular pat-
tern, an exact duplication of his movements from one trial to the
next. When this happens, however, it will rarely be difficult for
an experimenter to change that pattern and still observe ap-
proximately the same end effect. (Timing, however, may be
disturbed.)
In general, the variability of motor pattern in a hungry ani-
mal's reaching food becomes comprehensible if it is supposed
that the behavior is determined both by an expectancy of the
immediate result of each movement and by a continual percep-
tion by the animal of his own locus and posture at each stage
of his response. The problem has sometimes been discussed as
though the motor variability occurred despite a constant set of
stimulating conditions. Sometimes, indeed, it may, when there
are two or more equally efficient modes of solution available to
the animal. But the variability characteristically appears when
155 Motor Equivalence
there is a variation of the animal's initial posture, of his locus
with respect to the goal, or of intervening barriers.
The animal must perceive his own posture and his place in
relation to the goal; and his behavior must be affected by this
perception as much as by his anticipation, expectancy, or fore-
sight of the state of affairs in which food has been found in the
past (remember that we are discussing the variability that oc-
curs after learning is established). The monkey that must lift
the lid of a box to get food has an expectancy of the lid's rising
in contact with his hand; and he knows where his hand is at
the moment. The question then is, what neural mechanism will
account for a movement of the hand direct to the edge of the
lid?
Here we have an example of motor variability in a learned
response that is about as simple as can be. The muscular pat-
tern of response, when the hand moves direct to the lid, is obvi-
ously different for each initial position of the hand. We cannot,
I believe, accept the solution of this problem proposed by Hull
(1934), in his concept of the habit-family hierarchy, for he
assumes that the final muscular event in the sequence is con-
stant, though earlier ones can vary. Here even the final pattern
of contractions is variable, depending on the monkey's position
with respect to the box; and we know that the monkey will open
the box despite changes of distance from the retaining cage or
changes of the box's orientation.
But we can follow the general line of Hull's argument and
assume (I believe, necessarily) that an earlier generalized
learning of infancy provides the basis of response, together with
an expectancy by the animal of his own later movements. The
only thing that seems common to the various responses is a per-
ception of the hand in final contact with the lid of the box. In
the normal primate, this perception certainly involves vision.
I assume thus that the movement of the hand primarily follows
an eyemovement that fixates the point of contact. Part of the
learning of infancy must establish an ability to move either hand
or foot directly from any point to the line of vision. Given this,
the behavior is explained, my earlier schema of neural action
having provided for eyemovement in perception.
156 Processes Related to Learning
The schema has also shown how the cortical process control-
ling eyemovement might be subliminal for actual movement and
yet have a facilitating action on other cortical processes. The
first learning involving manipulation, in the normal primate,
would be accompanied by eyemovements; but later, in man at
least, there is a capacity to move the hand to the point one only
"thinks" of looking at. In this case it is assumed that the motor
process is subliminal for eyemovement, but still with an inte-
grating function for other cortical processes. My account is
explicitly in terms of visual processes, but I will add that the
arguments applied (in the earlier schema) to eyemovement
might as easily have been applied to movement of the hands; so
this does not mean that the variable motor behavior of the con-
genitally blind is unaccounted for, or different in principle from
that of the seeing. Further, there is plenty of evidence to show
that the space perception of the congenitally blind is radically
defective (Senden, 1932)which justifies an emphasis on the
role of vision in the manipulative activity of normal subjects.
In the specific example discussed above, the hungry monkey
might equally raise the lid of the box with either hand, with a
foot, or even with a stick: all that is necessary to account for
the behavior is that the foot, or the end of the stick, be seen in
its present position, that its end position in contact with the lid
be conceived of in advance, and that the monkey have learned
to move it from one position to another presently being looked at.
More complex patterns of manipulation require only a number
of such steps: a continual awareness of the present position of
the hand (or foot), and at each point in its movement an antici-
pation of the next point in space to which it must be moved.
At all times the behavior is under the control of these two
things; the muscular pattern of the half-grown or adult animal
is so variable simply because the subject is not learning now to
make specific movements (as he is earlier, perhaps, when he is
learning to bring the hand to the line of vision), but learning
a relationship, an association, between perceived environmental
events. This makes adult learning primarily perceptual; but, in
accordance with the developmental changes of learning dis-
cussed in the preceding chapter, it still leaves a great emphasis
157 Motor Equivalence
on the motor learning of infancy, which frequently can be re-
garded as purely sensori-motor.
In this section I have tried to present my solution of a
ubiquitous, fundamental psychological problem. It would be
wrong to leave the impression that the solution can be regarded
as complete. The solution says in short that motor learning at
maturity is perceptual learning in the first place. It also says
that a true motor learning of infancy is an essential element in
the equivalence of adult responses. The adult who learns to
operate a slot machine learns, that is, that a quarter's moving
into and falling down a slot will be followed by the appearance
of a package of cigarettes. He learns that the depression of a
certain lever starts the motor on his car. He does not learn to
make a certain series of muscular contractions; but, once the
perceptual relationship between depression of the lever and
starting the motor is learned, an earlier infant learning is enough
to provide for the adult's being able to put a hand or footor
a stickin contact with the lever and depress it. But all this
is in very general terms.
I believe that such conceptions will be a good basis for the
study of motor learning, but it should be emphasized that the
study is still neededthat nothing more has been provided than
a rational approach which may or may not be found adequate
in the long run. The discussion has said nothing about the
changes of skill with practice. What is their nature? Are they
to be comprised in the conception of changes in the cell-assembly
due to an increased frequency of activity? Does the subject
revert to his first clumsiness of performance when he is required
to use a new muscular pattern? Does his perception of the
situation change with practice, or his muscular timingor both?
Above all, is he still responding to the same cues?
We still have a great deal to learn about the modification of
motor performance with practice. It should be said explicitly
that this discussion has not attempted to cover the development
of motor skills involving a precisely timed series of movements,
nor such learning as that demonstrated by Sperry (1947), which
monkeys are capable of and rats are not.
158 Processes Related to Learning
INSIGHT AND HYPOTHESIS
A somewhat different aspect of the problem of learning, and
its phylogenetic correlations, brings us to the question of intel-
ligence or insight in learning, and the related question of
"hypotheses" in animals. These ideas have occasioned a good
deal of dispute, and some confusion, in the psychological litera-
ture.
The analysis of animal intelligence, by controlled experiment,
has been a slow process. Mainly, the trouble has been to make
the transition from an earlier subjective psychology (resorting
freely to the notion of "mind" or conscious awareness as an agent
in behavior) to an objective theory of neural action, without
oversimplifying the facts.
If an animal shows "intelligence" in his learning and problem-
solving (Hobhouse, 1915), "ideation" (Yerkes, 1916), or "insight"
(Kohler, 1925), can the fact be dealt with in a strictly monistic
theory with no hidden thread of animism? It was hard to find
a way of doing so. What psychologists really needed, of course,
was a physiological mechanism of thought to supplement their
frankly physiological treatment of afferent and efferent mecha-
nisms. The failure to suggest any intelligible mechanism of
insight, or ideation, led to two sharply opposed points of view.
One groupin general, those I have called learning theorists-
began by denying that the facts were what they seemed, and
argued that insight must be reducible to simple learning.
The other groupin general, the configurationistsproceeded
by urging the need for a better theory of learning, whether it
could be found at once or not, and continued a search for evi-
dence to show that the need existed; that animals do have ideas,
or insight, or intelligence.
As a result of this undertaking, psychological knowledge made
a considerable advance. I am going to argue that such notions
as insight have a limited value, and that something with more
precision will be needed for the further development of theory;
but it should be recognized that these ideas have led to an im-
proved analysis of animal behavior and its qualitative variations
at different phylogenetic levels.
159 Insight and Hypothesis
Besides Hobhouse, Yerkes, and Kohler, already mentioned,
Lashley (1929a, 1938&) and Krechevsky (1932) should be listed
as making major contributions to the difficult problem of analyz-
ing animal intelligence; and Tinklepaugh (1928), Tolman (1932),
and Cowles and Nissen (1937), as well, for demonstrating ex-
pectancy (a form of ideation) in animals and establishing the
value of this notion in the theory of behavior.
It is generally conceded by now that learning involves more
than the gradual reinforcement of responses originally made at
random. "Sophisticated" learning theory recognizes the fact as
much as configuration theory (Spence, 1938, 1940). Learning
is often discontinuous; error curves show sharp drops without
warning, and the kind of error that is made on one day may be
quite changed on the next. The great disagreement that still
continues in psychology does not concern the existence of some
factor producing these effects, but its nature: how it operates,
and how it is related to learning as such.
Let me make this a little clearer. In the first efforts to be rid
of anthropomorphism in animal psychology, "learning" theory
said something like this: When an animal is faced with a new
problem, we can avoid any reference to intelligence or compre-
hension and describe the cat's attempts to open the door of a
problem box (for example) as at first a series of random move-
ments. Some one of these movements will eventually open the
door, by chance. The tendency to make that movement is then
increased because it was rewarded (the "law of effect").
On each trial, the tendency is further increased. Eventually,
the cat opens the door immediatelynot because of any separate
process of intelligence or understanding, but because a specific
association has been built up. This is the "naive" theory of
learning by trial and error; it is no longer maintained, largely
because of the work of Hobhouse, Yerkes, Kohler, Lashley,
and Krechevsky, together with an important analysis by Adams
(1929) and the later work of Thorndike (1931). But it had its
value. It was a necessary corrective to the vagueness of other
theories, and though in development it has become rather vague
itself it still represents a line of thought that insists, rightly, that
theory must aim at definable conceptions and a clear statement
of its original assumptions.
160 Processes Related to Learning
The configurationist attack on the earlier, "naive" learning
theory proceeded as follows. A search was made for tasks to
set an animal, of just the right degree of difficulty for each spe-
cies to demonstrate the operation of intelligence. The task must
neither be so easy that the animal solves the problem at once,
thus not allowing one to analyze the solution; nor so hard that
the animal fails to solve it except by rote learning in a long series
of trials.
With a problem of such borderline difficulty, the solution may
appear out of a blue sky. There is a period first of fruitless
effort in one direction, or perhaps a series of attempted solu-
tions. Then suddenly there is a complete change in the direc-
tion of effort, and a cleancut solution of the task. This then is
the first criterion of the occurrence of insight. The behavior
cannot be described as a gradual accretion of learning; it is evi-
dent that something has happened in the animal at the moment
of solution. (What happens is another matter.)
The second, and essentially more important criterion, involves
a comparison of species. To the human observer, a monkey's
efforts to get at food may at times appear utterly stupid. It is
one of the great achievements of modern psychology to show
that the stupidity is always a matter of degree, and to make a
beginning at saying what intelligent behavior is, in its various
manifestations. The description of intelligence is a more diffi-
cult task than it sounds; this is one matter in which appearances
are really deceptive. The box-stacking of the chimpanzee, for
example, is a very clumsy performance indeed. The chimpanzee
has no notion of statics and must pile his boxes by trial and error
until finally they will stay piled and bear his weight (Kohler,
1925). But comparison with other species reveals a considerable
intelligence in this performance, nonetheless. The dog is not
low in the intellectual hierarchy, and yet he is unable to see the
possibility of moving even one box to a place where he wants to
climb. Physically, this is possible for him, though piling one
box on another may not be. The failure is thus intellectual.
The dog, again, shows his great superiority to rat or hen in
his skill at passing obstacles, and reaching a goal by the shortest
route (psychologically a more brilliant achievement, more in-
sightful, than it may sound).
161 Insight and Hypothesis
It is often said that most human beings never think, and go
by habit instead of intelligence. The dullest person who can
earn a living in this culture, however, continually functions at
a level well above the chimpanzee'scomplex as the chimpanzee
intelligence isand it becomes nonsense in view of this fact to
suppose that a man could ever get along solely by routine with-
out the continuous operation of whatever it is that we call intel-
ligence. This becomes even clearer when one looks at cases of
aphasia. As we shall see later, aphasia is by no means a mere
loss of speech; the patient has lost something else as well, an
intellectual capacity that no chimpanzee will ever have and one
that every normal person has and uses continuously (even if he
never sits down, puts his chin in his hand, and announces in so
many words that he is about to "think"). Without the study of
insight in animals, we should not have been able to see how
much is involved in ordinary, everyday behavior that seems ele-
mental in its simplicity but is not.
There is an unsuspected degree of intelligence in even the
simplest behavior of normal man. Krechevsky (1932) estab-
lished a similar point for the rat, showing that apparently simple
learning is more than an uncomplicated acquisition of stimulus-
response connections; it involves "hypotheses" as well.
Briefly, what Krechevsky found is this. A hungry rat is given
a repeated choice of two doors, one left and one right, one white
and one black (for example). The white door is sometimes on
the left, sometimes on the right; and food is to be found by
going through one of the doors (the other being locked). With
such a problem, the rat does not choose at random. At first he
is almost certain to persistently try the door on one particular
side, trial after trial; and later, to try the other side in the same
way. Still later, he may alternate, one left, one right, and so
on; or choose always the white door, whichever side it may be
on, until finally one of these systematic attempts at solution is
consistently rewarded. The rat is then said to have had a series
of hypotheses, and it is clear that his learning is not a gradual
strengthening of responses that were at first made at random.
The behavior is stupid enough, by primate standards, but it in-
volves intelligence of a kind and it throws a good deal of light
on the behavior of higher species.
162 Processes Related to Learning
DIFFICULTIES OF THE CONCEPTION OF INSIGHT
So much for the value that psychology has drawn from the
conceptions of insight and hypothesis (the latter evidently a
form of insight itself). We can next ask in what respect such
conceptions are unsatisfactory, and the inquiry will lead us to
recognize a complementary value in the "learning" theory that
has been consistently opposed to the theory of insight.
From the first the meaning of insight and hypothesis, as exact
scientific terms, has been left confused. Their connotations ob-
viously are anthropomorphic and unfortunate. This might have
been less important if the terms had been well defined, but they
were not.
"Hypothesis" has been defined as a form of behavior: a set of
responses that deviates systematically from a chance perform-
ance. This will not do at all, as Prentice (1946) has pointed
out; it confuses cause and effect. An hypothesis, it becomes
evident, is something postulated, hypothetical, inferred from
behavior.
Psychology has sometimes in the past been too free in postu-
lating special processes to explain behavior. Now it is doing
penance by going to another extreme, scientifically just as sinful.
This current misconception of the scientific method makes any
hypothetical entity somehow disreputable. Psychologists now
seem to feel that it is risky indeed to depart from statements of
fact; nothing should be mentioned that is not "operationally"
demonstrable. Thus in another field we find emotion defined as
a bodily reaction, which it is not; or neurosis as a set of neurotic
symptoms. At the same time, the context in which such defini-
tions appear makes it obvious that the writer has in mind some-
thing inferred from behavior, not the behavior itself, just as the
rat's hypothesis is inferred. It is completely necessary to keep
out of such confusion (in the first place; and in the second place,
to realize that neither operationalism nor logical positivism really
discourages inference or postulate but only seeks the rules by
which inference is to be made).
Insight, again, has on occasion been defined as a sudden
change in behavior, or as a conscious experience that one may
163 Difficulties of Insight
have in solving a problem. Neither definition gets at the real
meaning, which is that of a causal factor in behavior.
When this is clear, it becomes clear also that there is no need
to say that a rat has an hypothesis only when he makes a series
of systematic choices. The hypothesis might endure only for
one trial, or part of a trial; though it could not then be directly
detected, it would still exist. In the same way, it is not neces-
sary to limit insight to those occasions such as when a chim-
panzee, at first failing to reach a banana outside his cage, sud-
denly turns away, looks for a stick, and rakes in the food. These
are the occasions on which it is evident that something has
changed in the animal and has led to a solution. But if it is
insight that solves a problem following error, it is no less insight
when an animal makes no errors and sees the solution at once.
When the solution is seen at once, there must be at least as much
insight as when the solution is delayed.
In this sense "insight" is fully equivalent to "intelligence"
the term that Hobhouse used. The term insight therefore seems
to have two references, one to an enduring factor in behavior,
the other to a variable one. Both "insight" and "intelligence" are
probably unsatisfactory termsit will be seen in the following
chapter that they may be hard to distinguish from some uses of
"motivation" except as different aspects of the same thingbut
it may be suggested that, if they are to be used, "intelligence"
should refer to the enduring aptitude, the existence of a causal
factor in problem-solving that may or may not lead to the solu-
tion of a particular problem; "insight" to refer to its operation
in the particular instance.
There is still a difference of opinion as to what this causal fac-
tor in behavior is, and how it operates: is insight or hypothesis
or, in the broadest terms, intelligencesomething distinct from
the mechanism of association? And here, I think, we must take
a long step toward the position of those theorists who originally
denied the existence of insight but who now, like Hunter (1934)
or Spence (1940), are concerned to show that it can be derived
from a theory of learning.
Whatever insight is, we now know that it continually affects
the learning of the adult mammal. This has been maintained
by those who have emphasized the value of insight as a theo-
164 Processes Related to Learning
retical conception; if we grant it, the conception does not be-
come more valuable for theory, but less.
We find at once the need for more precise conceptions. The
original value of the notion of insight lay in the distinction be-
tween intelligent and rote learning. Now all behavior involves
intelligence, and the distinction is a matter of degree: how much
insight (or intelligence) is present? Further, if all learning is
insightful we can no longer speak of insight as if it were a sepa-
rate process and distinct from learning.
The conception of insight was a valuable first approximation,
but a first approximation only. Insight is not distinct from at-
tention or expectancy (which are also first approximations), just
as it is not wholly separate from rote learning. Postulate insight
to account for the discontinuity of a sudden improvement in
learning, and sooner or later we shall have to add other postu-
lates (temporary forgetfulness, frustration, loss of interest, and
so on) to account for the regressions and disruptions that also
occur in learning.
It has already been seen how insight, as a sudden perception
of new relationships, can result from the simultaneous activity
of two conceptual cycles in adult learning (pp. 1334). This
need not be elaborated further; the new relationship is a change
of the phase sequence, and each phase sequence has its distinc-
tive motor facilitation, so the change would be evident in a dis-
continuity of behaviorthe actual evidence from which the in-
sight is inferred.
Apart from the special case of a sudden solution following
failure, insight is essentially the phase sequence itself, the au-
tonomous central factor in behavior. It is that which makes
behavior directed, instead of random; the mechanism that is
both anticipatory, adjusting behavior to an obstacle before the
obstacle is reached, and recurrent, so that the chimpanzee, try-
ing to reach a banana, remembers the stick he saw a moment
before, goes and gets it, and rakes in the fruit.
Phylogenetic differences of insight are thus the differences in
the complexity of phase sequence that is possible in the brain
of a higher or lower species. The lack of insight in the earliest
learning of primates becomes comprehensible, since the phase
sequence must first be organized.
The Relation of Instinct to Learning 165
Also comprehensible is the slowness with which insightful
learning develops in these species, by the very mechanism that
makes their ultimate level of behavior more insightful than any
other. The larger association cortex of primates, both in abso-
lute terms and in A/S ratio (p. 124), means a slower establish-
ment of sensory control, and so a longer time to set up the phase
sequence and insight. At the same time, in a larger association
cortex the number of assemblies of cells that can function as
separate closed systems will be larger. This means that the
higher animal at maturity will see any particular situation in a
larger number of ways. The comparatively small sensory pro-
jection means also that the phase sequence can escape from
complete sensory control more frequently, and for longer pe-
riods, so that the insightful behavior of higher species is less a
response to the immediate environment and more an integration
of serial stimulation from situations that cannot be surveyed as
a whole.
Finally, all this explains how insight depends on experience
and yet is not a simple, direct result of learning, or the operation
of specific habits (Birch, 1945). The perception of new relation-
ships discussed above is usually not delayed, but immediate; the
animal put into a new situation will perceive its parts as familiar,
and as related in a meaningful way (that is, by the mechanisms
discussed on pp. 130-4). The behavior is then insightful, as an
organized set of responses to a new total situation. An organi-
zation that consists of a new temporal combination of elements
which themselves were organized by experience obviously de-
pends on learning, but is not learned. Heredity sets a limit to
the amount of development of the phase sequence and deter-
mines the principles of development, but experience is necessary
as well.
THE RELATION OF INSTINCT TO LEARNING
The problem of instinct is the correlative of that of intelli-
gence, or insight, and of learning. It has just been seen that
intelligence is not an entity that is quite distinct from learning;
and we may now see that instinct, also, is not to be cut off
sharply from either.
166 Processes Related to Learning
The term "instinctive" will be used here to refer to behavior,
other than reflex, in which innate factors play a predominant
part. Empirically, this is behavior in which the motor pattern
is variable but with an end result that is predictable from a
knowledge of the species, without knowing the history of the
individual animal. This class of behavior must be recognized,
and we are no farther forward by coining a new name for it.
"Instinct," however, is less useful. It suggests a special agent
in behavior, just as "intelligence," "insight," or "drive" may.
Essentially, we are involved here in the difficulties of the
constitutional-experiential dichotomy that are discussed later
(Chapter 10). We must distinguish, conceptually, the consti-
tutional factor in behavior from the experiential, but there is pre-
sumably no mammalian behavior that is uninfluenced either by
learning, or by the constitution that makes some learning easy
or inevitable. We need not consider (1) that instinctive be-
havior is sharply distinct from other behavior, or (2 ) that it in-
volves some special process that is radically different from those
involved in other behavior. Ultimately, our aim must be to find
out how the same fundamental neural principles determine all
behavior.
There is a means at hand for attempting this. Consider first
Lashley's (1938a) useful distinction between reflex and instinct.
A reflex may be defined as innately determined behavior that is
controlled by stimulation of a specific group of receptors and
takes the form of a predictable set of muscular contractions. In-
stinctive behavior, on the other hand, is more than a complica-
tion of reflexes, although of course it involves reflex elements.
It is not always controlled by a specific stimulation, and may
in fact be aroused by a perceptual lack ("reaction to deficit,"
Lashley, 19380; "vacuum activity," Lorenz, cited by Tinbergen,
1942). Instinctive behavior does not consist of a predetermined
sequence of muscular contractions, and yet it has a constant and
predictable end result. The spider of a given species will spin
a web highly specific in design, though the movements necessary
to produce it vary with the distance of the objects to which it is
attached; the kind of movements and number of trips made by
a bird in building a nest (in those cases where the design can
be foretold from the bird's species) depends on the kind of ma-
The Relation of Instinct to Learning 167
terial used and the shape and size of the crotch or ledge that
supports it; and so on.
All this means that the behavior, though it cannot be called
reflex, is still continually under sensory influence, more or less
direct. At each stage in the construction the muscular activity
varies with circumstances and is such as to produce a certain
perceptual effect. This indirect sensory control is demonstrated
whenever accident destroys part of the structure. The muscular
activity then does not continue in a predetermined sequence, but
reverts to an earlier stage, in such a way as to restore the missing
part. Since the behavior is continually responsive to such
events, it must be under afferent influence throughout.
Now we have already seen that the A/S ratio, or proportion
of association-area cortex to sensory cortex, affects the direct-
ness of sensory control over behavior and the promptness with
which it can be established. In order to make this relationship
meaningful for lower forms in which the cerebral cortex is absent
or insignificant, let me speak here of the proportion of inter-
nuncial to afferent cells (association/sensory cortex being only a
particular case of this more general relation). A massive affer-
ent system and a negligibly small internuncial one implies an im-
mediate and direct sensory control of behavior, and one in which
any particular sensory constellation would tend to have a single
effect on behavior. The larger the internuncial system, the
more slowly can a sensory control be set up. A sensory control
that is established only with experience, however, means that
there is an element of learning in the subsequent behaviorand
so an increased variability. Thus, with phylogenesis, the pre-
dictability of instinctive behavior would steadily decrease as the
ratio of afferent to internuncial tissues becomes smaller.
This does not of course explain why the instinctive behavior
of any one species takes the particular form it does; but at this
point we do have an explanation for the fact that instinct is most
evident in lower species, in which internuncial structures are
small, both absolutely and relative to afferent ones. Before
specific instincts can be explained, much more must be known
about the behavior in question and the finer nervous structures
that determine it. It is a reasonable expectation that the knowl-
168 Processes Related to Learning
edge will be obtained, and that the explanations when they are
found may be much simpler than one would suppose.
To show that the expectation is not chimerical, that this
does not simply dodge a crucial difficulty for the present theory
of behavior, let me cite the earlier discussion of the "spatial
hypothesis" or position habit. The position habit of course is
not called instinctive by psychologists, presumably because it
only shows up during the course of learning; and learning and
instinct are supposed to be incompatible. But it is so completely
predictable in the rat that it might well be called instinctive.
Before a means was found for analyzing the behavior (Hebb,
1938a, 1938&; Hebb and Williams, 1941) it almost appeared that
the rat had a thoroughly mysterious capacity for perceiving
compass direction. It turns out, however, that his behavior is
simply response to the most constant stimulation of his environ-
ment (pp. 91-5).
Eventually, explanation at a similar level can be expected for
other species-predictableor instinctivebehavior. It is almost
certain that the mysteriousness of the processes determining in-
stinct is generally overrated; when analysis has finally suc-
ceeded, the process turns out to be quite different from what it
appeared at first (Schneirla, 1948). The nest-building of the
pregnant rat may be the simple result of a disturbed tempera-
ture control (Kinder, 1927). The visual preference of the honey-
bee for complex flower forms appears to come from the fact that
these forms produce a greater flicker frequency in the bee's eye
(Wolf and Zerrahn-Wolf, 1937) and thus have a maximal stimu-
lating value. The preference is not at all an innately deter-
mined perception of a particular pattern as distinct from other
patterns. There is still an innate factor, but a much simpler one
than it seemed at first to be, and the evidence does not require
the assumption that some special and complex process is in-
volved in the behavior. Another example is the apparently pur-
poseful behavior of the newly hatched loggerhead turtle in
going directly to the sea (Daniel and Smith, 1947), which, how-
ever, is phototropically determined.
I have already suggested that "instinct" may fundamentally
involve some learning even at lower phylogenetic levels, if the
kind of learning is strictly limited and if it may occur in a matter
The Relation of Instinct to Learning 169
of seconds (pp. 113, 115). This is quite speculative at present;
but one thing is clear, at least, that there is no evidence to
justify the a priori assumption that the control of learned and
the control of instinctive behavior are quite separate processes
(Beach, 1937, 1939). In mammalian behavior, certainly, we
cannot separate out one group of responses that are innate,
others that are learned. Constitution (the factor of heredity)
may dominate in determining some properties of behavior, expe-
rience (the factor of learning) in others; but there must be a
gradation from the one type to the other.
In the same way, instinct is not a separate process from intel-
ligence or insight. It is intelligence, or insight, that is innately
limited in variety. An animal low in the scale must see a par-
ticular situation in a particular way because of the overwhelm-
ing mastery of afferent over internuncial activity, whereas a
higher animal with a more variable association-area activity,
under imperfect sensory control, may see it in a number of ways
and respond accordingly: now to one property of the stimula-
tion, now to another. But the mechanism of response is not
different in the two cases, except in complexity.
This should be more explicit in one respect. It has been
argued that expectancy is an essential explanation of motor varia-
bility in mammalian behavior. At the same time, it must be
clear that instinctive behavior does not involve an expectancy of
its remote biological results (this would be the worst sort of
teleology).
I have said that instinct and intelligence differ in complexity
of mechanism; and one of these differences is in the degree of
anticipatory central action that is called expectancy, and in the
temporal extent that expectancy can span. At the lowest levels
of instinctive behavior there can be no expectancy whatever,
where the internuncial activity is completely under sensory con-
trol, without intermission. At intermediate levels, there may be
an expectancy of the immediate consequences of the single act
in an instinctive sequence (the bird may have an anticipatory
concept of a twig placed in a certain part of the nest, before it
is actually laid there; the rat whose pups have been scattered
may possibly retrieve them under the influence of an expectancy
of perceiving them in the nest again). Even this degree of
170 Processes Related to Learning
expectancy, however, involves learning; expectancy must be an
association between a present state of affairs and one that has
followed that state of affairs (or similar ones) in the past. The
temporal span of expectancy in instinctive behavior must vary,
from zero to the human expectancy of pregnancy and birth nine
months after copulation (and the difficulties of rearing a child
for twenty years after that). To the extent that expectancy oc-
curs it must represent an influence of learning, not an inborn
knowledge of the future.
Problems of Motivation: 8.
Pain and Hunger
This chapter and the next aim at a broader picture of the
problem of motivation than is usually drawn when neural mecha-
nisms are to be discussed. The biologically primitive hunger,
pain, and sex drives are important in behavior, but these are not
all of motivation. In particular, there is an essential problem in
play, and in the changes of motivation that occur when bodily
condition (including humoral factors), and the environment in
all important respects, seem unchanged.
In the present chapter, motivation is dealt with first as a theo-
retical conception, particularly in relation to the theory of learn-
ing; we can then try to see how the more general ideas that re-
sult from this examination are to be applied to the specific prob-
lems of pain and hunger. The following chapter will be an
attempt to round out the picture by treating the spontaneous
variations of motivation that appear in such different aspects of
behavior as play and sleep.
THE CONCEPT OF MOTIVATION
The term motivation concerns two distinct questions: (1) why
is an animal active at all, and (2) why does the activity take a
particular form? Ostensibly, (1) is the fundamental question.
Actually, it is (2) that modern psychology is usually concerned
withthough the writer who uses the term often seems to be
talking about (1).
We have seen (Chapter 1) that early experiments with nerve-
muscle preparations made it easy to think of the nervous system
as inert, unless stirred up by some outside irritant. It then
171
172 Problems of Motivation: Pain and Hunger
seemed necessary to look to sensory stimulation for the cause of
any muscular contraction. Now, however, it is known that the
central nervous system is continuously active, and the necessity
has disappeared, as long as we do not ask about the patterning,
the timing, of muscular contractions.
In fact, most of the musculature itself is continuously active
(for whatever reason), and the chief problem the psychologist
is concerned with, when he speaks of motivation, is not an
arousal of activity but its patterning and direction. Little time
is spent discussing the source of an animal's tonic contractions
and random movements, in the first stage of a learning experi-
ment. What we have usually wanted to know is why the random
movements change and become directed. The main question
for psychology concerns the timing of muscular contractions, not
their arousal.
There seems still to be some confusion on this point. Psy-
chological theory sometimes seems to be still looking for some
extraneural irritant to account for maternal behavior, for ex-
ample. In an earlier day, it was thought that maternal behavior
must be aroused by the sensory stimulation of distended breasts,
male sexual behavior by distention of the seminal vesicles; the
search for food or water by stomach contractions or dryness of
the throat; and so on. Without exception, these ideas have
broken down. Desire to drink can occur when the throat is
moist or anesthetized, eating continues after the stomach has
been surgically removed. This point has been adequately dis-
cussed by Morgan (1943). The older manner of thinking per-
sisted, however. As contradictions accumulated, a tendency
appeared instead to postulate "central receptors" instead of
peripheral ones as motivating agents. By this device, it is still
possible to discuss motivation as a stimulation that arouses ac-
tivity of a particular kind.*
* The implied distinction between theories of motivation does not merely
concern use of the word "stimulation." It is more than a question of defi-
nition. The sex hormones, for example, are known to have a primary action
on the central nervous system. This action might be either (1) by excit-
ing specific cellsin which case it would be meaningful to speak of a
"central stimulus"or (2) by changing the excitability or frequency prop-
173 The Concept of Motivation
It has frequently been proposed that all behavior, and all
learning as well, must be motivated. The proposition may be
taken in either of two ways. First, we can treat it as a funda-
mental assumption. We then infer the existence of a motiva-
tion, known in no other way, when an apparently unmotivated
act is seen. (When the rat reaches food and, before eating, goes
off to some other part of the maze, he must have an exploratory
drive or motivation. He has a hoarding motivation when he
gathers pellets of food without eating them.) Such a use of
assumption is often legitimate; what it really does is define moti-
vation or drive, and such definitions have been an essential part
of the scientific method. But here the term motivation becomes
so broad, when all the deviations of behavior are considered, as
to have little usefulness, except as a convenient reference to the
fact that behavior in general is directed and organized. The
fate of the term becomes that of insight when all learning is said
to be insightful.
Or, secondly, one can define motivation independently, as an
extraneural bodily condition or an irritant that acts on specific
neural cells. The proposition that all behavior and learning are
motivated then becomes untrue, on the existing evidence, and
with it the proposition that learning is directed only by moti-
vation.
And yet there is an important problem here. In its simplest
terms, the question that has bothered psychologists is this: An
animal learning to solve a problem makes some right move-
ments, some wrong. What is it that reinforces the right ones
and discourages the wrong? Why are the right ones remem-
bered in later trials, the rest forgotten? This has been a persist-
ently baffling problem; and its solution is essential to the theory
of behavior.
The difficulty may seem trivial. If so, however, the reader
may be begging the question by saying to himself, in a common-
sense (and animistic) way, that the animal of course can see that
one of his movements has had the right effect, others not. But
if the question is asked, What neural processes constitute the
erties of a system of cells, without being able to induce activity in the
system directly. If the term "stimulus" is to have a determinate meaning, it
should not perhaps be overextended to include the second reference (2).
174 Problems of Motivation: Pain and Hunger
"seeing"? the solution is evidently as far off as ever. The sim-
plicity of the question of reinforcement, or selective retention of
some responses and not others, is like the simplicity of the fun-
damental postulates of mathematics that are taking centuries to
ravel out. A simple common-sense answer has no value what-
ever; the first thing one has to do is get away from the animism
that saturates common sense. Having done that, there remains
an uncommonly difficult scientific problem.
What it involves will perhaps be seen best in showing its re-
lationship to Pavlov's theory of conditioned reflexes. Pavlov's
failure to solve the problem is one of the decisive reasons why
psychologists have had to abandon or change the theory.
THE THEORY OF CONDITIONED REFLEXES
Pavlov (1927, 1928) formulated a simple rule for the occur-
rence of learning. Any stimulus that acts repeatedly at the
same time as a response will form a connection between the
cortical cells involved. Subsequently, the stimulus will be suffi-
cient to arouse the response. (The classic example is a bell rung
at feeding time: after a number of trials, ringing the bell alone
makes the animal's mouth water.) The original or uncondi-
tioned stimulus (food in the mouth) has an innate relationship
to the response. The substitute or conditioned stimulus (the
sound of the bell) has an acquired relationship. The relation-
ship is supposed to be set up in connections from sensory to
motor area in the cortex.
This assumption is the keystone of the theory. The theory as
a whole is of course much more elaborate, with a number of
other assumptions, but I have not oversimplified the point under
discussion. An afferent activity at one point in the cortex, and
an efferent activity at another point, at the same time or imme-
diately after the afferent one, is all that is needed (with enough
repetitions) to establish a sensori-motor connection; and all be-
havior is determined by such connections, innate or conditioned.
This means that all acts at the time of a particular stimulation
would be associated with it; and the act that has most frequently
occurred would be most strongly associated. Now in problem-
solving, or any learning situation in which a number of acts are
The Theory of Conditioned Reflexes 175
possible but only one rewarded with food, "mistakes" occur fre-
quently. Often, in the first trials of maze learning, the same
mistake is repeated much more frequently than the "correct"
behavior. Yet the net effect is that mistakes vanish, and leave
only the correct responses. Instead of being associated, the act
that occurred frequently is forgotten. Its elimination, also, can
hardly be ascribed to an inhibition. Pavlov defined inhibition
as producing a cessation of activity and leading directly to sleep.
But repeated mistakes do not tend to make the. animal drowsy
or motionless. They are often followed by an increased activity.
Maze learning, however, is not a very clearcut instance. For
the theorist, a more crucial argument has been found with
Pavlov's own procedure, and minor variations of it. This argu-
ment is in the fact that the conditioned stimulus does not evoke
the unconditioned response, as Pavlov implied (Zener, 1937; Hil-
gard and Marquis, 1940).
The conditioned response is not a duplicate of the uncondi-
tioned, but something new. All the elements of the uncondi-
tioned response may be there, but if so some of them are mini-
mal, and the total pattern is significantly changed. When the
dog hears the bell, he does not usually make eating move-
ments, with his nose in the food dish, until the food itself ap-
pears. Conditioned and unconditioned stimuli, then, are not
just two ways of arousing the same combination of effectors.
If the objection is made that no one would expect the dog to
chew and swallow until food shows up, the answer is that Pav-
lov's theory says he will. The conditioned stimulus is supposed
to set up connections with any following effector activity; and
eating movements follow the conditioned stimulus as much as
the secretion of saliva does. If the critic of this argument points
out that a dog is too intelligent to respond to a bell in such a
way, he is really criticizing the original theory. A separate in-
telligence, to tell the dog when his conditioned reflexes may
operate, would be a deus ex machina. One may suppose that
the dog is intelligent, but not that his intelligence is separate
from his neural activity; and we are examining the theory that
all higher neural activity consists only of conditioned reflexes.*
* The physiological and clinical literature suggests that a number of
workers still feel that Pavlov's conclusions are a factual statement of what
176 Problems of Motivation: Pain and Hunger
MOTIVATION IN CONDITIONED-REFLEX THEORY
Now let us see whether some principle of motivation can be
added to the theory and thus make it able to account for the
facts. Hull (1943) has made this addition explicitly to show
why "right" responses become conditioned, others not. Hull
proposes that those acts that are immediately followed by a
"need reduction" are retained; and he believes that with this
change Pavlov's main structure can stand, successfully reducing
behavior to a set of conditioned reflexes. That is, Hull has in-
troduced motivation as a supplementary principle. The idea
was used earlier by Thorndike, who called it the law of effect.
The law of effect is roughly this. An act that is closely fol-
lowed by satisfaction, pleasure, or comfort will be learned; one
that is followed by discomfort, forgotten or eliminated. Partly
because of the result of subsequent experiments, Thorndike's
(1931) later treatment is not in terms of pleasure and pain but
refers to a very vaguely defined condition of "belonging" as the
factor that makes for learning. The response that will be re-
is known to go on in the cerebrum. It is important to see the difference
between the facts he demonstrated and the theory based on the facts.
Pavlov has deservedly had a great influence on psychology, and his theory
has not been rejected because it is too physiological but because it does
not agree with experiment. Apparently, Pavlov isolated himself from the
contemporary literature, and his theory took no account of psychological
discussions after 1900. Criticism at present is based on later experiments,
physiological as well as psychological.
The main points can be listed briefly: (1) Simultaneous sensory and
motor activity is not enough to establish a conditioned reflex when the
motor activity is directly aroused by stimulation of motor cortex (Loucks,
1935, 1938). (2) The data published by Pavlov do not agree with his
theory of a slow irradiation of excitation across the cortex (Loucks); and
the strychninization method reveals a kind of cortical organization quite
unlike what Pavlov assumed (Hilgard and Marquis, 1940). (3) Irradia-
tion, by which Pavlov accounted for stimulus generalization, does not pro-
vide for generalization of stimulus intensity (Hovland, 1937; Hilgard and
Marquis), and'the theory also has considerable trouble with the kind of
pattern generalization illustrated in figure 1. (4) Finally, it does not seem
possible to translate Pavlov's system into terms of problem-solving or serial
learning without serious modification. This is the point discussed in the
text.
177 Motivation in CR Theory
membered and repeated is one that "belongs" in the total situa-
tion. Unfortunately for the older and simpler law, punishment
by electric shock for the right response may help the subject to
learn as much as punishing wrong ones. Mild punishment for
either may be better than no punishment, whether with animal
or human subjects (Muenzinger, 1934; Drew, 1938; McGeoch,
1942). It is therefore not possible to say that discomfort always
acts to eliminate a response. The vast literature on this issue,
and the facts that led Thorndike along with many others to
abandon the original law of effect, have been surveyed by Post-
man (1947).
To substitute belongingness for comfort and discomfort makes
the law of effect extremely vague and says nothing about physio-
logical processes. The theory presented here can be regarded
as an attempt to fill in the gap left by Thorndike's treatment of
this point. In psychological approach his ideas are in many
ways identical with those that were outlined earlier (Chapter 6).
They constitute, that is, a theory of association, as distinct from
a theory of conditioned responses, in which an association may
be between autonomous central processes instead of between
afferent and efferent processes.
Hull (1943) has attempted to deal with the problem in another
way, essentially by going back to the earlier law of effect with-
out the vagueness of Thorndike's later formulation. Unfortu-
nately, he has done this by retaining the idea of motivation as
stimulation, and of all learning as consisting fundamentally of
sensori-motor connections.
"Effect" becomes a reduction of need, and a need is defined as
follows: "When any of the commodities or conditions necessary
for individual or species survival are lacking, or when they de-
viate materially from the optimum, a state of primary need is
said to exist" (Hull, 1943, p. 17). Further, the need affects be-
havior by stimulating special receptors, either peripheral or in
the brain itself. (Hull actually starts out by saying that they
"probably" do [p. 18] but thereafter takes the point for granted;
his argument and diagrams deal with stimulations of particular
hypothetical cells.)
What the theory says in short is this. All behavior is deter-
mined by sensory stimulations. All learning consists of form-
178 Problems of Motivation: Pain and Hunger
ing or strengthening S-R connections. An S-R connection is
strengthened whenever the stimulus, S, and the response, R,
occur together and some need stimulation, JV, decreases at the
same time. The S-R connections amount to conditioned reflexes.
What Hull has done, in this modification of Pavlov's ideas, is to
postulate a special class of stimulus that (1) must be present
before learning can occur, and (2) must then decline. The de-
cline, and the decline only, determines the occurrence of new
connections. No way is suggested as to how the decrease of
stimulation promotes the formation of connections.
To Hull's system as a whole vigorous objections have been
made by other psychologists. Its weakest point, and clearest
departure from the facts, is in the treatment of motivation as
biological need.
In the first place, the apparently clear and precise definition
of a need will not hold water. A need occurs "when any of the
commodities or conditions for individual or species survival are
missing." One of these needs is "the occasional presence and
specialized reciprocal behavior of a mate." But the absence of
a mate cannot excite sensory receptors. By this definition, evi-
dently, another need is for testes in the male animal and simul-
taneously ovaries and uterus in the female. Species survival is
not possible otherwise. But lack of testes is not usually a moti-
vating condition, nor can it excite receptors. The injection of
male hormone would not affect the needs of the animal, as de-
fined, since he is still incapable of fertilizing a female and con-
tributing to survival of the species. Yet his behavior sexually is
that of a normal animal once the injection has been made.
The preciseness with which "need" has been defined is illusory.
It is a biological statement that does not coincide with the psy-
chological conception Hull had in mind when he wrote. First,
then, it must appear that Hull's reformulation of the law of effect
is not the precise postulate that it was meant to be. A system
of deductions that makes repeated use of it may have a funda-
mental flaw.
Secondly, the law of reinforcement, as it is now called, is open
to the same criticism that made Thorndike abandon the law of
effect in its earlier form. As we have seen, pain does not always
act to eliminate a response. Incidental learning, unaccompanied
179 Motivation in CR Theory
by reward or the avoidance of punishment, does occur. Accord-
ing to the theory, the rat in the maze should learn nothing about
it until one of his responses is accompanied by a decrease of
hunger or thirst, or escape from electric shock, or some similar
reward. In actual fact, when he is allowed to run in the maze
without reward or punishment the rat learns a good deal about
it. This is shown by the small number of errors he makes later,
when he is put into the maze hungry and allowed to find food
there (Blodgett, 1929, cited by Hilgard and Marquis, 1940).
The third point is the extreme narrowness of the range within
which needs as Hull has defined them must operate. This is a
return to the old idea that the only sources of action are such
things as hunger, thirst, pain and concupiscence, with mother-
love thrown in for good measure. Stimulation that is frequently
associated with one of these is supposed to acquire the same
kind of influence. It is not denied, therefore, that man or animal
may act without prospect of primary reward. But according to
the theory these secondary motivations can be set up only by
repeated coincidence with a primary one, and the connection
between the two must be frequently reinforced thereafter.
Emphasis on biological needs seems to limit animal motiva-
tion much too narrowly. It excludes conditions that clearly in-
fluence behavior and learning (Allport, 1946, p. 342), and it
may have the unfortunate effect of preventing the student who
takes the hypothesis seriously from seeing many of the facts
of behavior. For example, children spontaneously avoid dark
places even though no unpleasant event has been associated
with darkness (Chapter 10). They can learn what to do to
make their parents come into the bedroom and turn on the light.
But the "need" here is for more stimulation, not less. It is the
theoretical preoccupation with hunger, thirst, sex, and avoidance
of pain, as sources of motivation, that has led many child psy-
chologists to deny that there is any emotional response aroused
primarily by perceptual lack, such as darkness. Nevertheless,
the evidence is strong that such responses occur.
The point has been labored enough. As far as one can see at
present, it is unsatisfactory to equate motivation with biological
need. Theory built on this base has a definiteness that is very
attractive; but it may have been obtained at too great cost. To
180 Problems of Motivation: Pain and Hunger
follow this line seems to require either a persistent disregard of
facts such as those referred to above or the continual addition
of new assumptions as soon as one gets beyond the subject
matter of the classical experiments on rote learning in animals.
It is clear of course that the primitive drives of pain, hunger,
and sex are often of overwhelming importance. We need an
approach to motivation that neither minimizes these things nor
fails to provide for the unrewarded learning that also occurs
when the animal's belly is full and his sex drive satiated.
AN ALTERNATIVE TREATMENT OF MOTIVATION
The alternative treatment of motivation that is required by
the present theory is here introduced in general terms and is
made more precise in the following discussion of pain and
hunger. This will still have some of the essential points of con-
ditioning theory as developed originally by Pavlov and later by
Hull. The differentiation of two stages or levels of learning
(first, the assembly of cerebral cells, and second the establish-
ment of connections between assemblies) makes it possible, per-
haps, to treat motivation in a Pavlovian way without inserting
a special postulate in the theoretical structure. Pavlov neces-
sarily thought in terms of a sensori-motor conditioning; but, with
the data available from electrophysiology of the central nervous
system, one can conceive of conditioning as the establishment of
relations between cerebral neurons (in the first place) and be-
tween systems of neurons (in the second). The result, as I have
suggested, is a theory of associations; from another point of
view, it is a way of incorporating some of the advantages of asso-
ciation theory into a theory of conditioninga common ground
perhaps for Thorndike (1931) or Woodworth (1938), with Hull
(1943) or Guthrie (1946).
Learning according to the present ideas consists of a lasting
change of facilitations between the activities of specific neural
structures. The change results when two structures (single
pathways or assemblies) that have sufficient anatomical con-
nections are active at the same time. The first learning estab-
lishes an organization of individual transmission units into an
assembly. But later learning, the kind that is usually studied
Pain 181
experimentally, is a change of facilitation between organized
systems, a changed interaction between one complex of assembly
actions and another. In the terminology of Chapters 5 and 6,
learning is a change of relationship between cortical phase se-
quencesor conceptual activitiesbecause of their having oc-
curred together.
It then follows that the experimenter who wants an animal to
learn some particular thing must achieve some control of the
phase sequence. In a single environment, the phase sequence
may be extremely variable. Putting an animal into a particular
situation does not determine what particular conceptual activity
will result. Therefore, it cannot determine that a facilitation
will be established between the conceptual activities the experi-
menter is interested in. When the experimenter takes further
steps to limit the variety of conceptual activity that will occur
in an animal, he sets up a motivation. The term motivation then
refers (1) to the existence of an organized phase sequence, (2)
to its direction or content, and (3) to its persistence in a given
direction, or stability of content.
This definition means that "motivation" is not a distinctive
process, but a reference in another context to the same processes
to which "insight" refers; it also means that the waking, normal
adult animal always has some motivation (because of items 1
and 2 in the definition) though its persistence in any one direc-
tion (3) may not be great.
If a bodily need is found to limit the variability of central
neural action, it can be used experimentally to control learning;
but this of course is quite different from saying that the animal
will learn nothing unless he has first been deprived of food or
injured by electric shock. Without such means of control one
is less able to predict what will be learned in a given situation,
but one would still expect some learning to occur.
Now we have to ask how such bodily conditions can control
conceptual activities. First, pain.
PAIN
The relationship of pain to learning has already been taken
for granted, in the assumption that certain afferent processes act
182 Problems of Motivation: Pain and Hunger
primarily to disrupt central neural activities (p. 150). Disrup-
tive processes are a general classification that includes more
than pain. In most sensory modes there is an intensity limen
at which avoidance appears. Below this point, the stimulation
may be sought outthat is, it is "pleasant"; above it, the same
kind of stimulation produces avoidance and, if the avoidance is
unsuccessful, behavioral disturbance. With different kinds of
stimulation, the avoidance limen falls in a different part of the
intensity range. Pain is not unique, but may seem so partly be-
cause there is a great range of intensity above the limen of initial
avoidance. Pain, therefore, as a particular class of event, is
regarded here as a disturbance originating in the somesthetic
afferent system, though its mechanism may be like that of dis-
turbance originating from certain sounds, smells, or tastes.
It should be understood that we are not mainly concerned here
with the conscious experience of pain, and consequently that the
class of event discussed is not quite coterminous with pain as
discussed in the introspective literature. There is, for example,
what has been called "pleasant pain," which will not be discussed
apart from pointing out that the phenomenon may be exactly
parallel to "thrill" or pleasant fear (p. 233).
I class pain then as a disruptive somesthetic event. The dis-
ruptive action is usually limited, and is mainly at higher levels
in the brain stem or in the cortex. At the spinal level, tissue
injury may arouse only integrated reflex responses (Sherrington,
1906). At the thalamo-cortical level, the disturbance must still
be focalized, since of course pain is referred to specific parts of
the body. Its effects may be widespread, but if the disturbance
becomes really general it is emotion, not pain.
With these assumptions, it has been shown how pain might
control learning by disrupting one phase sequence, leaving
another unmolested. Essentially, this is a limitation of concep-
tual processes in a given environment. Pain then is a means of
channeling motivation, as motivation was defined in the preceding
section. It is not a source of learning, from this point of view,
but limits and directs it. However, pain does have an energizing
effect on behavior, and in some circumstances it appears to fa-
cilitate learning directly. To this possibility I shall return.
Now let us see what the assumptions mean concerning pain
Pain 183
itself. There have been in general two theories of pain (Morgan,
1943). In one, pain is not a special sensory mode but an effect
of overstimulation of receptors for heat, cold, or pressure. In
the other, pain is a sensory mode with its special receptors,
peripheral nerve fibers, and conduction paths in the spinal cord.
Each of these theories has its difficulties; each has something to
recommend it. The two seem to have been thought of as mu-
tually incompatible, and exhaustive: if one is proved wrong, the
other must be right. But following up a suggestion made by
Nafe (1934),* and making a comparatively small change in each
theory, we can find a way of combining them. The combination
has the best features of both, and covers certain facts that may
be equally destructive of either theory singly.
Against the first, peripheral-intensity theory, is the existence
of pain spots in the skin: here a weak stimulation can arouse
pain, although a much stronger stimulation elsewhere does not.
Pain seems also to depend mainly on the activity of C fibers in
the sensory nerve. This in itself denies that intensity is all that
matters in pain; but, even more, intensity in neural terms must
be a matter of impulses per second, and C fibers are the very
ones in which conduction is slowest and which deliver the fewest
number of impulses per second to any one point in the nervous
system.
However, there are strong arguments against the alternative,
sensory-mode theory. If pain is a sensory mode, it is known that
its receptors must be free nerve endings, and free nerve endings
are also known to mediate pressure. There may of course be
two kinds of free nerve ending, histologically indistinguishable.
However, very weak or very brief stimulation of pain spots or
pain fibers produces no pain; as Nafe has pointed out, this has
obliged supporters of the sensory-mode theory to hold that there
may be "nonpainful pain."
Also, it is not at all clear that pain can be permanently abol-
ished by clean surgical section of one or two of the tracts only,
* And see Weddel, Sinclair, and Feindel (1948), whose views in some
respects are identical with those elaborated here. Wortis, Stein, and
Jolliffe (1942) have also anticipated some of this treatment. They sug-
gested that the faster-running impulses in A fibers can prevent C impulses
from traversing certain synapses, so preventing pain.
184 Problems of Motivation: Pain and Hunger
in each side of the spinal cord: there are tracts that are mainly
involved in the production of pain, but are they solely involved
in it? Finally, there is no specialized cortical area for pain as
there is for touch, and apparently no specialized pain nuclei in
the diencephalon or lower brain stem: if specialized tracts for
the conduction of pain do exist, what do they conduct to, other
than the centers of somesthesis?
The suggestion of Nafe, to which I have already referred, is
that pain somehow concerns the patterning of central activities
aroused by the pain stimulus; and with this it seems that the
main contentions of both classical theories can be combined.
Let us forget peripheral intensity, as the essential determinant
of pain, and think instead of a central intensity that could, as
we shall see in a moment, be initiated by a weak stimulus in the
periphery as well as by a strong one. Let us also suppose that
the pain fibers do not inevitably mediate pain; that the slow
impulses, and the central connections of these fibers, are such
that they are peculiarly effective in producing sudden massive
discharges in somesthetic nuclei at thalamic or cortical level,
with a widespread effect. This would be pain. Isolated stimu-
lation of a pain spot then would not necessarily produce pain
but would usually do so.
It is a highly significant fact that pain is often the result of
conditions that must decrease sensory activity, including activity
in the pain fibers or their central connections. As Lashley
(1938c) has observed, pain is often an accompaniment of in-
jury to the somesthetic system all the way from peripheral
nerve to thalamus. In some cases, the pain may be due to irri-
tation; but in many cases there is little basis for the assumption,
if the facts can be explained without it. As the work of Adrian
(1934) and Weiss (1941fl) has shown, an essential function of
normal sensation is a regulation of central firing; without con-
trol, the firing is spasmodic or with a local hypersynchrony; with
it, the cells do not build up to the point of spontaneous dis-
charge but are fired diffusely in a frequently varying pattern.
If higher somesthetic nuclei are deafferented, everything we
know about neural function indicates that the cells must even-
tually fire, and fire more or less in synchrony, with massive jolts
to other nuclei connected with them. When it is supposed that
Pain 185
pain may be a disruptive activity at the cortical level, it must
be supposed also that a conceivable source of pain is a decrease
or absence of peripheral sensory activity, even in pain fibers.
This hypothesis does not specify what kind of disturbance con-
stitutes pain. Evidently there are deviations from normal sen-
sory function, including hypofunctions, that never involve pain;
so the hypothesis is certainly not complete. As for pain in
healthy persons, the predominantly slow impulses in the sensory
nerve may produce pain only (1) because of some peculiarity
in the pattern of thalamic activity they arouse, apart from total
frequency of impulses in a given part of the afferent tract (i.e.,
apart from intensity at this level). Alternatively, pain may occur
(2) because the slow impulses are peculiarly effective in arous-
ing excessive bursts of firing from the thalamus. Either expla-
nation will suit my more general thesis, though the second is the
one I have preferred. But the essential point is this: a theory
of pain must provide for the fact that an increase of pain often
results from hypofunction in afferent structures.
I have already referred to Lashley's observation, that pain
may follow lesions anywhere in the somesthetic system up to
the thalamic level. In Head's (1920) classic experiment on the
regeneration of cutaneous nerve, hyperalgesia was prominent.
According to Boring (1916) the hyperalgesia is not a lowering
of the pain threshold. Thresholds, including that for pain, are
raised; yet the pain when it does occur may be of extreme
severity. Wolff and Hardy (1947, p. 173) make the same point,
concerning patients with thalamic lesions. It seems obvious that
the rise of threshold means a decrease in the total amount of
afferent activity. The pain that is so often felt in the phantom
limb (Riddoch, 1941) also seems an obvious accompaniment of
a decreased sensory activity. After amputation, regenerating
nerves may form neuromas that are a source of sensory irritation.
But it is very unlikely that the total of this afferent activity can
be as great as would result from the use of the whole limb; and
why should pain fibers in the neuroma be more irritable than in
the normal limb? Neuralgia, finally, is usually thought of as an
irritation, as a hyperfunction that must be reduced by deaden-
ing the nerve. But only too often an injection of the nerve fails
to workjust as might be expected if the pain, in such cases, is
186 Problems of Motivation: Pain and Hunger
essentially related to a Tit/pofunction. The original source of the
neuralgia must have been a pathological process in the nerve or
the nuclei to which it leads. It is highly improbable that a
pathological process can increase the efficiency of any fiber, even
a pain fiber; is it not more likely that total afferent activity has
decreased, and that a disruption of central timing has resulted
from an incomplete peripheral control of central activity?
Return for a moment to the older theories of pain. The evi-
dence concerning hypofunction directly denies the theory of
peripheral intensity. It also makes difficulty for the sensory-
mode theory, but a more clearcut argument against the latter is
found in the fact that the pattern or combination of stimuli is
often the decisive factor for the occurrence of pain. Stimulation
of A produces pain; of A and B together, no pain. The clearest
example has been provided by Head (1920). Stimulating the
tip of the glans penis with water at 45 C. is painful; but the
pain totally disappears when the glans is immersed more deeply
in the water. Such facts might be accounted for if pain im-
pulses are "inhibited" by other sensory impulses, but this expla-
nation has really grave difficulties.
The difficulties of the inhibition idea are clearest if we con-
sider an organ, such as a sound tooth, that contains pain fibers
but has never been the site of pain. We can hardly suppose that
there are normal nerve fibers that have never once been active
over an individual's lifetime. If we suppose that they are active,
but at a low level that is not recognized as pain, then we have
conceded that the activity of the pain fiber does not essentially
mediate pain; and the alternative is to suppose that an inhibi-
tory action of other fibers has throughout a lifetime prevented
the occurrence of pain. It is very doubtful that an inhibitory
process can be so uniformly and continuously efficient (p. 212).
It seems in short that the pain receptors in a sound tooth, or
in the cornea of an eye that has had no grit in it for months,
must have some other functionwith weak or brief stimulations,
that they are pressure receptors as well. It is still reasonable to
think of them as pain receptors if, as it seems, the thalamus in-
nately has a very low tolerance for the activity they initiate; if
their action tends innately to result in a disruption of any be-
havior that is accompanied by tissue damage.
Pain 187
It may be objected that pain is a unique sensory experience,
and therefore that it must have its special receptors and central
connections which serve no other function. But competent in-
trospective observation has always found more than one kind of
pain experience, and Titchener (1920) gives four: prick, clear
pain, quick pain, and ache, each no more closely related to
FIGURE 18. Diagram from Titchener, 1920, showing the independence of
various kinds of pain, introspectively determined. By permission of the
American Journal of Psychology, Karl M. Dallenbach, Editor.
another according to his diagram (figure 18) than quick pain is
to bright pressure or to itch.* Clearly, the subjective evidence
in itself does not establish a separate modality of pain, internally
unified, and distinctive from others. How then have we arrived
at this impression of uniqueness?
I have already suggested that pain may seem unique because
the limen of avoidance in the intensity range of the pain
* Professor Dallenbach, in a personal communication, has pointed out
that other evidence may render this account of pain, by Titchener, out of
date. I cite it here, however, as showing that competent introspective
observation does not inevitably find pain to be a single entity, whatever
the conclusion from other experiments may be. Professor Dallenbach has
drawn my attention also to his ingenious experiments with Burns, Stone,
and Edes (Dallenbach, K. M., "Pain: history and present status," Amer. J.
Psychol., 1939, 52, 331-347), which controvert the theory that pain is in-
tensive stimulation of tactual receptors only, but do not appear sufficient to
show that it is a separate and coequal sensory mode.
188 Problems of Motivation: Pain and Hunger
stimulus is so low. It may also seem unique because the char-
acteristic event causing it, damage to body tissue, is not imme-
diately reversible. Once the damage is done the stimulation of
nerve continues and avoidance does not put an end to it. It is
common knowledge that there are noises that are almost unbear-
able, although the fact is hard to demonstrate in the laboratory
(Miller, 1947). But one can move away from the noise or cover
the ears, at once decreasing the intensity of stimulation. These
conditions must tend to determine a special kind of learning
related to pain, with a more strongly marked avoidance and a
greater emotional element, even though the mechanism of the
disturbance when it does occur is the same in different sensory
systems.
Let me say again that this hypothesis of pain is incomplete.
Hypofunction often occurs without pain; syringomyelia, a de-
generative disease of the cord, provides an outstanding example.
The hypothesis is of value in suggesting how pain may result
from hypofunction, and in reconciling divergent views of normal
pain. Intensity is still a factor, but at the cortical or thalamic
level; and the arousal of C fibers may still have a special sen-
sory quale, not disruptive of behavior at the very lowest level
of intensity or duration, but promptly becoming disruptive un-
less they are part of a much more extensive stimulation that has
a thoroughly well-organized central action. This central organi-
zation would be a diffuse firing, and would make impossible what-
ever pattern of local synchronization it is, in the ventral nucleus
of the thalamus, that leads to local disruption in the cortical
somesthetic area and so constitutes pain.
It may be proposed that the optimal condition for the arousal
of pain without gross bodily injury is a punctate stimulation of
the relatively inefficient C fibersbeing inefficient, they do not
readily set up a well-organized central action; and, accordingly,
that it is not a complete hypofunction of afferent systems that
conduces toward pain, but a decreased efficiency. Higher somes-
thetic nuclei then lack their normal degree of sensory control
and become hyper-responsive, and yet are still exposed to an in-
termittent "triggering" from peripheral stimulation. A complete
deafferenting consequently would fully eliminate pain, while
pathological changes without complete destruction, up to and per-
Pain 189
haps including the nucleus ventralis thalami, would increase it.
Specifically, lesions of peripheral nerve would then be con-
sidered to have reduced the efficiency of A fibers to that of C
fibersthat is, make them into pain fibers; higher lesions would
have a similar effect by reducing the efficiency of A-fiber con-
trol of thalamic and cortical tissue. It would thus be conceiv-
able also that "quick pain" could be aroused by undamaged A
fibers, though less readily, since their control is more efficient
than that of the C fiber.
It is possible now to complete the picture of pain as related to
learning, and especially as having a positive value in learning.
The psychological entity pain must be a disruptive action at the
cortical level, but the pain stimulus has other effects. In the
spinal animal, it may arouse only a well-coordinated reflex avoid-
ance (Sherrington, 1906). In the diencephalic animal, it is
likely to produce sham rage, with an integrated action of the
whole organism (Bard, 1934). It has been seen in the preced-
ing paragraphs that the stimulus that by itself arouses pain may
have no such effect when combined with other stimulation.
The pain stimulus arouses pain when its central action is un-
organized and interferes with other, organized action, but not
when it enters into and forms part of an organized neural action.
Pavlov (1928) and Liddell (1944) have shown that the dis-
turbing effects of cutting and burning the dog's skin disappear
completely when the pain stimulus is regularly followed by feed-
ing, in the conditioned-reflex procedure. The transformation
from typical pain responses to feeding responses was gradual;
once it was established, there was immediate relapse when the
site of pain stimulation was changed to new parts of the body.
This means that there were synaptic changes in the cortical
region in which the focal pain disturbance originally occurred,
leading to the disappearance of pain from one region of the
body, not from others. In that one region, the pain stimulation
now aroused organized cortical activity. This conclusion is
meaningful for the observation of Muenzinger (1934) and Drew
(1938) that giving a rat electric shock for the correct response
(followed of course by food) increases the rate of learning. In
Drew's experiments, when the food itself was electrified so that
the rat got a shock with every bite, eating was voracious. In
190 Problems of Motivation: Pain and Hunger
such a case, once the central effects of the pain stimulus have
become organized, we must regard the stimulus as essentially
motivating: just as a blow of the whip may be motivating and
energizing to a racehorse that appears already to be exerting
himself to the utmosta further stimulation that adds to the
effectiveness of a pre-existent cerebral control of behavior.
HUNGER
The treatment by Morgan (1943) is taken here as a basis for
discussing the problem of hunger, with two questions to be kept
in mind: how the approach to motivation outlined in this
chapter accords with the experimental facts of hunger, and
whether it can be of any further value for understanding the
behavior involved.
The fundamental problem is to know what controls the initia-
tion and termination of eating (or drinking). Hunger is defined
here as the tendency to eat; conscious awareness or verbal report
of hunger is important only as a convenient index of that tend-
ency. Eating is not necessarily dependent on the occurrence of
stomach contractions, since the surgical removal or denervation
of the stomach does not prevent or decrease it. It is not neces-
sarily dependent on blood-sugar level, since hunger is reported
when the blood-sugar level has not changed (Scott, Scott, and
Luckhardt, 1938). Both stomach contractions and blood sugar
are related to hunger, but in no simple way.
The termination of eating also has no simple control. The
animal that has been starved eats more than one that is well fed,
but the termination of eating comes before the food can have
much effect on bodily tissues. A thirsty animal will drink ap-
proximately the amount of his water deficit, and stops drinking
before the water has been absorbed. The experiment can be
done in fact with a dog with a fistula in his throat, so that little
if any of the water is absorbed, yet "satiation" occurs (Bellows,
1939).
The problem is complicated but not fundamentally changed
by the demonstration that there is not one but a number of
hungers (Richter, Holt, and Barelare, 1938; Young, 1941). The
rat lacking any one of a dozen or so substances will choose a
Hunger 191
diet containing that substance, selectively correcting a specific
deficiency. We still have the same problem: what is it that ini-
tiates the eating of salt or sugar, or drinking water; and what
terminates it? Appetite and satiation are selective, but it is de-
sirable if we can to avoid postulating fifteen or twenty mecha-
nisms (allowing for future discovery of other hungers besides
those we already know), all in control of normal eating. If pos-
sible, we must find a single mechanism to comprehend the facts.
The neural mechanism that determines such motivated be-
havior as eating or sex activity Morgan has called the "central
motive state." This, for the theory presented in this book, evi-
dently is a phase sequence or conceptual activity. It does not
require that we postulate a special process to deal with hunger,
but only that the process already postulated have some distinc-
tive feature for each of the separate hungers. Conceptual series
A, B, and C might correspond to one hunger; A, B, and D to
another, and so on.
The total process would fit Morgan's specifications for the cen-
tral motive state, since: (1) the phase sequence is independent
of input from any specific sensory mode; (2) it has a central
priming action that reinforces some sensory stimulations and not
others; and (3) it evokes specific forms of behavior, since each
phase in the phase sequence may have its specific motor facili-
tation. Morgan's final specification (4) is that the central mo-
tive state arouses general activity or restlessness: for the present
theory, this is qualified by saying that the restlessness must occur
only when the environment fails to supply the specific stimula-
tion that the central motive state would reinforcethat is, when
the hungry animal is not exposed to food, or the sexually moti-
vated animal does not have a sex partner at hand.
This point is of some importance. We must distinguish be-
tween an integrative, motivating effect of a lack of food, and a
disintegrative one. The integrative action appears in any pur-
poseful search for food, and in eating. That it occurs, in the
experienced animal and in a familiar environment, is an evident
fact; it is the real fact with which most discussions of hunger
have been concerned. But the picture will not be complete if
we recognize only this aspect of the problem, and assume im-
192 Problems of Motivation: Pain and Hunger
plicitly that the physiological need of food equals hunger (that
is, the tendency to eat).
There are two reasons for denying such an assumption. One
is that learning is involved in hunger, as we shall see in a mo-
ment. The other is that, until this learning has occurred, the
lack of food is apt to be disruptive of behavior. The effect of
the lack appears in restlessness, which is not directed behavior
but implies an instability of direction: restlessness as such is not
a purposeful seeking but activity without purpose. With more
experience, the restlessness becomes a directed search. Bio-
logically, it has served a purpose, having made a certain kind
of learning possible; but this does not make it, from the psycho-
logical point of view, purposefuland the point here is that the
first effect of a lack of food is not an increased directedness in
behavior (unless the mouth happens to be in actual contact with
the nipple, when there is directedness at the reflex level).
Even in experienced subjects the need of food has disintegrat-
ing effects. The relation of hunger to emotional disturbance is
notorious, and Goodenough (1931) has shown it clearly in a
study of anger in children. Lack of food, or of some particular
element in the diet, such as salt (Saphir, 1945), can produce
profound and lasting emotional disturbances.
Furthermore, hunger itselfthe tendency to eatmay be dis-
rupted by a lack of food. Hunger practically disappears about
the fourth day of starvation in man, and there may even be a
"repugnance" to food (Carlson, 1916). In the rat, the rate of
eating is markedly lowered by the fourth day, and the amount
eaten is also low (Bousfield, 1935). Physical changes in the ali-
mentary canal would account for the smaller total intake, and
even perhaps for a 50 per cent decline in the rate, but they would
not account for a total loss of human appetite if the neural
mechanism of hunger were undisturbed.
In chronic starvation, however, the hunger mechanism seems
undisturbed even though the degree of malnutrition may be
great. Is it possible that the difference here is due to learning,
in the same way that learning changes the restlessness of the
first need of food into a directed search? Let us look at some
evidence concerning the relation of learning to hunger, before
Hunger 193
returning to the question of how a lack of food might disorganize
behavior.
It is common knowledge that a person eating on an irregular
schedule when he is used to a regular one, and so without his
usual environmental cues for eating, may be "hungry without
knowing it." With a deficiency sufficient to produce headache
and irritability, he may find out what is wrong only when food
tastes unusually good or when his symptoms disappear as a
consequence of eating. The fact obviously suggests that eating
is partly a learned response to certain environmental events.
Hunger, defined as the excitation of a neural mechanism that
controls eating, is not a simple direct product of a need for food.
There are cognitive factors to be considered when hunger is
studied in the mature animal.
It does not affect the value of Richter's study of food prefer-
ences (Richter et al, 1938) to recognize that his method clearly
permits of a short-term learning as a factor in preference. Young
(1944) and Beebe-Center, Coffin, and MacLeod * in fact have
shown that the learning has a strong influence.
Also, it is well known that tame laboratory rats deprived of
food, taken from their cages and offered food in a strange place,
will for some time eat little or nothing. Before he will run in
a maze for food reward, the rat must be given the routine "pre-
liminary training," extending over a period of days. Inanition,
meanwhile, may become extreme. It has been generally sup-
posed that the failure to eat is due to an independent emotional
disturbance, but some preliminary experiments (to be reported
in more detail elsewhere) suggest strongly that this is not the
whole explanation.
Six rats were thoroughly habituated to an enclosed field, con-
taining a number of movable objects, over an 8-day period.
Then they were deprived of food for 24 hours, and put into the
field with food and water present. The amount of time each
rat spent eating, in a period of 120 seconds, was recorded. The
procedure was repeated a number of times as one would carry
out a learning experiment, to see whether the amount eaten
would be affected by "practice"whether it would be small at
* Personal communication from Dr. J. G. Beebe-Center.
194 Problems of Motivation: Pain and Hunger
first, and increase on subsequent days; whether "forgetting"
would occur when practice was interrupted; and whether
changes in the environment might change the response, but
without inducing emotional disturbance to account for failure
to eat.
The amount that a rat can eat in 120 seconds is not nearly
enough to maintain health, but the average time spent eating
in the first test was only 79 seconds, one rat eating for as short
a time as 35 seconds. On subsequent days, the amount in-
creased; dropped whenever practice was interrupted (i.e., after
the rats had been left for a few days in their cages with food
and water present, before being tested again) and rose once
more when practice was resumed; and dropped with each
change in the arrangement of objects in the experimental envi-
ronment. The changes induced exploration but no evident signs
of emotion. It is also important to note that there was no tend-
ency to take a bite of food and then explore, or even to take
a quick bite whenever exploration brought the animal near the
food (this shows up later, as the rat becomes more sophisti-
cated ).
A second experiment was done to check on these results, and
also to see what would happen when exploration was eliminated.
In this experiment 10 rats (from another colony) were tested
individually in a living cage, being allowed food for 300 seconds
after a 24-hour deprivation, and again after a 48-hour depriva-
tion. The average time spent eating was again low on the first
test41 seconds: one animal ate for as short a time as 5 seconds.
On the second test, the average rose to 140 seconds. But the
total amount eaten was still estimated to be not more than one-
tenth of what would be necessary to maintain even a depleted
body-weight.
It seems therefore that the rat must learn to eat the amount
he needs; and "satiation"cessation of eating when no external
event has occurred to interrupt itoccurs before the physio-
logical needs of the unsophisticated animal have been met.
These results agree with the indications that human hunger is
partly dependent on being aroused by events associated with
eating. They also show that a theory of hunger does not neces-
sarily have to account for an automatic regulation of the amount
Metabolic Factors Disturbing Behavior 195
eaten, that will just equal the amount needed. The termination
of eating or drinking may be determined by experience, in part
at least (but partly as the termination of other pleasurable ac-
tivities is determined, by a kind of fatigue: Chapter 9). This
learning process is prolonged. I have the impression that the
rat, offered food for short periods only, at 24-hour intervals, will
take from 3 to 5 weeks before he gets to the point of promptly
eating the amount of his food deficit as soon as food is made
available.
Must there not be something of the same sort of learning in
the hunger of the dog and cat? The experiment suggests that
when hunger in the dog is to be studied it will be necessary to
know the animal's history or, better, to rear the animal in the
laboratory with a systematic control of his experience of the
need of food. Data obtained when this has not been done may
make hunger appear more an innately established process than
it actually is.
METABOLIC FACTORS DISTURBING BEHAVIOR
The evidence discussed in the preceding section has indicated,
first, that lack of food tends to disrupt behavior, producing rest-
lessness, discomfort, irritability, and, in the extreme degree, emo-
tional apathy. Secondly, a learning process seems to be involved
in transforming this primitive disturbance into hunger as we
know it in the adult animal or human being.
We may now ask what the source of the original disturbance
is before considering (in the next section) the processes that
transform it. The present discussion, primarily of the origins
of hunger, has a broader implication and will be relevant also to
the problem of emotional disturbance due to metabolic disorders.
It is well established that the need for food affects the central
nervous system directly, through changes in the blood content
(Morgan, 1943). (There may of course be peripheral effects as
well.) The effect of chemical changes in the blood content
would influence the rate of firing of the nerve cell. A nutritive
deficiency may be thought of as usually lowering the rate,
though a calcium deficiency might raise it (Lehmann, 1937fl;
Bronk, 1939), the important thing being the change of rate
196 Problems of Motivation: Pain and Hunger
rather than the direction of change. It must also be assumed
that any particular chemical influence would not affect all cells
equally. There is a considerable variety in the make-up of cen-
tral neural cells and their reaction to changes of blood content.
This is shown in the selective susceptibility of particular parts
of the central nervous system to particular diseases and toxins;
it is also shown in the differential effects of anoxia and of various
anesthesias. The firing of some cells may be especially changed
by a lack of thiamin; others by hypochloremia, and still others
by hypoglycemia.
Now it has been seen that neural integration is fundamentally
a question of timing, quite apart from the particular theory of
integration that has been developed in these pages. Metabolic
changes, by altering time relations in neural firing, must tend to
disrupt behaviornot merely slow it up, but disorganize it.
In the present theory, timing has its effect in the functioning
of the cell-assembly and the interrelation of assemblies: diffuse,
anatomically irregular structures that function briefly as closed
systems, and do so only by virtue of the time relations in the
firing of constituent cells. Synaptic changes are necessary to
the setting up of an assembly, but these act by coordinating
the action of two or more cells. The firing of one cell imme-
diately after another is not determined by synaptic knobs alone
but also by what is going on in some other cell or cells. Synaptic
knobs alone cannot determine that a particular system will func-
tion as such.
Furthermore, an individual cell or transmission unit may enter
into more than one assembly, at different times. Which it will
form part of, at any moment, depends on timing in other cells;
and to enter into any assembly requires that its frequency ac-
cord with the time properties of the active system. If a meta-
bolic change affects one cell much more than other cells in the
system, it must drop out; if enough cells drop out, the assembly
does not function.
If, instead, all the cells in an assembly are affected to about
the same extent, the system would continue to function; but the
facilitation delivered to other assemblies might be modified in
a way that would disturb, or redirect, the phase sequence (that
is, the sequence of assembly actions).
The Mechanism of Learning in Hunger 197
At any one moment, the action of an assembly may be con-
sidered to be on an all-or-none basis; the system functions as
such, or it does not. But an affected assembly might function
at one time and not at another, in agreement with the statistical
approach of Chapter 4. Suppose for example that the function-
ing of an assembly requires the activity of n pathways in parallel
at some point in the system, and that n -f 4 are anatomically avail-
able. At any time, some pathways are refractory but there is
normally a margin of 4 at this point, maintaining the operation
of the system. If now a change of blood content lengthens the
recovery cycle of some of these cells, the average number of
refractory pathways might rise from 2 to 3, say, and the upper
limit of the number from 4 to 6. When 5 or 6 are refractory, the
number of available pathways would fall to n 1 or n 2, and
the assembly does not function for the moment. A half second
later, it might again be responsive.
One might assume that longer-established assemblies would
have in general greater safety margins. With the greater devel-
opment of synaptic knobs, due to more frequent activity, the
number n might decrease (p. 66). This would imply that older
habits, and longer-established memories, would be most resistant
to disruption by metabolic changes of blood content.
In general terms, these ideas provide a rationale for the ac-
tually observed disturbance of behavior that may result from
changes of blood content, whether in hunger or otherwise. In
hunger itself, the blood change might be a lack, as in hypo-
glycemia, or it might be the addition of the "hunger hormone"
of Luckhardt and Carlson (1915) and Morgan (1943). This
would be a substance liberated by depleted body tissues, with
a selective effect on the firing of neural cells.
THE MECHANISM OF LEARNING IN HUNGER
The facts discussed earlier, showing an element of learning
in hunger, leave room at present for a considerable difference
of opinion as to how large a part it plays. In order to further
define the problem for research, let us assume for the moment
that the innate factor consists only of the reflexes of sucking,
chewing, and swallowing, of spitting out noxious substances, and
198 Problems of Motivation: Pain and Hunger
of the activities of the gastrointestinal tractthat all else is
learned.* The deficiencies of this approach will help us to dis-
cover what facts are needed for a better understanding.
Stimulation of the lips of the newborn infant elicits sucking,
as a reflex activity. The same stimulation stops hunger contrac-
tions, and the effect is increased by the stimulation of taste buds
and by swallowing (Carlson, 1916). In the adult at least the
effect of stomach contractions is related to pain ("hunger is an
uncomfortable pain sensation caused by stomach contractions":
Carlson, p. 97)mild, but painful. In the infant, also, the be-
havior accompanying strong stomach contractions is in general
not to be distinguished from the effects of a loose safety-pin in
the diaper.
Consequently, hunger contractions should have the same role
as pain in their effect on learning. As the contractions begin
and get stronger, some time after suckling, there would be an in-
creasing disturbance of the intrinsic organization of activity in the
infant's cerebrum (p. 121). This organization it will be recalled
makes for inactivity in behavior, other than reflex behavior; as
it is broken up, an increasing restlessness should be expected.
As a result, the moving mouth is likely to make contact with
the nipple. Reflex sucking follows, with a further stimulation of
taste buds and swallowing: stomach contractions cease, and the
restlessness disappears.
Later, as the infant becomes more mobile and explores more
of his environment, other things are investigated and enter the
mouth, with similar effects. Some things produce, perhaps, too
strong a stimulation of salt, sour, or bitter receptors and are spit
* In correcting proof I realize how misleading this statement may be.
Obviously, I assume that the "innate factor" also includes all the mechanism
that determines what kind of learning will take place. It includes for ex-
ample the susceptibility of cortical organization to disturbance by hunger
pangs or lowered blood nutrients.
Throughout this book the reader may feel that the emphasis is too much
on learning. But this is only verbal. Whenever one could think of several
ways in which some particular learning might happen, but recognizes that
the facts of anatomy or physiology allow only one of them, one is recognizing
the existence of the innate peculiarities of the central nervous system. All
mammalian behavior involves learning; but the innate factor is always there,
determining what learning occurs and how.
The Mechanism of Learning in Hunger 199
out. Others fall below the avoidance limen, reduce to paste and
are swallowed. These edible substances, not only stimulating
taste receptors but also being retained in the mouth and even-
tually swallowed, are evidently the ones that most effectively
interrupt stomach contractions: they provide more of the inhib-
iting stimulation, and more-prolonged stimulation.
The conditions necessary for learning are provided in this
sequence of events. As the sequence is repeated, the sensory
stimulationvisual, olfactory, gustatory, and kinestheticorgan-
izes assemblies of cells with a motor facilitation of the various
activities of eating. This brings eating farther and farther under
cortical control, and so increases the possible influence of learn-
ing. The sight, smell, and taste of food, the tactual stimulation
of mouth and throat, and kinesthesis from chewing and swallow-
ing are the stimuli that arouse this assembly action. To a cer-
tain extent, the sensation from stomach contractions itself would
become conditioned to arouse the same activity. Though pri-
marily disruptive, like pain, this sensation should like pain be
capable also of being conditioned to eating responses. It has
been seen that the potentially painful stimulation is not disrup-
tive (that is, it is not pain) when it forms part of a stimulation
that has well-organized central effects. Finally, the development
of conceptual processes controlling eating makes possible an
association of eating with other conceptual processes.
Schematic and oversimplified, this discussion still shows the
way toward avoiding oversimplification. It points out three
ways in which hungeran organized activity in the cortico-
diencephalic systemcan be aroused: (1) by the sight and smell
of food and the sensation of eating itself, (2) by stomach con-
tractions, and (3) conceptually, as the sequel to a train of
thought. These agencies may summate. One, on occasion, may
not be effective by itself, but becomes effective when another is
added to it. Hunger can then be aroused or further increased
by an attractive appearance of food, or by its smell and taste.
Such effects are familiar; but since they do not fit into the
rubric of hunger as a bodily need, or alternatively as sensations
from the stomach, we are apt to forget them. Consider the
salted-nut phenomenon. Ordinarily, one can take salted nuts or
let them aloneuntil one has eaten a mouthful, when it becomes
200 Problems of Motivation: Pain and Hunger
much harder to let them alone. Hunger has increased: but how?
A lack of food cannot be increased by eating something, and
stomach contractions are stopped by chewing and swallowing.
If, however, we consider hunger to be neither a particular con-
dition of the body, nor a set of sensations from the stomach, but
an organized neural activity that can be aroused (like any other
conceptual process) in several ways, the puzzle disappears. The
schematizing of the preceding paragraphs, then, has developed
some of the further possibilities of Morgan's conception of the
central motive state.
The puzzle just referred to has been dealt with in the past by
a distinction of hunger from appetite. ("A taste of salted nuts
does not increase hunger; it only increases appetite.") The solu-
tion is not a good one because no one has been very clear as to
what appetite is, if it is not hunger. It seems in fact that hunger
cannot exist without appetite; and "appetite" seems in practice
to be a term that is applied to the tendency to eat when eating
is not biologically desirable or valuable, "hunger" when it is
desirable. Such a distinction has no place in discussion of the
physiological control of eating.
If the explanation proposed in this section is right, cutting the
vagus in the newborn infant will retard the normal development
of hunger, by preventing the postulated role of hunger pangs
in the learning process. Cutting the vagus does not stop eating
in the adult, but it may have more effect at an earlier stage.
This would provide a direct experimental test of my schematiz-
ing. It is not predicted, however, that no learning could then
take place, since there is still a mechanism of addiction to food
to be discussed, which would also provide a basis for learning.
The explanation predicts that cutting the vagus would retard the
development of hunger, perhaps seriously, but would not pre-
vent it entirely.
HUNGER AS AN ADDICTION
I have already discussed the metabolic changes due to lack of
food as a disturbing influence on behavior. The treatment of
hunger can now be completed by considering the relation of
these changes to the learning process: a relation that makes eat-
201 Hunger as an Addiction
ing, in the mature animal, equivalent to an addiction, as Carlson
(1916) long ago suggested.
It may be objected that hunger cannot be related to metabolic
changes, because of the experiment by Scott, Scott, and Luck-
hardt (1938) in which it was found that hunger can occur in
human subjects with no change of blood-sugar level. This ex-
periment is valuable, but in the light of present knowledge, and
the discussion of the preceding pages, it needs two things to be
complete:
1. In the first place, we need to know about other blood com-
ponents besides glucose. Glucose is only one of a number of
substances involved in hunger: if there are nine specific hungers
besides those for oxygen and water (Morgan, 1943), there are
nine substances whose concentration in the blood must be exam-
ined before it can be concluded that hunger is not determined
directly by the blood content.
2. In the second place, since we know also that there is an
important element of learning in hunger, the experiment should
be repeated with subjects in early infancy. Only then will it
be possible to say that stomach contractions are not determined,
originally, by changes of blood content, or to conclude that the
development of hunger is independent of such changes. The
obvious experiment, dealing with both points 1 and 2, is to de-
termine the levels of thiamin, chlorides, and so on, in addition
to the level of glucose, at the time when a need of food begins
to disturb infant behavior or when the adult actually reports
hunger.
Until such an experiment is done we can only speculate, but
a little speculation may help to define the problem. Return to
the learning process of the last section. In general, the cell-
assemblies whose activity is hunger would be organized in a
condition of mild deficiency. We do not know whether this
condition is enough to produce a lowering of blood nutrients,
or secretion of the hunger hormone; but, if it is, observe that a
mild deficiency would be the optimal condition for the function-
ing of the assemblies in question. It is not a low blood sugar,
or low chlorides, that would be disruptive of an assembly, but
a change from the level at which the assembly was organized.
If hunger is established originally in the presence of a slight
202 Problems of Motivation: Pain and Hunger
deficiency, the return of blood sugar and so on to normal would
tend to disrupt it, and equally, a more extreme lowering of the
blood-sugar level. This accounts for a loss of appetite in starva-
tion, and is a first step toward accounting for satiation.
Suppose that an experienced, adult animal is exposed for the
first time to a severe lack of food. Both the increased hunger
pangs and lowered blood nutrients contribute to a disturbance
of cortical organization and so of behavior. If the disintegration
of the hunger mechanism itself has gone so far that there is in-
difference to the thought of food, or if the animal put where he
can find food makes no effort to seek it outthat is, if the hunger
mechanism is not excitable conceptuallythe sight and smell of
food may be more effective; or food in the mouth may be. This
sort of thing of course has been observed in starvation. The
animal is apathetic about food until he has eaten a mouthful,
and then shows an increased hunger (the "salted-nut phe-
nomenon" again). Once swallowed, the food is absorbed and
reverses the slow fall of blood nutrients, increasing the possi-
bility of actual hunger as an organized cerebral action.
Now observe that each occurrence of such a sequence of
events in chronic starvation would tend to set up new cell-
assemblies (or modify old ones) by the same process that set
up assemblies in the normal infant. The new assemblies, how-
ever, would be such as to function best in a condition of low
blood nutrients. The learning then would establish a hunger
that could occur strongly in the presence of starvation, and not
as before only with a minimal need of food.
Hunger established in the presence of lowered blood nutrients,
and having the effect (through eating) of raising them, would
be physiologically the equivalent of an addictionbiologically
valuable, but still an addiction. There has occasionally been
confusion in distinguishing between addiction and habit. Some
writers apparently have seen no difference (except possibly in
strength) between a habit of drinking milk at breakfast and a
habit of drinking coffee. But an addiction involves more than
a relationship of overt response to certain sensory stimulations:
otherwise Sanka, with the taste and smell of coffee but without
its caffeine, would satisfy every need that coffee does and could
be substituted for it without the victim's knowing that anything
203 Hunger as an Addiction
had happened. Unfortunately, it cannot: the addict may not
know what has happened, but he will know that something is
wrong. An addiction, therefore, besides involving habit, also
acts so as to maintain a blood concentration of some substance
that is necessary to stable neural functioning, or whose absence
results in emotional disturbance, restlessness, and discomfort or
pain. This is a trite statement, but one of its implications should
be considered. What has happened to prevent integrated neural
function, in the absence of the drug?
We know, first, that there is always a period of "learning" or
apprenticeship, in addiction; and we know also that an addiction
lasts longer than the overt withdrawal symptoms of physiological
disturbance. It may be proposed, therefore, that the learning
is a genuine learning; and that it consists in part of the change
of assembly action discussed above in relation to starvation. It
follows that an addiction must be considered to be more than
a metabolic reorganization: it includes a neural reorganization
also, and one that tends to persist, since (as any tobacco addict
or alcoholic knows) a "cure" is apt to be unreliable and the
addiction can be reinstated much more promptly than acquired
for the first time.
It is this very learning in addiction that we must know more
about, to understand hunger. The learning can be described as
an association between a sensory event and the subsequent
euphoria: being at peace with the world follows a full belly, or
an injection of morphine. The reference to "association," how-
ever, is misleading if it suggests that we understand the mecha-
nism involved. We do not.
The problem of a learned preference, either for a drug or a
needed food substance, can be seen better in the light of Spragg's
(1940) study of morphine addiction in chimpanzees. For ex-
ample, his experiment has demonstrated the difference between
physiological dependence on a substance, and addiction to it.
Daily injections established a dependence on morphine, demon-
strated by withdrawal symptoms when the drug was withheld.
But his chimpanzees still took a considerable timefrom three
weeks to three monthsbefore they discovered the connection
between (1) injection of the drug and (2) feeling better. This
exemplifies an intermediate stage of addiction. Eventually, the
204 Problems of Motivation: Pain and Hunger
chimpanzee in need of morphine tried hard to get an injection
(by struggling to get to the accustomed room, getting the hypo-
dermic needle and giving it to the experimenter, and so on).
The learning that leads to this final stage of addiction presents
us with our problem, the same one that is involved in hunger.*
It has sometimes been said that the rat in need of salt learns to
eat it, by "experiencing its beneficial effects." Spragg has shown
that an association of this kind can occur, even if its mechanism
is not at all clear, and his data should be kept in mind in deal-
ing with the problem of the self-selection of diet.
When a rat is suddenly put on a salt-free diet, he may be
considered to be in the intermediate stage of addiction. His
behavior shows that he has a physiological dependence on salt,
but he may still have to learn to look for salt-containing foods
and eat them selectively.
The immediacy and reliability with which an animal will cor-
rect a dietary deficiency seems to have been overestimatedat
the least, overemphasized. I have already cited Young (1944)
and Beebe-Center (personal communication) as showing that
the rat's preference for a needed food is not immediate, but
learned. With a low concentration of the needed substance,
preference is established slowly; it may require as long as 7
days. With stronger concentrations, the preference appears
early. But showing that a rat already has a complete prefer-
ence at the end of 24 hours does not show that no learning has
occurred. We badly need a detailed study of what learning may
occur in the first 10 minutes, or first hour, of a rat's exposure
to a choice of foods.
The Editors of Nutrition Review (1944) in a summary of the
history of this topic, and a paper by Pilgrim and Patton (1947),
have drawn attention again to the unreliability of the self-
selection method in many circumstances. The failures must be
kept in mind as well as the successes, if selective hungers are
* Dr. Austin H. Riesen has pointed out, in a personal communication, that
another example is the slowness with which air crews learn what to do
about oxygen deficiencies, when oxygen is available.
Obviously the law of effect would be the very thing to invoke to account
for such learning, but I can still find no satisfactory way of fitting it into
the present set of physiological conceptions.
205 Hunger as an Addiction
to be explained. Man is notoriously capable of malnutrition
while freely choosing his own diet. This might be put down to
bad habits, interfering with an innately sound mechanism of
choice; but such an explanation would hardly account for hypo-
chloremia, promptly relieved by taking common salt (Saphir,
1945), in men who had access to salt-shakers. Their symptoms
were severe, and their condition once more can be properly
likened to that of Spragg's chimpanzees when they had with-
drawal symptoms without seeking a morphine injection. Man
must often learn to take more salt when he moves to a tropical
climate, and the learning is not automatic.
Until such learning has occurred, it appears that the only
direct effect of a need of some particular food, or of food in
general, is restlessness, emotional disturbance, and malaise. The
lack of specificity in this disturbance may well be a clue to the
nature of pathological eating, arising from emotional disturb-
ance, or of the alcoholic's tendency to eat a less and less ade-
quate diet. Clinically, there is an inescapable relation between
drug addictions, food habits, and chronic emotional disturbance.
The fact has some meaning on the assumption that the need of
food or drug, and emotional disorder, have a common element
in a disruption of timing in the firing of central neural cells, and
on the further assumption that this disruption is transformed
into an organized action by a process of learning.
An hypothesis dealing with part of this learning process has
been set forth in the present chapter, but there remains a part
about which the hypothesis must be so vague as to have no
present value. Once more, Spragg's experiment makes the
matter clearest. He found it possible, once a full addiction was
established, to abolish withdrawal symptoms by an injection of
saline onlybriefly, but nonetheless effectively (Spragg, 1940,
pp. 95-96). Similar effects are known in human addicts, and
"explained" by a reference to suggestion. Here we have a sen-
sory and perceptual event substituted effectively for the physio-
logical presence of a substance in the blood stream. The ana-
logue in hunger is the occurrence of satiation long before a
meal has been absorbed from the stomach and intestine. It has
been argued that withdrawal symptoms, or the restlessness and
discomfort of hunger, are the result of a disturbance of central
206 Problems of Motivation: Pain and Hunger
organization. If so, the sensory stimulation of seeing the hypo-
dermic needle and feeling its jab, or the sensations of eating,
must have some rather direct action in restoring the normal time
relations in the firing of central neural cells. Some such possi-
bility has actually been envisaged in supposing that sensory proc-
esses may actively support and direct a given cerebral organiza-
tion. This of course is in extremely general terms, and no details
relevant to the present problem can be proposed. A more spe-
cific hypothesis may be possible later, when some of the experi-
mental evidence is available whose need has been shown in this
chapter.
9.
The Problem of
Motivational Drift
The two primitive motivations of pain and hunger can be re-
garded as mainly determined by specific bodily conditions-
tissue damage and lack of nourishmentalthough, as we saw in
the preceding chapter, neither of these relationships is simple
and neither is as yet really understood. Hunger, particularly,
must be a very complex process.
A third motivation, biologically primitive, is that of sex. The
sex drive is also tied to a definite physiological condition, the
presence of certain hormones in the blood stream. But it would
once more be an oversimplification to make this the only factor:
like eating, sex behavior is not reducible to any simple formula
(Beach, 1947a), and it serves especially to make clear a further
complexity of the problem of motivation. This concerns the
time relations in the organism's responsiveness to a particular
class of stimulation. In sex behavior, the problem appears in
the comparatively slow build-up of interest (that is, responsive-
ness to or seeking out of genital stimulation) and its frequently
abrupt decline, following orgasm: humoral conditions presum-
ably remaining constant. A roughly parallel course of events is
to be seen in other behavior, particularly in the alternation of
sleep and waking, but also in almost any behavior that is a
source of pleasure. It should therefore have an important place
in the theory of motivation.
A constant property of mammalian behavior is a variability of
responsiveness to the environment that cannot be ascribed to
conditions originating outside the nervous systemit appears,
that is, to reflect something intrinsic to the organization of cere-
bral activity. I have already emphasized in Chapter 7 this
207
208 The Problem of Motivational Drift
variability of attitude (or interest), and I propose now to look
closer at its neural basis. The reader will recall, further, that
"motivation" has been treated not as a distinct process, separate
from learning, insight, or attitude, but as a useful first approxi-
mation whose main reference is to the temporal organization of
cerebral events (Chapter 8). The present discussion accordingly
deals with the slow change or "drift" of motivation, but inter-
prets motivation in a rather broad sense.
THE PROBLEM OF A LASTING EXCITATION, FATIGUE,
OR INHIBITION
The problem of this chapter is the relationship between
cyclical changes of responsiveness in the whole animal and
events in the nervous system. A permanent change of respon-
siveness, of course, is called learning, and does not concern us
here; but one that involves a continuing alternation of respon-
siveness to some stimulation, each phase lasting perhaps for
hours before reverting predictably to the other, is a different
matter. In describing the behavior, such transient conditions
may be called excitation or fatigue, which is a commonly ac-
cepted and useful terminology. It has, however, certain dan-
gers and may conceal certain complications for the theory of
behavior. "Excitation," "inhibition," and "fatigue," as prolonged
neural conditions * determining the behavior of the whole ani-
mal, are conceptions which perhaps are not a primary concern
for the neurophysiologist but which certainly need examination
in psychological theory.
When an animal is "excited" what happenssimply an in-
crease of total cortical activity? What changes take place when
excitability, or responsiveness to a particular kind of stimulation,
* It will be evident that a discussion of fatigue on this basis is not very
closely related to the undertaking of Hartley and Chute (1947), who define
fatigue as a form of experience that is the outcome of mental conflict. They
have not been mainly concerned with the neural mechanism of this con-
flict, or with the nature of the "loss of interest" which contributes to it.
Their emphasis on the complexities of fatigue, however, strengthens the
present argument that it should not be lightly invoked as an explanatory
conception.
Excitation, Fatigue, and Inhibition 209
increases over a period of hours, days, or weeks? Is there simply
a decrease of cortical firing in "mental fatigue"? When the
cortex inhibits a certain hypothalamic pattern of response, is one
group of cells preventing the firing of another group? Presum-
ably, these questions cannot be definitely answered at present;
but it should be seen that the answers are not obvious, and that,
when the temporal characteristics of excitation are taken into
account, the question of its neural basis becomes a crucial one,
for physiological psychology at least. And so with inhibition
and fatigue.
Consider excitation first. There are two aspects of the prob-
lem: one is simply the duration of the excitation following brief
stimulation. It seems generally recognized now that an excita-
tion which endures for minutes or longer at once raises a ques-
tion of the special mechanism involved. The all-or-none excita-
tion in the nerve cell is almost infinitesimal in duration, bearing
little relation to behavioral excitations that take an appreciable
time to develop. The widespread theoretical interest in rever-
beratory or re-entrant pathways as the basis of a continuing ex-
citation in the central nervous system shows that this side of the
problem has been recognized (unlike the problem of a continu-
ing inhibition, as we shall see in a moment).
But another aspect of the problem of excitation needs more
attention. A behavioral excitation, an increase in some bodily
activity, is not necessarily a sign of an increased neural activity
either in the brain as a whole or in some one part of it. The
point is well illustrated by the process of getting drunk. A small
amount of alcohol may be an excitantsocially, and in its imme-
diate net effect on behaviorbut this of course does not prove
that alcohol is a neural excitant; it does not even prove that
lower centers are being released from cortical inhibition.
There is a fallacy here which is most likely to be made when
one is thinking of different kinds of behavior as controlled by
different parts of the brain: "There is an increase in certain
behavior, therefore some part of the brain is more active, there-
foresupposing that alcohol is not a neural excitant but a de-
pressanta higher inhibition has been removed."
Perhaps this is what happens. The point, however, is that it
is not logically inescapable. A change of behavior does not
210 The Problem of Motivational Drift
necessarily mean that some neural units, active before, have be-
come inactive, and that inactive ones have become active (ex-
cept at the level of the final common path, and very often not
even here since the same effectors may enter into different re-
sponses). The same neural cells may still be active, but in new
combinations, when a different response occurs. This is one of
the profoundly important implications of recognizing the role of
timing in neural transmission and of such conceptions as that
of "optional transmission" (Lorente de No, 1939). Stimulate
AB and CD and the excitation follows one set of pathways;
stimulate AC and BD and the excitation may follow different
efferent pathways and determine a totally different response
although the same internuncial cells are active.
The behavior of the drunk may be produced because alcohol,
depressing all neural cells, depresses some more rapidly than
others and so changes the pattern of firing throughout the cere-
brum. The cortex of course may be more affected than the rest
of the brain; but this is different from saying that the activity
of the brain stem actually increases with the administration of
alcohol. The idea that cortical function can be almost elimi-
nated before lower centers are affected is scarcely tenable; and
in fact the drunk, up to the moment of collapse, does not in
any way resemble a decorticate preparation but continues to
manifest behavior that we know is fully dependent on the
cerebral cortex. The behavior changes and may be less adap-
tive, but is it less "cortical" and more "hypothalamic" or more
"mesencephalic"?
The theory that drunken behavior is due to loss of cortical
inhibition is open to these objections: first, it implies that the
inhibition, during long periods of sobriety, has an astonishing
duration and efficacya point to which we shall return; and
secondly, it seems to imply that all drunks should act much
more alike than they do, perhaps approaching the picture of
sham rage before passing into stupor. The other possibility,
suggested above, is that alcohol from the first knocks out cells
in the hypothalamus and striatum as well as in the cortex, and
slows up firing rates, thus reducing the complexity of cerebral
organization but leaving until the last some cortical as well as
subcortical function. Whether this is so or not, it is enough to
Excitation, Fatigue, and Inhibition 211
show that there is no logical necessity of regarding drunken
rudeness or hyperactivity as a loss of inhibition that permits an
absolute increase of neural activity in some part of the brain.
Two points have been made concerning a lasting excitation in
behavior: the long duration must be accounted for, and an in-
crease of some kind of behavior does not always or necessarily
mean an increased activity in some region of the central nervous
system. The same two points apply to inhibition and fatigue.
If a decrease of bodily activity that lasts more than seconds is
to be explained by a neural inhibition or neural fatigue, some
explanation is needed of the duration; and secondly, the decrease
of bodily activity is not a priori proof that such an inhibition or
fatigue exists. There is risk in assuming that any prolonged
change in some aspect of behavior means a parallel change in
some one part of the brain: that mental fatigue is a neural fa-
tigue, or that a suppression of anger means that some one set of
cells (preferably cortical) suppresses the firing of some other set
(preferably hypothalamic). A decrease of bodily activity does
not always mean a decrease of neural activity: sleep may consist
only of a change of the combinations in which cortical cells
firei.e., may essentially be hypersynchrony, with no decrease of
total activity; and the EEC offers evidence that immobility and
unresponsiveness can accompany a considerable cortical activity
(Jasper, 1941).
It is commonplace to say that an organic whole has properties
other than the sum of the properties of its parts in isolation. The
necessary corollary is this: do not take for granted that the parts
of a neural complexor that any one partmust act in parallel
with the action of the whole. The fatigue that leads to pro-
longed sleep, or the disappearance of sexual responsiveness after
ejaculationoften for hours afterwardis not proof of fatigue in
cortical cells, or of an inhibition of sexual centers, for corre-
sponding periods of time.
Fatigue as demonstrated physiologically in the neuron (the
refractory period) has a duration measured in milliseconds.
"Mental fatigue" may be a phenomenon of behavior, but its
duration gives us no warrant for identifying it with a primary
neural fatigue. Similarly, the inhibition that is studied in lower
neural centers (where a fairly direct inference is possible con-
212 The Problem of Motivational Drift
cerning what happens in the single neuron) is a briefly lasting
process that is promptly followed by excitation. In a matter of
seconds it becomes possible for any strong excitation to break
through. Consequently, to speak of an inhibition that lasts in-
definitely is not just making use of the established conception of
spinal reflex inhibition but introducing a different one. The
conception may be quite valid, but it is not necessarily the c.i.s.
of Sherrington (1925).
This point applies to the idea that pain fibers in a sound tooth
can be continuously prevented from firing (p. 186); that the
sham-rage center in a good-natured, normal cat is continuously
inhibited by the cortex; or that a conditioned inhibition (Pavlov,
1928) can account for schizophrenic catatonia or for long-stand-
ing Freudian repressions. Some of these ideas assume first that
each form of behavior is determined by a separate part of the
central nervous system, and all of them assume secondly that
inhibition may have an extraordinary degree of persistence.
When for example damage to the frontal lobe is followed by
an increased appetite, or increased sexuality, or increased re-
sponsiveness to carotid-sinus stimulation, some clinical writers
are prone to take for granted that this is of necessity evidence
that the frontal lobe normally inhibits whatever lower centers
are involved. Not at all; the normal action of the frontal lobe
(and of the rest of the cortex) may be only to stimulate lower
centers, but determining at the same time a particular timing of
action in those centers. The behavior that results is a function
of this timing; removal of the stimulation (not inhibition) from
the frontal lobe means a different pattern of firing and so a
change in behavior. It must be realized that when a patient
says nasty things to his friends after lobotomy he uses the same
effectors as when he is politebut in a different firing combina-
tion; and the appearance of rudeness following brain damage
does not show that there is a rudeness center in the brain, nor-
mally inhibited by the region that has now been damaged.
Decerebrate rigidity, again, may imply either (1) that the
cerebrum acts primarily to diminish the activity of ventral-horn
cells, or (2) that its action stimulates only, and in stimulating
continually modifies the activity of the intrinsic circuits of the
spinal cord which if left to themselves would simply raise the
Excitation, Fatigue, and Inhibition * 213
level of tonic contraction. It is legitimate (though possibly con-
fusing) to speak of the cerebrum here as "inhibiting" the action
of these circuits; my point is to draw attention to the differ-
ence in meaning between such usage, and the usage when one
speaks of inhibition as the suppression, by one cell, of activity in
another. I have argued that the physiological evidence does not
justify the idea that pain fibers, for example, may be indefinitely
inhibited. It is a long step of inference from the prolonged in-
hibitory action of one complex set of neurons upon another to
an inhibition of the individual cell.
As a last example of this problem, consider the "suppressor
areas" of the cerebral cortex (von Bonin, Carol, and McCulloch,
1944; McCulloch, 1944a, 19446).
These areas, in or near areas 2, 4, 8, 19, and 24 of Brodmann,
are widely spaced in the cerebral cortex and appear superficially
to have a primary inhibiting action. Stimulation of one of them
in the anesthetized animal may have two effects: relaxation of
the musculature on the opposite side of the body, and disap-
pearance of electrical potentials throughout the rest of the cortex.
This might appear to be a direct suppression of neural activity,
as a positive action by one set of cells in the cortex that pre-
vents the activity of other cells. It might appear to justify, once
and for all, the conception of a primary inhibitory function of
the cerebral cortex.
The experimental evidence, certainly, has made it clear that
stimulation of a suppressor area may produce a decrement of
activity. But this leaves two questions: is this inhibition, in the
sense that the vagus inhibits heart muscle, the same inhibition
that presumably must operate in Lloyd's two-neuron spinal reflex
arc (Fulton, 1943)? and does it occur as a result of the normal
action of suppressor-area cells, or only as a function of massive
nonphysiological stimulation by electrical current or strychnini-
zation?
Suppose for a moment that the normal action of the suppressor
area is a facilitation of other cortical activity; that the function
of areas 4s, 19, and so on, is like the function of the reticular
substance in its facilitation of oculo-motor reflexes (Lorente de
No, 1939). On the lines of the theory of the preceding chapters,
integrated behavior requires a diffuse firing in the cortex. But
214 The Problem of Motivational Drift
a massive electrical stimulation would fire all the cells in one
region together and prevent their entering into the diffuse pat-
terns of firing necessary to the normal cortical control of be-
havior. That is, the "inhibition" would be a loss of essential
facilitation only.
The evidence presented by McCulloch (1944&) fits into such
a picture. Suppressor areas are directly connected with the
caudate nucleus, and direct stimulation of the caudate produces
"large, long voltages" in thalamic nuclei. McCulloch, accord-
ingly, has suggested that the action of the caudate, in "suppres-
sion," is not a positive inhibition but the induction of rhythms
of thalamic activity that are incompatible with normal thalamo-
cortical reverberation.
Finally, it is important to see that the suppression cannot be
a direct inhibitory action from one part of the cortex on another,
both because the strychninized suppressor area does not appear
to fire directly to other areas in the cortex and because of the
observed time relations: The first effect of suppressor-area stimu-
lation is the appearance of spikes in the immediate neighbor-
hood, but with no simultaneous loss of potentials elsewhere.
Next, a spreading extinction of potentials moves outward from
the point stimulated, affecting the suppressor area before affect-
ing others. If the cells in this area were inhibitory, a high level
of firing here should coincide in time with a low level elsewhere;
but this is not what happens. The suppressor area suppresses
itself first, and then, after it has itself become inactive, makes
others inactive. Obviously, this does not fit in with the idea that
suppressor-area cells are inhibiting other cells but does fit in
with the idea of a slowly spreading loss of facilitationthe im-
plication of McCulloch's remarks.
It is still early for final interpretation of much of the work that
has been done in this field. The nature of even spinal inhibi-
tion is still a disputed question, and there are serious dangers
in applying the conception directly to the complexities of cere-
bral function, especially when the time properties of a decrement
of activity are found to be markedly different. If it were safe
to assume a true long-term inhibition (or, equally, fatigue) in
cerebral action, it would greatly simplify the task of the psy-
chologist. As matters stand, however, this is not justified. It is
The Time Course of Sexual Motivation 215
necessary to recognize the problems for behavioral theory that
result from this fact.
Now let us see more specifically what some of these prob-
lems are.
THE TIME COURSE OF SEXUAL MOTIVATION
What we are interested in here, essentially, is the nature of a
behavioral excitation, and this perhaps would present no par-
ticular problem if it were not for the way in which excitability
and excitation develop and decline. The difficulty is clearest in
sexual behavior.
The review of studies of mammalian sex behavior by Beach
(1947a, 1947&, 1948) makes for emphasis on an interaction of
sensory and central factors. It has been clearly shown that the
arousal of sex behavior does not depend on any particular sen-
sory stimulationon olfactory, visual, or auditory stimuli, nor on
somesthetic stimulation from the genitalia or from other regions
of the body. If gonadal hormones are also considered to be
stimuli, arousing receptor organs in the brain, even this stimula-
tion is not always essential to sexual arousal (since gonadectomy
frequently does not put an end to copulation). Nevertheless,
all these factors normally play a part in the arousal. Their effect
is additive.
The presence of gonadal hormones in the blood stream does
not often produce continuous sexual excitement, and it appears
that a normal level of hormone facilitates certain effects of sen-
sory stimulation, selectively, rather than itself producing any
specific pattern of coordinated responses. When essential stimu-
lation is lacking, the motor pattern of copulation is apt to deviate
from the norm or not to be completed. Also, "There is evi-
dence that the cortex is involved in the arousal and mainte-
nance of the male's sexual excitement; and reduction in suscep-
tibility to sexual arousal subsequent to cortical injury appears to
be proportional to the amount of neopallium removed" (Beach,
1942, p. 187). The evidence thus gives us a picture of sexual
excitation as being produced by the close collaboration of sen-
sory and central mechanisms.
During courtship and during copulation up to the point of
216 The Problem of Motivational Drift
orgasm, activity in the "central excitatory mechanism" (Beach,
1942) summates, cooperating closely with sensory stimulation
and producing forms of behavior which increase that stimula-
tion, until the joint effect of sensation and central facilitation
reaches a final limen of response. Then, abruptly, the central
excitatory mechanism is somehow discharged; or, alternatively,
an inhibition or fatigue not evident before suddenly assumes
maximal effectiveness and may last without interruption for
hours.
It is, I believe, the first of the alternativesa slow growth of
excitation followed by a condenserlike dischargethat should be
explored theoretically. It has already been said that it would
be very desirable for psychological theory to be able to postulate
an inhibition or fatigue in such cases to explain the transient
disappearance of responsiveness. But there are difficulties about
doing so, as we have seen, and for the present there is no ad-
vantage in postulating a special inhibition in sexual behavioras
if we understood the excitation but not its absence. Really, we
do not understand either, and a transient disappearance of
excitability may be inherent in the nature of the behavioral ex-
citation itself. Nothing that has been said in this discussion has
made it impossible to suppose that the inhibition referred to
above does exist; but if this is assumed, the essential problem
of the time relations remains. The inhibition must be such as
to last for days, on occasion, waning slowly in the older subject
who may be capable of copulation only after days of recupera-
tion from an earlier copulation (Kinsey, Pomeroy, and Martin,
1948). This is no easier to explain than an equally slow rise
of excitability.
Let us turn now to the similar problem of sleep, before asking
how this problem might be solved.
THE ALTERNATION OF SLEEP AND WAKING
The comparatively well-defined course of sexual arousal pro-
vides a clear example of a generally unsolved problem concern-
ing cyclical changes of excitability. It cannot be asserted that
fluctuations in sexual excitability, and the diurnal fluctuation of
general excitability called the sleep cycle, must have the same
The Alternation of Sleep and Waking 217
mechanism; but the two have important similarities, and it may
be profitable to proceed on the working hypothesis that the
problems are in essence the same. The course of sexual motiva-
tion suggests that a cyclical change may be inherent in the
process of excitation, so that an extended period of arousal must
eventually be followed by an extended period of unresponsive-
ness (though the reason is hard to see); and this of course is
exactly what happens in the alternation of sleep and waking.
Physiological study of sleep has centered about two questions:
(1) whether sleep may be determined by one of several condi-
tions external to the nervous system, such as muscular fatigue,
the accumulation of toxins, and so on; (2) whether it may be
due to the action of a sleep center, or inaction of a waking
center, within the nervous system itself (Kleitman, 1939; Nauta,
1946). Research has emphasized the importance of a waking
center in the caudal hypothalamus; its action evidently has much
to do with regulation of the sleep cycle. Further knowledge on
these two points must be obtained by physiological methods,
and the present discussion will not contribute to it.
But there is also (3) a psychological factor of "interest" or
"boredom" in sleep (Kleitman, 1939). With this we are directly
concerned. The theory developed in the preceding chapters has
definite implications about sleep; sleep has in fact been referred
to repeatedly because it seems not to be an isolated phe-
nomenon but an aspect of processes fundamentally involved in
learning, for example, or volition or emotional disturbanceand
these implications make it possible to deal with the "cortical" or
"higher" components in sleep.
Return to the earlier discussions of spontaneous neural activity,
and its relation to sensation and consciousness (pp. 9, 121-4,
146). We have, to start with, (1) the suggestion by Bartley and
Bishop (1933) and Adrian and Matthews (1934) that uncon-
sciousness may be the result of a lack of afferent activity, and
(2) that this excitation normally has the function of breaking
up the synchronous spontaneous firing of central nervous tissue.
Add to this (3) that the EEG in sleep shows a marked hyper-
synchrony and (4) the general conclusion of Jasper (1941) that
hypersynchrony may be opposed to adaptive behavior. We then
have (5) a picture of sleep as a state in which sensation has lost
218 The Problem of Motivational Drift
its usual control of central neural firing, and in which a large seg-
ment of the cerebrum has drifted, from the diffuse firing neces-
sary to adaptive behavior, into the hypersynchrony of uncon-
sciousness.
From such ideas it is a short step to thinking of sleep as a
functional deafferentation of the cerebral cortex (Kleitman,
1939). But this phrase should be used cautiously. It may sug-
gest for example that consciousness is a function of the cortex
alone, or that the cortex is an isolable unit physiologically.
Even in the anesthetized animal, cortical action is closely re-
lated to activity in cortico-diencephalic circuits (Morison and
Dempsey, 1943; McCulloch, 19446; Murphy and Gellhorn, 1945).
There is probably a continuous interaction with the striatum as
well (McCulloch, 19446). Also, the term deafferentation may
overemphasize the role of sensation in the sleep cycle of the
normal animal. Important though it is, the level of sensory ac-
tivity does not alone determine sleep or waking. This of course
is the point made by Kleitman's "wakefulness of choice" as well
as by his emphasis on boredom as a source of sleep. Wakeful-
ness of choice is mainly a characteristic of higher animals: when
a point is reached in the cycle at which sleep may occur, whether
it does or does not depends on certain cerebral activities (re-
ferred to as conscious in man), as well as on the subject's
environment.
With some qualification, however, the idea of a functional
deafferentation is valuable. How would it occur? Structural
connections with the sense organs are, presumably, unchanged
as one goes to sleep. The waking center in the posterior hypo-
thalamus, whose destruction leads to a continuing sleep, is not
the portal by which afferent excitation reaches the cortex. It
cannot therefore simply act like a valve, periodically cutting off
the cortex from stimulation. However, it could be an essential
link in a system "priming" the cortex for the reception of af-
ferent impulses (Morgan, 1943). This would imply that the
diffuse firing of the cortex cannot be maintained by afferent
impulses alone; in terms of the present theory, it implies either
that one cell-assembly cannot arouse the next, in a "phase se-
quence," without supporting facilitation from the waking system,
or else that some significantly large fraction of assemblies incor-
The Alternation of Sleep and Waking 219
porate cells from the waking system. Damage to this system,
or synchrony of firing in it, would then interfere with assembly
action and consciousness.
This does not localize consciousness or the waking state but
assumes that the kind of activity throughout the cerebrum which
we call consciousness is dependent on activity within a more
limited region: primarily, the waking center in the mammillary
bodies, or in the neighboring posterior hypothalamus and an-
terior mesencephalon (Nauta, 1946); and the intralaminar and
midline nuclei of the thalamus (Penfield and Jasper, 1946; Jasper
and Fortuyn, 1946). The essential structures in waking may
also include the anterior cingular region (Brodmann's area 24,
one of the suppressor areas). The cingular area is connected
with the waking center via the mammillothalamic tract and the
anterior nuclei of the thalamus, and Smith (1945) showed that
stimulation of the area produces a transient condition apparently
equal to sleep, except in duration.
At first glance, this would mean that the cingular area is a
sleep center, not a waking center; but the function of a waking
center is to maintain a diffuse firing, and a strong electrical
stimulation firing a large number of cells in the area all at once
would prevent it from having this functionthe same point that
was made in discussing the action of other suppressor areas.
If thus the activity of the posterior hypothalamus is necessary
to the maintenance of diffuse firing, the anterior hypothalamus
may have a similar function, a pacemaker for the synchronous
cortical firing of sleep. It may, that is, be a sleep center, as
Nauta (1946) appears to have demonstrated experimentally. I
shall urge in a moment that sleep is not a negative thing, not a
mere lack of the organized activity of consciousness but the pres-
ence of an alternative organization, and the existence of a sleep
center fits into this conception.
From this point of view, we might consider that the cerebrum
deafferents itself to produce sleep: by failing to provide the
facilitation that is necessary for environmental stimulation to
maintain any control of cerebral firing. The cerebrum also de-
afferents itself indirectly by ceasing to initiate motor activity,
which at once means a decrease in sensory stimulation. But we
must go farther than this. The conditions that produce sleep
220 The Problem of Motivational Drift
are not merely negative: the fact is that some stimulations con-
tribute positively to sleep, and sleep is reinforced by learning.
Physiologically, also, it is evident that hypersynchrony is not a
random affair but organized, so that going to sleep means the
positive substitution of one organization for another. This was
referred to in an earlier chapter as the intrinsic organization of
cerebral action, and it was suggested there how a prenatal learn-
ing process might contribute to it, if the organization is not
simply due to the action of inherited pacemakers in the cere-
brum (footnote, pp. 121-2).
On the same assumptions that justified the conception of the
formation of cell-assemblies (Chapter 4), an afferent activity
that coincides frequently with a synchronous firing will estab-
lish synaptic knobs between the afferent fibers and any members
of the synchronously firing pool with which they come into con-
tact. Repetition of that afferent excitation will then tend to
support the synchrony, or re-establish it, which means a mecha-
nism of learning to go to sleep. One may have to learn to sleep
easily in the daylight or in unusual conditions; an habitual pos-
ture often helps one to get to sleep; and even habitual noises
may do so, the lack of familiar stimulation delaying the onset of
sleep orif it occurs in the nightwaking the sleeper. Kleitman
(1939) considers that the sleep cycle of the adult is a function
of learning; and in old age, when other learned behavior is de-
teriorating, we find the cycle also tending to return to a more
infantile stage. Thus an accustomed stimulation may actively
support hypersynchrony, as a learned relationship that is affected
(just as other learning may be) by cortical atrophy. All this
involves a further qualification of the "deafferentation" idea;
one might suggest perhaps that sleep is more a de-e/ferentation.
The rationale of the de-efferentation should perhaps be made
explicit here. It is simply that normal waking behavior is a
highly selective set of contractions of muscle, in which it is
usually quite essential that antagonists are not both at peak ac-
tivity simultaneously, but in alternation, and in which a par-
ticular contraction must have a particular strength and endure
only for a particular length of time. It is this that requires a
dispersed firing in the cerebral cortex. I have referred to Jas-
per's conclusion that hypersynchrony per se is opposed to adap-
The Alternation of Sleep and Waking 221
tive behavior; it is evidently opposed also to the process of
thinking, which tends to break up even the moderate synchro-
nization of the alpha rhythm. Similarly, Kleitman has remarked
the negative correlation between (1) hypersynchrony in sleep
and (2) the responsiveness of the sleeper or the extent to which
his dreams are organized and rememberable. Finally, Loucks
(1938) found that a conditioned response to direct stimulation
of the visual cortex could not be elicited in the sleeping animal:
here, clearly, deafferentation of the cortex is not an explanation
of the failure of stimulation to elicit a response.*
These facts appear to mean that the pyramidal-tract control of
ventral-horn cells and cells of the motor nuclei of the brain stem
depends on a dispersed firing in the efferent fibers from the
cortex, except when a very high degree of synchronization pro-
duces an outright convulsion. It can be supposed, then, that the
degree of synchrony in sleep (and in petit mal) is too great for
the temporally integrated control of muscle and too low to pro-
duce convulsions.
One consequence of such ideas is the conclusion that sleep
produces relaxation as much as relaxation produces sleep. This
makes possible a vicious circle, or cumulative reciprocal influ-
ence, that partly explains the gradualness of change from waking
to sleep and vice versa. The drift from a dispersed firing to
synchrony might be initiated within the cerebrum; if there is a
sleep center in the anterior hypothalamus, cells in this region
would start to recruit others and might assume control of local
cortical areas, which in turn would mean a decrease in the
number of cells available for assembly action and a decrease in
muscle tonus. The consequent lowering of afferent excitation
would further the same drift into hypersynchrony.
In waking, on the other hand, sensory stimulation would be
considered to wean transmission units away from the pools of
synchronous firing; each accession to this asynchronous assembly
* This one experimental fact alone might justify the treatment of sleep that
is made in these pages. The cortical transmission that would be involved
in a conditioned reflex, according to the phase-sequence hypothesis, depends
on the series of assembly actions that is replaced in sleep by a local syn-
chrony of firing, so even when a stimulus is applied directly to the cortex
it could not elicit an adaptive response (in deep sleep).
222 The Problem of Motivational Drift
action would increase muscle action and increase afferent exci-
tation; and so on.
The experimental evidence has shown that a particular region
has a crucial role in such changes; there is a waking center, and
Nauta has given reason for thinking that there is a sleep center
as well. But such centers may be necessary only in coordinat-
ing the action of the relatively large mammalian cerebrum, and
the same kind of process might occur in a smaller brain without
special pacemaking centers. Also, even in man, waking or sleep
may not be necessarily initiated by changes in such centers.
Sleep is produced by cortical lesions (Davison and Demuth,
1945) and may be interrupted by nightmaresa "conscious" ac-
tivity. It may be profitable to consider sleep and waking as an
alternation of excitability to which any central nervous tissue is
prone, but which can be coordinated throughout the whole
cerebrum only by the regulating action of special structures in
the hypothalamus. When the action of these structures is im-
paired, disorganization of neural firing is possible: producing
emotional disturbance. Interference with the sleep mechanism
is accompanied by irritability (Nauta, 1946); and the irritability
commonly seen when sleep is abruptly cut short, but which
nevertheless disappears soon, as the subject becomes fully awake,
indicates that between the intrinsic organization of sleep and
the conscious organization of waking there may again be an
interval of disorganization of cortical firing.
The role of psychological influence in determining the rate of
drift from the diffuse organization to the intrinsic organization
appears to mean that some phase sequences are more firmly
established than others: that is, are more capable of perpetuat-
ing themselves. The level of stimulation from sense organs
affects the persistence, and a phase sequence may perpetuate
itself by raising this level, as when one knits his brow and
pinches himself to stay awake.
Other phase sequences doom themselves to a short career by
lowering the level of stimulation, except for stimuli that make
for hypersynchrony. When such a phase sequence is in com-
mand, we say that the subject "tries to sleep," seeking relaxation,
darkness, and quiet. But as we have seen there is a further
cerebral factor, which becomes our next concern.
The Alternation of Sleep and Waking 223
This is the factor of interest or motivation, which is provi-
sionally translated into the stability and persistence of the phase
sequence. It appears first in the fact that no amount of muscular
relaxation and quiet is likely to put the subject to sleep in the
morning just after he has slept through the night; or, in other
circumstances, in the fact that a good light, a loud voice, and a
hard pew are not enough to keep one's eyes open. Again, if the
reader cares to make a little experiment and turns from this
book to a detective story, costume thriller, or sentimental family
piecewhatever his preference may behe will find himself wide
awake again though he sits in the same chair, with the same
light and the same degree of quiet, with the same degree of
eyestrain in reading the same 26 letters of the alphabet (but
happily contrived in other words). The problem raised here
concerns the relationship between sensory input and the phase
sequence; what characteristic it is, in this relationship, that tends
to prolong the phase sequence and avert sleep.
Another facet of the relationship can be added at once: no
matter how exciting a first reading of the detective story may
have been, it will have lost this property on rereading unless
considerable time has elapsed. To common sense, the reason is
obvious. The fun is gone when you know exactly what happens
next. But it is not obvious theoreticallyif the book kept one
awake yesterday, why not today? If a particular sequence of
sensory events tones up the waking center, keeps hypersynchrony
to a minimum, and maintains pyramidal-tract control of the
muscles, why should it not have the same physiological effects
a second time?
In this, as I shall try to show, there is a clue to the sources of
motivation, and "pleasure." In view of the definition of motiva-
tion proposed in the preceding chapter, that it consists of the
directedness and persistence of the phase sequence, we must
consider sleep to be the extreme case of a loss of motivation.
From the discussion in the present chapter, it has appeared that
sleep is not an isolated topicnot an isolated phenomenon de-
pendent only on certain nuclei in the hypothalamus and amount-
ing only to an intermittent cessation of cortical function. The
frequently slow transition from unresponsiveness to full respon-
siveness (that is, from sleep to being wide awake) presents the
224 The Problem of Motivational Drift
same difficulty for theory that the slow development of sexual
responsiveness does; and in his conception of "wakefulness of
choice" Kleitman has evidently related waking and sleep to a
much wider problem of motivation. Let us see what approach
can be found in the commonplace, but theoretically peculiar, fact
that an event which is exciting and interesting on its first occur-
rence is likely to be dull and sleep-provoking on its second.
EFFECTS OF MONOTONY ON LEARNED BEHAVIOR
Certain changes in the form of learned behavior, resulting
from repeated or prolonged stimulation, do not seem to agree
with the fact that repetition is usually what established that
particular form of response in the first place, and is also neces-
sary to maintain it. Dunlap (1932) has shown how the delib-
erate repetition of a response may eliminate, instead of strength-
ening, it. Hovland (1936) has described an "inhibition of rein-
forcement" from too many successive practice trials of a condi-
tioned response. As Hilgard and Marquis have remarked, Pavlov
(1928) found it advisable to space trials widely, in setting up a
difficult discrimination; and it is well known that in ordinary
learning massed practice does not have the same effect as dis-
tributed practice. Since a few seconds between trials should be
enough to avert fatigue in the neural cell, one might think that
longer intervals would only promote forgetting; but this does not
seem to be true.
Presumably related to these effects of repetition is the "fading"
discussed by Gardner and Nissen.* In prolonged training, there
are often periods when practice seems to have a negative effect.
The more he tries the worse the subject does. In man this is
apt to go with a motivational disturbance, or loss of interest.
One thinks little of it when repeated failure reduces motivation;
but after all, why should it? Also, even if the reason for this
were self-evident, a deterioration in performance can take place
when the human subject is still "trying as hard as he can," or in
an animal, working for a food reward, that continues to be
* L. Pearl Gardner and Henry W. Nissen, "Simple discrimination be-
havior of young chimpanzees: comparisons with human aments and do-
mestic animals" (in press, /. Genet. Psychol.).
225 Effects of Monotony
genuinely in need of food and apparently makes every effort to
get it. In almost any discrimination training there are periods
of deterioration of performance which are usually disregarded
as due to chance but which, as Gardner and Nissen point out,
seem to require some further explanation.
All these things might suggest that the connections involved
in learning are (1) somehow weakened by being activated, and
need a period of recovery before they can function well a second
time; but (2) are strengthened, instead, when the period of re-
covery has been permitted. Hull (1943) has shown the explana-
tory value of such an idea, though he has put the weakening
influence as a separate inhibitory process ("reactive inhibition").
How are we to deal with the apparent weakening in the present
theory?
Two possibilities may be consideredthat the cell-assembly
tends to be disrupted by its activation, or that the relationship
between assemblies changes and affects the phase sequence.
There are indications that both things may occur, and can be
distinguished one from the other by behavioral signs; and, also,
that the effect we are interested in is primarily an effect on the
phase sequence, not the assembly. The argument is as follows:
The assumptions made in Chapter 4 about the formation of
assemblies implied, first, that self-maintained activity in the as-
sembly would stop after a very short period (up to half a
second). If nonetheless the arousing sensory stimulation per-
sists without intermission, forcing a continued activity, the tend-
ency would be to induce a change of frequency properties in
the assembly. Some fractionation and recruitment would thus
be expected (p. 76). The facilitation delivered to other cere-
bral systems would then be changed, which means some change
in perception. Theoretically, this appears to be exactly what
happens in the notorious effect of prolonged staring at a word
or repeating it over and over to oneself: the familiar perception
tends to become something strange and unrecognizable. Such
effects can be regarded, then, as evidence of deterioration in the
assembly due to continuous activation.
But the assumptions of Chapter 4 also implied, secondly, that
when an assembly has become inactive it might be aroused
again in a matter of a second or so, and after the same short
226 The Problem of Motivational Drift
interval aroused once more: and so on, indefinitely. The be-
havioral evidence seems to support this, indicating that the single
assembly or small group of assemblies can be repeatedly aroused
when some other activity intervenes: in vision, for example, the
perception of vertical line must occur thousands of times an
hour; in conversation, the word "the" must be perceived and
uttered with very high frequency; and so on.
According to the theory, then, the deteriorative effects of re-
peating the same long series of words over again (loss of in-
terest on rereading a novel), or of repeating the same group of
attempted solutions (loss of motivation in problem-solving),
should be in the phase sequence instead of the assemblies that
compose it: that is, in the relation between constituent parts,
not in the parts themselves. The same conclusion is indicated
by the fact that the individual words of the reread story, or the
details of the unsolved problem, do not appear different (as the
stared-at word does) but have an annoying sameness. Also, a
very ingenious experiment briefly reported by Lewin (1938)
strengthens the argument:
The subject was required to write a word over and over until
the task became intolerable. The experimenter accepted his
refusal to write the word again, agreed to stop the experiment,
but casually requested the subject to write the word on the back
of one of the work sheets, to identify it. The subject did so
without difficulty. Thus a change of set, or intention, banished
fatigue. The assemblies directly necessary to writing the word
may have been somewhat impaired, but not to the point at
which they could not function: and the mental fatigue must
have mainly concerned the accompanying assembly activities
which constitute the subject's attitude or motivation. The re-
covery in this experiment seems closely related to the increased
output of industrial workers when a change is made in their
attitude toward a monotonous task (Roethlisberger and Dickson,
1939). The theoretical significance thus is the same as that of
the revitalizing effect of "insight" (a new combination of assem-
bliesp. 134) on interest and motivation. This effect is greatest
on the first occurrence of the new assembly combination; on
repetition, the stimulating value declines, just as when a joke or
a story is heard for the second time.
227 An Hypothesis of Excitability
These considerations imply that the phase sequence continu-
ally needs new content to maintain its organization and persist-
ence.* It appears that the individual assembly may be rearoused
indefinitely at short intervals, but that the phase sequence as a
whole cannot beit must continue to develop if motivation is to
be maintained. "Develop" here means to lead to new combina-
tions of assemblies, which in turn means new perceptions, new
insights, new ideas. As we know, such events are exciting. But
why should new combinations be continually necessary to main-
tain the phase sequence, and with it a normal level of excita-
bility? An answer can now be proposed which, in its wider sig-
nificance, will also concern the nature of pleasure.
AN HYPOTHESIS OF THE RELATION OF EXCITABILITY
TO MEMORY
To the psychologically naive there is no riddle about the fact
that hearing a story for the second time makes poor entertain-
mentthe better you remember the details of the first telling, the
less your interest in the second. Common sense thus ascribes
the change of excitatory value to a change in memory; and I
shall try next to show how this idea may be utilized theoretically.
Our question is why it is not possible to repeat a prolonged
* This conclusion is not contradicted by the fact that long periods of
monotonous work are possible in the factory, where no change of insight
or the like is possible. Habituation to monotony is apt to be accompanied
by complex daydreaming, which means the development of a parallel
phase-sequence activity, so the total pattern of cerebral activity is not at all
repetitive. The work of Roethlisberger and Dickson, referred to above,
also shows that the behavior is related to the ancillary cerebral activity.
Another possible objection can also be considered here. According to
the hypothesis presented in the following section, the infant's capacity for
an endless repetition of some action or set of words is possible because the
phase sequence is not yet well enough consolidated to lead to short-circuit-
ing. The following section modifies the conclusions of this section in one
respect, suggesting that it is only the highly organized phase sequence that
cannot be repeated in detail; if the sequence loses in degree of organiza-
tion with the passage of time, it can occur again. Accordingly, the infant's
repetitive action is considered to occur during the period when organiza-
tion is being established. With completion of that process, the repetition
disappears.
228 The Problem of Motivational Drift
phase sequence in its entiretywhy it must always lead to new
assembly combinations, or else give way to some other cerebral
activity. The answer may be that the mere occurrence of a par-
ticular phase sequence, once, induces changes at the synapse
(memory) that make it impossible for exactly the same sequence
to occur again, unless the synaptic changes have disappeared
with time. This can be made more specific, and made to com-
prise an explanation for the slow development of excitability
that has been discussed in this chapter. This is done by slightly
elaborating the assumptions about the memory trace that were
made in Chapter 4. The implication is that an immediate re-
arousal of a phase sequence is quite possible, but that the syn-
aptic changes of memory then tend to make the sequence short-
circuit * and run off so quickly that it cannot hold the field long;
if it is to endure, and remain coordinated with environmental
stimulation, new elements not so well integrated with it must
enter to delay its course.
Nowhere in the preceding chapters has there been any as-
sumption as to the permanence of the synaptic connections estab-
lished by learning (though it has been concluded that early
learning has, in one way or another, permanent effects on later
learning). This is because until now, at least, there has been
no decisive reason to assume either that synaptic knobs once
formed are permanent or that they deteriorate with disuse. If
(1) they are permanent, one has accounted directly for the per-
manent effects of early learning, and may assume that forgetting
occurs as an addition of new learning: the establishment of
further connections changing the time properties of assembly
function and so changing the facilitation delivered to other cere-
* The reader will recall that the phase sequence is "recurrent" and
"anticipatory," containing cyclical conceptual activities schematized as
A-B-A-C-B-D-E-F-D-E-G-F-H, etc.
"Short-circuiting" might cut such a sequence down to
A-B-D-H, etc.
That is, on repetition the sequence might touch only the high spots; after
some synaptic knobs have deteriorated, however, D might be no longer able
to arouse H directlyonly when E, F, and G are also aroused.
229 An Hypothesis of Excitability
bral systems. The connections originally established would still
be there, but the original habit would disappear, replaced by
a new one. Alternatively, (2) one might assume that a knob
disappears with disuse, which accounts directly for forgetting;
and, since we have seen reason (in Chapter 6) to think that new
learning incorporates earlier learning as a "transfer" effect, the
persistence of early memory can be accounted for because most
early connections are maintained by the exercise they receive as
they take part in the learning of maturity. Either of these two
assumptions might thus be made without disturbing the main
structure of the theory.
But for the problem now being considered, a solution may be
found if we can adopt assumption 2, that an unused connection
decays; or, better yet, if we can assume that a very frequently
and long-used connection becomes permanent, but that there is
a stage in the development of synaptic knobs before which the
development is reversible. The old, long-established memory
would then last, not reversible except with pathological proc-
esses in the brain; less strongly established memories would
gradually disappear unless reinforced.
Now: suppose a man picks up the book he read last week.
The title and the first paragraphs recall the storythe phase
sequence runs off quickly, short-circuited, leaving many of the
cell-assemblies briefly refractory before they can be aroused by
the rest of the printed matter. No cooperation is then possible
between sensation and central facilitation; the latter is too far
ahead. Something of this sort would account for the continued
need of some degree of novelty to maintain wakefulness of
choice. The thoroughly familiar arouses a well-organized phase
sequence; the very fact that it is well organized means that it
runs its course promptly, leaving the field for less well-established
sequences: and so, from this theoretical point of view, one would
find behavior dominated always by the thought process that is
not fully organizedone that is achieving a new organization or
one in which synaptic decay makes it necessary that organiza-
tion be reachieved.
This means that the subject will not linger over the thoroughly
familiar event even though this is the one that is most capable
230 The Problem of Motivational Drift
of arousing and controlling central processes; the control may
be momentary only. The totally unfamiliar does not arouse a
phase sequence, so "interest" and "motivation" are likely to be
preoccupied by whatever is new in the combination of familiar
events, and by events that produce some frustration or some
fear (which tend to break up the phase sequenceChapter 10).
Rereading a half-forgotten novel would thus be possible; imme-
diately after the conclusion of any exciting or interesting series
of events a repetition would not be exciting (or not as exciting,
depending on the degree to which synaptic changes have pro-
duced short-circuiting); in the following minutes, hours, or days
the subject's responsiveness to that sequence of events would
steadily rise as the synaptic changes regressed. Similarly, the
psychological function of a night's sleep may be to permit a cer-
tain decay in the particular connections that were established or
strengthened by the previous day's behavior, so that the behavior
may recur again as a cooperation between sensory and central
processes.
If it should be true that such reversible growth processes
occur, the problem of the slow changes of excitability might
thus be solved. Is the idea at all plausible? One is accustomed
to thinking of the nerve cell as a static structure (as it is after
being fixed and stained), and of growth as a very slow process.
Yet it may be recalled that Cajal (among others) conjectured
that the change at the synapse in learning is an ameboid out-
growth of the cell (see, e.g., Freeman, 1934, footnote p. 23),
which might need very little time for its occurrence.
Cajal's proposal has been disregarded by psychologists and no
longer figures in discussions of learning. Presumably, this is
partly because the whole theory of synaptic connections and
resistances fell into disrepute some time ago. But it has been
seen (Chapter 1) that modern neurophysiology makes it pos-
sible to reopen this entire question, and so Cajal's idea cannot
be at once dismissed as fantastic.
As to the time needed for an outgrowth of the cell wall, we
may note (1) that the actual distance involved may be very
short indeed, and (2) that reverberatory activity in the assembly
would serve to prolong the time in which a structural change
231 An Hypothesis of Excitability
could be completed. An ameboid outgrowth would be rever-
sible; but the mere absence of the electrochemical influence
which produces the outgrowth might not act as forcefully and
promptly as its presence. So it is possible to assume, in the light
of the psychological evidence, that synaptic decay occurs slowly
and perhaps is never quite complete. Thus each repetition of
the ameboid change at a particular point might leave a higher
residual level, and with a frequent occurrence the outgrowth
might become permanent. These assumptions then would cor-
relate directly with the fact that memory can be quickly estab-
lished but declines slowly, and with the fact that there is apt to
be little or no forgetting of the long-practiced response.
It is proposed therefore that some such postulate should at the
least be considered seriously, for it would help us to deal not
only with (1) the facts of learning, but also with the formidable
difficulty of understanding (2) the reversals of motivation dis-
cussed in this chapter, (3) the length of time over which excita-
bility (rather than excitation) may on occasion increase, and
(4) mental fatigue without a fatigue of the neurons and effectors
that enter directly into the response. The assumptions discussed
above do not help particularly to deal with a behavioral excita-
tion (except to suggest that its time course may depend on the
rate of establishment of newand reversibleconnections); but
they do appear to account directly for long-term changes in excit-
ability, and in general for the psychological component in sleep
and waking, in sexual responsiveness and so on, and in the
effects of novelty on these conditions. It is of course evident
that some degree of novelty, combined with what is predomi-
nantly familiar, is stimulating and exciting over a wide range of
activitiesfrom sexual responsiveness (Beach, 19470; Kinsey,
Pomeroy, and Martin, 1948) to the appreciation of painting or
music, or the pleasures of exploration.
I do not propose to discuss the neurological assumption fur-
ther, but only point out once more that the decay of memory
allows what has been experienced to become relatively novel
again and, therefore, exciting; so that the necessity of a lapse of
time between repetitions of some activities can now be fitted
into a more general statement of the conditions of pleasure.
232 The Problem of Motivational Drift
THE NATURE OF PLEASURE
The theory that has been developed implies that pleasure is
not the activity of particular structures in the nervous system,
and not even a particular kind or pattern of cerebral organiza-
tion, but fundamentally a directed growth or development in
cerebral organization. It is thus necessarily a transient state of
affairs in which a conflict is being reduced, an incipient disor-
ganization being dissipated, or a new synthesis in assembly ac-
tion being achieved. The preceding section has tried to explain
why such states of affairs should dominate cerebral processes; in
other words, to explain why pleasure should have the peculiar
place in behavior that it has, without covert recourse to an ani-
mistic consciousness; just as the discussion of emotional disturb-
ance and pain, in other chapters, has tried to provide mechani-
cally or deterministically for the avoidance of certain activities
and situations.
Those sensory conditions are called pleasant, then, which con-
tribute to the current development in the cerebrum, or which
consist of the decline of a sensory process that interferes with
development. If going to sleep is, as has been suggested, the
establishment of an "intrinsic" organization, the process accord-
ing to the definition above should be pleasantas in fact it is;
and to the drowsy subject those stimuli are pleasant which ac-
tively support the development of hypersynchrony. But some
eight hours later, as the subject is waking up, the pleasant sights
and sounds are those that "catch his interest" and promote
another kind of organization, without hypersynchrony. The
same stimulus is thus pleasant at one time, not at another;
when the event that could be interesting later occurs too soon
in the waking-up process, and disrupts hypersynchrony without
promptly establishing a phase sequence, it is the very reverse of
pleasant.
In view of the discussion in the preceding section of the
effect of short-circuiting on motivation, there is a paradoxical
implication here. Sensory events should not support the phase
sequence too strongly, so that it runs its course immediately and
cannot continue to dominate association-area activity. The well-
233 The Nature of Pleasure
developed phase sequence depends, for a continued existence,
on repeated checks from the environment. Hence, as we saw
in the preceding section, the preoccupation with what is new
but not too new, with the mildly frustrating or the mildly fear-
provoking. Many adult sports of course depend for their
pleasure on the presence of some danger; and both Woodworth
(1921) and Valentine (1930) have remarked on the tendency
of children to seek out the frightening situation in which the
degree of emotional stimulation remains under control and can
be terminated at will. This seems wholly parallel to an adult's
reading of "thrillers" and ghost stories.
Similarly, problem-solving involves frustration, and many of
the activities that are sources of pleasure actually depend on
problem-solvingnot only in games like bridge or chess, but
also, as Woodworth and others have remarked, in art apprecia-
tion and novel reading.
In music, again, the dissonances that are harsh and disruptive
at first become pleasant as they become more familiar, but finally
are dull and boring (Chandler, 1934). The course of events is:
first, a dissonance with too much conflict to elicit an organized
cerebral action; secondly, with the establishment of new assem-
blies, or modification of existing ones (due to repeated stimula-
tion by the new tone combination), an organized activity is
aroused, in which, however, some conflict remains; and, finally,
organization reaches a point at which the sensory stimulation no
longer offers any check to the phase sequence, and pleasure has
disappeared. This phenomenon in music might be subsumed
quite well under Woodworth's generalization concerning the
role of problem-solving in art appreciation, but still more easily
by the present formulation. What I have sought here is a gen-
eralization that would subsume not only the problem-solving
factor, but also the effects of novelty and of a mild degree of
fear, in contributing to pleasure.
In its relation to the rest of the theory presented in this book,
the discussion of pleasure has had two significances. First, it is
important to show that the theory of emotional disturbance in
the following chapter does not require that all conflict is un-
pleasant and grossly disruptive of behavior; on the contrary,
2S4 The Problem of Motivational Drift
some degree of conflict is stimulating and necessary to the main-
tenance of normal responsiveness to the environment.
Secondly, whether it is eventually found to be adequate or
not, this discussion of pleasure together with the hypotheses of
the nature of pain and emotional disturbance represents my at-
tempt to be rid once and for all of the little man inside the skull
who approves of some sensory events relayed to him by the
nervous system, disapproves of others, and guides behavior ac-
cordingly: who encourages stimulation of the afferents from the
genitalia in his crafty and teleological aim to perpetuate the
species, and who for the same purposes becomes violent when
C fibers are stimulated or when he foresees that they are going
to be. By some such approach as the one suggested, it may
become possible to understand the directedness and order in
behavior, and the variability of motivation, as produced by
neural functioning alone.
l10 Emotional Disturbances
The discussion of emotion has been about as confused as that
of any topic in psychology, partly because the terminology is
often equivocal and partly because tradition carries great weight
in this part of the field and it is hard to keep a modern point
of view consistently. The present chapter deals with emotional
disturbance instead of emotion, for reasons to be made plain;
and the discussion is limited in other respects. It is not con-
cerned mainly with sham rage and the hypothalamus, or with
emotion as a kind of awareness, or with emotion as it may exist
in states of quiet affection and the like. "Emotional disturbance"
here is used to refer to the violent and unpleasant emotions,
roughly, and to the transient irritabilities and anxieties of ordi-
nary persons as well as to neurotic or psychotic disorder. Let
me offer what justification I can for such an arrangement of
topics.
EMOTION AS DISTINCT FROM SHAM RAGE
Since the work of Bard (1934), the phenomenon of sham
rage and its dependence on the posterior hypothalamus have
held the center of the stage in the theory of emotion. But the
analyses of Lashley (1938c) and Masserman (1942), particu-
larly where they have dealt with the clinical data that were
supposed to establish the diencephalon as the seat of human
emotion, have limited the emotional functions of this region to
a motor integration. An equally significant modification of his
earlier work has been made by Bard (1942) himself: He has
reported that rage can be evoked from cats in which the pos-
235
236 Emotional Disturbances
terior hypothalamus only has been destroyeda most important
observation.
Thus, "rage" is not localized in the hypothalamus. Also, the
rage of the decorticate cat is markedly different from the normal,
so the name "sham rage" was properly chosen and should be
taken literally. Sham rage is different in its time properties,
lacking duration or after-discharge following stimulation; it is
different in the rigidity of its pattern, having almost none of the
normal variability of motor expression; and it differs radically
in the limited range of stimulation that is adequate to elicit it.
Also, Spiegel, Miller, and Oppenheimer (1940) have reported
the production of rage by lesions of (1) tuberculum olfactorium,
(2) hippocampus and fornix, and (3) amygdaloid complex; and
slight, transient signs of rage from damage to (4) the pyriform
cortex.
Apart from these considerations, there are other emotional
processes besides those that produce overt rage behavior. It is
true that fear as well as rage has been evoked from the decorti-
cate cat, but apparently only by auditory stimulation and with
considerable difficulty. Further, strong emotional disturbance
can exist in the intact animal without producing either clearcut
attack or cowering and flight. The problem of emotion, in short,
is not simply the problem of the motor pattern of sham rage (or
"sham fear"). It is of course essential to have discovered the
unexpected extent to which rage and fear patterns may appear
in the decorticate animal. Bard's work was one of the most
important single contributions to understanding emotion. But,
given that knowledge, we must go on next to ask how the hypo-
thalamic expression is modified by the presence of a large
thalamus and cortex, how learning affects emotional behavior,
and the relationship of emotion to perception and thought.
EMOTION NOT A STATE OF CONSCIOUSNESS
Traditionally, emotion is an awareness, an event in conscious-
ness. Here, perhaps more than anywhere else in psychology, a
traditional interactionism (which is animism) tends to persist.
The afferent excitation is thought to produce a feeling or aware-
ness, and that feeling then acts on the nervous systemit must
Emotion Not a State of Consciousness 237
do so, according to such ideas, for it is the feeling that makes
the subject sweat or tremble or run away, and the sweat glands
and the legs are controlled by nerve fibers.
Just such an inconsistency of thought has led to an endless,
and pointless, debate on the James-Lange theory of emotion.
What James and Lange were accounting for, again, was that
emotional feeling or awareness; they postulated that the aware-
ness is a set of sensations, that the awareness follows and does
not cause emotional behavior ("I see the bear, I run, I feel
afraid"). Their critics proceeded to show that emotional expres-
sion is still there in the dog whose sensory processes have been
interfered with, and thought this a refutation of the theory. But
such an argument is totally irrelevant; James did not say that
emotional behavior depends on sensations from the limbs and
viscera.
It is equally irrelevant to show that emotional expression is
centrally organized, in the hypothalamus. If James had raised
the question at all, he must have assumed something of the sort;
remember that what he denied was that consciousness intervenes
between stimulus and response. Therefore, showing that the
"higher" centers of the cortex are not needed for the response is
clearly support for his position. The extraordinary non sequitur
of such criticisms, made by some very distinguished critics in-
deed, would not be possible if there were not the immutable
idea that only emotional awareness or feeling can produce emo-
tional response. If the response is there, the feeling must be
also. Such illogic, assuming James to be wrong first, in order to
prove him wrong, is the clearest evidence of the hold traditional
ideas have on psychological thought.
But we must get rid of the tradition both for scientific con-
sistency and because psychological observers have been unani-
mous in denying that there is any special, fundamental category
of consciousness that can be called emotion. "Emotional experi-
ence is a highly variable state [and] often partakes of the
complicated nature of a judgment" (Landis and Hunt, 1932). I
have reviewed the evidence on this point elsewhere (Hebb,
1946a) and need not go into it again. Since it seems that the
term emotion does not refer to a special kind of event in con-
sciousness, and since in any case we must not slip into the in-
238 Emotional Disturbances
consistency of treating an immaterial awareness as a causal
agency, the term is not very useful in its traditional significance.
At the same time, we must postulate that the disturbances of
emotional behavior have a neural origin; and the term emotion
still can be useful to refer to the neural processes that produce
emotional behavior.
It is important to be clear that in this discussion "emotion" is
a reference to the hypothetical neural processes that produce
emotional behavior; explicitly, it refers neither to an immaterial
state of consciousness nor to the observable pattern of emotional
behavior.
EMOTION AS A DISORGANIZING INFLUENCE
The third point to be clarified is the distinction between emo-
tional disturbance and those processes (also called emotional)
which are inherently organizing and motivating. A paper by
Leeper (1948) has urged that all emotion be considered to have
such an integrating function. He has expressed a commonly
held point of view, and has at the same time I believe clouded
the issue with which he dealt. His position demands consid-
eration.
Leeper has shown effectively that regarding emotion as disor-
ganizing has led some of us into an inconsistency, but he has
proposed another inconsistency to take its place. He points out
that others have first defined emotion as a disorganization of
behavior and have then gone on to give mother-love as one
example of emotion. This is certainly inconsistent. But how is
it different logically from defining emotion as something that
makes behavior more efficient and at the same time giving, as
examples, the stage fright that ruins a performance, the anger
that makes a boxer less skillful, and the mental depression that
makes a man unable to earn a living? I must add at once that
Leeper anticipated this criticism. He answered it by saying that,
though emotion may sometimes disorganize, the disorganization
is not characteristic because it occurs only (1) when emotion
is extreme, or (2) when the emotion conflicts with some other
motivation. As far as one can see, however, the first of these
Emotion as a Disorganizing Influence 239
explanations depends on a mistaken use of analogy, the second
begs the question.
1. It might be doubted whether emotion is disorganizing only
when extreme. A touch of shyness or of self-consciousness, for
example, may sometimes ruin one's conversational skills. Leave
this to one side, however, and accept the assumption that only
strong emotion disrupts behavior.
The idea that such an effect does not indicate the true nature
of emotion depends on this analogy: An excess of salt or oxygen
has toxic effects, "but we do not use this as a means of deter-
mining the normal functions of such products. The same logic,
then, should apply to emotional processes" (Leeper, 1948, p.
15). But we must really use the same logic. We must, that is,
ask the same question in both cases. The question does not
concern the "normal function" of emotion, but what emotion is.
No one doubts that fear has a useful function, aiding survival;
the question is, how? Might fear not be a simple disrupter of
behavior and still be useful, if the disruption is mainly of the
behavior that gets us in trouble? Emotion could then be disor-
ganization and still have its survival value, because the disor-
ganization is selective, eliminating some actions and allowing
others to take their place.
If we accept the analogy between emotion and salt we still
can say that emotion is disruptive of behavior, that it generally
has a useful function when it occurs in moderate degree and
bad effects when there is too much of itexactly as with salt or
oxygen.
2. In discussing the stage fright that keeps a pianist from
giving a good performance, Leeper says the emotion is not a
disorganizer of behavior because, by itself, it would produce
only flighta well-organized pattern of behavior. This seems to
beg the question, since the pianist neither runs away nor stays
to play well. Leeper says it is not the emotion that produces his
incoordinations, but the conflict. But if there were no conflict
at all, would there be any emotion? Are the conflict and the
emotion quite independent?
The difficulty here is clearest by Leeper's own criterion of
"organization." He points out first that any directed behavior
means the suppression of conflicting tendencies; so conflict is not
240 Emotional Disturbances
ipso facto disorganization. The test is "whether [an] interfer-
ence is relatively chaotic and haphazard, or whether the sup-
pressions and changes of subordinate activities are harmonious
with some main function that is being served." The trembling,
palmar sweating, disturbance of breathing, and incoordinations
of the pianist's fingers, obviously, are not harmonious with the
main function being serveduntil he actually runs away. Could
we regard his emotion as solely organizing even then? See what
this implies. If we are really consistent in saying that any dis-
organization in emotion is accidental, that emotion is essentially
an organizer only, then a well-organized avoidance is at least as
much an instance of fear as an ill-organized one. Consequently,
the pianist who anticipates the possibility of stage fright, and
calmly decides to keep away from the stage, provides as good
an example of emotion as the one that trembles before an audi-
ence. We have deliberately got rid of any criterion of emotion
except that it "arouses, sustains, and directs" behavior. We have
equated avoidance and fear, aggression and anger. But these
surely are not identical.
The upshot of such an argument is to broaden the category
of emotion so much that it includes all psychological processes.
We should then have to find another name for the distinctive
event thatat presentis called emotional.
How are we to find some halfway point between this position
and the one that Leeper has criticized so effectively? I think it
is evident that Munn, Young, Dockeray, Woodworth, Landis,
and so on (authors criticized for treating emotion as disorgani-
zation), were after all not as absurd as Leeper thought them;
at the same time, he has made untenable the proposition that
"emotion" simply means a disruption of behavior.
A way out, I believe, can be found if we do two things: first,
stop talking about emotions as a single, fundamental kind of
unitary psychological process, and separate (1) those in which
the tendency is to maintain or increase the original stimulating
conditions (pleasurable or integrative emotions) from (2) those
in which the tendency is to abolish or decrease the stimulus
(rage, fear, disgust), including, however, depressions in which
the organism may discover no way of escape from the condition
giving rise to the emotion. Secondly, if it is assumed that
241 Rage and Fear in Chimpanzees
stimuli in class 2 above are essentially disintegrative, it must also
be recognized that the disintegration in rage or fear is often
incipient or potential and likely to be successfully averted by the
aggression or avoidance of the subject.
As to the first of these points, we have no need to bring to-
gether pleasure-seeking and melancholia, mother-love and temper
tantrums, always in the same single category"emotion." I have,
in the preceding chapter, discussed pleasure as a special prob-
lem; and I propose in this one to try to deal with a quite dif-
ferent sort of process. Since "emotion" undoubtedly will often
continue to mean affection or pride as well as anxiety or shame,
"emotional disturbance" is a better way of designating the sub-
ject matter of the present chapter.
RAGE AND FEAR IN CHIMPANZEES
The theory of emotion proposed by Watson (1924) is by now
classical. It held that there are three innate emotions, rage,
fear, and love: rage aroused by a restriction of physical move-
ment, fear by a loud noise or sudden loss of support (and pre-
sumably, pain), love by stimulation of the genitalia. These
stimuli may then serve to condition others: if a child fears ani-
mals, it is because he has been scratched by a cat or knocked
over, perhaps, by a friendly dog while learning to walk. As soon
as one tries to apply this theory to the emotional disturbances
that are actually observed in children, serious difficulty shows
up; it is not really plausible, and it has been subjected to pow-
erful criticism by Dennis (1940) on rage, by Jones and Jones
(1928) on fear, and by Valentine (1930), also on fear. But it
has been hard to document the criticism with experimental evi-
dence, for two reasons. First, it is not safe or socially permitted
to arouse a strong degree of emotion in human subjects for
experimental study. Important as the topic is, consequently, we
have surprisingly little exact information about the causes and
consequences of human emotionabove all, of adult emotion.
Secondly, when one studies emotion as it occurs socially, outside
the laboratory, it is almost impossible to know the antecedent
conditions as they must be known if one is to understand the
present responses of the subject.
242 Emotional Disturbances
But neither of these objections need apply to the study of
emotion in animals, and the chimpanzee fortunately has an emo-
tional repertoire that is very like man's. The things that annoy
us are apt to annoy him, and he shows his annoyance much as
we do; his fears in some respects are astonishingly like man's
in what is feared, in complexity of the causes of fear, in the re-
lation of fear to intellectual development, in resistance to extinc-
tion or reconditioning, and in the marked individual differences
between one subject and the next. Even more, there are data
on what were with practical certainty two cases of neurosis or
psychosis in mature chimpanzees whose histories were known
from birth or very early infancy, one born in the Yerkes Labora-
tories of Primate Biology and one captured in the first year of
life. These animals of the Yerkes colony can be studied as
human subjects cannot. Their histories are known in detail,
being recorded in individual diaries. With these data, one is
in a position to see just what complexity of response must be
provided for in a theory of emotion, chronic as well as acute.
The causes of rage may have nothing to do with physical re-
straint of movement. When the chimpanzee Dita, in heat, would
sit where he could watch her from the next cage but one, Don
seemed calm (if not content); but he had a temper tantrum
repeatedly whenever she left the outer cage for an inner room
where he could not see her. When Mona had a noisy temper
tantrum because Pan had stolen a peanut from her, Pan was
finally enraged to the point of beating her up. A chimpanzee
may be angered by a reproof, by being startled, or by being
obliged to look at something unpleasant such as a model of a
snake. These causes act "spontaneously"; that is, the emotional
disturbance does not have to be learned, any more than a human
being must learn specially to be angry at disturbance of the
peace, at being shown something nasty, or at being momen-
tarily scared by a practical joker (Hebb, 1945a)although, as
we shall see, a certain kind of learning, in infancy, may have to
have been established first.
The causes of fear include not only loud noise or sudden fall,
but snakes with which an animal has never had contact (this
can be known certainly with some chimpanzees reared in the
nursery), solitude, and the strange or mysterious. McCulloch
243 Rage and Fear in Chimpanzees
and Haslerud (1939) have analyzed such fears in a young chim-
panzee reared quite apart from others, and were thus able to
show that they are not necessarily acquired by "social condition-
ing" or imitation. I discovered accidentally that some of the
chimpanzees of the Yerkes colony might have a paroxysm of
terror at being shown a model of a human or chimpanzee head
detached from the body; young infants showed no fear, increas-
ing excitement was evident in the older (half-grown) animals,
and those adults that were not frankly terrified were still con-
siderably excited. These individual differences among adults,
and the difference of response at different ages, are quite like
the human differences in attitude toward snakes, the frequency
and strength of fear increasing up to the age of seventeen or so
in persons who had never been injured by a snake (Jones and
Jones, 1928). The increase fits in with the conception that many
fears depend on some degree of intellectual development, and
so with the idea that the range of such fears is characteristic of
"higher" animals like man and chimpanzee (McBride and Hebb,
1948).
The observations using detached heads were then followed up
by others, using various stimuli: an isolated eye and eyebrow, a
cast of a chimpanzee's face, a skull, a cured chimpanzee hide,
an anesthetized chimpanzee.
Two things appeared: first, that such stimulation may be pri-
marily a source of profound excitation, rather than specifically
fear-producing; the excitation was usually followed by avoid-
ance, but sometimes by aggression, and sometimes it produced
a marked autonomic reaction combined with apparent friendli-
ness. On occasion, a confusing mixture of all these reactions
was seen.
Secondly, this primitive excitation appeared to be fully parallel
to the human emotional disturbance that may occur at the sight
of a badly deformed face, in watching a major operation, dis-
section, or autopsy for the first time, or as a result of contact
with a dead body (Hebb, 1946&). This human disturbance,
like the chimpanzee's, is not specificit may take the form of
nausea, of vascular disturbance and dizziness, of disgust or fear,
or may even facilitate anger (e.g., at "desecration" of the dead).
It is important to recognize individual differences of sensitivity
244 Emotional Disturbances
to such experiences, and the loss of sensitivity with repeated
exposure (which is presumably an effect of learning); but this
does not mean that the responses are wholly learned. The evi-
dence is very definitely to the contrary: the degree of excitation
is likely to be strongest on the first exposure, provided certain
other experiences have preceded. On the other hand, the dis-
appearance of response with repetition must be an effect of
learning, and I believe we must assume that it is learning also
that establishes any integrated response to such stimuli, trans-
forming a nonspecific emotional disturbance into an organized
flight (disturbance plus a tendency to flight then constituting
fear) or aggression (constituting rage).
The complex origin of emotional disturbance may be clearest
in the infant fear of strangers. About the age of four months
the chimpanzee reared in the nursery, with daily care from three
or four persons only and seeing few others, begins to show an
emotional disturbance at the approach of a stranger (Hebb and
Riesen, 1943). The disturbance increases in degree in the fol-
lowing months. This is "shyness" but may become much more
violent than that term would usually imply. It commonly shows
up in the human baby about the sixth or seventh month. (The
age difference is in keeping with the fact that chimpanzee de-
velopment generally takes only about two-thirds of the time for
the corresponding human development.) Everyday experience
tells us that shyness does not occur in the human child that is
continually in contact with strangers; and that it is very strong,
as a distrust or fear of "foreigners" (even in adults) in isolated
communities. We might then suppose that the excitation out of
which the fear develops is aroused innately at the sight of a
stranger, but extinguished if enough new faces are seen with no
accompanying injury. But this supposition is wrong: there is
definite evidence that the response is not innately established.
Dennis (1934) pointed out that human patients, congenitally
blind and operated on to restore vision, have shown no emotional
disturbance at the first sight of a human face. Chimpanzees
reared in darkness, and brought into the light at an age when
the response normally would be at its strongest, show not the
slightest disturbance at the sight of either friend or stranger.
But some time later, after a certain amount of visual learning
245 Mental Illness in Chimpanzees
has gone on, the disturbance begins to appear exactly as in other
animals.
So it appears that the emotional disturbance is neither learned
nor innate: a certain learning must have preceded, but given
that learning the disturbance is complete on the first appearance
of certain stimulus combinations. Dennis (1940) has also made
it clear that temper tantrums are not learned, and still require
that other things have been learned first. We are accustomed
to think of any particular response as either learned or innate,
which is apt to be a source of confusion in thinking about such
things as far apart as an insightful act or neurotic behavioris
the response inherited, or acquired? The answer is, Neither:
either Yes or No would be very misleading. The irrational emo-
tional disturbances of man and chimpanzee are fully dependent
on learning, but are not learned in the usual sense of the term.
On the other hand, in order to account for the specific proc-
esses of rage and fear, and the integrated, adaptive responses of
aggression and flight, we must I believe give learning its role.
This point will be returned to after broadening the field of our
inquiry by considering two cases of mental illness in chim-
panzees.
MENTAL ILLNESS IN CHIMPANZEES
Now, before turning to the mechanism of these responses, let
me report two cases of apparent neurosis or psychosis in the
chimpanzee. Here again we have the advantage of knowing
more of the previous experience of the patients than is possible
with most human patients in the clinic. We have no verbal
recollections to go by (recollection in clinical cases, after all, is
apt to be unreliable); but the animals' experience, in cage life,
is much less varied than a man's would be andonce morethe
chimpanzees of the Yerkes colony have been under daily obser-
vation by trained observers who have faithfully recorded at the
time of its occurrence any observed illness, unusual experience,
or peculiarity of behavior.
We have further, in the particular cases to be discussed, the
invaluable asset of control subjects. It happens that four chim-
panzees were reared together for a large part of the develop-
246 Emotional Disturbances
mental period, in the same circumstances: Alpha, Kambi, Bula,
and Bimba. Alpha and Bula were born in captivity, Kambi and
Bimba were captured wild, at about nine months of age. Noth-
ing was recorded, and nothing could later be recalled, by the
trained workers who reared them, that would account for any
great differences in the adult personalities of these four chim-
panzees. Yet, by the time of adolescence, one of the four
(Kambi) was a hypochondriacas far as behavior could show;
and one other (Alpha) in full maturity suddenly developed an
extreme phobia. Let me give these histories in more detail.
Alpha was born in 1930, the first chimpanzee to be born in the
Yerkes colony. She was taken from her mother immediately and
reared apart from other chimpanzees until she was one year old
(Jacobsen, Jacobsen, and Yoshioka, 1932). She then was brought
into the laboratory, as one of the group of four to which I have
already referred. The menarche was at age eight, and her sexual
behavior with males has always been entirely normal. The con-
census of the staff was that Alpha, as she became mature, had
a strong liking for human company, was not particularly bright
and was very obstinate but still a good experimental subject.
Otherwise she was just a chimpanzee. Chimpanzees, like men,
have their individual peculiarities, and no great prognostic sig-
nificance was attached to some small peculiarities of Alpha's
although, in the light of her later psychiatric problem, some of
the peculiarities were peculiarly Freudian. Alpha would defi-
nitely not have anything to do with any carrot with a forked
root, nor usually with any long cylindrical vegetable such as
okra, though when carrot or okra was cut up before her eyes she
would accept and eat the pieces immediately. Finally, she had
a marked tendency to masturbate, frequently against the water
faucet projecting from the wall of her cage, whenever a human
being came in sight.
In 1942, when Alpha was twelve (corresponding roughly to
full maturity), a serious disturbance of behavior appeared. With
no warning signs, she suddenly refused all solid foods, despite
being very hungry. It then appeared that she would eat at once
if food was cut small for her; and a number of tests showed con-
clusively that she was violently afraid of contact with any piece
of food more than an inch or so in length or thickness. Con-
247 Mental Illness in Chimpanzees
siderations given elsewhere (Hebb, 1947) lead me to believe
that this behavior cannot be sufficiently explained as due to in-
jury, as from a thorn or stinging insect in the food, though such
event may well have been the precipitant. The fears fluctuated
considerably, but waned gradually until they had completely dis-
appeared two years after the original onset; then, after another
year, they reappeared with equal suddenness and severity (this
second episode clearing up more promptly, however).
A further phenomenon, of the greatest importance for showing
the parallel with clinical cases, was observed during the first of
these two neurotic or psychotic episodes. Four months after the
first onset, when the fear of food had greatly decreased, Alpha
out of a blue sky developed a strong fear of me, later a violent
aggression, and then (having been soaked with water from a
hose for her persistent attacks) became quite friendly, at the
same time reverting to her fear of food.
Two friendly critics, on the occasion of the original report of
this case, were able (with serious intent) to diagnose the case
and explain the cause of illness. Alpha, in short, was in love
with me; the earlier fear of forked carrot and of cylindrical ob-
jects, and the masturbation against a projecting water faucet,
were not the only Freudian features of the case. The aggression
occurred because Alpha had had to repress her unmaidenly de-
sires, with an inevitable result. This is an interesting idea; liking
chimpanzees (or some chimpanzees), it would please me to
think them capable of such delicacy, such capacity for repressing
vulgar desirein short, to think of chimpanzees as having Super-
egos.
But if Alpha had a Superego its operation was rather hit-and-
miss, and in many of her dealings with other chimpanzees (or
with the staff), and in some of her personal habits, there was
little sign of self-imposed restrictions. If Alpha was in love, it
must have been with someone else; and her emotional disturb-
ance had no relation to fluctuating level of sex drive, which in
chimpanzees is directly correlated with the menstrual cycle. It
might be amusing to try to psychoanalyze Alpha, but it seems it
would not be very profitable either as therapy or as a contribu-
tion to knowledge.
The second history is quite different.
248 Emotional Disturbances
Kambi was captured at the estimated age of nine months and
reared constantly in the company of other females of her own
age. From eighteen months of age onward, she was one of the
group of fourAlpha, Kambi, Bula, Bimba. Morphine addiction
was briefly established at the age of eight (Spragg, 1940). She
showed almost none of the usual sex play before menarche (at
age nine) and copulated once only after italthough she showed
the physical signs of coming into heat periodically, with the
menstrual cycle. She was spayed at the age of twelve for experi-
mental purposes, with no apparent change in over-all personality.
To summarize Kambi's behavioral disturbance first in clinically
intelligible language, she gave the impression from early child-
hood of being introverted, with a very strong preference for
human company, and subject to periods of depression. During
adolescence, the depression when it occurred was extreme. The
attacks continued after maturity and often lasted for months at
a time. Her depression was sometimes initiated by a somatic
illness; when it did begin so, it did not clear up with the illness
but continued long after.*
The actual evidence of depression is as follows. Ordinarily
Kambi was easily annoyed by failure to get something she
wanted, but she was always responsive to (and acted in a way
that was likely to get) attention from the human staff; definitely
more interested in such attention than in food, but ordinarily
getting enough to eat. But then for a period up to six or eight
months in length she would sit day after day with her back
hunched up against the wall of her cage, never seeking attention
and rarely responding to human overtures, and almost totally
uninterested in food so that if she were not caged alone she
might have starved: definitely a different animal.
Once more the case shows a Freudian element, in Kambi's
* I am indebted, for some of the observations in this summary, to Dr.
Henry W. Nissen, who knew Kambi thoroughly in her infancy and at ma-
turity. It was he who obtained her in Africa and brought her to the
Yerkes colony with Bimba and a number of other young animals, and later
had personal charge of their rearing. I should emphasize that the descrip-
tion is a convenient way of reporting behavior and does not necessarily
imply anything about conscious states. Kambi in short acted as a depressed
introverted human might act.
249 Mental Illness in Chimpanzees
lack of sexual motivation (which has, however, been observed in
other animals at the Yerkes laboratories, with no sign of psy-
chotic involvement). But no traumatic sexual event can be ad-
duced to account for this. The one copulation that was observed
was quite normal, and Kambi's lack of sexual responsiveness had
appeared much earlier. Disturbed ideas about sex are common
in human mental illness, but so are disturbances of eating and
sleep, and there is usually no evidence to show that sexual con-
flict occurs first and is the cause of the more profound disturb-
ance of which it is part. In Kambi's case there was no evidence
of sexual conflict at allsimply a complete lack of apparent
interest.
The emphasis in these two histories of chimpanzee psychosis,
or neurosis, is on the absence of any particular experience to
which the behavioral breakdown can be ascribed. But I do not
suggest for a moment that experience is not a factor; I shall urge
later that experience is a factor in almost any emotional break-
down. The point here is that if it was the main factor, in these
two cases, it must have operated in a way that we do not at
present understand at all, and quite unlike the mechanism of the
"experimental neurosis."
The fact to be kept in mind is that we have two chimpanzees
as control subjects (in an unplanned experiment, since no one
foresaw the later course of development in Alpha and Kambi,
but still an excellent experiment). After capture, Bimba had
the same general experiences as Kambi (both captured at the
estimated age of nine or ten months), in every respect that our
present knowledge would suggest is important. There is no
basis for postulating some special trauma for Kambi before she
came under psychological observation, for the diaries of the two
animals show that for the first three years of captivity it was
Bimba, not Kambi, that was less well adapted to her environ-
ment: more fearful, less trusting, less cooperative in an experi-
mentuntil the age of four, when the roles were gradually re-
versed.
Again, Alpha and Bula were both born in captivity; and
though Bula was left with her mother until the age of fourteen
months whereas Alpha was removed at birth, there is little sign
250 Emotional Disturbances
in the diaries of the subsequent four or five years to show that
this had any effect on their later adjustments.
Experience, and experience alone, may have been the decisive
factor in the breakdown of two of these four animals at maturity;
but, if so, it operated in some way that is quite mysterious in the
light of our present knowledge of behavior. Freudian theory
has the credit of recognizing the existence of a kind of learning
that causes, apparently, no immediate emotional disturbance
and yet may contribute to one much later. Such learning un-
doubtedly occurs, and may well have been involved in Alpha's
or Kambi's illness, but if this is to be postulated we must find
some way of incorporating it into other learning theorynot as
an ad hoc assumption specially made to deal with mental
illness.
MECHANISMS OF EMOTIONAL RESPONSE
What I believe to be the mechanisms of primary emotional
disturbance have been discussed in preceding chapters, together
with a mechanism of emotional learning. It remains to pull the
earlier discussions together and to try to show how a mechanism
of learning might transform emotional disturbance, as a break-
down of adaptation, into the adaptive responses of attack, flight,
and so on: that is, how anger or fear might develop out of an
undifferentiated emotional disturbance, and how a failure to
develop an adaptive mode of response might constitute neurosis.
The aspects of emotional disturbance that mainly determine
this treatment of the problem can be summarized as follows: (1)
the great variety of causes of disturbance, ranging from an un-
familiar combination of familiar things (fear of the strange) or
an interruption of sleep, to hunger, nutritional deficiency, or
withdrawal of a drug from the addict; (2) the fact that a single
cause may produce anger, fear, or nausea and faintness, in the
same subject at different times, or in different subjects; (3) the
great variety of expression even of a single emotion; and (4) the
different ways in which the expression of emotion changes, as the
subject is habituated to the stimulating conditions.
To account for item 1, it has been assumed that the emotion
is a disturbance in the timing of thalamo-cortical firinga dis-
Mechanisms of Emotional Response 251
ruption of the phase sequence when the subject is awake, of the
intrinsic organization of cerebral activity when he is asleep. The
disruption may be slight and brief, in this hypothesis, or exten-
sive and prolonged; it may be produced by a conflict of phase
sequences, by a lack of sensory support for the phase sequence,
or by metabolic changes.
Now the question is, how far a second assumption can account
for items 2, 3, and 4, listed above. This assumption was made
earlier (p. 150), to account for the learning by a rat to choose
a lighter door and avoid a darker one, in discrimination train-
ing. It is, in short, that a disruption of thalamo-cortical timing
tends to prevent the recurrence of the phase sequence that led up
to that disruption on previous occasions, because of the cyclical
(anticipatory and recurrent) organization of the phase sequence.
This is a mechanism of learning, and it is proposed now that it
would account for the adaptive features of rage (emotional dis-
turbance plus attack) or fear (the same thing plus avoidance),
as learned behavior that is determined by the accompanying
emotional disturbance. Particularly, it would account for the
variability of the behavior and for the marked tendency of the
emotional element itself to disappear as the originally disturbing
situation is repeated.
Above and beyond the question of the intrinsic plausibility of
the idea that flight, in fear, is a learned response (it may not be
as implausible as it seems at first glance), is the apparent neces-
sity of my first assumption, that emotional disturbance is a dis-
ruption of cerebral timing. If this is necessary, we must then
find some way of accounting for the other aspects of emotional
behavior. Let us look again, then, at the causes of emotional dis-
turbance, which may be arranged in three classes.
Class I. Take as a starting point the infant chimpanzee's fear
of the strange. It has already been seen that this is not innate,
not a fear of what is totally strange, since the animal reared in
darkness does not show it until vision in general has begun to
have some "meaning" for him (until it arouses phase sequences).
Consider further that a familiar attendant A, wearing the equally
familiar coat B of another attendant, may arouse the fear just as
a complete stranger would. A causes no disturbance; B causes
none; A and B together cause a violent emotional reaction. It
252 Emotional Disturbances
is this sort of fact that makes it necessary to suppose that two
phase sequences may interfere with one another. The first class
of emotional disturbance thus is one caused by a conflict (which
may be extended to the disruptive effect of pain stimuli and
other "unpleasant" sensory events).
Class II. Now a second case. It was for long heresy to sup-
pose that there could be a fear of darkness. Darkness is a lack
of stimulation (at least after the first few seconds in the dark);
how can it arouse an emotional response? Fear of the dark is
not likely to appear until the age of three or later in human chil-
dren (Jersild and Holmes, 1935) and so the skeptic who thought
that his child developed the fear without cause could be an-
swered by the old appeal to ignorance: how can he be certain,
with the varied unsupervised experiences a child must have in
a three-year period? However, the violent avoidance of solitude
by young chimpanzees (Kohler, 1925) is at least as marked as
the human fear of the dark, and can be shown to be due only
to the lack of the perception of companions. This then implies
a second class of emotional disturbance: fear of the dark, fear
of solitude, fear aroused by loss of support, and the emotional
changes observed by Ramsdell * following prolonged absence of
the normal auditory stimulation. It would comprise also emo-
tional disturbance due to contact with a dead body (which lacks
warmth and the usual responsiveness); anger at the lack of social
response in another person ("inattentiveness"), or a monkey's
anger at not finding the particular kind of food reward he ex-
pected (Tinklepaugh, 1928); grief; homesickness, and so on.
* Personal communication from Dr. Donald A. Ramsdell, who was con-
cerned to understand the high frequency of marked emotional changes
which he observed in veterans following traumatic loss of hearing. He
arranged experimentally to deafen himself, effectively eliminating airborne
sounds for a period of three days. The results are described by Dr.
Ramsdell as follows, particularly with regard to the question whether the
deafness would contribute to any neurotic or psychotic tendency:
"In my experience there was a definite increase of irritability and a de-
sire to either withdraw from the field or, if held in the field, to 'charge' into
it. This would be comparable to being irritated with a friend, and either
withdrawing without saying anything or 'letting him have it.' Such be-
havior would not be characterized by ideas of persecution nor suspicion of
the friend. However, you might say that the desire to 'charge' into a situa-
tion is such stuff as paranoia is made of, but for a true paranoid idea of
Mechanisms of Emotional Response 253
That a conflict of phase sequences, or a lack of sensory sup-
port for the phase sequence, should produce any extensive dis-
ruption in the timing of cortical action is not a logically inescap-
able deduction from the schema of neural action presented in
Chapters 4 and 5. What the schematizing actually did was
suggest the possibility, as something that would account for a
similar effect of two so dissimilar causes and make the relation-
ship between perception and emotional disturbance intelligible.
The idea then seemed to gain strong support when it drew atten-
tion to another set of facts that is usually forgotten in discussing
emotional theory. These facts concern the relationship of emo-
tional disturbance to metabolic changes.
Class III. The third class of the causes of fear, rage, and so
on, though it was actually arrived at last in formulating these
ideas, comes nearer than either of the others to being a neces-
sary consequence of the neural schema. To the earlier discus-
sion of the effects of chemical changes in the nutrient fluids that
bathe the neural cell (p. 196), showing that a disturbance of
timing must result, I need add only that a large intracranial
tumor, compressing blood vessels and interfering with blood
flow, obviously could have similar effects; and also that asthma,
or vascular disease, or an antigen acting to produce vascular
spasm in intracranial vesselsany of thesemight produce emo-
tional disturbance directly, by affecting the amount of oxygen
in the blood or the amount of blood that is supplied to neural
tissues. Emotional changes are in fact frequently associated
persecution, the capacity for interpersonal relationship must be so weak that
real suspicion can develop. Assuming then interpersonal relationships to be
secure, I would say that deafness is accompanied by a feeling of perplexity
and disorientation analogous to the fear of the dark or fear of solitude.
Such a disorientation in one's social milieu need not necessarily involve
paranoid disturbances any more than being in the dark would induce one
to hear footsteps at his back. In my opinion, the feelings of personal in-
adequacies (impairment of the body image); irritability, and an exagger-
ated response to stimuli; and the wish to leave the field or, if held in the
field, to 'charge' into it, characterize the experience of deafness."
I should like to draw attention also to Dr. Ramsdell's suggestion that the
emotional condition might lead to more than one overt pattern of response.
This is relevant to the later discussion of the way in which different
"emotions" might develop from the same primitive disturbance.
254 Emotional Disturbances
with all such conditions; and the one assumption that seems to
provide a common ground for such varied sources of disturbance
(including not only the metabolic changes in Class III but also
the perceptual conflicts and perceptual deficits of Classes I and
II) is the assumption that emotional disturbance is in the first
place a disruption of the timing of neuronal activity in the cere-
brum. This also accounts directly for the incoordinations of
emotion. Until some better guess is made, we must see what
can be done with the assumption; and this imposes the burden
of accounting for the integrated and coordinated aspects of
emotional behavior (Leeper, 1948).
Now this coordinated part of the behavior, in the unpleasant
emotions, has one constant function. It is always such as to
tend to put an end to the original stimulation (in the pleasant
emotions, of course, this tendency is reversed). Aggression, in
rage, tends to change the irritating behavior of another animal
by cowing, driving off, or killing the annoyer. Flight, in fear,
tends to prevent or terminate noxious stimulation. Avoidance-
turning the head away from an unpleasant sight, covering the
ears, holding the nose, or withdrawing the hand, as well as ac-
tually running awaytends to stop the stimulation that arouses
either an irrational fear of some inert object or the practically
equivalent disgust. Fawning may be a manifestation of fear as
much as flight, and so may the desperate attack of a cornered
animal: and each is a possible way of putting an end to the
emotion-arousing behavior of another organism.
Failure to respond, in shame, is a special case, in which one's
own behavior is the emotion-provoking stimulus; the effect of
the emotional disturbance is negative, preventing the recurrence
of that behavior, but this is also in line with the present thesis,
namely: All these effects may be achieved because strong emo-
tional disturbance tends to prevent the repetition of any line of
thought that leads up to it, and to eliminate the corresponding
behavior. In some situations, a primitive, undifferentiated dis-
turbance would develop into fear, where withdrawal is the only
thing that ends the disturbance; in others, anger, when attack is
the only effective behavior; in still others, either fear or anger
might result, when either behavior would be effective, and
which result happens would be determined by the animal's past
Mechanisms of Emotional Response 255
experience as well as by his own inherited peculiarities. But in
some situations, and for some subjects, no behavior might be
discovered that would abolish the emotional disturbance; and the
effect of learning in this case might be to reinforce the visceral
disturbance and the incoordinations of the skeletal musculature,
andabove allthe disorganization of the phase sequence, which
is a disorganization of thought.
This is the point at which neurosis and psychosis enter the
theoretical structure: they are (1) conditions of chronic emo-
tional disturbance, or (2) ones in which emotional disturbance,
in the past, has effected a lasting modification of the thought
process (see footnote, p. 259).
The first objection that might be made to such an approach is
this. As the hypothesis has been presented so far, it might sug-
gest that aggression, in anger, appears only by defaultonly be-
cause other forms of behavior have been eliminatedand to
common sense this is not reasonable. Aggression appears to
have a much more intimate and necessary relation to the emo-
tional disturbance of anger. And so with avoidance, in fear.
My first answer is that this may be perfectly true, in the adult;
even in the four-year-old. But this proves nothing about the
origin of the relationship. Emotional disturbance and the ag-
gression may seem inseparable to us for exactly the same reason
that nodding the head and Yes are inseparablein this culture:
because the two have been tied in together by long practice and
experience.
It is instructive to observe just what happens in emotional
learning, when we do have a chance to observe it. Bridges
(cited by Jones, 1933) has shown that children's emotional pat-
terns develop out of a primitive undifferentiated excitation. This
fact may not be crucial, for the change may depend on physical
maturation and not on a learning process. But when chimpan-
zees are exposed for the first time to certain emotion-provoking
situations, the first effect again appears not to be a specific pat-
tern of adaptation but a generalized excitation. When Alfalfa,
one of the chimpanzees reared in darkness by Riesen (1947),
was brought out into the light she did not show any emotional
disturbance at the sight of strangers; as I have already pointed
out, the emotion requires a certain amount of visual experience
256 Emotional Disturbances
before it occurs. When signs of the disturbance appeared some
time later, they seemed completely unspecific: a raising of the
general level of excitation, with signs of autonomic action and
undirected movements. I have remarked elsewhere (Hebb,
1946&) on the lack of a consistent mode of response of some
adult chimpanzees at the first exposure to a disguised person,
wearing a mask and acting in an aggressive manner which they
were not used to encountering in human beings; and also on the
gradual emergence of a consistent mode of emotional response,
but different in form for different animals.
The implication of my hypothesis, then, is that adult intro-
spection cannot be an argument against its validity, since the
essential learning would be thoroughly established in early in-
fancy.
Observe further the conditions in which the adaptive behavior
of adult emotion occurs. Apart from the spinal reflexes aroused
by noxious stimulation, a fully coordinated emotional response
appears always to be a response to premonitory cues, not in
themselves disruptive, but associated with disruptive stimulation.
When the primitively disruptive event occurs suddenly, without
any warning, the response is never a smooth and economical
cooperation of effector organs but is likely to involve startle,
trembling, sweating, vocalization, vascular change, and so on,
and is then followed by a coordinated response which differs
according to the subject's perception of the total situation. If
the adaptive behavior of fear (that is, avoidance) is indeed a
response to events that are not directly disruptive but premoni-
tory cues associated with disruption, the behavior is by defini-
tion a learned response. It is then intelligible why there is
usually less emotional activity at maturity. As experience in-
creases, rage and fear would tend to disappear in the familiar
environment, for the utilization of premonitory cues, to avoid
disturbance that is foreseen but not yet in existence, would be-
come more and more efficient. The integrative part of emotional
behavior and the disintegrative are inversely related: as one in-
creases, the other decreases.
The objection that we have just considered is that the con-
nection between aggression and emotional disturbance, in rage,
is not an accidental oneestablished merely by association or
Mechanisms of Emotional Response 257
by the elimination of alternative modes of reaction. The objec-
tion may be taken further. It cannot really be a matter of
chance whether fear or anger will result, even in situations in
which either avoidance or attack would have the same effect
of changing a disruptive stimulation. Characteristically, some
situations evoke rage, others fear, and so on. The objection car-
ries weight, and must be considered: I should say at once that
I do not believe that it is sufficiently answered by what follows,
and there are, with reasonable certainty, important constitutional
factors determining the mode of emotional response which I
cannot deal with. It is also possible that milder degrees of dis-
ruption innately tend to lead to aggression, stronger ones to
avoidance.
In addition, however, factors of learning are present that
should not be overlooked. One thing that determines whether
fear or anger will result when the phase sequence is disrupted
may be the particular phase sequence itself, and the effect of
earlier learning in similar situations, which determines what
alternative sequences are available to take over when the dis-
rupted sequence leaves the field open. With this in mind, ob-
serve that anger is characteristically aroused by the behavior of
others (i.e., by social stimulation) when the behavior is not con-
sistent with what one is proposing to do, when one expects re-
sistance in having the behavior changed, and finally when one
thinks one may be able to change it.
It is a humiliating fact that one tends not to get angry at
prize-fighters, or at the criticism of the higher-up whose praise
or approval is important. This is far from being absolutely true,
but it seems that anger is most easily aroused by opposition
where one thinks one could prevail by force: where the oppo-
nent ought to see that you can get your own way, and give in
gracefully.
I believe it is also true that, the more authoritarian the regime
in which a child grows up, and the more frequent and severe
the punishment he suffers for disobedience, the less likely he is,
as an adult, to get angry at higher authority and the more likely
to be angered by insubordination. If so, it means that anger
does not occur in situations where disruptive stimulation has
regularly broken up the phase sequence that determines aggres-
258 Emotional Disturbances
sion; that it does occur where aggressive behavior has not usually
been disrupted before. (That such evidence should have to be
resorted toI know of none apart from such folkloreis the best
indication of a lack of scientific knowledge, on a very important
topic in psychology, which is due to the difficulty of experi-
menting with any but the mildest of human emotions.)
Now a final objection to the hypothesis that learning is what
mainly differentiates the various emotions. Nothing has been
said so far to provide for the exceptional vigor that an emotional
response is apt to have. This I believe is the greatest defect in
the hypothesis. Several things might be suggested to remedy
it. One possibility is that when disorganization occurs it leaves
transmission units in the cerebrum ready to fire with any subse-
quent activity. These units would ordinarily be involved in the
assembly action that is now prevented; they may then tend to
be recruited by whatever assemblies do act subsequently: in-
creasing the strength of the facilitation delivered to motor cen-
ters but also tending to make the motor response less precise
(since the recruitment of extra cells would tend to change the
frequency properties of the individual assembly). This might
also be applied to the pleasurable activity, discussed in the pre-
ceding chapter, that depends on first arousing apprehension,
then dispelling it. It may offer some cue also to the nature of
laughter, which may be most efficiently aroused, perhaps, when
there is a combination of phase sequences that produces a minor
disturbance of timing immediately followed by a further com-
bination of assemblies that have no conflict. There would still
remain the fragments, so to speak, of assemblies that were dis-
rupted, which tend to induce some motor activity. In some
cases this may induce anger or fear or embarrassment, if the dis-
ruption is extensive; but in other cases a cortically controlled
overactivation of the breathing mechanisms may serve to pick
up the fragmentswhich appears as laughter. First a minor
conflict, incongruity or apprehension, then a new perception of
the situation that leaves no need of avoidance, nothing to be
changed in the behavior of other persons or in the total situa-
tion, but still leaves a condition of excitation that must issue in
one form of behavior or another.
259 The Nature of Mental Illness
THE NATURE OF MENTAL ILLNESS
It has already been proposed that mental illness consists either
of a chronic disturbance of time relations in the cerebrum, or a
lasting distortion of the thought process from such a disturbance
at an earlier time.* This is a rather general proposition, and
we must now consider the theoretical problems of mental illness
in more detail. The general approach that has been proposed
may have value only in helping to maintain a point of view that
is hard to maintain, and yet essential. The point of view is this:
neurosis or psychosis is a product neither of experience nor of
constitution, but a joint product of both.
Tradition, once more, is the villain of the piece. In this field
we are still dominated by the old dichotomy: is the illness mental
or physical, functional or organic, and now, under new names, is
it psychosomatic, or just somatic?
It may make sense to ask whether a breakdown is primarily
experiential, implying that with an ordinary degree of strain it
would not have occurred, or primarily constitutional, implying
that the mildest of strains would have caused the breakdown,
that the patient's ability to withstand strain was low. But this
cannot give us two kinds of mental illness, any more than, in a
completely parallel situation, there are two kinds of tuberculosis.
If there are not two kinds of mental illness, we have no basis
for dismissing the mental illness that goes with pellagra or some
glandular dysfunction as not being a "true" psychosis; no basis
for thinking of a special category of chronic emotional disturb-
ance, of which experience is the sole cause, and psychotherapy
in some form the sole method of cure. Logic and the whole
clinical literature require that we remember the role of the con-
* This distinction is not further elaborated in the text. It is primarily
based on theoretical considerations, but it seems to agree with the fact that
the neurotic or psychotic frequently appears to have found a way of think-
ing and acting that obviates emotional disturbance. The illness is perhaps
either (1) a continued emotional disturbance, with anxiety or the like, or
(2) changes in thought (new phase sequences) that are not characteristic
of the majority of the population but which avoid some conflict that makes
for major disruption of assembly or phase sequence.
260 Emotional Disturbances
stitution in any mental illness, just as we must remember the role
of experience. It is hard to avoid going to extremes when this
topic is discussed; it is very hard to keep from thinking in terms
of the old dichotomy; but I believe we must keep a middle way.
Until quite recently, the task of the psychologist and psychiatrist
was to keep the medical world from thinking of mental illness
as always due to some lesion of the central nervous systema
lesion that usually could not be found. Now the pendulum has
swung to the other extreme, and the medical world may be in
some danger of overestimating the role of psychological causa-
tion in both mental illness and illnesses that used to be called
physical. The task is still to find a middle way.
I believe there is no longer a danger that psychological factors
in illness will not be recognized: there is hardly a popular journal
in the western hemisphere that does not publish articles explain-
ing how illness may be due to worry or to a desire to get out of
some sort of trouble. In medical practice it may happen that
failure to make a definite diagnosis can automatically determine
that the patient is psychoneurotic. This tendency has been pro-
tested against by physicians concerned with the welfare of the
patient (Bowman, 1946; Wilson and Rupp, 1947). But it also
represents a mode of thinking by the psychologist or psychiatrist
himself that may hinder the development of theory. It is as
much incumbent on the psychologist to demonstrate the limits
of psychological influence in illness as to demonstrate its extent;
as matters stand today, the first need for research in the field
may be to ask what we do not know about psychological in-
fluence.
Psychology's ignorance about a number of these matters is
bad, but it is worse when we substitute dogmatism for scientific
knowledge. We still need an Ajax to stand up and defy the
lightning and ask, What is the evidence? when some authority
informs the public that believing in Santa Glaus is bad for chil-
dren, that comic books lead to psychological degeneracy, that
asthma is due to a hidden mental illness, or that a child's bed-
wetting must be treated by psychoanalysis, and psychoanalysis
alone, instead of the simple and effective method of condition-
ing described by Mowrer and Mowrer (1938).
Let me then present four propositions for serious considera-
261 Mental Illness and Constitution
tion, even though they may sound ridiculous: (1) that there is
no separate class of mental illness caused by psychological fac-
tors aloneby what was seen and heard in childhood; (2) that
we know little about "mental hygiene" or how to achieve it; (3)
that it has not been shown that worry or the like can by itself
cause any bodily condition such as asthma or stomach ulcers,
though it aggravates such conditions; and (4) that it has not
been shown that any specialized psychotherapy, such as psycho-
analysis, has any special value in mental illness. The first of
these I think is logically inescapable; propositions 2, 3, and 4
may not be as absurd as they appear. What I am talking about
is proof; one may know that psychoanalysis is of value and yet
not have the evidence in a form that would establish this as a
scientific fact. Obviously, in a scientific argument one must
accept the scientific canons; and putting the evidence in logical
shape should ultimately lead to improved medical procedures as
well as more precise theory.
If we are to improve therapy, or if prophylaxis is to be pos-
sible at all, we must improve theory first and know more exactly
what leads to what. As long as such questions as those raised
above are answered on a priori opinion alone, theory will get
nowhere, for the experiments necessary to it will not be done.
Evidently the current development of psychological ideas in
medicine has a profound necessity behind it: let us see what
kind of solid factual evidence we need in order to keep this
development on the rails and to extend its influence, not only in
the clinical but also in the social and educational fields.
MENTAL ILLNESS AS A FUNCTION OF THE
CONSTITUTION
The first question raised was whether there is any class of
mental illness that is due to experience alone. My point of view
here has already been stated: that such a class does not exist,
since any behavioral condition must be a function both of expe-
rience and of the constitution that does the experiencing. We
may take it, however, that the relative importance of experience
and constitution may vary in mental illness. At one extreme, it
might be assumed that there is no human constitution that will
262 Emotional Disturbances
not break down when emotional disturbance is sufficiently severe
and prolonged (Swank and Marchand, 1946); at the other, that
metabolic changes alone may produce emotional disturbance
that will persist as long as metabolic factors remain variable and
unstable. But the theory that has been developed indicates that
the essential nature of the resulting condition may be the same
in both cases. There have been an increasing number of reports
in the literature showing that trained diagnosticians have not
been able to distinguish mental illness precipitated by a physio-
logical disturbance from the mental illness for which no such
cause can be discovered, and reports as well showing that the
characteristic "functional" disturbance is accompanied, to a more
than chance extent, by constitutional deviations.
Typical neurotic or psychotic disorder may be found in brucel-
losis (Harris, 1944); in pancreatic disorder (Romano and Coon,
1942); in adrenal disorder (Allen and Broster, 1945; Greene,
Paterson, and Pile, 1945); in pernicious anemia (Bowman, 1935;
Ferraro, Arieti, and English, 1945); in rheumatic brain disease,
following rheumatic fever (Bruetsch, 1947); in hypochloremia
(Saphir, 1945); in vitamin-B deficiencies (Williams, Mason, and
Smith, 1939, cited by Jolliffe; Jolliffe, 1942; Matthews, 1938;
Egana et al., 1942; Spies, Aring, Gelperin, and Bean, 1938).
Certain types of neurosis are associated to a more than chance
degree with inherited vascular abnormality (Hauptmann, 1946;
Cobb, Cohen, and Badal, 1946). Dewan and Owen (1945) have
reported twelve from a much larger number of cases of schizo-
phrenia, manic state, depression, psychoneurosis, and so on,
which were so representative of the usual picture of mental ill-
ness that no physiological or pathological source was even sus-
pected; but the causes eventually were found to include brain
tumor, diabetes, uremia, and so on: in all cases, a major cause
in the "organic" class.
It may be impossible to distinguish neurosis following head
injury from other neurosis (Gillespie, 1944). It is common, as
soon as some such "organic" cause is found, to reclassify the ill-
ness; but such reports as those cited make it impossible to con-
sider that there is any real psychological discontinuity between
organic and functional ills, as these terms are commonly used.
This first point is made both on logical grounds and on the basis
263 Mental Illness and Constitution
of the actual clinical facts. It is further reinforced by the fac-
tual evidence and discussion of Cobb (1944), Hoagland (1947),
Landis (1947), and Bellak and Willson (1947).
Secondly, these considerations make it possible that a con-
tributing organic cause will be found for all those cases that are
now called purely functional. The history of psychiatric knowl-
edge shows one syndrome after another being removed from the
functional class, as pathological and physiological knowledge has
increased. There are still gaps in our knowledge of the complex
chain of events in the nutrition of the neural cell; and it is still
possible that all well-defined cases of mental disorder will even-
tually be found to depend, to a major extent, on a predisposing
constitutional conditionhereditary or acquired.
Consider for example Kallman's work, the main results of
which have been summarized by Garrison (1947). Among other
statistics are the following: among the relatives of patients with
schizophrenia, the incidence of schizophrenia for brothers and
sisters is 14.3 per cent; for fraternal twins, with the same consti-
tutional similarity as any brother or sister, 14.7 per cent; for iden-
tical twins, 85.8 per cent. For stepbrothers and sisters, growing
up in the same home, the figure is 1.8 per cent. There is no
apparent way in which these data can be reconciled with the
idea that mental illness is unrelated to heredity, that it is due
alone to the way a child is brought up. One can argue that two
children of different ages have different environments though
they are brought up in the same home. An oldest child cer-
tainly has a different social environment from that of a youngest
child, and of course any one child in a family may have some
bad experience that another does not. This might explain why
an ordinary sibling of the schizophrenic (born at a separate
time) is less likely to have schizophrenia than a twin. Butthis
difference, predicted by the environmentalist hypothesis, is not
observed: siblings and fraternal twins have the same frequency,
exactly as would be expected on the genetic hypothesis. And
there seems to be no way at all of accounting for the difference
between the figures of 14 per cent for fraternal twins, and 85 per
cent for identical twins. As we shall see in the following chapter,
the data from twin studies have not shown a great deal about
264 Emotional Disturbances
the relation of heredity to the IQ; but they could hardly be much
clearer as far as schizophrenia is concerned.
There is, however, still the other side of the coin. Eighty-five
per cent is not perfect coincidence; it does not make heredity
the sole determinant even of schizophrenia, and still less the sole
determinant of some other mental ills. A paper by Hobbs
(1941) makes the point perfectly: one of a pair of identical
twins, brought up with very strict moral standards, broke train-
ing as an adult and a psychotic episode followed. The other
twin showed no ill effects; heredity provided only a susceptible
constitution and did not alone determine that the psychosis
would occur. A similar case is referred to by Bellak and Willson
(1947).
The same conclusion can be drawn from the very significant
experiments of Pavlov (1928), Gantt (1944), Liddell (1938),
Masserman (1943), and others on the "conditioned neurosis" or
"experimental neurosis." I have suggested (Hebb, 1947) that
we should be cautious in identifying this with a full-fledged
neurosis in man, since the experimental neurosis is apparently
specific to the situation in which it was established. But there
can be no doubt that this work is very important for the under-
standing of clinical neurosis, showing how anxiety (at the least)
can be established in a way that would not otherwise have been
suspected, since no actual injury or threat may be involved.
Anxiety is one source of mental illness, and Pavlov showed that
this chronic emotional disturbance can be created in some dogs
much more readily than in others: apparently, a constitutional
difference between dogs (Gantt, 1938). But here, again, there
is cooperation between constitution and experience, for it takes
a very special experience indeed to produce the emotional break-
down in most animals, even when they are definitely susceptible.
THE PRESENT SIGNIFICANCE OF MENTAL HYGIENE
The second point of our present ignorance about mental ill-
ness is the way in which early experience may be related to a
disorder at maturity. The mental-hygiene conception was estab-
lished some years ago, when it was possible to think of mental
illness as simply a set of bad emotional habits. It seemed then
265 Mental Hygiene
that if one could avoid such habits in childhood one could be
sure of mental health at maturity, whatever the constitution one
may have inherited.
Now this may still be partly true; the unfortunate fact, how-
ever, is that we do not have any good basis for being sure how
children should be brought up, in order to minimize their
chances of getting into a state hospital. The necessary experi-
ments have not been done. What may be necessary to decide
the question is to bring up some children in one way, some in
others, and see what leads to what. One cannot assert that too
much coddling produces mental illness at maturity, or that too
little does. For all we know, a lot of spanking, or unsympathetic
parents, may only help to prepare the child for the trials of
maturity. It cannot be assumed that what produces emotional
disturbance in the infant is badnor that it is good: the long-
term effects may be the same as the short-term, unrelated, or
opposed. One must make some decision when a practical prob-
lem has to be dealt with, and in these circumstances the only
thing to do is make the best guess one can; but opinions should
be presented as opinion, and not as scientifically established
fact.
In such matters, our ignorance is virtually complete and vitally
in need of being replaced by definite knowledge. We do not
know whether sexy comic books at the age of ten have any dele-
terious effect at twenty; we do not know that play with toy
guns and soldiers makes a war-loving nation; we do not know
that juvenile delinquency, associated with broken homes, is due
to the home environment and not just as much to the inherit-
ance of the same emotional instability that broke up the home.
It has sometimes seemed that answers to such questions might
be found by anthropological study; when different primitive
peoples treat their children differently, what are the effects at
maturity? In one tribe, for example, the children are treated
generously and the adults are found to be easy-going and
friendly; in another, child training is harsh, and the adults are
found to be full of hostilities. But, as Klineberg (1940, p. 434)
has pointed out, one cannot single out cause and effect here,
any more than in the case of the broken home. Heredity, in
the two tribes, may differ significantly, producing different adult
266 Emotional Disturbances
personalities; or some other variable such as the history of the
tribe may be the source of the differencesand the reported facts
may mean only that emotionally stable people treat their chil-
dren well.*
The mental-hygiene program first needs research, then more
research, so that we can find out how to rear children to make
them as healthy as possible, emotionally, at maturity. It is im-
portant to apply what knowledge we have now as best we can;
it is equally important to remember how incomplete the knowl-
edge is. We are not now in a position to do more than apply
the conceptions of one or other theory of mental illness, that
becomes insufficient as soon as one tests it by facts. I have
already cited the cases of Alpha and Kambi, chimpanzees of the
Yerkes colony, to show that, if early experience is a main cause
of mental illness at maturity, we have no real information as to
how it works. The theory that has been presented here is much
too vague, or general, to provide any useful guide in the prac-
tical problems of child education; it may be useful, however, if
it leads to more of the obviously necessary research.
PSYCHOSOMATICS AND PSYCHOTHERAPY
The third and fourth points of ignorance that have been listed
can be considered together: the questions are whether worry or
the like can produce bodily lesions, and whether psychotherapy
can cure mental illness. Though the questions may sound ridicu-
lous, I think it may be of some value to consider them seriously
and ask oneself just what factual evidence one could cite to jus-
tify one's answers.
Just where can one look for the specific evidence on which to
base the psychosomatic thesis? Clearly, it is not enough to show
that neurotic disturbance and pathological changes in the body
go together,! or even that when the first clears up the second
* I.e., reverse the statements above: "In one tribe the adults are easy-
going, and so children are treated well; in another the adults are aggres-
sive, and children are treated harshly"as one might expect.
f Obviously, it is also not enough to show that emotion is accompanied
by transient somatic changes in normal or psychopathic persons. The very
important studies of Wolff (1943) and Wolf and Wolff (1946), for example,
267 Psychosomatics and Psychotherapy
does alsounless the recovery is due to psychotherapy. If some
metabolic disorder contributed to the neurosis, it may have de-
termined the pathological changes also. Then mental and phys-
ical difficulties would appear together, and disappear together.
If one could show that asthma, or stomach ulcers, or skin lesions
could be cured by talking to the patientthat is, by psycho-
therapythen we would have the desired proof. But, as we
shall see shortly, there are great difficulties in the way of show-
ing that psychotherapy is effective (chiefly, the difficulty of
showing that it produces more cures than are accounted for by
spontaneous remission). Despite its general acceptance, in
which I concur, the hypothesis that psychological factors affect
the recovery of the body from any physical illness has not been
established scientifically. It may appear to the reader that it is
not necessary to do so; but there must be much that should be
found out about the psychological influencein what conditions
it operates best, what its limits are, and so on; and these things
will never be discovered until we start examining our assump-
tions, and start the slow, plodding process of building up a solid
body of facts, separate from opinion, upon which the theory of
treatment can be firmly established.
Not all cases in which a "psychosomatic" relationship is seen
are ones in which the original trouble was emotional. The fol-
lowing case of early asthma offered an exceptional opportunity
for separating cause and effect. It is rare that one can have an
effective control of asthma, and banish it or not, at willbut
when this is possible, new light is cast on the disease.
A child of twenty-one months developed acute, severe attacks
of asthma after having had a head cold some months earlier and
a persistent nasal infection. The attacks became less frequent
several months later, during the summer, but recurred with in-
creasing severity and frequency as winter set in. Eventually it
do not show that chronic gastric and cardiac lesions result from emotional
disturbance alone, without contribution from a preceding physiological dis-
order, although, if it is assumed that "normal" emotion has such effects,
these studies help to show us how the lesions are established. It should be
clear that what is discussed in the text is the question of scientific proof,
not of reasonable belief. Belief in the psychosomatic thesis is reasonable,
perhaps inevitable; but the very importance and value of that thesis require
that its foundations be inspected and tested rigorouslyand periodically.
268 Emotional Disturbances
appeared that a running nose preceded each attack, and it oc-
curred to the physician in charge that a nasal infection might be
the source of the allergen. He found a heavy chronic bacillary
infection, and recommended chemotherapy. The results were
dramatic. The asthma disappeared completely, and a remark-
able change took place in the child's behavior. During an attack,
the child had been almost completely incapacitated; but between
attacks, there had been no difficulty in breathing and the child
appeared quite normalexcept that she was always emotionally
unstable, easily giving way to tears and being rather hard to deal
with. Actually, this behavior had developed so gradually dur-
ing the period of illness, and was so constant, that the parents
had not realized that there was any connection between the
emotional picture and the intermittent asthma. Now the patient
became a different child.
Two further observations made the interpretation of these
facts even clearer. It was expected that the allergic sensitivity
would gradually disappear as the child grew older; so on two
subsequent occasions when a nasal infection began the treatment
was omitted at first, in the hope that asthma would not develop.
On each occasion the same sequence of events was observed.
The first symptom of disturbance was circulatorythe child's
hands and feet became very cold. This was followed by the
same emotional disturbance that had been chronic earlier, at
least a day before the first symptoms of difficulty in breathing;
and finally, a definite asthma appeared. This and the emotional
changes were then promptly (in six to twelve hours) banished
by chemotherapy. Such a sequence of events can only mean
that the allergen affected the vascular system firstthe cold
hands and feet were not due to lung congestion and decrease of
oxygen in the blood stream. On one of the two occasions on
which treatment was withheld, moreover, a placebo was used.
This served as a control of the effects of suggestion, showing
that it was in fact the action of the drug which, when it was
actually used, produced the recovery from asthma and the dis-
appearance of whining, crying, and the need of continual atten-
tion. Only one conclusion seems possible: that the allergen
acted more or less directly on the vasculature, including the
cerebral vasculature, or on the central nervous system, thus caus-
269 Psychosomatics and Psychotherapy
ing both emotional and physical symptoms. Ordinarily, with
multitudinous sensitivities and no possibility of such cleancut
"experimental" manipulation of allergens, cause and effect can-
not be separated. The case described might have been a typical
psychosomatic picture: first emotional disturbance is evident,
then physical symptoms follow. So the first must cause the
second. But in this case such an interpretation is not possible.
Because asthma and emotional disturbances go together, or
even because an emotional disturbance can precipitate an attack
of asthma, it does not follow either that emotion caused the
original difficulty, or that psychotherapy can do more than alle-
viate an illness that may have quite a different basis. It is still
possible that some cases of asthma, or some other bodily illness,
are psychogenic and can be cured by psychotherapy; but evi-
dence of this has not yet been provided in a form in which it
can be evaluated. Future research may well show that the
causal relationships in the case I have described are the excep-
tion and not the rule (i.e., may show that allergies are usually
psychogenic). But this research has not yet been done.
It is implied throughout this discussion that, if psychotherapy
can be shown to be an effective treatment of asthma or mental
illness, the argument above would be reversed. It seems gen-
erally believed that the validity of psychoanalytic theory, for
example, is well established by the fact that it has led to curing
mental illnesses that would not otherwise have been cured.
If one looks for the evidence, however, one cannot find it.
There is no meaning in the fact that a particular patient was
psychoanalyzed, and recovered. We do not know that he would
not have recovered without the treatment.
The point is clear when it is recalled how difficult it was to
show that the insulin-shock method had any value. Psychiatrists
and physicians in general were skeptical. Why? Because some
not-very-exactly-determined proportion of psychotic patients re-
cover without treatmentperhaps 30, 35, or 40 per cent. The
figure varies widely from one series of patients to another; and
it took five to ten years of work, by a large number of investi-
gators, to finally establish the fact that insulin therapy signifi-
cantly increases the proportion of patients that recover from
schizophrenia. A few patients recovering after lobotomy would
270 Emotional Disturbances
mean nothing: Freeman and Watts' (1942) triumph was in being
able to report not only 63 per cent good results in a long series
of chronic cases of mental illness, but much higher percentages
with certain types of case. These figures had an immediate
meaning, since the research on shock treatments had already
made known what the spontaneous recovery rate would be,
roughly, in such cases. No such data, so far as I can discover,
have ever been provided as a test of any form of psychotherapy
that is, the frequency of recovery in an adequately chosen, un-
treated control group is not available to compare with the fre-
quency after psychotherapy is applied. The one exception to
this statement is the paper by Denker (1946), and his conclu-
sions do not show any special value in formal psychotherapy.
Denker's paper is based on admittedly unsatisfactory data, and
obviously is not a final answer to the question it has raised; yet
it stands as a very important paper, simply because it is the only
one, apparently, that has attempted an essential task.
The data reported by Denker are the rates of recovery in a
series of 500 patients with severe incapacitating neurotic dis-
order, treated by general practitioners throughout the United
States. These patients were all receiving disability payments
from an insurance companya fact that might delay recovery, at
least as compared with the recovery of patients who were pay-
ing for treatment. At the end of five years, the recovery rate
was 90 per cent. Denker also reviewed the reported recovery
rates under more specialized treatment. The average rate of
recovery under psychoanalysis is roughly 62 per cent; under psy-
chiatric treatment (which may have included psychoanalysis),
roughly 70 per cent. These rates are markedly lower than that
obtained by the general practitioner. However, if we assume
that the results by specialized treatment are for a shorter period,
perhaps two years, we find that the general practitioner's results
are about the same: 72 per cent. The conclusion is that rest,
sedation when necessary, reassurance, and time are what pro-
duce recovery from neurotic disorderwhen recovery occurs.
Obviously, this conclusion is not final. We do not know that
the patients treated by psychoanalyst, psychiatrist, and general
practitioner, in Denker's analysis, were really comparable (how-
ever, the one respect in which it is certain that they were not
271 Shock Treatment and Lobotomy
comparablethe economic factoris such as to strengthen Den-
ker's argument). Denker's conclusions may be quite wrong; but
the real point is that his are the only data * we have by which
to evaluate psychotherapy.
Such a result cries for research; if nothing else has induced the
psychoanalyst to give us some control data, some evidence for
the validity of his theory and treatment of mental illness, this
should do so. In the meantime, there is no body of fact to show
that psychotherapy is valuable; there are no data to prove that
psychotherapy can cure lesions of the gastrointestinal tract, der-
matitis, and so on; and none to prove that there is a class of
"functional" mental illness that is produced by emotional dis-
turbance alone and susceptible of cure only by talking to the
patient or allowing the patient to talk. I think any psychologist
(or physician) must believe that psychotherapy is important, as
an adjunct of other treatment, and even by itself; but if we are
to increase its efficiency, we must learn more about it, and the
first thing to establish is the definite fact that it works at all.
SHOCK TREATMENT AND LOBOTOMY
The final concern of this chapter is a rationale for the shock
and surgical treatments that revolutionized the field of psy-
chiatry.
Lobotomy, a surgical operation that severs most of the con-
nections between the frontal pole and the rest of the brain, and
* An earlier discussion by Landis, however, based in part on the same
data, has clearly led to the same conclusion and must be regarded as rein-
forcing Denker's conclusions (C. Landis, A statistical evaluation of psy-
chotherapeutic methods, in L. E. Hinsie, Concepts and problems of psycho-
therapy, New York: Columbia Univ. Press, 1937, pp. 155-169). Landis
first determined the reported rate of recovery from psychoneurosis, apart
from specialized methods of treatment, as varying around a central figure
of about 68 per cent. Speaking of another such figure (for all mental ill-
nesses), he went on to say that, although the figures given were not strictly
speaking those for spontaneous remission, "still any therapeutic method must
show an appreciably greater size than this to be seriously considered."
Finally, he reported a figure of 58 per cent for psychoanalytic treatment of
the neuroses. The inference, not made explicit by Landis, is the same as
Denker's.
272 Emotional Disturbances
electroshock treatment, in which an electric current passed
through the skull induces a brief convulsion, both appear to
work best with patients suffering from emotional depression, and
both frequently have the effect of banishing the depression. In
trying to understand how a cure is effected, the first question
might be, What do such different procedures have in common?
The surgical treatment (Freeman and Watts, 1942, 1946)
severs a large number of connecting fibers in the brain and
leads to retrograde degeneration of a particular nucleus in the
thalamus (the nucleus medialis dorsalis). A number of investi-
gators have expressed the opinion that it is the loss of this tissue
that decreases the strength of the patient's emotion and allows
him to recover. But no such gross destruction of tissue results
from the passage of the electric current through the head. Is
there any other effect of the surgery that might produce emo-
tional changes?
A suggestion comes from the study of frontal lobe removals.
A comparatively common operation, prefrental lobectomy, is
usually done so that it severs the same connections that are
severed in lobotomybut severs all other connections of the
same area, and actually removes the tissue. In these operations,
consequently, the medial dorsal nucleus of the thalamus must
degenerate. But the kind of emotional picture drawn by Free-
man and Watts (1942), as representing the results of frontal
lobotomy, is sometimes definitely not present (Nichols and Hunt,
1940; Mixter, Tillotson, and Wies, 1941; Hebb and Penfield, 1940).
On the basis of this fact it was suggested (Hebb, 19452?) that
the effect of lobotomy might be due not to a simple loss of some
function resident in the frontal lobes but to a physiological dis-
turbance set up in that area by an operation that cannot help
but be scar-producing. This would account for the fact, also,
that the recovery from depression, in lobotomy, is reversible
although the degeneration of the thalamo-frontal connections
and of the medial dorsal nucleus is not reversible.
A physiological disturbance in the cerebrum is usually de-
tected in the electroencephalogram, or EEG, by the presence
of large potentials (hypersynchrony). My proposal, therefore,
amounted to this: that in those patients that recover from de-
pression following lobotomy the EEG will show a hypersyn-
273 Shock Treatment and Lobotomy
chrony; and that if relapse occurs the hypersynchrony will have
disappeared or decreased. Cohn (1945) has reported persistent
disturbances in the EEG "in most subjects" of Freeman and
Watts' series, and delta waves in ten out of fifteen in one group
of patients examined four to six years after operation. This is
almost exactly the reported frequency (63 per cent) of chronic
recovery from depression. Unfortunately, I have not been able
to discover whether those with "bad" EEC's were ones with
good clinical results (i.e., lasting recovery), as my hypothesis
would predict.
However, a direct confirmation of the hypothesis, up to a cer-
tain point, has been provided by Hoagland, Malamud, Kaufman,
and Pincus (1946). They report a lowered incidence of fast
waves (14 per second or faster) in patients showing remission
from depression following electroshock treatment, and the return
of fast waves at the time of relapse. This is directly in the line
with the hypothesis; although it may be a strain on the term
"hypersynchrony" to say that it may appear in a statistical de-
crease of fast waves, the change is actually in the same direction
as if delta waves had been observed.
If the hypothesis should be further confirmed, the effects of
electroshock and lobotomy both could be understood on the fol-
lowing basis. It has been proposed here that emotional disturb-
ance consists of a disorganized firing of cerebral neurons. The
establishment of hypersynchrony, due to any cause, would pro-
vide a pacemaker, as it were, that would tend to pick up any
neurons that are not incorporated in assembly activities and so
substitute the "intrinsic" organization of cortical activity for the
diffuse organization of the normal adult waking state. The
presence of the hypersynchrony would tend to interfere with the
assembly and to lower intellectual competence at the same time
that it decreases emotional disturbance. Further, this would ac-
count for the lowered pain sensitivity, and the value of lobotomy
in cases of intractable pain (Watts and Freeman, 1946). It will
be recalled that pain was treated (in Chapter 8) as the equiva-
lent of a local emotional disturbance, in the somesthetic system,
and so can be regarded as being affected in the same way by
a focus of hypersynchrony.
274 Emotional Disturbances
It is not necessarily implied that scar formation is what pro-
duces the hypersynchrony in lobotomy. The effect of deaffer-
enting an extensive cortical region would be to permit it to go
into spontaneous hypersynchrony (Adrian, 1934). This, or this
plus any effect of scar formation, may be the source of the re-
covery from mental depression.
/ /. The Growth and Decline
of Intelligence
This discussion began originally, in Chapter 1, with the puzzle
of the high intelligence-test scores that are sometimes found
after a surgeon has destroyed a large part of the human brain.
A tentative explanation was suggested, that these scores are due
to a conceptual development which brain damage does not
easily reverse. The chapters that followed have tried to make
this intelligible physiologically (besides showing what the im-
plications are for other problems). "Conceptual development,"
"insight," "thought," and so on, have been given a physiological
meaning, and a possible basis has been provided for seeing how
capacities could be retained, once developed, despite brain in-
jury which if it had occurred in infancy would have prevented
the development. This chapter returns to the original problem
of the effects of brain operations on intelligence. It presents
an interpretation which fundamentally concerns the nature of
intelligence, its normal development, and its later decline with
old age.
This chapter will be concerned also with certain problems of
method. Clinical research on intelligence has difficulties as a
blackberry-bush has thorns, and yet is an important source of in-
formation for the psychologist, and can be expected to become
more important. If the difficulties are to be surmounted, we
must know what they are. These matters of method will be
taken up first.
IDENTIFYING INTELLECTUAL DETERIORATION
We feel, most of us, that we can make a reasonably good esti-
mate of another's intellectual ability by conversation with him;
275
276 The Growth and Decline of Intelligence
and we are apt to feel, too, that we have in general a fairly
good idea of what sort of problem can be solved by the healthy,
"average," "normal" people in our own communities. Both con-
victions are frequently misleading. The first point on method
is that the detection of intellectual defect, following brain in-
jury, requires the test procedure and (in some form or other)
the use of a control group.
Ordinarily, the person who is clear in conversation, quick to
understand what is said to him, and coherent in his reply, is
"intelligent": that is, conversational ability is correlated with
other abilities, and is not a bad basis of prediction. But these
correlations sometimes break down with pathological changes in
the brain, and then the prediction may be false. Also, one is
apt to judge capacity by the patient's alertness and manner,
apart from what he says. The physician is familiar with de-
terioration in the form of unresponsiveness, apathy, slowness, or
inconstancy of purpose. Accordingly, when such signs are
absent the physician may feel certain that intelligence is not
impaired.
It is reasonably sure that this has been a major source of the
disagreement about intellectual impairment in aphasia. Because
the aphasic always tries to respond, is fully attentive, and gives
the impression of being quite clear mentally, it is easy to con-
clude that he has lost nothing but the power of verbal recogni-
tion and expression. There are cases on record in which it has
been "proved" that aphasia does not affect intelligence, on the
ground that the patient nodded and smiled when the physician
greeted him, that he knew whether it was morning or afternoon,
or that he was able to let an attendant know when he needed
the bedpan.
Tests, on the other hand, show that the aphasic usually can-
not fit blocks together to make a simple pattern, that he does not
recognize absurdities in pictures, cannot make a simple draw-
ingeven though he clearly knows what he is to do. In such
performances the aphasic is almost certain to be at a much lower
level than a normal individual of the same background and
277 Identifying Deterioration
original ability (Weisenburg and McBride, 1935; Hebb, 1942a).
The aphasic patient is apt to have trouble in finding his way
about, even in a familiar region, and to have trouble in manipu-
lating simple mechanical contrivances. There is no sense in
arguing over terminology, and each writer can define that
equivocal term "intelligence" to suit himself (though one might
expect some consistency in its subsequent use); but because the
aphasic is alert, and persistent in trying to discover what 2 and
2 add up to, or to discover what is missing in a picture of a dog
with three legs, it does not follow that he has lost only the
capacity for communication. There is also a change in what
would usually be called intelligence.
To detect deterioration, one cannot depend on impression
alone but must also use test methods; and this requires a normal
control group. It is most dangerous to assume knowledge of
what the "average" person knows and can do. Let me give
some examples, based on data obtained in the standardization of
some adult tests (Hebb, 1942&; Hebb and Morton, 1943). Some-
where in the neighborhood of half the adult English-speaking
population cannot complete the analogy foot is to ankle as hand
is to what? or under : overdown : ? (though almost everyone
can do dog : barkcat : ?). About half the adult population
think that priceless means "of no value"; nearly the same number
are unaware that brunette refers to a dark complexion, and many
think it a synonym for "blonde"meaning, any lively young
woman. Figure 19 shows some simple problemsapparently
simplethat cannot be solved by some fraction, between 10 and
25 per cent, of the adult population (Hebb, 1945&).
The subjects in these tests were able to earn a living and con-
duct their own affairs with ordinary prudence, and would not
be classed as anything but normal in the clinic. If such a person
happened to have a brain injury, and the examiner did not know
what other persons of the same class would do with such tests,
he would be bound to conclude that the brain injury was the
reason for the subject's incompetence. No matter how simple
one's tests may seem to be, the brain-injured patient's perform-
ance must be interpreted in the light of the scores of a normal
control group. The professional man's idea of the average intel-
278 The Growth and Decline of Intelligence
ligence is colored by his preoccupation with the abstract and
academic, and by the fact that his intimacy is mainly with others
in the professional community whose preoccupations, interests,
and abilities are of the same kind as his own.
How is a normal control group to be chosen? Despite some
recent discussion making this an esoteric question, almost impos-
FIGURE 19. Problems from a performance test (top line) and attempted
solutions (bottom line) by a person not in any way classed as feebleminded.
A pattern in cardboard was placed before the subject, and left there while
he tried to duplicate it, using two or more identical wooden triangles.
Each of the hatched diagrams shows a subject's placing of the blocks in an
attempt to duplicate the pattern immediately above (by B.G., female, 30
years of age, a "normal control subject" convalescent after appendicitis).
A fairly large proportion of the adult population will make such errors, and
have difficulty in seeing how to change the placing of the blocks when told
that the solution is not correct.
sible to understand, the desiderata are comparatively simple.
There might be no need of a control group if one could compare
the patient's scores after brain operation with his scores before
anything was wrong with his brain; all one can get, however, is
a pre- and post-operative comparison, which must be a different
matter. The condition for which he is being operated on must
itself affect his scores, and in fact the operation quite often
causes a rise in score. The purpose of the control group, then, is
to let us know as far as possible what the brain-operated patient
would have done if his brain had remained healthy and undam-
aged. This tells us what to look for in a control group. The
object is to find persons who are like the patients being studied
in every way that, as far as one knows, affects intelligence-test
279 Identifying Deterioration
scores, except in the one respect, that the normal control subjects
have undamaged brains.
Now test scores are related to education, social background,
and occupationa concatenation that can be summarized as the
factor of sophistication. Scores are also affected by age, rising
to a peak in the teens and declining (for almost all tests) after
the twenties. These are the correlates of intelligence in normal
persons, as far as they are known at present; if they are con-
trolled, one can estimate, by comparing clinical and control
groups, the degree and kind of impairment that results from
brain damage; but if they are not controlled this is not possible.
It is not possible to show that brain damage affects intelligence,
by the use of a control group that is made up of hospital em-
ployees, or college students, or clerical workers, when the clin-
ical group includes truckdrivers, farmers, and unskilled laborers-
persons who are apt to have been long away from clerical work,
with a lower average level of education. It is not possible to com-
pare a man who went to high school ten years ago with a stu-
dent now in high school. If the clinical subjects are older adults,
or less sophisticated, they are bound to have lower scores than
the normal subjects.
These considerations are what should determine the nature of
the normal control group. One must then go farther, and use
clinical controls, if one wishes to know not only that a certain
kind of injury produces intellectual defects but also whether it
produces more defect than some other kind of injury. If one
wanted to show that frontal-lobe damage is more serious than
damage elsewhere in the brain, one would use two groups, a
frontal-lobe group and a parietal, temporal, occipital group. But
note this: everything that was said above about making the
normal group equivalent to the clinical group is equally neces-
sary in the comparison of two clinical groups. They must be
the same in age, in sophistication, and, in addition, in the size
and type of lesion.
Halstead (1947) has criticized Rylander (1939) on the ground
that his normal control group was not a clinical control group,
and me (Hebb, 19452?) for being taken in by Rylander. Ac-
tually, Rylander's normal control group was an excellent one, the
only good one that has been used in the study of the frontal
280 The Growth and Decline of Intelligence
lobe. His first object was to show that frontal-lobe damage pro-
duces intellectual defects, and he showed this clearly. (Whether
the defects are greater than with parietal lesions, or whether
they are due to the surgical lesion or to an accompanying dys-
function, would be another question.) Halstead's control data,
both normal and clinical, do not meet the elementary require-
ments set forth above: his normal group was younger (the
average was 26 years) than either of his clinical groups and
had a high proportion of persons with clerical and academic
sophistication. His parietal, temporal, occipital group (33 years)
was younger than his frontal-lobe group (40 years), the data
supplied as to the relative extent of frontal and nonfrontal brain
damage (direct or indirect) are insufficient, and so too are the
data as to occupation or sophistication (the designation "house-
wife" should be supplemented either by educational level or
husband's occupation).
For some reason, frontal-lobe patients in the clinic sometimes
tend to be older than others and may have a larger total destruc-
tion when the effects of pressure in the rest of the brain are con-
sidered. This may be because the first signs of a frontal-lobe
tumor are less specific than those of a tumor elsewhere, so the
patient does not seek treatment as promptly. Frontal tumors
tend also to produce bilateral damage of corresponding struc-
tures, by pressure, more than other tumors do. Consequently,
it is very difficult indeed to find cases in which there is, for
example, comparable damage of one frontal and one temporal
lobe, or of both frontal and both temporal lobes.
There is nothing mysterious about the logic of the control-
group method. If one wants to estimate the pre-operative intel-
ligence of a group of barbers with frontal-lobe injury, or of high-
school teachers, one would look for a group of barbers or high-
school teachers of about the same age and education, and pref-
erably from the same part of the country, since occupational
standards vary. The trouble of course is that man is a suspi-
cious, sensitive, and uncooperative animal who objects to having
his intelligence tested and is not usually interested in helping a
scientific investigation at such a price. To obtain test data for
a good sample of the general population, even if only a sample
Anatomical and Pathological Data 281
of fifty or sixty subjects, is a very large undertaking (Weisen-
burg, Roe, and McBride, 1936).
The problem of normal control data has been reviewed else-
where (Hebb, 1945&). The conclusions arrived at can be stated
again: Only two studies have utilized an adequate normal con-
trol group in psychological study of the effect of brain damage:
Rylander (1939) and Weisenburg and McBride (1935). These
two consequently are of outstanding importance (though they
are defective in not having adequate anatomical data, and one
cannot be quite sure that the defects were due to a simple loss
of tissue). Rylander's norms, of course, since they are for
another culture, are not applicable directly to English-speaking
patients; but Weisenburg, Roe, and McBride's (1936) careful,
though apparently little-known, analysis of intelligence in a rep-
resentative group of adult English-speaking subjects still stands
as the best basis for the study of intellectual defect. With cau-
tion, their data may still be used if one cannot undertake the
tremendous task of obtaining new normal-control data for each
new study of brain damage.
Dr. Morton and I have also supplemented their test battery
with some new tests designed especially for adult use (Hebb,
1942&; Hebb and Morton, 1943).* Such norms, however, be-
come less and less accurate with the passage of time, and as
more individual tests, of the homogeneous type, become avail-
able it would be desirable for someone to attempt a new stand-
ardization of a battery of homogeneous tests, each one measur-
ing a single kind of ability and so making it more possible to
analyze the intellectual changes that occur in cases of brain
damage.
ANATOMICAL AND PATHOLOGICAL DATA
The second major aspect of the problem of method in such
studies is the difficulty of getting satisfactory anatomical and
physiological data. In an animal study, the only data one trusts
* A recent restandardization of one of these tests, in a short form that is
more convenient for clinical use, has been done by W. A. Hunt, E. G.
French, and I. N. Mensh. An abbreviated form of the McGill Verbal
Situation Test, J. Clin. Psychol, 1948, 4, 406-411.
282 The Growth and Decline of Intelligence
concerning the locus and extent of lesion are obtained by killing
the animal and removing the brain for anatomical study. Human
patients, after brain operation, have a regrettable way of living
on indefinitelyinstead of first making a good recovery, to permit
thorough testing, instructing their relatives to permit an autopsy,
and then dying of pneumonia.
So one is obliged to make the best guess one can, instead of
having exact data. It is not possible at operation to know
exactly what tissue is destroyed, particularly when the operation
is for a tumor which displaces the brainas tumors mustand
in displacing it is more than likely to produce a diffuse damage
throughout parts that may be at some distance from the scene
of operation. In operations for tumor, actually, one is practi-
cally always dealing with two kinds of lesion: the clean surgical
lesion that can often be accurately described, and the diffuse,
irregular lesions that result from pressure and have no clearcut
boundary. Brain operations seemed to open a new era in the
study of the localization of function, as they became common
about 1920. Why? Because the effect of pathological lesions is
almost impossible to interpret. One cannot say where the dif-
fuse lesion begins and where it ends; attempting to interpret
such data, between 1870 and 1920, had led to an endless series
of arguments. But this difficulty applies equally to interpreting
the effect of a brain operation plus a pathological lesion, so in
most cases of tumor the fact that the brain has been operated on
does not actually make the case suitable for the localization of
function.
Such a point would mean little if the effect of the added
pathological lesion were slight. But it has been discovered that
a comparatively small region of pathological change, producing
hypersynchrony, may have a much more deteriorating effect on
behavior than a complete loss of the tissue concerned. An idea
of this sort was proposed by Jefferson (1937) and Stookey,
Scarff, and Teitelbaum (1941). A direct demonstration on the
point has been provided in two cases in which the removal of
frontal-lobe tissue, in an area of scarring, changed a marked
deterioration into behavior indistinguishable from the normal
(Hebb and Penfield, 1940; Mixter, Tillotson, and Wies, 1941).
A region in which the blood supply is interfered with, but not
Anatomical and Pathological Data 283
entirely shut off, usually shows some loss of cells and a number
of remaining cells whose staining properties are changed. This
indicates a change in the chemical properties of the cell, which
in turn implies a change in frequency properties and obviously
may account for the existence of a hypersynchrony which inter-
feres with the functioning of the cell-assembly. A focus of
hypersynchrony must act as a pacemaker that tends to wean
transmission units away from the assembly. When hypersyn-
chrony is not great, it would allow some assemblies to function
(particularly those that are long established: p. 197) but would
tend to interfere with recent memory, decrease responsiveness,
and interfere with complex intellectual activities. When it is
more extensive, it would prevent all higher functions.
Hypersynchrony thus extends the influence of a pathological
lesion beyond the borders of the lesion itself and may make it
impossible to draw any conclusion about the localization of the
functions that are affected after a frontal pole, for example, has
been removed. When one is interested in an attempt at localiza-
tion of function, the question must always be asked whether the
sequelae of operation are due (1) to the surgical removal; (2)
to permanent damage in the remaining parts of the brain; (3)
to the transient physiological disturbance that always occurs
along the margins of the surgical lesion, that may take from one
to six months to subside, and that, since it is reversible, has de-
ceived more than one worker into thinking that recovery is an
evidence of vicarious functioning; or (4) to a chronic physio-
logical disturbance, from scar formation or from diffuse patho-
logical lesions throughout the rest of the brain. If a loss of func-
tion is due to the surgical lesion, the function depends on the
tissue that was removed; if due to dysfunction, it depends on
tissue that still remains. All this makes the question of localiza-
tion complex indeed. It explains why we are little farther for-
ward in the localization of function, in the human brain, than
Broca and Jackson werethe "experiments of nature" are poorly
designed, and produce disturbance, as well as a simple loss, of
function.
The EEC may be a relatively crude index of what is going
on in the cerebrum, with no direct relation to the firing of the
individual cell; but it is a practical index of the physiological
284 The Growth and Decline of Intelligence
status of the different parts of the brain, and as such it has the
greatest importance in any attempt to study localization of func-
tion from a patient's behavior after recovery from brain opera-
tion. For such a purpose, electroencephalographic report is not
merely desirable but essential. With it, one has at least some
chance of knowing when defects of behavior can be attributed
to a surgical lesion and not to a pathological dysfunction
instead.
Animal experiments, anatomical studies, and studies in which
the brain of the fully conscious human patient is stimulated
electrically have clearly localized a number of sensory and motor
areas. But, because of the difficulties of method that have just
been considered, about the only localization of a higher function
that has so far been achieved is that of the so-called speech area:
the middle regions of one hemisphere, usually the "dominant"
one (the left hemisphere in a right-handed person), appear to
have some sort of executive and organizing role that is essential
to speech and some other higher functions. No other localiza-
tion of function in the human cerebrum has been established.
I am aware that this statement is not in accord with much of the
literature, but critical evaluation of the evidence presented in
support of the various claims concerning localization permits no
other conclusion, so far as I can see.
A paper by Klebanoff (1945), very useful as a review and bib-
liography of most of the psychological work in this field up to
1941, did not sufficiently consider the twin problems of normal
control data, and anatomical and pathological data. Failure to
solve one or both of these problems tends to invalidate the
papers that try to prove that some higher function is resident in
some particular part of the human cortex, in the work reviewed
by Klebanoff. If one considers the normal control data used in
these studies, or the anatomical data, with the same critical eye
that would be turned at once on an animal experiment, one will
find none of them justifying more than an opinion, a guess, con-
cerning the relation of higher functions to the cerebrum.
Even with regard to the speech area, it is none too clear either
what the boundaries are, or that a limited surgical lesion in any
part of the region will disturb speech. Aphasia may be pro-
duced only by a hypersynchrony whose focus is strategically
Anatomical and Pathological Data 285
placed here, in a position to disturb function throughout the
whole hemisphere. We have no evidence on this, for the sur-
geon naturally declines to touch this area if he can help doing
so; if he must, he disturbs as little of it as possible.
However, the case of left hemidecortication reported by Zol-
Hnger (1935) shows that a clean surgical extirpation can pro-
duce aphasia. Again, anterior lesions in the speech area may
produce more difficulty in speaking and writing, posterior lesions
more difficulty in understanding speech and printed matter; but
no real anatomical evidence to this effect seems to exist, and
Weisenburg and McBride's (1935) cautious distinction between
a predominantly expressive and a predominantly receptive
aphasia should not be extended into a sharp dichotomy of
"motor" and "sensory" aphasias, since it seems highly likely that
both aspects of the disorder are always present.
A number of clinical writers have disagreed with this conclu-
sion, arrived at by Weisenburg and McBride, Head, and others;
but they seem never to have established satisfactory evidence.
An important paper by Kennedy and Wolf (1936), for example,
reports a "pure alexia" and a pure motor aphasia, with no other
involvement. But the case reports are typically incomplete: to
show that alexia is aphasic, it should be shown that there is suffi-
cient acuity for reading, and that the perception of other (non-
symbolic) visual patterns is undisturbedelse one has simply a
loss of vision or a visual agnosia. In the second case of motor
"aphasia," what is reported by Kennedy and Wolf is a loss of
the capacity for drawing, with no disturbance of anything re-
lated to speech: a rare and valuable observation, but not appar-
ently relevant to the question of aphasia. Again, Kennedy and
Wolf have provided no details of sensory and motor tests, and
these details must be given if it is to be shown that such a thing
as a pure alexia or agraphia can exist. Neither, certainly, has
yet been shown to result from damage to any limited part of the
speech area. It is hardly likely that they will be; such concep-
tions arise from an older type of theory, no longer entertained
in psychological thought, and it is highly probable that such
questions (concerning what peculiar contribution each part of
the cortex makes to behavior) will remain unanswered until we
can find out how to ask better questions.
286 The Growth and Decline of Intelligence
There have been a great many interesting and important ob-
servations showing a relation of complex functions to the parietal
and temporal lobes in man; but in all the reported cases, of
various agnosias, disturbances of the body image, and so on,
there is presumptive evidence of the presence of pathological
dysfunction as well as of a simple loss of tissue, and we are not
yet in a position to interpret such data as far as the localization
of function is concerned. Von Monakow wrote that he did not
know of a single case in which a purely traumatic lesion in the
occipital lobes of an otherwise sound brain had resulted in a
persistent visual agnosia.* This, implying that agnosia is an
effect of dysfunction and not of an uncomplicated loss of cells,
is directly in line with the present discussion (though it should
be noted that a traumatic lesion, accompanied by scar formation,
may also produce chronic dysfunction). There is little that can
be definitely concluded at present concerning localization of the
functions disturbed in agnosia, and so on, until the disturbance
is found to be produced by a clean surgical lesion, unaccom-
panied by delta waves in the EEG.
Only one thing need be added: although the frontal lobe is
the favorite place in which to localize higher functions when one
is speculating about these matters, it is still true that there is no
proof that any single higher function depends on this part of the
brain (Hebb, 1945&). At least as good a case might be made
out for the parietal and temporal lobes as the seat of man's
distinctive psychological characteristicsif these in fact depend
on one part of the brain more than another (see, for example,
Kubitschek, 1928; Keschner, Bender, and Strauss, 1938).
Because of the enormous amount of space that has been de-
voted to the effort to show that the prefrontal lobe is the seat
and habitation of some higher function or other (there are few
mental processes that have not been ascribed to it), the basis of
these negative conclusions must be made more detailed. Doing
so will also illustrate other points dealt with in the preceding
pages and make them more specific.
* Personal communication from Dr. K. S. Lashley. My interpretation of
this observation is based on the fact that a traumatic lesion is less likely to
produce chronic dysfunction than a tumor or abscess.
Anatomical and Pathological Data 287
The first and most effective piece of evidence is the history of
a patient both of whose prefrontal lobes were removed by Dr.
Wilder Penfield, with results that radically changed the possi-
bilities of surgical therapy in the frontal area. Before this time,
it was universally accepted that loss of both frontal poles must
be followed by gross deterioration of personality and intellect.
Dr. Penfield's patient completely reversed the prediction, by
being psychotic before operation (when the damage was rela-
tively small) and quite indistinguishable from a normal person
afterward (when the total destruction was increased in size
with the simultaneous removal of a source of physiological dis-
turbance ).
The history is that of a boy of sixteen whose skull was frac-
tured, with damage to both frontal lobes, and who in the fol-
lowing ten years developed severe epilepsy and became (much
of the time) an irresponsible and rather dangerous charge. The
condition varied, not only with the occurrence of attacks but also
between attacksapparently owing to subclinical seizures: the
IQ, on the basis of an uncompleted testing on one day, was esti-
mated to be below 70; next day a completed test gave an IQ of
84. After operation, the IQ was consistently close to 96, no evi-
dence of intellectual loss could be found in any one test or in
the pattern of test scores as a whole, and the psychotic behavior
completely disappeared. Clinically and socially he was "normal"
and was able to take full control of his own affairs: on the last
examination he had been doing so for more than five years. He
enlisted in the army and served satisfactorily overseas for ten
months until he had another fitat which time the medical of-
ficers caught up with him and he was discharged. In his sub-
sequent behavior it is possible that there were deficiencies in
planning for the distant future, but it is certain that he had more
initiative and planned foresight than many persons with frontal
lobes intact.
There must of course have been some loss of function in this
patient. Logically, it seems impossible to suppose that behavior
can be unchanged after bilateral loss of any area of the brain.
My conclusion is not that the frontal lobes have no function, but
either (1) that we have not yet guessed at the real nature of
288 The Growth and Decline of Intelligence
this functionhave not asked the proper questions in such inves-
tigations, or (2) that once the development of behavior has been
completed the prefrontal lobes have no great further importance.
Surely it is the first of these, 1, that is the answer, not 2; but
there is nothing in the evidence at present that justifies the idea
that the frontal lobes have some special relation to higher func-
tions in human behavior.
This general conclusion does not actually rest on the one case
that I have described but can also be established independently.
The evidence that has been supposed to show the importance of
the frontal lobes does not withstand scrutiny (it has been re-
viewed in detail elsewhereHebb, 1945&). What the Penfield
case did was, first, to demonstrate conclusively that an area in
which there is dysfunction (the area of scarring that produced
both epilepsy and deterioration of behavior) can be far more
deteriorating than complete removal of the area; and, second, it
raised the question, How are we to account for the astonishing
difference between this case and other, apparently similar, cases,
in which, however, profound deterioration remained after sur-
gical operation on the frontal lobes?
Examination of the other case reports at once showed strong
reason to believe that the deterioration they described was due
to dysfunction, or diffuse destruction in the rest of the brain re-
sulting from the earlier pressure of a tumor, or both. It thus
appeared that the behavior changes were not due to the sur-
gical injury, and therefore that such cases are not suitable for
studying the localization of function. The extent of the total
lesion in each case was not known, and the loss of tissue was
complicated by additional factors. Dysfunction is to be expected
also in frontal lobotomy (p. 272), and so cases of lobotomy also
are not a source of interpretable data for the localization of func-
tion, especially since we do not know at all what the effect
would be of a similar bilateral incision in other regions of the
brain.
This line of argument then obtained confirmation from the re-
ports of bilateral prefrontal lobectomy by Mixter, Tillotson, and
Wies (1941) and Heath and Pool (1948). Equally important,
it now appears that monkey or chimpanzee may lose both frontal
289 Early and Late Brain Injury
poles without evident impairment, though earlier work seemed
to have established the contrary.*
We have at present, then, no reason to think that a clean sur-
gical removal of both frontal association areas is necessarily fol-
lowed by observable, clinically evident defects of behavior; nor
have we as yet any real understanding of the functions of this
part of the brain.
DIFFERENCES BETWEEN EARLY AND LATE
BRAIN INJURY
The study of brain operations in human patients has not so
far provided the data that would allow one to be very definite
about the way in which higher functions are related to special
parts of the cerebral cortex; but it has raised one problem that
gives us a valuable lead concerning the nature of intelligence.
The problem has been referred to alreadyit arises from the dis-
covery that an IQ of 160 or higher is possible after the removal
of one prefrontal lobe (Hebb, 1939) or an IQ of 115 after hemi-
decortication (Rowe, 1937), and from the frequency with which
above-average scores are reported after cortical injury in any
except the speech area. It appears possible, however, that
similar damage to the infant brain has a much greater effect on
the subsequent IQ. This possibility suggests a clue to the nature
of adult intelligence and suggests a distinction between two
quite different meanings of the term "intelligence"a distinction
that may help to resolve current theoretical disagreements.
The average IQ after brain operation, in all cases in which an
IQ was reported up to the end of 1941, was 108at least 8 points
above the normal adult average, which is probably well below
100 according to the U. S. Army data of 1917 and Weisenburg,
Roe, and McBride's (1936) data. When brain damage is trau-
matic or pathological, instead of surgical, there are also frequent
indications that on certain kinds of test (on which the Stanford
Binet IQ is mainly based) the patient's score is not much lower
than it would have been had the brain damage not occurred.
These data are given in more detail elsewhere (Hebb, 1942a).
* See T. Spaet and H. F. Harlow, /. Exp. PsychoL, 1943, 32, 424-434;
R. A. Blum, Amer. Psychol, 1948, 3, 237-238.
290 The Growth and Decline of Intelligence
But in practically all such cases the use of other tests shows that
there is a definite intellectual loss. That is, some tests such as
the Stanford Binet show little sign of a loss of intelligence, others
show much greater loss. There are then two kinds of test mate-
rial, or two extremes in test content that shade into one another.
The same picture appears when older adult subjects are exam-
ined: some (Binet type) tests show no deterioration due to
advancing age; others show a marked deterioration; and most
tests fall somewhere between the two extremes (Jones and
Conrad, 1933; Wechsler, 1939).
The Binet IQ is determined by the use of a composite test,
and it amounts to an average level of performance in a miscel-
lany of tasks. It is mainly verbal, however, and mainly involves
a kind of problem to which the answer is rather obvious, to the
subject who can solve it at all. Tests of intelligence may also
consist of nonverbal tasks, and puzzle-like tasksones that re-
quire an effort and have to be worked out logically. Analysis of
the test scores made after brain operation, when aphasia is not
present, indicates that it is the first of these two kinds of task
that is little affected by brain damage, the second that is much
affected (Weisenburg and McBride, 1935; Rylander, 1939; Hebb,
1942a). The IQ shows little effect of brain damage outside the
speech areas because it is mainly based on "Binet type" tasks.
These consist particularly of tests of vocabulary, information,
digit repetition, digits backward, understanding of the meaning
of fables or proverbs, and certain problems of a familiar kind,
dependent usually on common sense (in the form of a cultural
sophistication).
Now, for normal persons, vocabulary and information tests
are among the most valid tests of intelligence, and the Binet-
type test in general is what best predicts the subject's level of
problem-solving outside the laboratory or clinic, in a wide range
of situations. This is what presents us with our problem: how
can they also be the ones that show the least effect of brain
operation, and the degenerative changes of senescence?
But we must consider at once a further fact: that in certain
essential respects, "intelligence" does not decrease after the age
of twenty or thirty, and the brain-operated patient very fre-
quently demonstrates an unimpaired level of functioning in his
291 Early and Late Brain Injury
daily occupations. The patient with an IQ of 160 after pre-
frontal lobectomy (Hebb, 1939) graduated with honors in his
arts course and went on to do well in medical school; the pa-
tients with half the cortex removed (Dandy, 1933; Rowe, 1937;
Karnosh and Gardner, 1941) clearly did not act in any respect
like the feebleminded, in their daily affairs. Certain intellectual
capacities are well retained by the patient; in some respects, the
prediction made by the Binet-type test is valid. The puzzle is
not why the Binet suddenly loses its diagnostic value, after brain
injury: but why an important part of intelligence is little affected
by the injury. The fact that it is not, at once suggests that this
part of intelligence is not directly correlated with the integrity
of the brain; and, since the problems in which "intelligence" is
unimpaired are in general those with which the patient is fa-
miliar, a further suggestion is that part of intelligence is a
product of experience.
An idea of this sort led to the examination of the effects of
early brain injury (with the help of Dr. Heinz Werner); the re-
sults, in short, indicated that any damage to the infant brain
would affect later performance on vocabulary tests, information,
and the like at least as much as performance on other tests
(Hebb, 1942a). The situation apparently is this. Unless every
known case of large damage to the infant cortex is a case of
speech-area damage (a most improbable situation), destruction
of tissue outside the speech areas will prevent the development
of verbal abilities, but the same destruction may not greatly
affect these abilities once development has occurred.
The evidence for this interpretation is not perfect, but it is
rather strong. The one flaw appears to be the possibility that
the known cortical-type birth-injury population is made up pre-
dominantly or entirely of patients with speech-area lesions. It
is conceivable that when palsy is not present only speech-area
cases are detectedthat all the "exogenous" cases of mental de-
ficiency (Werner and Strauss, 1939) are discovered because of
damage to this particular cortical area alone. But against this
possibility is the fact that cases of palsy, presumably detected
because of it, also have the low verbal test scores characteristic
of other birth-injury cases. The twelve cases reported by Doll,
Phelps, and Melcher (1932) showed marked vocabulary retarda-
292 The Growth and Decline of Intelligence
tion. The low vocabulary score of the birth-injured is not due
to deficient opportunity to learn the meaning of words (Hebb,
1942a)a conclusion supported by the almost identical retarda-
tions of vocabulary and digit-repetition scores reported by Doll,
Phelps, and Melcher. Further, Doll (1933) reported that there
is no correlation of speech defect and IQ in the birth-injured.
It appears, therefore, that an early injury may prevent the de-
velopment of some intellectual capacities that an equally exten-
sive injury, at maturity, would not have destroyed. To complete
the picture, it should be said again that this relationship does
not holdat least not to the same degreefor all intellectual
capacities; and sensory and motor capacities after damage to the
infant brain tend to reach a higher level than that attained after
destruction of the same regions at maturity.
How are we to understand the first type of capacity in the
adult, the one that is not greatly affected by brain damage at
maturity? Psychologically, the matter might be put this way.
The actual modifications of behavior that occur in intellectual
development are from one point of view qualitative: faced with
a complex situation, the more intelligent subject sees it in a dif-
ferent way, and makes a different responsenot more responses,
or responses that take more effort. The chief intellectual effort,
in such behavior, is demanded on the first occasion on which the
new way of seeing the situation occurs. Learning to solve a
problem usually demands more effort than solving other prob-
lems of the same kind on later occasions when one has discov-
ered the principle involved. The same is often true of percep-
tion, whether one is solving a formal problem or not; once one
has seen a certain relationship in a picture, or heard it in music,
it may take no effort to find it again in similar pictures or music.
This perception of relationships is essential in problem-solving,
and one would steadily accumulate an increasing stock of such
ways of seeing and attacking problem situations, as age ad-
vances; so, in a familiar field, the fifty-year-old may be definitely
more "intelligent" (as defined by the problems he can solve)
than a twenty-year-old. At the same time, the problem that
requires a radically new approach, new ways of perceiving, is
not likely to be solved by the older man with his shrunken brain.
The difference in the demands that are made by different pro-
293 Early and Late Brain Injury
fessions, or different fields of investigation, would account for
the fact that some fields are characteristically the domain of
young men, others of the middle-aged. We cannot simply gen-
eralize, and say that intellectual capacity, objectively defined,
decreases from the age of twenty-five onward. Some capacities
do; others do not.
Physiologically, the matter may be put as follows: some types
of behavior that require a large amount of brain tissue for their
first establishment can then persist when the amount of available
tissue has been decreased. This of course is consistent with the
theory of cell-assemblies that has been presented in this mono-
graph. It has been postulated that, with the enlargement of
synaptiq knobs, the number of fibers necessary for transmission
at the synapse decreases (p. 66). In the first establishment of
an assembly, then, more fibers are necessary than for its later
functioning.
Also, it has been said that there is a possibility of "short-
circuiting" in the phase sequence (p. 228) so that, after a phase
sequence has been well established, some assemblies may no
longer be necessary to it. If brain injury occurs in a limited
region of the brain, it would presumably remove a small number
of transmission paths from each of a very large number of as-
semblies. These assemblies might still function, though they
would have smaller safety margins (p. 197) and therefore would
function less reliably. In other assemblies, the loss might be
great enough to prevent any functioning; but a phase sequence
in which some of these assemblies were originally essential might
later short-circuit them, and produce the same ultimate insight
or understanding of the situation, though some of the steps of
inference would have been omitted.
Visual stimulation may have been involved in the formation
of some concept, and yet not be necessary to its later arousal
(p. 128). Thus the congenitally blind would be incapable of
insights and understanding possessed by those who have become
blind after having had vision. In the same way, an area-18
lesion might prevent certain "visual" assemblies from function-
ing. These may have been originally necessary, as a connecting
link, in the establishment of connections between other assem-
blies, constituting a particular conceptual activity; but once the
294 The Growth and Decline of Intelligence
connections are established, the connecting link may not be
needed any longer. The same area-18 lesion in infancy thus
might prevent a conceptual development but not destroy it at
maturity.
THE TWO MEANINGS OF "INTELLIGENCE"
The clinical evidence has indicated, in effect, that there are
two components in intelligence-test performance and in any in-
telligent behavior. One is diminished immediately by damage
to the brain, and amounts to a factor of heredity; one is related
more to experience, consisting of permanent changes in the or-
ganization of pathways in the cerebrum (in the present theory,
these changes are the establishment first of assemblies of cells,
and secondly of interfacilitation between assemblies). The
hereditary factor is essentially the capacity for elaborating per-
ceptions and conceptual activities; the experiential factor is the
degree to which such elaboration has occurred (and particularly,
when we speak of intelligence, the conceptual elaboration that
is not specific to one occupation or situation, but that enters into
many human activities; concepts of number, of causal relations,
of common human behavior, and so on).
From this point of view it appears that the word "intelligence"
has two valuable meanings. One is (A) an innate potential, the
capacity for development, a fully innate property that amounts
to the possession of a good brain and a good neural metabolism.
The second is (B) the functioning of a brain in which develop-
ment has gone on, determining an average level of perform-
ance or comprehension by the partly grown or mature person.
Neither, of course, is observed directly; but intelligence B, a
hypothetical level of development in brain function, is a much
more direct inference from behavior than intelligence A, the
original potential. (I emphasize that these are not two parallel
kinds of intelligence, coexistent, but two different meanings of
"intelligence.") It is true that estimating intelligence B requires
a summation of observations of behavior in many different situa-
tions, at different times; however, if we assume that such an esti-
mate is possible, what we actually know about an intelligence-
The Two Meanings of "Intelligence" 295
test score is that it is primarily related to intelligence B rather
than intelligence A. The relationship to A is less direct.
Most of the disagreement in recent years over the nature of
"intelligence" concerns the relation of A, innate potential, to B,
the estimated level of functioning at maturity. If A determines
B fully, "intelligence" is a matter of heredity and maturation
only; the IQ is not dependent on experience. But if intelligence
A is only one of the conditions of intelligence B, not the sole
determinant, what then? Intelligence A is still hereditary, and
it would not be true to say that "intelligence" (without qualifi-
cation) is influenced by experience: only intelligence B is so
affected.
The dispute in the current literature has arisen, I believe,
partly because of the double reference of the term "intelligence"
and partly because it has not been realized that if the effects of
early experience are more or less generalized and permanent
one can concede a major effect of experience on the IQ, and still
leave the IQ its constancy and validity as an index of future
performance.
An innate potential for development is not logically a guar-
antee that the development will occur. There is in fact an over-
whelming body of evidence to show that experience is essential
to the development (the original potentiality is of course equally
essential). Binet himself assumed the necessity of adequate ex-
perience in the subject to be examined by his method; the con-
tent of intelligence tests still makes this a necessary precondi-
tion; and all psychologists recognize that one cannot compare
the innate intelligence of subjects from twc different cultures:
experience affects their IQ's to an unknown degree. Why then
should we object to the idea that enriching an inadequate envi-
ronment will raise the IQ, as Stoddard and Wellman (1940) and
others have urged?
Some of the objection, certainly, may be occasioned by a
faulty use of evidence and be due to the impression that the case
for an environmental influence on the IQ has been overstated.
But the roots of the objection, I believe, are to be found in cer-
tain ideas about the nature of learning. It is apt to be assumed
that learning is more or less specific to particular situations; that
what is learned can always be forgotten; or that if something
296 The Growth and Decline of Intelligence
can be learned in infancy it can also be learned at maturity.
Assuming the truth of such propositions, one might then think
as follows:
"If the IQ has any great dependence on learning it must fluc-
tuate easily; it might be raised from 100 to 150 in the adult, or
might fall from 100 to 70 if a child is removed from school and
other stimulating influences. But the IQ does not in fact be-
have sotherefore learning does not contribute to it. It must be
hereditary, because it is so stable."
However, there are definite limits to the "constancy" of the
IQ; and most of the facts that seem to show that it is heredi-
tarily determined would be accounted for if we assume that
heredity only sets a limit to the IQ, and assume also that early
learning tends to be permanent, that it has generalized transfer
effects, and that it is not specific to particular situations. That
early learning has such properties has already been suggested,
in the first part of Chapter 6. I wish now to report some pre-
liminary experiments with the rat, which seem to show that some
aspects of this animal's intelligence at maturity are a function
of his early experience. If this principle can be established by
means of animal experiment, we can then see how it would be
applied to the known facts of human intelligence and intelli-
gence-test performance.
EXPERIMENT: INFANT EXPERIENCE AND ADULT
PROBLEM-SOLVING
When we test human intelligence, we always assume that the
subject has had a certain body of experience common to all
"normal" children in the community, the experience whose lack
may invalidate comparison of the innate potential of subjects
from different communities. Apart entirely from controversy
about the effect of special experiences on the IQ, we are all
agreed, in this sense, on the effect of a certain kind of experience
(which is not, however, well defined).
One purpose of the experimental work briefly described here
was to find out more about how the ordinary experience of in-
fancy affects mature behavior. It was not certain that such
effects could be found in rat behavior, even if they exist for man;
297 Experiment: Infant Experience
but if one could find some trace of them in rat behavior their
nature could be investigated much more easily in a species that
reaches maturity in three months than in one that takes fifteen
years. Also, the clinical evidence reviewed in this chapter has
indicated that the effect of early experience is not found equally
in all forms of behavior. Accordingly, these were exploratory
experiments, with the main object of establishing the existence
of some lasting effect of early experience on the later problem-
solving of the rat. The first explorations gave wholly negative
results, but they have the value of showing that some behavior
is not affected by early experience; that is, they serve to delimit
the effect.
In the first part of the study, one group of 7 animals were
blinded in infancy, and a group of 7 littermates were blinded at
maturity. Both groups were handled frequently, and were al-
lowed daily to run outside their cages in a large space in which
there were a number of objects. The question asked was this.
Would the group with vision during growth learn something
about finding their way around that they would retain after
being blinded at maturity? Would this affect their behavior
permanently, in other situations than those in which their visual
experience occurred?
At the age of five monthstwo months after the second group
were blindedthe rats were trained to find food in one of four
containers in an open field. Two hundred trials were given, and
the learning was not complete for most of the animals in this
timeso this presumably is to be classed as rote learning rather
than insightful. A second period of training in a similar task
also appeared to give rote learning, and in neither could any sig-
nificant difference be found between the two groups, early- and
late-blinded.
At this point, however, a chance observation suggested the
possibility that an "intelligence test" for the rat (Hebb and Wil-
liams, 1946) might reveal differences between the groups. Al-
though the test had been meant for normal animals, it might
work with the blind as well.
The test showed a clear difference between the groups. By
this time the experiment was being done with only 3 late-blinded
animals and 4 early-blinded, a small number for statistical sig-
298 The Growth and Decline of Intelligence
nificance. However, there was no overlap of scores in the two
groupsall 3 late-blinded were better than all 4 early-blinded,
which is certainly significant.*
Before repeating and confirming this result, it appeared profit-
able to explore further. The effect of vision during develop-
ment, on the behavior of rats after being blinded at maturity,
was great; would other variations of experience also have effects
that could be detected by thisapparently sensitivemethod of
testing? The obvious experiment was to compare rats reared in
ordinary small cages with others that had had the run of a wider
environment. Two litters were taken home to be reared as pets,f
a first group of 3 (after 1 mortality at home) and a second
group of 4. They were out of their cages a good deal of the
time and running about the house. While this was being done,
25 cage-reared rats from the same colony were tested.
When the pet group were tested, all 7 scored in the top third
of the total distribution for cage-reared and pets. More im-
portant still, the pets improved their relative standing in the last
10 days of testing, following 10 days of preliminary training, and
11 days of testing (a total of 21 tests was used). One explana-
tion of the better scores of the pets is just that they were tamer,
more used to handling, and less disturbed by testing. But, if
this were so, the longer the cage-reared animals were worked
with the closer they would come to the pet group, as the cage-
reared became tamer with prolonged handling. On the contrary,
the pets improved more than the cage-reared. This means that
the richer experience of the pet group during development made
* The experiments described in this section are now being repeated and
carried further by B. Hymovitch and H. Lansdell, and will be reported in
detail later. The method has been considerably improved since the first
report of Hebb and Williams; it will be described in detail in a separate
publication. The essential features of the method are: (1) 10 to 14 days'
preliminary training gets the animal used to handling, to the test situation,
and to daily changes in the route he must take to food, and teaches him
where the food is to be found (its position is constant throughout); (2)
the test itself is a series of 20 to 24 separate problems, each relatively easy
for the seeing rat; he is given between 6 and 10 trials with each, and his
score is the total number of entries into error zones (see figure 17, p. 137).
f Grateful acknowledgments are due to the Misses Jane and Ellen Hebb,
ages seven and five, for their enthusiastic assistance in the investigation.
The Nature and Nurture of Intelligence 299
them better able to profit by new experiences at maturityone
of the characteristics of the "intelligent" human being. Further-
more, a measure of motivation and tameness in the cage-reared
was available, and the correlation of this with the test score was
negligible (0.13 0.19).
These results show a permanent effect of early experience on
problem-solving at maturity: "permanent," because, in the first
experiment with early- and late-blinded, visual experience that
ended at three months of age had marked effects at eight months
of age (a long period in the life of a rat); and because, in the
second experiment, the pet group improved its standing com-
pared to the others as experience increased. Differences of
early experience can produce differences in adult problem-
solving that further experience does not erase.
These preliminary results are already being confirmed by more
elaborate experiments now going on, and a principle is estab-
lished that, first, is fully in accord with other evidence showing
lasting and generalized effects of early experience (Chapter 6)
and, secondly, clarifies the interpretation of existing data con-
cerning human intelligence-test performance.
THE NATURE AND NURTURE OF INTELLIGENCE
Actually, some of the data on normal human intelligence (as
well as the clinical data) allow only one interpretation, namely,
that there is a major effect of experience on the IQ. One diffi-
culty in accepting this interpretation is that it may seem to deny
the validity of the intelligence-test method. But, if the principle
of a lasting and generalized effect of early learning is accepted,
not only on the IQ but also on the everyday problem-solving of
the adult, the validity of the intelligence-test method may be
extended, not reduced. It is common to say that an intelligence
test is not valid when given to a foreigner or a Negro. Not valid
in what sense? as an estimate of innate potentiality, of intelli-
gence A. It may be quite valid, on a purely empirical footing,
for estimating intelligence Bthe actual level of comprehension,
learning, and problem-solving in this culture. Separating these
two meanings of "intelligence" allows one to. show where the test
is valid, as well as where it is invalid.
300 The Growth and Decline of Intelligence
Look closer at this question of the relationship between test
score and culture. It is agreed by psychologists that the IQ's of
different peoples should not be compared, since intelligence tests
are culture-loaded and not equally fair to subjects from different
cultures. Now it is quite clear from the context of such discus-
sion that the "unfairness" lies in estimating a subject's hereditary
endowment. Negroes living in the United States make lower
average scores on intelligence tests than whites do, but we can-
not conclude that the Negro has a poorer brain than the white.
Why? Negro and white speak the same language, are taught in
the same curriculum and sometimes in the same schools, work
at the same plants, and on the whole intermingle very freely.
Why then does the Negro's IQ not have the same meaning as
a white man's? Because Negro and white do not have the op-
portunity to learn to speak the language with equal range and
accuracy, are not usually taught in equally good schools, and do
not have equally good jobs or equal exposure to cultural influ-
ences that usually require a fairly good income.
All this we can accept; but when we do so we must recognize
that we have completely undermined the argument that differ-
ences of IQ among white, native Americans are determined by
heredity.
In what respect does the argument, concerning the Negro's
IQ and his innate endowment, differ for the poor white in the
South and the white tenement dweller in the North? They do
not have opportunity to acquire the same vocabulary, are often
taught in poor schools, do not get good jobs or have good sala-
ries. The extent of this influence of experience is quite un-
known; conceivably, it is small; but one cannot argue that Negro
and white IQ's cannot be compared, and at the same time that
those of white and white can be, when the white subjects have
different social backgrounds. The IQ can be trusted as an index
of intelligence A only when the social backgrounds of the sub-
jects compared are identical; and this adds up to the proposition
that we cannot in any rigorous sense measure a subject's innate
endowment, for no two social backgrounds are identical and we
do not know what the important environmental variables are in
the development of intellectual functions. Intelligence A may
The Nature and Nurture of Intelligence 301
sometimes be estimated, but it cannot be measured (Hebb and
Morton, 1944).
Intelligence in sense B is a different matter. We know, be-
yond dispute, that the adolescent with generally low intelligence-
test scores, whether Negro, poor white American, or foreigner,
is a poor prospect for college training, or training as a mechanic,
or Army officer, or dress designer. The inability to determine
intelligence A from a test score should not blind us to the fact
that a foreigner's intelligence B can be estimated, as far as its
operation in this culture is concerned.
To be a bank manager, an airplane pilot, a mathematician, a
secretary, or a surgeon requires a certain common conceptual
development that must occur in this or a closely related culture,
and in childhood mainly; and intelligence tests on the whole can
provide a rather good index of the extent to which that devel-
opment has occurred. No other interpretation seems possible of
the differences, both in educability and in intelligence-test scores,
of first- and second-generation immigrants. Supposing that both
have an adequate nutrition, father and son must have on the
average the same intelligence A; the son very often has a much
higher intelligence B.
The nature of the cultural environment that is necessary to
this conceptual development cannot be described accurately. It
does not necessarily consist of a formal schooling, and it may be
present in spite of poverty. In general, one might guess, it con-
sists of an exposure to ideas, to books, and to intelligent conver-
sation; the opportunity to acquire common technical knowledge
and skills; and exposure to persons with social skills, who are
good at getting along with other persons. Besides being a guess,
of course, such a statement is pretty vague and shows rather
clearly how much we cannot say about the matter. Also, we
have no way of knowing what ceiling there is on this environ-
mental influence, at what age it is greatest, and so on. Such
questions can be answered only by further research.
Our information at present is scanty, almost entirely natural-
istic rather than experimental in origin. Neff's (1938) review
will show the interested reader how definite the evidence is
that environment has a major effect on the IQ. Identical-twin
data, for example, are commonly supposed to have shown that
302 The Growth and Decline of Intelligence
heredity is the only major determinant of "intelligence." In none
of the studied pairs, however, was one twin brought up in an
entirely favorable environment, one in an entirely unfavorable
environment; usually they were brought up in about the same
stratum of society, so that similar IQ's may be the result of
similar environment, a similar heredity, or both.
What we really want, as evidence on this point, is one twin
brought up in a good home, with books, toys, kindergarten
training, and plenty of exposure to intelligent adults; the other
brought up by illiterate, poverty-stricken, anti-social mountain-
eers with low IQ's: then test their intelligences at maturity, and
try making doctors, or politicians, of both twins. When this is
done, and when it has been shown that the twin with a poor
environment (for the first fifteen years of life) can with time
become as good a diagnostician or committeeman as his geneti-
cally identical brother, it will then be in order to say that the
factual evidence shows that "intelligence" and the IQ are with-
out any major effect from experience; not before.
If the Iowa studies (Stoddard and Wellman, 1940) are dis-
trusted, the studies of Gordon, Asher, and Jordan, among others
reviewed by Neff, all tell the same story. The constancy of the
IQ is the main argument for its being determined by heredity;
but the fact is that the IQ is not constantit is stable, and
changes slowly, and more and more slowly as maturity is ap-
proached (Anderson, 1939), so that the IQ of the adult may be
constant (though there are few data on long-term comparisons
even here). But prediction of the adult IQ, in infancy, is more
accurate on the basis of the parents' IQ's than on that of the
infant himself; from an IQ at the age of five, prediction of the
IQ at twelve is about 20 per cent better than chance prediction,
very little better than can be done from knowing what kind of
home the child is growing up in.
There are then two determinants of intellectual growth: a
completely necessary innate potential (intelligence A), and a
completely necessary stimulating environment. It is not to the
point to ask which is more important; hypothetically, we might
suppose that intelligence will rise to the limit set by heredity or
environment, whichever is lower. Given a perfect environment,
The Nature and Nurture of Intelligence 303
the inherited constitution will set the pace; given the heredity
of a genius, the environment will do so.
The essentials of this environmental influence cannot be speci-
fied. Though we know that wealth, prolonged schooling, or
"intelligent" parents (that is, with intelligence B) are not essen-
tial, these things all may contribute. Since the guess has been
made that the essential is exposure to intelligence B, it is pre-
sumably true that the child must either have intelligent parents
or intelligent acquaintances and teachers. Schooling also is be-
coming more and more necessary to an understanding of adult
problems in this society; and a certain amount of wealth, of free-
dom from economic pressure, may be quite necessary to full intel-
lectual development. The fact is, however, that we know almost
nothing specific about the matter. The country may be full of
potential geniuses, for all we know, and it should be a pressing
concern for psychology to discover the conditions that will de-
velop whatever potentialities a child may have.
This page intentionally left blank
Bibliography
Adams, D. K. 1929. Experimental studies of adaptive behavior in cats.
Comp. Psychol. Monog., 6, No. 1.
Adrian, E. D. 1931. Potential changes in the isolated nervous system of
Dytiscus marginalia. J. Physiol., 72, 132-151.
Adrian, E. D. 1934. Electrical activity of the nervous system. Arch.
Neurol. Psychiat., 32, 1125-1136.
Adrian, E. D., and Buytendijk, F. J. J. 1931. Potential changes in the iso-
lated brain stem of the goldfish. J. Physiol, 71, 121-135.
Adrian, E. D., and Matthews, B. H. C. 1934. The interpretation of po-
tential waves in the cortex. J. Physiol., 81, 440471.
Allen, C., and Broster, L. R. 1945. A further case of paranoid psychosis
successfully treated by adrenalectomy. Brit. Med. J., No. 4402, 696-698.
Allport, G. W. 1946. Effect: a secondary principle of learning. Psychol.
Rev., 53, 335-347.
Anderson, J. E. 1939. The limitations of infant and preschool tests in the
measurement of intelligence. /. Psychol., 8, 351379.
Arvanitaki, A. 1942. Effects evoked in an axon by the activity of a con-
tiguous one. /. Neurophysiol., 5, 89108.
Bard, P. 1934. On emotional expression after decortication with some re-
marks on certain theoretical views. Psychol. Rev., 41, 309-329.
Bard, P. 1942. Neural mechanisms in emotional and sexual behavior.
Psychosom. Med., 4, 171-172.
Bartley, S. H., and Bishop, G. H. 1933. Factors determining the form of
the electrical response from the optic cortex of the rabbit. Amer. J.
Physiol., 103, 173-184.
Bartley, S. H., and Chute, E. 1947. Fatigue and impairment in man.
New York: McGraw-Hill.
Beach, F. A. 1937. The neural basis of innate behavior. I. Effects of
cortical lesions upon the maternal behavior pattern in the rat. J. Comp.
Psychol, 24, 393-439.
Beach, F. A. 1939. The neural basis of innate behavior. III. Compari-
son of learning ability and instinctive behavior in the rat. 7. Comp.
Psychol, 28, 225-262.
Beach, F. A. 1942. Analysis of factors involved in the arousal, mainte-
nance and manifestation of sexual excitement in male animals. Psy-
chosom. Med., 4, 173-198.
Beach, F. A. 1947a. A review of physiological and psychological studies
of sexual behavior in mammals. Physiol. Rev., 27, 240-307.
305
306 Bibliography
Beach, F. A. 1947&. Evolutionary changes in the physiological control of
mating behavior in mammals. Psychol. Rev., 54, 297-315.
Beach, F. A. 1948. Hormones and behavior. New York: Hoeber.
Bellak, L., and Willson, E. 1947. On the etiology of dementia praecox
.... /. Nerv. Ment. Dis., 105, 1-24.
Bellows, R. T. 1939. Time factors in water drinking in dogs. Amer. J.
Physiol, 125, 87-97.
Birch, H. G. 1945. The relation of previous experience to insightful
problem-solving. /. Comp. Psychol., 38, 367-383.
Bishop, G. H, 1946. Nerve and synaptic conduction. Ann. Rev. Physiol.,
8, 355-374.
v. Bonin, G., Carol, H. W., and McCulloch, W. S. 1942. The functional
organization of the occipital lobe. In Kliiver, H., Visual mechanisms.
Biol. Symp., 7, 165-192.
Boring, E. G. 1916. Cutaneous sensation after nerve-division. Quart. /.
Exp. Physiol, 10, 1-95.
Boring, E. G. 1930. A new ambiguous figure. Amer. J. Psychol., 42,
444-445.
Boring, E. G. 1933. The physical dimensions of consciousness. New
York: Century.
Boring, E. G. 1946. Mind and mechanism. Amer.. J. Psychol, 59, 173-
192.
Bousfield, W. A. 1935. Quantitative indices of the effects of fasting on
eating-behavior. /. Genet. Psychol, 46, 476-479.
Bowman, K. M. 1935. Psychoses with pernicious anemia. Amer. J.
Psychiat., 92, 371-396.
Bowman, K. M. 1946. Modern concept of the neuroses. /. Amer. Med.
Assoc., 132, 555-557.
Bridgman, C. S., and Smith, K. U. 1945. Bilateral neural integration in
visual perception after section of the corpus callosum. /. Comp. Neurol,
83, 57-68.
Bronk, D. W. 1939. Synaptic mechanisms in sympathetic ganglia.
J. Neurophysiol, 2, 380-401.
Brown, Warner. 1932. Spatial integrations in a human maze. Univ.
Calif. Publ. Psychol, 5, 123-134.
Bruetsch, W. L. 1947. Rheumatic brain disease: late sequel of rheumatic
fever. J. Amer. Med.Assoc., 134, 450-454.
Carlson, A. J. 1916. The control of hunger in health and disease. Chi-
cago: Univ. Chic. Press.
Carmichael, L., Hogan, H. P., and Walter, A. A. 1932. An experimental
study of the effect of language on the reproduction of visually perceived
form. J. Exp. Psychol, 15, 73-86.
Chandler, A. R. 1934. Beauty and human nature. New York: Appleton-
Century.
Clark, G., and Lashley, K. S. 1947. Visual disturbances following frontal
ablations in the monkey. Anat. Rec., 97, 326.
Bibliography 307
Cobb, S. 1944. Personality as affected by lesions of the brain. In Hunt,
J. McV.,Personality and the behavior disorders. New York: Ronald,
Vol. I, pp. 550-581.
Cobb, S., Cohen, M. E., and Badal, D. W. 1946. Capillaries of the nail
fold in patients with neurocirculatory asthenia (effort syndrome, anxiety
neurosis). Arch. Neurol. Psychiat., 56, 643-650.
Cohn, R. 1945. Electroencephalographic study of prefrontal lobotomy.
Arch. Neurol. Psychiat., 53, 283-288.
Cowles, J. T., and Nissen, H. W. 1937. Reward-expectancy in delayed
responses of chimpanzees. /. Comp. Psychol., 24, 345358.
Dandy, W. E. 1933. Physiologic studies following extirpation of the right
cerebral hemisphere in man. Bull. Johns Hopkins Hosp., 53, 3151.
Daniel, R. S., and Smith, K. U. 1947. The sea-approach behavior of the
neonate loggerhead turtle. J. Comp. Physiol. Psychol., 40, 413-420.
Dashiell, J. F. 1928. Are there any native emotions? Psychol. Rev., 35,
319-327.
Davison, C., and Demuth, E. L. 1945. Disturbances in sleep mechanism:
a clinico-pathologic study. II. Lesions at the corticodiencephalic level.
Arch. Neurol. Psychiat., 54, 241-255.
Denker, P. G. 1946. Results of treatment of psychoneuroses by the gen-
eral practitioner: a follow-up study of 500 cases. N. Y. State J. Med.,
46, 2164-2166.
Dennis, W. 1934. Congenital cataract and unlearned behavior. J. Genet.
Psychol, 44, 340-350.
Dennis, W. 1940. Infant reaction to restraint: an evaluation of Watson's
theory. Trans. N. Y. Acad. Set., Ser. 2., 2, No. 8, 202-218.
Denny-Brown, D. 1932. Theoretical deductions from the physiology of
the cerebral cortex. J. Neurol. Psychopathol, 13, 52-67.
Dewan, J. G., and Owen, T. 1945. Mental illness and the principles of
medicine. Canad. Med. Ass. J., 52, 349-357.
Doll, E. A. 1933. Psychological significance of cerebral birth lesions.
Amer. J. Psychol, 45, 444-452.
Doll, E. A., Phelps, W. M., and Melcher, R. T. 1932. Mental deficiency
due to birth injuries. New York: Macmillan.
Drew, G. C. 1938. The function of punishment in learning. /. Genet.
Psychol, 52, 257-267.
Dubner, H. H., and Gerard, R. W. 1939. Factors controlling brain po-
tentials in the cat. /. Neurophysiol, 2, 142-152.
Dunlap, K. 1932. Habits: their making and remaking. New York:
Liveright.
Editors, Nutrition Reviews. 1944. Self-selection of diets. Nutrition Rev.,
2, 199-203.
Egana, E., Johnson, R. E., Bloomfield, R., Brouha, L., et al. 1942. The
effects of a diet deficient in the vitamin B complex on sedentary men.
Amer. J. Physiol, 137, 731-741.
308 Bibliography
Erlanger, J. 1939. The initiation of impulses in axons. /. Neurophysfol.,
2, 370-379.
Ferraro, A., Arieti, .S., and English, W. H. 1945. Cerebral changes in
the course of pernicious anemia and their relationship to psychic symp-
toms. /. Neuropathol Exp. Neurol, 4, 217-239.
Fields, P. E. 1932. Studies in concept formation. I. Comp. Psychol.
Monog., 9, No. 2.
Forbes, A. 1939. Problems of synaptic function. J. Neurophysiol, 2,
465-472.
Freeman, G. L. 1934. Introduction to physiological psychology. New
York: Ronald.
Freeman, W., and Watts, J. W. 1942. Psychosurgery: intelligence, emo-
tion and social behavior following prefrontal lobotomy for mental dis-
orders. Springfield: Thomas.
Freeman, W., and Watts, J. W. 1946. Psychosurgery. In Spiegel, E. A.,
Progress in neurology and psychiatry: an annual review. New York:
Grune and Stratton, pp. 649-661.
Fuchs, W. 1920. Untersuchungen iiber das Sehen der Hemianopiker und
Hemiamblyopiker: II. In Gelb, A., and Goldstein, K., Psychologischen
Analysen hirnpathologischer Falle. Leipzig: Earth, pp. 419-561.
Fulton, J. F. 1943. Physiology of the nervous system. 2nd Ed., New
York: Oxford Univ. Press.
Gantt, W. H. 1938. Extension of a conflict based upon food to other
physiological systems and its reciprocal relations with sexual functions.
Amer. J. Physiol, 123, 73-74.
Gantt, W. H. 1944. Experimental basis for neurotic behavior: origin and
development of artificially produced disturbances of behavior in dogs.
Psychosom. Med. Monog., 3, Nos. 3 and 4.
Garrison, M. 1947. The genetics of schizophrenia. /. Abn. Soc. Psychol.,
42, 122-124.
Gasser, H. S. 1937. The control of excitation in the nervous system.
Harvey Led., pp. 169-193.
Gellerman, L. W. 1933. Form discrimination in chimpanzees and two-
year-old children: I. Form (triangularity) per se. J. Genet. Psychol,
42, 3-27.
Gibbs, F. A. 1945. Electrical activity of the brain. Ann. Rev. Physiol,
7, 427-454.
Gibson, J. J. 1929. The reproduction of visually perceived forms.
J. Exp. Psychol, 12, 1-39.
Gibson, J. J. 1941. A critical review of the concept of set in contempo-
rary experimental psychology. Psychol. Bull, 38, 781-817.
Gibson, J. J., and Crooks, L. E. 1938. A theoretical field-analysis of
automobile driving. Amer. J. Psychol, 51, 453-471.
Gillespie, W. H. 1944. The psychoneuroses. J. Ment. Sci., 90, 287-306.
Goodenough, F. L. 1931. Anger in young children. Minneapolis: Univ.
Minnesota Press.
Bibliography 309
Greene, R., Paterson, A. S., and Pile, G. C. S. 1945. Hypertrichosis with
mental changes: the effect of adrenalectomy. Brit. Med. /., No. 4402,
698-699.-
Guthrie, E. R. 1946. Psychological facts and psychological theory.
Psychol. Bull, 43, 1-20.
Halstead, W. C. 1947. Brain and intelligence. Chicago: Univ. Chic.
Press.
Hanawalt, N. G. 1937. Memory trace for figures in recall and recogni-
tion. Arch. Psychol, No. 216, 1-89.
Harris, H. J. 1944. Brucellosis: a case report illustrating a psychosomatic
problem. Psychosom. Med., 6, 334-335.
Hauptman, A. 1946. Capillaries in the finger nail fold in patients with
neurosis, epilepsy, and migraine. Arch. Neurol. Psychiat., 56, 631-642.
Head, H. 1920. Studies in neurology. London: Frowde, Hodder and
Stoughton.
Heath, R. G., and Pool, J. L. 1948. Bilateral fractional resection of
frontal cortex for the treatment of psychoses. /. Nerv. Ment. Dis., 107,
411^29.
Hebb, D. O. 1937a. The innate organization of visual activity: I. Per-
ception of figures by rats reared in total darkness. J. Genet. Psychol,
51, 101-126.
Hebb, D. O. 1937&. The innate organization of visual activity: II.
Transfer of response in the discrimination of brightness and size by rats
reared in total darkness. J. Comp. Psychol, 24, 277-299.
Hebb, D. O. 1938o. Studies of the organization of behavior: I. Behavior
of the rat in a field orientation. J. Comp. Psychol, 25, 333-352.
Hebb, D. O. 1938k. Studies of the organization of behavior: II.
Changes in the field orientation of the rat after cortical destruction.
J. Comp. Psychol, 26, 427-444.
Hebb, D. O. 1939. Intelligence in man after large removals of cerebral
tissue: report of four left frontal lobe cases. J. Gen. Psychol, 21, 73-87.
Hebb, D. O. 1942a. The effect of early and late brain injury upon test
scores, and the nature of normal adult intelligence. Proc. Amer. Phil.
Soc., 85, 275-292.
Hebb, D. O. 1942i>. Verbal test material independent of special vocabu-
lary difficulty. J. Educ. Psychol, 33, 691-696.
Hebb, D. O. 1945a. The forms and conditions of chimpanzee anger.
Bull Canad. Psychol Ass., 5, 32-35.
Hebb, D. O. 1945&. Man's frontal lobes: A critical review. Arch.
Neurol Psychiat., 54, 10-24.
Hebb, D. O. 1946c. Emotion in man and animal: an analysis of the
intuitive processes of recognition. Psychol. Rev., 53, 88-106.
Hebb, D. O. 1946&. On the nature of fear. Psychol Rev., 53, 259-276.
Hebb, D. O. 1947. Spontaneous neurosis in chimpanzees: theoretical re-
lations with clinical and experimental phenomena. Psychosom. Med.,
9, 3-16.
310 Bibliography
Hebb, D. O., and Foord, E. N. 1945. Errors of visual recognition and
the nature of the trace. J. Exp. Psychol, 35, 335-348.
Hebb, D. O., and Morton, N. W. 1943. The McGill Adult Comprehen-
sion Examination: "Verbal Situation" and "Picture Anomaly" Series.
J. Educ. Psychol, 34, 16-25.
Hebb, D. O., and Morton, N. W. 1944. Note on the measurement of
adult intelligence. 7. Gen.Psychol, 30, 217-223.
Hebb, D. O., and Penfield, W. 1940. Human behavior after extensive
bilateral removal from the frontal lobes. Arch. Neural Psychiat., 44,
421-438.
Hebb, D. O., and Riesen, A. H. 1943. The genesis of irrational fears.
Bull Canad. Psychol Ass., 3, 49-50.
Hebb, D. O., and Williams, K. 1941. Experimental control of cues de-
termining the rat's orientation. Bull Canad. Psychol Ass., 1, 22-23.
Hebb, D. O., and Williams, K. 1946. A method of rating animal intel-
ligence. /. Gen. Psychol, 34, 59-65.
Herrick, C. J. 1929. The thinking machine. Chicago: Univ. Chic. Press.
Hilgard, E. R., and Marquis, D. G. 1940. Conditioning and learning.
New York: Appleton-Century.
Hoagland, H. 1947. Enzyme kinetics and the dynamics of behavior.
7. Comp. Physiol Psychol, 40, 107-127.
Hoagland, H., Malamud, W., Kaufman, I. C., and Pincus, G. 1946.
Changes in the electroencephalogram and in the excretion of 17-ketoster-
oids accompanying electro-shock therapy of agitated depression. Psy-
chosom. Med., 8, 246-251.
Hobbs, G. E. 1941. Mental disorder in one of a pair of identical twins.
Amer. J. Psychiat., 98, 447-450.
Hobhouse, L. T. 1915. Mind in evolution. 2nd Ed. London: Mac-
millan.
Hovland, C. I. 1936. "Inhibition of reinforcement" and phenomena of
experimental extinction. Proc. Nat. Acad. Sci., Wash., 22, 430-433.
(Quoted by Hilgard and Marquis, 1940, p. 146.)
Hovland, C. I. 1937. The generalization of conditioned responses. II.
The sensory generalization of conditioned responses with varying inten-
sities of tone. 7. Genet. Psychol, 51, 279-291.
Hull, C. L. 1934. The concept of the habit-family hierarchy and maze
learning. Psychol Rev., 41, 33-54; 134-152.
Hull, C. L. 1943. Principles of behavior: an introduction to behavior
theory. New York: Appleton-Century.
Hull, C. L. 1945. The discrimination of stimulus configurations and the
hypothesis of afferent neural interaction. Psychol Rev., 52, 133-142.
Humphrey, G. 1940. The problem of the direction of thought. Brit. J.
Psychol, 30, 183-196.
Hunt, J. McV. 1941. The effects of infant feeding-frustration upon adult
hoarding in the albino rat. 7. Abn. Soc. Psychol, 36, 338-360.
Bibliography 811
Hunter, W. S. 1934. Learning: IV. Experimental studies of learning.
In Murchison, C., Handbook of general experimental psychology.
Worcester, Mass.: Clark Univ. Press, pp. 497-570.
Jackson, T. A. 1942. Use of the stick as a tool by young chimpanzees.
/. Comp. Psychol, 34, 223-235.
Jacobsen, C. F., Jacobsen, M. M., and Yoshioka, J. G. 1932. Develop-
ment of an infant chimpanzee during her first year. Comp. Psychol.
Monog., 9, 1-94.
James, W. 1910. Principles of psychology. New York: Holt.
Jasper, H. H. 1937. Electrical signs of cortical activity. Psychol. Bull,
34, 411-481.
Jasper, H. H. 1941. Electroencephalography. In Penfield, W., and
Erickson, T. C., Epilepsy and cerebral localization. Springfield: Thomas,
pp. 380^454.
Jasper, H. H., and Fortuyn, J. D. 1946. Experimental studies on the
functional anatomy of petit mal epilepsy. Publ. Ass. Res. Nerv. Ment.
Dis., 26, 272-298.
Jefferson, G. 1937. Removal of right or left frontal lobes in man. Brit.
Med. J., 2, 199-206.
Jersild, A. T., and Holmes, F. B. 1935. Children's fears. New York:
Teach. Coll. Bur. Publ.
Jolliffe, N. 1942. The neuropsychiatric manifestations of vitamin defi-
ciencies. J. Mt. Sinai Hasp., 8, 658-667.
Jones, H. E., and Conrad, H. S. 1933. The growth and decline of intel-
ligence: a study of a homogeneous group between the ages of ten and
sixty. Genet. Psychol. Monog., 13, No. 3.
Jones, H. E., and Jones, M. C. 1928. A study of fear. Childhood Educ.,
5, 136-143.
Jones, M. C. 1933. Emotional development. In Murchison, C., A hand-
book of child psychology. 2nd Ed. Worcester, Mass.: Clark Univ. Press,
pp. 271-302.
Kappers, C. U. A., Huber, G. C., and Crosby, E. C. 1936. The compara-
tive anatomy of the nervous system of vertebrates, including man. New
York: Macmillan, Vol. I.
Karnosh, L. J., and Gardner, W. J. 1941. An evaluation of the physical
and mental capabilities following removal of the right cerebral hemi-
sphere. Cleveland Clin. Quart., 8, 94-106.
Kennard, M. A. 1939. Alterations in response to visual stimuli following
lesions of frontal lobe in monkeys. Arch. Neurol. Psychiat., 41, 1153-
1165.
Kennard, M. A., and Ectors, L. 1938. Forced circling in monkeys follow-
ing lesions of the frontal lobes. /. Neurophysiol, 1, 45-54.
Kennedy, F., and Wolf, A. 1936. The relationship of intellect to speech
defect in aphasic patients. /. Nerv. Ment. Dis., 84, 125-145; 293-311.
Keschner, M., Bender, M., and Strauss, I. 1938. Mental symptoms asso-
312 Bibliography
ciated with brain tumor: a study of 530 verified cases. /. Amer. Med.
Ass., 110, 714-718.
Kinder, E. F. 1927. A study of the nest-building activity of the albino
rat. /. Exp. Zool, 47, 117-161.
Kinsey, A. C., Pomeroy, W. B., and Martin, C. E. 1948. Sexual behavior
in the human male. Philadelphia: Saunders.
Klebanoff, S. G. 1945. Psychological changes in organic brain lesions and
ablations. Psychol Bull, 42, 585-623.
Kleitman, N. 1939. Sleep and wakefulness. Chicago: Univ. Chic. Press.
Klineberg, O. 1940. Social psychology. New York: Holt.
Koffka, K. 1924. The growth of the mind. New York: Harcourt, Brace.
Koffka, K. 1935. Principles of Gestalt psychology. New York: Harcourt,
Brace.
Kohler, W. 1925. The mentality of apes. New York: Harcourt, Brace.
Kohler, W. 1929. Gestalt psychology. New York: Liveright.
Kohler, W. 1940. Dynamics in psychology. New York: Liveright.
Kohler, W., and Wallach, H. 1944. Figural after-effects: an investigation
of visual processes. Proc. Amer. Phil. Soc., 88, 269-357.
Krechevsky, I. 1932. "Hypotheses" versus "chance" in the pre-solution
period in sensory discrimination-learning. Univ. Calif. Publ. Psychol.,
6, 27-44.
Krechevsky, I. 1938. An experimental investigation of the principle of
proximity in the visual perception of the rat. /. Exp. Psychol., 22,
497-523.
Kubitschek, P. E. 1928. The symptomatology of tumors of the frontal
lobe based on a series of twenty-two cases. Arch. Neural. Psychiat., 20,
559-579.
Landis, C. 1947. A modern dynamic psychology. J. Comp. Physiol.
Psychol, 40, 135-141.
Landis, C., and Hunt, W. A. 1932. Adrenalin and emotion. Psychol
Rev., 39, 467-485.
Lashley, K. S. 1929a. Brain mechanisms and intelligence: a quantitative
study of injuries to the brain. Chicago: Univ. Chic. Press.
Lashley, K. S. 1929&. Nervous mechanisms in learning. In Murchi-
son, C., The foundations of experimental psychology. Worcester: Clark
Univ. Press, pp. 524-563.
Lashley, K. S. 1930. Basic neural mechanisms in behavior. Psychol
Rev., 37, 1-24.
Lashley, K. S. 1934. The mechanism of vision. VIII. The projection of
the retina upon the cerebral cortex of the rat. /. Comp. Neural, 60,
57-79.
Lashley, K. S. 1937. Functional determinants of cerebral localization.
Arch. Neural Psychiat., 38, 371-387.
Lashley, K. S. 1938a. Experimental analysis of instinctive behavior.
Psychol Rev., 45, 445-471.
Bibliography 313
Lashley, K. S. 1938&. The mechanism of vision: XV. Preliminary studies
of the rat's capacity for detail vision. 7. Gen. Psychol, 18, 123-193.
Lashley, K. S. 1938c. The thalamus and emotion. Psychol. Rev., 45,
42-61.
Lashley, K. S. 1941. Patterns of cerebral integration indicated by the
scotomas of migraine. Arch, Neurol. Psychiat., 46, 331-339.
Lashley, K. S. 1942a. The problem of cerebral organization in vision.
In Kliiver, H., Visual mechanisms. Biol. Sympos., 7, 301-322.
Lashley, K. S. 1942fe. An examination of the "continuity theory" as ap-
plied to discrimination learning. 7. Gen. Psychol., 26, 241-265.
Lashley, K. S. 1944. Studies of cerebral function in learning: XIII.
Apparent absence of transcortical association in maze learning. 7. Comp.
Neurol, 80, 257-281.
Lashley, K. S., and Clark, G. 1946. The cytoarchitecture of the cerebral
cortex of Ateles: a critical examination of architectonic studies. J. Comp.
Neurol, 85, 223-306.
Lashley, K. S., and Wade, M. 1946. The Pavlovian theory of generaliza-
tion. Psychol Rev., 53, 72-87.
Leeper, R. W. 1935. A study of a neglected portion of the field of
learningthe development of sensory organization. J. Genet. Psychol, 46,
41-75.
Leeper, R. W. 1948. A motivational theory of emotion to replace "Emo-
tion as disorganized response." Psychol. Rev., 55, 5-21.
Lehmann, J. E. 1937fl. The effect of changes in the potassium-calcium
balance on the action of mammalian A nerve fibers. Amer. 7. Physiol,
118, 613-619.
Lehmann, J. E. 1937&. The effect of changes in pH on the action of
mammalian A nerve fibers. Amer. J. Physiol, 118, 600612.
Levine, J. 1945a. Studies in the interrelations of central nervous struc-
tures in binocular vision: I. The lack of bilateral transfer of visual dis-
criminative habits acquired monocularly by the pigeon. 7. Genet. Psy-
chol, 67, 105-129.
Levine, J. 1945fo. Studies in the interrelations of central nervous struc-
tures in binocular vision: II. The conditions under which interocular
transfer of discriminative habits takes place in the pigeon. J. Genet.
Psychol, 67, 131-142.
Lewin, K. 1938. Will and needs. In Ellis, W. D., A source book of
Gestalt psychology. London: Kegan Paul, Trench, Trubner, pp. 283-299.
Libet, B., and Gerard, R. W. 1939. Control of the potential rhythm of
the isolated frog brain. 7. Neurophysiol, 2, 153-169.
Liddell, H. S. 1938. The experimental neurosis and the problem of
mental disorder. Amer. J. Psychiat., 94, 1035-1041.
Liddell, H. S. 1944. Animal behavior studies bearing on the problem of
pain. Psychosom. Med., 6, 261-263.
Lorente de N6, R. 1938o. Synaptic stimulation of motoneurons as a local
process. 7. Neurophysiol, 1, 195-206.
314 Bibliography
Lorente de No, R. 1938&. Analysis of the activity of the chains of inter-
nuncial neurons. J. NeurophysioL, 1, 207-244.
Lorente de No, R. 1939. Transmission of impulses through cranial motor
nuclei. /. Neurophysiol, 2, 402-464.
Lorente de No, R. 1943. Cerebral cortex: architecture. In Fulton, J. F.,
Physiology of the nervous system, 2nd Ed. New York: Oxford Univ.
Press, pp. 274-301.
Lorenz, K. 1935. Der Kurnpan in der Umwelt des Vogels. /. Ornith., 83,
137-213; 289-413.
Loucks, R. B. 1935. Experimental delimitation of neural structures essen-
tial for learning: The attempt to condition striped muscle responses with
faradization of the sigmoid gyri. /. PsychoL, 1, 5-44.
Loucks, R. B. 1938. Studies of neural structures essential for learning.
II. The conditioning of salivary and striped muscle responses to faradi-
zation of cortical sensory elements, and the action of sleep upon such
mechanisms. /. Comp. PsychoL, 25, 315-332.
Luckhardt, A. B., and Carlson, A. J. 1915. Contributions to the physi-
ology of the stomach. XVII. On the chemical control of the gastric
hunger mechanism. Amer. J. Physiol., 36, 37-46.
McBride, A. F., and Hebb, D. O. 1948. Behavior of the captive bottle-
nose dolphin, Tursiops truncatus. /. Comp. Physiol. PsychoL, 41, 111-
123.
McCulIoch, T. L., and Haslerud, G. M. 1939. Affective responses of an
infant chimpanzee reared in isolation from its kind. J. Comp. PsychoL,
28, 437-445.
McCulloch, W. S. 1944a. Cortico-cortical connections. In Bucy, P., The
precentral motor cortex. Urbana, 111.: Univ. Illinois Press, pp. 213-242.
McCulloch, W. S. 19446. The functional organization of the cerebral
cortex. Physiol. Rev., 24, 390-407.
McGeoch, J. A. 1942. The psychology of human learning. New York:
Longmans, Green.
Maier, N. R. F., and Schneirla, T. C. 1935. Principles of animal psy-
chology. New York: McGraw-Hill.
Marshall, W. H., and Talbot, S. A. 1942. Recent evidence for neural
mechanisms in vision leading to a general theory of sensory acuity. In
Kliiver, H., Visual mechanisms. Biol. Symp., 7, 117-164.
Masserman, J. H. 1942. The hypothalamus in psychiatry. Amer. J.
Psychiat., 98, 633-637.
Masserman, J. H. 1943. Behavior and neurosis: An experimental psycho-
analytic approach to psychobiologic principles. Chicago: Univ. Chic.
Press.
Matthews, R. S. 1938. Pellagra and nicotinic acid. J. Amer. Med. Ass.,
I I I , 1148-1153.
Miller, G. A. 1947. The masking of speech. PsychoL Bull., 44, 105-129.
Miner, J. B. 1905. A case of vision acquired in adult life. PsychoL Rev.
Monog. SuppL, 6, No. 5, 103-118.
Bibliography 315
Mixter, W. J., Tillotson, K. J., and Wies, D. 1941. Reports of partial
frontal lobectomy and frontal lobotomy performed on three patients: one
chronic epileptic and two cases of chronic agitated depression. Psy-
chosom. Med., 3, 26-37.
Morgan, C. T. 1943. Physiological psychology. New York: McGraw-Hill.
Morison, R. S., and Dempsey, E. W. 1943. Mechanism of thalamocortical
augmentation and repetition. Amer. J. Physiol., 138, 297-308.
Moss, F. A. (Ed.). 1942. Comparative psychology. Rev. Ed. New York:
Prentice-Hall.
Mowrer, O. H. 1941. Motivation and learning in relation to the national
emergency. Psychol. Bull, 38, 421-431.
Mowrer, O. H., and Mowrer, W. M. 1938. Enuresisa method for its
study and treatment. Amer. J. Orthopsychiat., 8, 436-459.
Muenzinger, K. F. 1934. Motivation in learning. I. Electric shock for
correct response in the visual discrimination habit. J. Comp. Psychol.,
17, 267-277.
Murphy, J. P., and Gellhorn, E. 1945. Further investigations on dien-
cephalic-cortical relations and their significance for the problem of emo-
tion. /. Neurophysiol, 8, 431-447.
Nafe, J. P. 1934. The pressure, pain, and temperature senses. In Murchi-
son, C., Handbook of general experimental psychology. Worcester,
Mass.: Clark Univ. Press, pp. 1037-1087.
Nauta, W. J. H. 1946. Hypothalamic regulation of sleep in rats: An ex-
perimental study. J. Neurophysiol, 9, 285-316.
Neet, C. C. 1933. Visual pattern discrimination in the Macacus rhesus
monkey. J. Genet. Psychol, 43, 163-196.
Neff, Walter S. 1938. Socioeconomic status and intelligence: a critical
survey. Psychol. Bull, 35, 727-757.
Nichols, I., and Hunt, J. McV. 1940. A case of partial bilateral frontal
lobectomy: a psychopathological study. Amer. J. Psychiat., 96, 1063-
1083.
Nissen, H. W., Machover, S., and Kinder, E. F. 1935. A study of per-
formance tests given to a group of native African negro children. Brit.
J. Psychol, 25, 308-355.
Pavlov, I. P. 1927. Conditioned reflexes. Oxford: Humphrey Milford.
Pavlov, I. P. 1928. Lectures on conditioned reflexes. New York: Inter-
national.
Pavlov, I. P. 1932. The reply of a physiologist to psychologists. Psychol
Rev., 39, 91-126.
Penfield, W., and Jasper, H. 1946. Highest level seizures. Res. Publ
Ass. Nerv. Ment. Dis., 26, 252-271.
Pennington, L. A. 1938. The function of the brain in auditory localiza-
tion. IV. Method of training and control experiments. /. Comp.
Psychol, 25, 195-211.
Pilgrim, F. J., and Patton, R. A. 1947. Patterns of self-selection of puri-
316 Bibliography
fied dietary components by the rat. /. Comp. Physiol. Psychol., 40,
343-348.
Pillsbury, W. B. 1913. "Fluctuations of attention" and the refractory
period. J. Phil. Psychol. Sci. Meth., 10, 181-185.
Polyak, S. L. 1941. The retina. Chicago: Univ. Chic. Press.
Postman, L. 1947. The history and present status of the law of effect.
Psychol. Bull, 44, 489-563.
Pratt, C. C. 1939. The logic of modern psychology. New York: Mac-
millan.
Prentice, W. C. H. 1946. Operationism and psychological theory: a note.
Psychol. Rev., S3, 247-249.
Prosser, C. L. 1934. Action potentials in the nervous system of the cray-
fish: I. Spontaneous impulses. /. Cell. Comp. Physiol., 4, 185-209.
Richter, C. P., Holt, L. E., and Barelare, B. 1938. Nutritional require-
ments for normal growth and reproduction in rats studied by the self-
selection method. Amer. J. Physiol, 122, 734-744.
Riddoch, G. 1941. Phantom limbs and body shape. Brain, 64, 197-222.
Riesen, A. H. 1947. The development of visual perception in man and
chimpanzee. Science, 106, 107-108.
Riesen, A. H., and Nissen, H. W. 1942. Non-spatial delayed response by
the matching technique. J. Comp. Psychol., 34, 307-313.
Roethlisburger, F. J., and Dickson, W. J. 1939. Management and the
worker. Cambridge: Harvard Univ. Press.
Romano, J., and Coon, G. P. 1942. Physiologic and psychologic studies
in spontaneous hypoglycemia. Psychosom. Med., 4, 283-300.
Rowe, S. N. 1937. Mental changes following the removal of the right
cerebral hemisphere for brain tumor. Amer. J. Psychiat., 94, 605-614.
Rubin, E. 1921. Visuell wahrgenommene Figuren: Studien in psycho-
logischer Analyse. Teil I. Berlin: Gyldendalske Boghandel.
Rylander, G. 1939. Personality changes after operations on the frontal
lobes: a clinical study of 32 cases. London: Humphrey Milford.
Saphir, W. 1945. Chronic hypochloremia simulating psychoneurosis.
J. Amer. Med. Ass., 129, 510-512.
Schneirla, T. C. 1948. Psychology, comparative. Encycl. Brit.
Scott, W. W., Scott, C. C., and Luckhardt, A. B. 1938. Observations on
the blood sugar level before, during, and after hunger periods in humans.
Amer. J. Physiol, 123, 243-247.
Senden, M. v. 1932. Raum- und Gestaltauffassung bei operierten Blind-
geborenen vor und nach der Operation. Leipzig: Earth.
Sherrington, C. S. 1906. Integrative action of the nervous system. New
York: Scribner.
Sherrington, C. S. 1925. Remarks on some aspects of reflex inhibition.
Proc. Roy. Soc., 97B, 519-545.
Sherrington, C. S. 1941. Man on his nature. New York: Macmillan.
Skinner, B. F. 1938. The behavior of organisms: an experimental analysis.
New York: Appleton-Century.
Bibliography 317
Smith, D. E. 1939. Cerebral localization in somesthetic discrimination in
the rat. J. Comp. Psychol, 28, 161-188.
Smith, K. U. 1936. Visual discrimination in the cat: III. The relative
effect of paired and unpaired stimuli in the discriminative behavior of
the cat. /. Genet. Psychol, 48, 29-57.
Smith, W. K. 1945. The functional significance of the rostral cingular
gyrus as revealed by its responses to electrical excitation. /. Neuro-
physiol., 8, 241-255.
Spearman, C. 1927. The abilities of man. New York: Macmillan.
Spence, K. W. 1938. Gradual versus sudden solution of discrimination
problems by chimpanzees. /. Comp. Psychol., 25, 213-224.
Spence, K. W. 1940. Continuous versus non-continuous interpretations of
discrimination learning. Psychol. Rev., 47, 271-288.
Sperry, R. W. 1943. Visuomotor coordination in the newt (Triturus
viridescens) after regeneration of the optic nerve. J. Comp. Neurol., 79,
33-55.
Sperry, R. W. 1947. Effect of crossing nerves to antagonistic limb mus-
cles in the monkey. Arch. Neurol. Psychiat., 58, 452-473.
Spiegel, E. A., Miller, H. R., and Oppenheimer, M. J. 1940. Forebrain
and rage reactions. J. Neurophysiol., 3, 539548.
Spies, T. D., Aring, C. D., Gelperin, J., and Bean, W. B. 1938. The
mental symptoms of pellagra: Their relief with nicotinic acid. Amer. J.
Med. Set., 196, 461-475.
Spragg, S. D. S. 1940. Morphine addiction in chimpanzees. Comp.
Psychol. Monog., 15, No. 7.
Stoddard, G. D., and Wellman, B. L. 1940. Environment and the IQ.
Yearb. Nat. Soc.Stud. Educ., 39 (I) , 405-442.
Stookey, B., Scarff, J., and Teitelbaum, M. 1941. Frontal lobectomy in
the treatment of brain tumors. Ann. Surg., 113, 161-169.
Swank, R. L., and Marchand, W. E. 1946. Combat neuroses: develop-
ment of combat exhaustion. Arch. Neurol. Psychiat., 55, 236-247.
Thorndike, E. L. 1931. Human learning. New York: Century.
Thurstone, L. L. 1935. The vectors of mind. Chicago: Univ. Chic. Press.
Tinbergen, N. 1942. An objectivistic study of the innate behavior of
animals. Bibl. Biotheoret, Leiden, 1, 39-98.
Tinklepaugh, O. L. 1928. An experimental study of representative fac-
tors in monkeys. J. Comp. Psychol., 8, 197-236.
Titchener, E. B. 1920. Notes from the psychological laboratory of Cornell
University. Amer. J. Psychol, 31, 212-214.
Tolman, E. C. 1932. Purposive behavior in animals and men. New
York: Century.
Tryon, R. C. 1939. Studies in individual differences in maze learning: VI.
Disproof of sensory components: experimental effects of stimulus varia-
tion. /. Comp. Psychol, 28, 361-415.
Valentine, C. W. 1930. The innate bases of fear. /. Genet. Psychol, 37,
394-419.
318 Bibliography
Walker, A. E., and Weaver, T. A. 1940. Ocular movements from the
occipital lobe in the monkey. J. Neurophysiol., 3, 353-357.
Watson, J. B. 1924. Behaviorism. New York: Norton.
Watts, J. W., and Freeman, W. 1946. Psychosurgery for the relief of in-
tractable pain. /. Int. Coll Surg., 9, 679-683.
Wechsler, D. 1939. The measurement of adult intelligence. Baltimore:
Williams and Wilkins.
Weddel, G., Sinclair, D. C., and Feindel, W. H. 1948. An anatomical
basis for alterations in quality of pain sensibility. J. Neurophysiol., 11,
99-109.
Weisenburg, T., and McBride, K. E. 1935. Aphasia: a clinical and psy-
chological study. New York: Commonwealth Fund.
Weisenburg, T., Roe, A., and McBride, K. E. 1936. Adult intelligence:
a psychological study of test performances. New York: Commonwealth
Fund.
Weiss, P. 1941a. Autonomous versus reflexogenous activity of the central
nervous system. Proc. Amer. Phil. Soc., 84, 5364.
Weiss, P. 1941&. Nerve patterns: The mechanics of nerve growth.
Growth (Third Growth Symposium), 5, 163-203.
Werner, H., and Strauss, A. 1939. Types of visuo-motor activity in their
relation to low and high performance ages. Proc. Amer. Ass. Ment.
Defic., 44, 163-168.
Wilson, G., and Rupp, C. 1947. Present trends in the practice of neu-
rology. J. Amer. Med. Ass., 133, 509-511.
Wolf, E., and Zerrahn-Wolf, G. 1937. Flicker and the reactions of bees
to flowers. /. Gen.Physiol, 20, 511-518.
Wolf, G. A., and Wolff, H. G. 1946. Studies on the nature~of certain
symptoms associated with cardiovascular disorders. Psychosom. Med., 8,
293-319.
Wolff, H. G. 1943. Emotions and gastric function. Science, 98, 481-484.
Wolff, H. G., and Hardy, J. D. 1947. On the nature of pain. Physiol.
Rev., 27, 167-199.
Woodrow, H. 1927. The effect of type of training on transference.
J. Educ. Psychol, 18, 160-171.
Woodworth, R. S. 1921. Psychology. New York: Holt.
Woodworth, R. S. 1938. Experimental psychology. New York: Holt.
Wortis, H., Stein, M. H., and Jolliffe, N. 1942. Fiber dissociation in
peripheral neuropathy. Arch. Int. Med., 69, 222-237.
Yerkes, R. M. 1916. The mental life of monkeys and apes: a study of
ideational behavior. Behavior Monog., 3, No. 1.
Young, P. T. 1941. The experimental analysis of appetite. Psychol. Bull.,
38, 129-164.
Young, P. T. 1944. Studies of food preference, appetite and dietary habit.
I. Running activity and dietary habit of the rat in relation to food pref-
erence. /. Comp. Psychol, 37, 327-370.
Bibliography 319
Zangwill, O. L. 1937. A study of the significance of attitude in recogni-
tion. Brit. J. Psychol, 28, 12-17.
Zener, K. 1937. The significance of behavior accompanying conditioned
salivary secretion for theories of the conditioned response. Amer. J.
Psychol, 50, 384-403.
Zollinger, R. 1935. Removal of left cerebral hemisphere: report of a case.
Arch. Neurol. Psychiat., 34, 1055-1064.
This page intentionally left blank
Index
Abstraction, 104f
Adams, D. K., 159
Addiction, 200-206
distinction from habit, 202f
intermediate stage of, 203
learning in, 203f
relation to food habits, emotion,
205
Adrian, E. D., 8f, 124, 184, 217, 274
"Afferent neural interaction," 15, 56,
79
Agnosia, 285f
Agraphia, 285
Alcohol as removing inhibition, 209ff
Alexia, 285
Allen, C., 262
Allport, G. W., 179
Analogy in scientific thought, 118f
Anderson, J. E., 302
Anger (irritation, rage), see Rage
Angles in perception, 47, 82f
Animism, xiii, xviii, 4, 158, 173, 232,
234, 236
Anxiety, causes of, 264
Aphasia, intelligence in, 161, 276f
locus of lesion in, 284f
Appetite, 200
Area 17, motor effects from, 82, 86
Arieti, S., 262
Aring, C. D., 262
Art appreciation, 233
Arvanitaki, A., 65
A/S ratio, 124f
and attitude, 144
and instinct, 167
and intelligence, 126, 165
Asher, E. J., 302
Assembly, see "Cell-assembly"
Association, varying ease of, 77f, 127
Association areas, sensory control of,
123ff, 146, 165, 167
Association cortex, size of, see A/S
ratio
Assumption, concerning EEG, 8
of sensory dominance, 3-6
of structural trace, 12-16, 60, 62-
66, 228-231
of synaptic resistances, 60, 6266,
228-231
Asthma, 260
relation to emotion, neurosis, 267ff
Attention, as factor in learning, 151-
153
defined empirically, 4
defined theoretically, 87, 102
lack of unity in, 152
relation to expectancy, 87
Attitude, affecting monotonous work,
226f
problem of, 141-146, 207f
"Autonomous central process," 4ff
as attention, 4, 87, 102ff, 151-153
as attitude, 141-146, 207f, 227
as expectancy, xviii, 87, 102
as insight or hypothesis, 133f, 158-
165
as intention, 151-153
as interest, 6, 223
as set, 102ff, 107f, 144
as spontaneous neural action, 7
control of, as motivation, 180f
enduring selectivity of, 141-146
interaction with sensation, 146-
157
time period of, 73f, 142ff
SSI
S22 Index
Backtracking in maze learning, 135 Bridgman, C. S., 96
Badal, D. W., 262 Brightness, perception of, 4144
Bard, P., 189, 235f Broca, P., 283
Barelare, B., 190, 193 Brodmann (cortical areas of) , 67,
Bartley, S. H., 9, 208, 217 213, 219
Beach, F. A., 5f, 169, 207, 215f, 231 Brogden, W. J., 70
Bean, W. B., 262 Bronk, D. W., 8, 195
Bedwetting, as neurotic, 260 Broster, L. R., 262
Beebe-Center, J. G., 193, 204 Brouha, L., 262
Bellak, L., 263f Brown, W., 135
Bellows, R. T., 190 Bruetsch, W. L., 262
"Belongingness," 176f Burns, M., 187
Berger, H., xvi, 8 Buytendijk, F. J. J., 9
Binet, A., 295
Binet-type intelligence test, If, 290f Cajal, S. R., 230
Birch, H. G., 117, 165 Carlson, A. J., 192, 197f
Birth injury and IQ, 2, 275, 291f Carmichael, L., 26, 111
Bishop, G. H., xi, 9, 11, 217 Catatonia not an inhibition, 212
Black, perception of, 105f "Cell-assembly," alternating paths in,
Blodgett, H. G., 179 73
Blood content, effect on neural fir- as unstable equilibrium, 100
ing, 148, 196f bilateral development of, 96
in emotional disturbance, 148, disruption of, by blood content,
195-200, 253f 196f
Bloomfield, R., 262 by long activation, 225
Blum, R. A., 289 by sensory relationships, 148f,
Body image, disturbance of, 286 182, 189, 250-253
v. Bonin, G., 67, 70, 96, 213 duration of activity in, 74, 143
Border (contour) in perception, 82 first stage of growth, 69-73
Boredom and sleep, 217, 223 fixed vs. variable order of stimuli,
Boring, E. G., 15, 22, 25, 74, 78, 185 89
Bousfield, W. A., 192 "growth" of, 76f
Bowman, K. M., 260, 262 statistical factors in, 7478
Brain injury, anatomical data in, in the relative perception, 106
281-289 interference between assemblies,
effects of, increased by local syn- 88
chrony, 283 in emotion, 251f
estimates of intelligence following, motor facilitation from, 87, 99ff,
276ff 107
hemidecortication, 2, 285, 289, 291 properties of, after brain injury,
infant vs. adult, 289-294 293f
IQ following, If, 289-292 in metabolic disorder, 197
mode of retention of, 292ff role of synaptic knobs in, 72, 197
Brain size and learning, 124f safety margin in functioning of,
and attitudes, 144 197, 293
and intelligence, 126, 165 "Central excitatory mechanism," 5f,
Bridges, K., 255 216
Index 323
"Central motive state," 6
in hunger, 191, 200
Chandler, A. R., 233
Chimpanzee, intelligence of, 160f
mental illness in, 245-250
morphine addiction in, 203ff
nonspatial delayed response by, 95
one-trial learning by, 112, 114
perception in, 29, 33
rage and fear in, 241-245
reared in darkness, 32, 108, 112
recognition by observer, 27, 116
Chute, E., 208
Cingular region and consciousness,
219
Clark, G., 34f, 67, 86
Cobb, S., 262f
Coffin, T., 193
Cohen, M. E., 262
Cohn, R., 273
Coma and EEC, 122
Comic books, 260, 265
Completion, as associative, 103
in hemianopia, 51
Concept, "core" vs. "fringe" of, 131-
134
in post-operative intelligence, 2
not solely verbal, 118, 133
variability in, 99, 133
"Condenser" discharge of excitation,
216
Conditioned reflex, not elicited in
sleep, 221
theory of, 174-180
central formulation of, 180
in motivation, 176-180
Conduction from visual area, 51, 67
"Configuration theory," xvii, 39ff
evidence from hemianopia, 51
summary of critique of, 58f
vs. learning theory, 15, 59, 114f,
158-161
Congenitally blind, memory for color,
29, 32
memory for patterns, 28f, 32, 114
perception of vertical and horizon-
tal, 80f
Congenitally blind, perceptual gen-
eralization, 28f, 50
recognition of faces, 114
slow learning not due to motiva-
tion, 11 If
study of, 18ff
time to gain pattern vision, 32f
Connections, neural, as the control
of behavior, xvii
conditioned by time factors, lOf,
72, 196f
Hull's theory of, 176-180
in present theory, xix, 61, 72,
107f, 180
in prestriate area, 67-69
infant vs. adult, 107f
needs as promoting, 177f
Pavlov's theory of, 174
Conrad, H. S., 290
Consciousness, 144-146
cortical organization in, 122f
diffusity of firing in, 145, 219,
221
mode of drift to intrinsic organ-
ization, 22 If
emotion as kind of, 147, 236f
not function of cortex alone, 218
not localized, 219
relation to anatomical structures,
219
relation to EEC, sensation, 9f, 123,
145f
Constitution, and IQ, 295f, 300-303
and schizophrenia, 263f
vs. experience, basis of false di-
chotomy, 166, 198, 245, 259,
261-264
conceptually distinguished, 166-
169, 259
Contours in perception, 82
Control group, choice of, 278-281
normal vs. clinical, 279
use in rating intelligence, 276-281
Convergence, in association, 129f
in growth of assembly, 67ff, 72f,
75
Coon, G. P., 262
324 Index
Cortex, not a functional unit, 218
organization of, in infancy, 121-
123
in sleep, 122, 217-222
in the waking adult, 122f, 141-
146, 219ff, 234
intrinsic, 121-123
mode of change in, 221f
Cowles, J. T., 159
Crooks, L. E., 21
Crosby, E. C., 63
Culbertson, J. T., xi
Dallenbach, K. M., 187
Daniel, R. S., 168
Dashiell, J. F., 147
Davison, C., 222
Deafferentation, functional, in sleep,
218
vs. de-efferentation, 220
rationale of, 220f
Decerebrate rigidity, 212f
Dempsey, E. W., 218
Demuth, E. L., 222
Denker, P. G., 270f
Dennis, W., 241, 244f
Denny-Brown, D., 7
Deterioration, clinical identification
of, 275-281
of learning ("fading"), 224f
Determinism in psychology, xiii
See also Animism
Dewan, J. G., 262
Dickson, W. J., 226f
Diet, self-selection of, 190f, 193, 204f
Diffusity, of conduction from visual
cortex, 51, 67
of firing, essential to consciousness,
145, 221
essential to motor control, 220f
Dockeray, F. C., 240
Dog, intelligence of, 160
Doll, E. A., 291f
Drawings and perceptual content,
91f
Drew, G. C., 177, 189
Dubner, H. H., 8
Dunlap, K., 224
Duration of reverberatory activity,
73f, 143
Dusser de Barenne, J. G., xvi
Dysfunction, complicating brain in-
jury, 282-288
Ectors, L., 34f
Edes, B., 187
EEG, see Electroencephalogram
Egana, E., 262
Electroencephalogram, after electro-
shock, 273
after lobotomy, 272f
assumption concerning, 8
crude index, 122, 283
essential in study of brain injury,
283f
origin of, 8f
related to sensation, consciousness,
9, 122f, 211
Emotion, and animism, 236f
as disorganizer, 238-241
defined, 148, 238
James-Lange theory of, 237
more than sham rage, 235f
not conscious state, 147, 238
two classes of, 240f
vs. "emotional disturbance," 238-
241
Watson's theory of, 241
Emotional disturbance, as guiding
learning, 138f, I50f
causes of, 87, 179, 242-244, 250-
254, 267ff
classified, 251-254
due to conflict, 87, 148f, 251f
due to metabolic change, etc., 148,
195-197, 205, 253f, 267ff
due to sensory lack, 148, 179, 252f
due to strange environment, 87,
149, 251f
in asthma, 267ff
mechanisms of, 250-258
mental depression, 248f
nonspecific first stage in, 243f, 255f
Index 325
Emotional disturbance, possible vas-
cular origin of, 253, 268f
related to addiction, food habits,
205
related to learning, 138f, 147-151,
242, 245, 254-258
vigor of response in, 258
with interruption of sleep, 222
English, W. H., 262
Enuresis, 260
Equipotentiality, and tachistoscopic
experiments, 4650
in the cell-assembly, 74
meaning of, 39f
sensory, 39, 43
Equivalence of stimuli, 14, 39, 83f
Erlanger, J., 65
Excitability, affected by frustration,
conflict, fear, 230, 233
affected by memory, 227-229, 233
hypothesis of, 227-231
Excitation, discharge of, 216
problem of, 208-211
Expectancy, as a form of ideation,
xviii
factor in attitude, 141f
in emotional disturbance, 87, 149
in instinctive behavior, 169f
in voluntary action, 144, 154-156
relation to set, attention, 5, 87, 102
"Experimental neurosis," 249, 264
Eyeblink, learned, 107f
Eyemovement, and contours, 82
and straight lines, 48, 82f
in motor equivalence, 155f
in perception, 33-37, 45, 85-87
Factor analysis, xi
"Fading" during learning process,
224f
Fatigue, mental, problem of, 208f,
211
Fear, causes of, classified, 251-254
in children, 179, 241, 244
in chimpanzees, 241-245
mild, factor in pleasure, 230-233
of darkness, 179
Fear, of mutilated or dead body,
243f
of snakes, 242f
of solitude, 252
of the strange, 242ff
pleasant ("thrill"), 182, 233
Feindel, W. H., 183
Ferraro, A., 262
Fessard, A., 8
Fields, P. E., 29
Figural after-effects, 54-58
Figure-ground relation, 19-26, 83
fluctuation of, 25f, 99f
mechanism of, 99f
Finch, G., 114
Foord, E. N., 13
Forbes, A., 75
Forgays, D., 49
Fortuyn, J. D., 219
Fractionation, 76, 89, 97
Freeman, G. L., 230
Freeman, W., 270, 272f
French, E. G., 281
Freud, S., xii
Freudian theory of learning, 250
Frontal lobe, lobectomy vs. lobotomy,
272
removal, bilateral, in man, 287
in monkey, chimpanzee, 288f
stimulation vs. inhibition from, 212
supposed functions of, 272, 286-
289
Frustration, effect on learning, 138f
maintaining excitability, 230, 233
source of pleasure, 233
Fuchs, W., 51
Fulton, J. F., 213
Gantt, W. H., 264
Gardner, L. P., 224f
Gardner, W. J., 291
Carol, H. W., 67, 70, 96, 213
Garrison, M., 263
Gasser, H. S., 10
Gellerman, L. W., 29, 33
Gellhorn, E., 218
326 Index
Gelperin, J., 262
Generalization, perceptual, 12-16,
27, 88, 104
in congenitally blind, 28f, 32, 50
with relative values, 41-44
Generalization gradient, 59, 79
Gerard, R. W., 8
Gestalt theory, assumptions of, xvii,
3f, 13, 17f, 23, 41, 55, 114
difficulty for, with figural after-
effects, 56-58
with hemianopia, 51
with memory, 15
of intelligence, 158-165
sensory emphasis in, 3f, 22ff
source of current ideas, 59
summary of critique of, 58f
Gibbs, F. A., 8
Gibson, J. J., 5f, 8, 21, 111
Gillespie, W. H., 262
Goodenough, F. L., 192
Gordon, H., 302
Gottschaldt, K., 23
Greene, R., 262
Growth, in cortical organization, 76f,
122
in perception, 76ff
synaptic, 60, 62-66, 228-231
Gun-shy dog, 120
Guthrie, E. R., 180
"Habit-family hierarchy," 155
Halstead, W. C., 279f
Hanawalt, N. G., 13
Hardy, J. D., 185
Harlow, H. F., 114, 289
Harris, H. J., 262
Haslerud, G. M., 243
Hauptmann, A., 262
Head, H., 185, 285
Heath, R. G., 288
Hemianopia, 51f
after frontal injury, 34f
Hemianopic completions, 51
Heredity, and IQ, 295f, 300-303
and schizophrenia, 263f
Heredity, vs. environment, basis of
false dichotomy in behavior,
166, 198, 245, 259, 261-264
conceptually distinct factors,
166-169, 259
Herrick, C. J., xiv
Hilgard, E. R., 5f, 8, 61, 145, 175f,
179, 224
Hinsey, L. E., 271
Hoagland, H., 263, 273
Hobbs, G. E., 264
Hobhouse, L. T., 158f, 163
Hogan, H. P., 26, 111
Holmes, F. B., 252
Holt, L. E., 190, 193
Householder, A. S., xi
Hovland, C. I., 176, 224
Huber, G. C., 63
Hull, C. L., 5, 7, 15f, 38, 41, 56, 58,
79, 154f, 176ff, 180, 224
Humphrey, G., xix, 134
Hunger, 190-206
and emotional disturbance, 192f
arousal of, 199f, 202
as addiction, 200-206
chronic vs. acute effects, 192
defined, 190
for specific substances, 190f, 204f
innate factors in, 197f
learning effects in, 192-195
mechanism of, 197-206
metabolic factors in, 195-197,
201-206
satiation of, 190, 194, 202, 205
vs. appetite, 200
Hunger hormone, 197
Hunt, J. McV., 109, 120, 272
Hunt, W. A., 237, 281
Hunter, W. S., 163
Hymovitch, B., 298
Hypersynchrony, see Local syn-
chrony
Hypotheses in rats, 158, 161-163
difficulties of conception of, 162
spatial, 91-95
Index 827
Idealism, philosophic, xiv Intelligence, experience as factor in,
Ideation in animals, xviii, 158f 110, 117, 119f, 291f, 295-303
"Identity," 26-31 human vs. animal, 161
Image, a forbidden notion, xviii not separate process from learn-
assembly as simplest instance of, ing, 164f
60 of aphasics, 161, 276f
eyemovement a factor in, 34 of chimpanzees, 114, 160f
illusion of, 36 of dogs, 160
Incidental learning, in lower species, of identical twins, 301f
115f of primitive peoples, 119f
memory, 151-153, 181 of rats, 161, 296-299
Infant brain, effects of injury to, retention of, after brain injury,
289-294 276f, 279f, 286-294
intrinsic organization in, 121-123 in old age, 290ff
learning in, 111-116, 120-126, neural mechanisms of, 293f
228f, 296 tests of, affected by sophistication,
Inhibition, 208-215 age, 279f
as loss of facilitation, 213f two types of, 290
vs. excitation, from frontal lobe, two meanings of, 294-296
212 use of control group in rating,
from suppressor areas, 213f 275-281
"Inhibition of reinforcement," 224 Intention, 151-153
Innate, see Constitution Interactionism, philosophic, xiv, 236f
Insect, efficient first learning in, 113
Interest, factor in sleep, 6, 223
Insight (intelligence), 158-165
variability of, 141, 207f
after brain injury, 293f
Interference-pattern hypothesis, 15,
as function of experience, 110, 117,
47, 52
165
Intrinsic organization in cortex, 121f
definition of, 162f
behavioral manifestations of, 122,
difficulties of conception of, 162
221
164
mode of drift to diffuse organiza-
in early vs. later learning, 115
tion, 221f
mechanism of, 133-134, 164f, 293f
IQ, after brain injury, 289-292
not a separate process, 164f
and heredity, 295f, 300-303
in identical twins, 30 If
related to motivation, 181
limited constancy of, 296, 302
two criteria of, 160
of foreigner or Negro, invalid for
Instinctive behavior, 113, 165-170
intelligence A, 299f
Insulin shock, 269
valid for intelligence B, 295,
Intelligence, after birth injury, 29If
299ff
after hemidecortication, 2, 289,
of Negro vs. poor white, 120, 300
291
Isomorphism, 52ff
as factor in learning, 158-165
clinical estimates of, 276ff
Jackson, Hughlings, 283
development of, neurologically,
Jackson, T. A., 117
293
Jacobsen, C. F., 246
psychologically, 292 Jacobsen, M. M., 246
328
Index
James, W., 110, 237
James-Lange theory, 237
Jasper, H. H., 8f, 211, 217, 219f
Jefferson, G., 282
Jersild, A. T., 252
Johnson, R. E., 262
Jolliffe, N., 183, 262
Jones, H. E., 241, 243, 290
Jones, M. C., 241, 243, 255
Jordan, A. M., 302
Kallman, F., 263
Kappers, C. U. A., 63
Karnosh, L. J., 291
Kaufman, I. C., 273
Kennard, M. A., 34f
Kennedy, F., 285
Kinder, E. F., 119, 168
Kinsey, A. C., 216, 231
Klebanoff, S. G., 284
Kleitman, N., 6, 217f, 220f, 224
Klineberg, O., 265
Koffka, K., 10, 12f, 23, 51, 111
Kohler, W., xiv, 10, 12, 15, 23ff,
38ff, 54ff, 80, 117, 158ff, 252
Krechevsky, I., 92, 111, 159, 161
Landahl, H. D., xi
Landis, C., 237, 240, 263, 271
Lange, C., 237
Language as set of conditioned re-
sponses, 117
Lansdell, H., 298
Lashley, K. S., 10, 12ff, 19, 27, 29ff,
38ff, 67, 80f, 83f, 86, 90, 93,
99, 115, 117, 145, 154, 159,
166, 184f, 235, 286
Law of effect, 159, 176ff
Learning, adult, and insight, 110,
115, 158-165
as function of early learning,
109-111, 119f, 228f
as reinforcing existing connec-
tions, 127, 130
mechanism of, 126-139
phase sequence a pre-condition
of, 128
Learning, adult, role of meaning in,
117, 128, 130-134
constitutional factor ubiquitous,
198
contiguity not sufficient for, 127
early, channeling effect of, 109,
119
generalized transfer from, 109-
111
possible permanence of, 109,
228-231, 296, 299
effect of, on emotion, 148f, 244f,
250f, 254-258
on hunger, 136f, 192-195, 197-
206
on intelligence (insight), 110,
117, 119f, 291f, 295-303
on sleep, 220
effect of emotion on, 138f, 150,
198
effect of sleep on, 221
"free" of specific cues, 129
incidental or unmotivated, 151-
153, 181
in lower species, 115f
inconstancy of ("fading"), 140,
224f
mistaken for instinct, 113
motor, 98, lOlff, 107f, 134-139,
153-157
one-trial, in chimpanzee, 114
primary stage of, 120-126
as establishing sensory control,
125
slowness in, and motivation,
lllf
relation to A/S ratio, brain size,
124-126
relation to phylogenesis, 112-116
relation to set, attention, 108f
theory of, continuity, noncontinu-
ity, 114f
not based on adult data only,
110
vs. heredity, see Heredity vs. en-
vironment
Index 329
"Learning theory," as attack on ani-
mism, 4
nai've, 4, 159
sophisticated, 59, 159, 177
vs. configurationism, 15, 59, 114f,
158-161
Leeper, R. W., 22, 111, 238ff, 254
Left and right in psychology, 59
Left-right distinction, paradigm of
adult learning, 117f
Lehmann, J. E., 8, 195
Letter-square, illusion of image, 36
Leucotomy, see Lobotomy
Levine, J., 45
Lewin, K., 226
Libet, B., 8
Liddell, H. S., 189, 264
Lloyd, D. P. C., 213
Lobotomy, 271-274
hypothesis of effect of, 272f
not suited to localization, 288
Local synchrony, after lobotomy,
electroshock, 272f
assumption concerning EEC, 8
complicating brain injury, 283f
reinforced by learning, 12 If, 220
in infancy, sleep, 122
in sleep, 217ff
negating function, rationale of,
213f, 220f
Localization of function in human
brain, 282-289
limited knowledge of, 284
tumor cases not suited to, 282, 288
Lorente de No, R., xvi, 10, 61, 63ff,
72f, 75, 142, 210, 213
Lorenz, K., 120, 166
Loucks, R. B., 176, 221
Luckhardt, A. B., 190, 197, 201
McBride, A. F., 243
McBride, K. E., 277, 281, 285, 289f
McCulloch, T. L., 242f
McCulloch, W. S., xi, 67, 70, 75, 96,
213f, 218
McGeoch, J. A., 109f, 128, 131, 177
MacLeod, J. F., 193
Machover, S., 119
Malamud, W., 273
Mammillothalamic tract, 219
Marchand, W. E., 262
Marquis, D. G., 5f, 8, 61, 145, 175f,
179, 224
Marshall, W. H., xix, 43, 48, 53, 58,
82
Martin, C. E., 216, 231
Mason, H. L., 262
Massed vs. distributed practice, 224
Masserman, J. H., 235, 264
Matthews, B. H. C., 9, 217
Matthews, R. S., 262
Maze learning, 134-139
Meaning, 128, 130f, 133, 165
"Means-end-readiness," 145f, 153
157
Melcher, R. T., 291f
Memory trace, assumptions concern-
ing, 12-16, 62-66, 228-231
dual mechanism of, 61f
related to generalization, 13ff
reverberatory, 12f, 61f
time factor in production of, 62,
230f
Mensh, I. N., 281
Mental hygiene, 261, 264-266
Mental illness, accompanying other
conditions, 262
causes of, 261-271
ignorance about, 249f, 260f,
265f
in chimpanzees, 245250
nature of, 259-264
no separate psychogenic type of,
259, 26 Iff, 271
not specifically sexual, 249
recovery without special treatment,
267, 269-271
treatment by general practitioner,
270f
Metabolic factors in emotion, 195-
197, 202, 205, 253f
Migraine scotoma, 51
Miller, G. A., 188
Miller, H. R., 236
330 Index
Mind as neural activity, xiv, 79
See also Animism, Consciousness
Miner, J. B., 114
Mishkin, M., 49
Mixter, W. J., 272, 282, 288
Mnemonic trace, see Memory trace
v. Monakow, C., 286
Morgan, C. T., xviii, 6, 8, 137, 172,
183, 190f, 195, 197, 200f, 218
Morison, R. S., 218
Morton, N. W., 277, 301
Motivation, as bodily stimulation,
172f
as theoretical conception, 163,
171-174, 181
channeled by pain, 182
defined as aspect of phase se-
quence, 181
in CR theory, 176-180
loss of, due to monotony, 224-227
following error, 224
neural vs. animistic, 234
related to insight, 181
Motor control and diffuse firing, 220f
Motor effects from visual area, 82, 86
Motor equivalence, 153-157
and perception, 155f
and volition, 145f
in instinctive behavior, 166-169
Motor facilitation from assembly,
86f, 98, 102
Motor learning, 98, lOlff, 107-109
equivalence in, 153-157
in the maze, 134-139
Motor links in perception, 35-37, 98f
Mowrer, O. H., 109, 260
Mowrer, W. M., 260
Muenzinger, K. F., 177, 189
Munn, N. L., 240
Murphy, J. P., 218
Music and pleasure, 233
Nafe, J. P., 183f
Nauta, W. J. H., 217, 219, 222
"Need" in learning theory, 176-180
Neet, C. C., 29
Neff, W. S., 301f
Neologism, difficulty of recall, 118
Neuralgia, 185f
Neurobiotaxis, 63, 65
Neurosis, see Mental illness
"Neurosis, experimental," 249, 264
Nichols, L, 272
Nissen, H. W., 95, 119f, 159, 224f,
248
Nonsense syllables, learning, 128
"Nonsensory" figure, 21-26
Old age, intelligence in, 2, 290, 292f
memory in, 228f
sleep in, 220
Operationalism, 162
Oppenheimer, M. J., 236
Opposites in children's learning, 117f
"Optional transmission," 10, 72
"Oscillation principle," 7
Owen, T., 262
Pain, 181-190
as guiding learning, 150
as strengthening response, 177,
189f
classical theories of, 183
definition of, 182
in glans penis, 186
increased by hypofunction, 184ff,
188f
inhibition of, 186, 212
pleasant, 182
relation to C fibers, 183
reason for, 188f
relation to emotion, 148, 182
relief by lobotomy, hypothesis of,
273
Parallelism, philosophic, xiv
Paterson, A. S., 262
Pattern of stimulation vs. locus, 12-
15, 39-41, 44-50
Patton, R. A., 204
Pavlov, I. P., xii, 5, 38, 41, 80, 174ff,
180, 189, 212, 224, 264
Pechstein, L. A., 115
Penfield, W., 219, 272, 282, 287ff
Pennington, L. A., 92
Index 331
Perception, as action in specific cells,
15, 17, 42, 44, 48
as function of experience, 22-24,
26f, 31-35, 111, 119f
duration of content of, 25, 74
equipotentiality in, 39-41, 44-50,
74
growth in, 77
innate organization in, 23f, 36f,
42f
integration of, prior to association,
77
involving conceptual activity, 9 If
of black, 105f
of brightness, 4144
by rats reared in darkness, 42
hypothesis of, 43f
of nearby vs. remote objects, 91-
95
of pattern, anatomical features in,
52-54, 57f
and isomorphism, 52-54, 57f
as summative process, 18, 32,
34, 36f
changes with prolonged stare,
225
contours in, 82
distortions in cortex, 5254, 57f
eyemovement in, 34, 36f, 99
102
figural after-effects, 54-58
"identity," 26-31
identity a function of experi-
ence, 3 Iff
illusory simplicity of, 36
integration of parts, 95-98, 103
locus vs. pattern, 12-15, 39-41,
44-50
"nonsensory figure," 2126
part vs. whole in, 24f, 99f
"primitive unity," 19-21
recognition of words, 49f
right vs. left visual field, 50
with varied angle of regard, 9If
with varied distance, 90f
of pitch, size, 4 If, 44, 106
phylogenetic differences in, 29-31
Perception, recurrent or cyclical, 25,
99f
relative, 41-44, 105f
Perceptual elements, 83f
in association of ideas, 130
Perceptual generalization, see Gen-
eralization
Phantom limb, pain in, 185
"Phase cycle," duration of, 143
in perception, 99f, 130f
"Phase sequence," 98-100
conditions of persistence of, 142f,
222, 227, 229f, 232
conflicts of, in emotion, 148151,
251-253
dominance of partly organized,
229, 232
flexibility of, 87, 149
in motivation, 181, 191
in simple perceptions, 99f
relation to waking center, 218f
self-destruction or perpetuation,
222
sensory disturbance of, 148-151,
251-253
in pain, 150, 181-190
sensory guidance of, 153-157, 181
sensory lack, effect of, 148, 252f
short-circuiting in, 228, 293
Phelps, W. M., 29If
Phylogenetic differences, in A/S ratio,
brain size, 124-126, 167
in attitude, 141
in insight, 125f, 160f, 164f
in learning, 112116
in perception, 29-31
Physiologizing in psychology, xiv
Physiology, relation of, to psychol-
ogy, xv
Pigeon, interocular transfer in, 45
Pile, G. C. S., 262
Pilgrim, F. J., 204
Pillsbury, W. B., 25, 74
Pincus, G., 273
Pitts, W., xi
Pleasant fear ("thrill"), 182, 233
Pleasant pain, 182
332 Index
Pleasure, 227-234
in art, music, sports, 233
relation to memory, 227-231
theory of, substitute for animism,
234
Polyak, S. L., 52f
Pomeroy, W. B., 216, 231
Pool, J. L., 288
Position habit, 92, 161, 168
Positivism, xiv, 162
Postman, L., 177
Pratt, C. C., xiv
Prentice, W. C. H., 162
"Primitive unity," 19-21
Probability of association, 127
in perceptual integration, 77
Prosser, C. L., 8
Psychoanalysis, xii, 247, 261, 269-271
Psychosis, see Mental illness
Psychosomatics, 266-269
Psychotherapy, 261, 266-271
Purpose, 144-146
Rage (anger, irritation), causes of,
148, 252, 257f
in children, 192, 241
in chimpanzees, 241245
not dependent on physical re-
straint, 242
not localized in hypothalamus,
235f
related to hunger, 192
sham, 235f
Ramsdell, D., 252f
Rashevsky, N., xi
Rat, hunger in, 193-195
intelligence of, 137, 161, 296-299
"intelligence test" for, 135, 137,
297f
perception in, different from man's,
30f, 95
of direction, place, 92ff
reared in darkness, brightness dis-
crimination in, 42
figure-ground perception in, 19f
slow pattern learning in, 81, 113
Rat, visual concepts after blinding,
129, 297f
"Reaction to deficit," 166
"Reactive inhibition," 224
Reading speed, 49
Recognition of faces, 27, 114, 116,
127f
Recruitment in growth of assembly,
76, 97
in effects of lobotomy, 273
in sleep, 221
strengthening intrinsic organiza-
tion, 121f
Relative perceptions, 41-44, 105f
Reverberation, brief in simple cir-
cuits, 73, 142f
Richter, C. P., 190, 193
Riddoch, G., 185
Riesen, A. H., 18, 32, 78, 95, 108,
112, 117, 204, 244, 255
Roe, A., 281, 289
Roethlisberger, F. J., 226f
Romano, J., 262
Rowe, S. N., 2, 289, 291
Rubin, E., 19
Rupp, C., 260
Rylander, G., 279flF, 290
"Salted-nut phenomenon," 199f, 202
Santa Glaus vs. mental hygiene, 260
Saphir, W., 192, 205, 262
Satiation, of hunger, 190, 194f, 202,
205
without sufficient food, 194, 205
of sex motivation, as neural dis-
charge, 216
"Satiation" of visual cortex, 54-58
Scarff, J., 282
Schema of perception, 69-74, 84-
91, 95-100
"Schema with correction," 46, 111
Schneirla, T. C., 168
Scientific method, xiv
Scientific thought, xiv, 118f
Scott, C. C., 190, 201
Scott, W. W., 190, 201
Self-selection of diet, 190f, 193, 204f
Index 333
Senden, M. v., 18ff, 28f, 31f, 50, 78,
80f, 11 If, 117
Senescence, intelligence in, 2, 290,
292f
memory in, 228f
sleep in, 220
Sensori-motor connections, criterion
of, 108
linear theory of, xvii, 4, lOf, 40f
modification of idea of, 180
sophisticated theory of, 59, 159,
177
vs. relative perceptions, 15f, 4 If
Sensori-sensory association, 70
Sensory activity, cumulative effect of,
7, 123f, 126
in motivation, 172f, 177-179, 181
in sex behavior, 215f
in sleep, 217f, 220, 223
in voluntary acts, consciousness,
9f, 123, 145f, 153-157
lack of, determining response, 166,
179, 252f
Sensory-central interaction, modes of,
87, 146-157, 232f, 250-253
Sensory control, of cortex, 9f, 122
127, 145f, 232f
and A/S ratio, 125
as steering phase sequence, 87,
138f, 150f, 153-157, 181
established in primary learning,
125
of motor activity, 107, 154-157,
167, 220f
variations in directness of, 107
Sensory dynamics in Gestalt theory,
3f, 19-25, 51, 54-58
Set, 4f, 102ff
in voluntary act, 107f, 144
Sex behavior, as problem of motiva-
tion, 207
effect of novelty on, 231
time course in, 215f
Sherrington, C. S., 79, 182, 189, 212
Shock treatment, electrical, 271-274
insulin, 269
Short circuiting in phase sequence,
228
aid to retention of intelligence,
293
Shyness, 244
Similarity, 26f, 104
association by, 132
Sinclair, D. C., 183
Size, perception of, 4 If, 106
Skinner, B. F., xiv, 146
Sleep, 216-224
as causing relaxation, 221
as extreme loss of motivation, 223
as intrinsic organization, 122f
effect of, on learning, 221
effect of learning on, 220
induced by cortical lesions, 222
interest a factor in, 217
mode of drift to waking state, 22 If
not a lack of organization, 219f
not a lowered cortical activity, 211
not only sensorily determined, 218
relation of, to sensation, 9f, 122f,
146, 218, 220, 223
similarity to infancy and patho-
logical conditions, 122
Sleep center, 217, 219
Smith, B. F., 262
Smith, D. E., 92
Smith, K. U., 29, 96, 168
Smith, W. K., 219
Sophisticated learning theory, 59,
159, 177
Spaet, T., 289
Spatial hypothesis, in rats, 91-95,
161, 168
Spearman, C., xi
Speech area, localization of, 284f
Spence, K. W., 41, 159, 163
Sperry, R. W., 63, 157
Spiegel, E. A., 236
Spies, T. D., 262
Spontaneous neural activity, 8f
as problem in learning, 121-126
as regulated by sensation, 9
relation to motivation, 171f
Spragg, S. D. S., 203ff, 248
334 Index
Stein, M. H., 183
Stimulation, physiological, as dis-
rupting normal function, 213f,
219
Stimulus, in motivation theory, 172
pattern of, vs. locus, 12-15, 39-
41, 44-50
Stimulus-response theory, see
"Learning theory"
Stoddard, G. D., 295, 302
Stone, L. J., 187
Stookey, B., 282
Straight line, and eyemovement, 48,
82f
learning to perceive, 81
physiological properties of, 82f
Strauss, A., 291
Superordinate perceptions, 95-98
Suppressor areas, 213f
Swank, R. L., 262
Synaptic knobs, and probability of
transmission, 66, 76, 197, 293
as affecting neural control, 72
assumptions concerning, 60-66,
228-231
formation of, 65f, 228-231
in one-trial learning, 130
offsetting brain damage, 74, 293f
offsetting metabolic disorder, 197
permanence of, 228-231
vs. time factor, lOf, 72, 196f
Synchrony of firing, see Local syn-
chrony
Tachistoscopic vision, and equipo-
tentiality, 46-50
essentials of, 47
number of habits necessary for, 48
Talbot, S. A., xix, 43, 48, 53, 58, 82
Teitelbaum, M., 282
Thalamus and consciousness, 219
Thirst, see Hunger
Thorndike, E. L., 26ff, 159, 176f, 180
Thought, as key problem, xvixviii
restructuring of, 134
See also Consciousness; Phase se-
quence
Thurstone, L. L., xi
Tillotson, K. J., 272, 282, 288
Tinbergen, N., 109, 113, 120, 166
Tinklepaugh, O. L., 159, 252
Titchener, E. B., 41, 187
Tolman, E. C., 59, 146, 154, 159
Train of thought as phase sequence,
xix, 100, 130-134
Transfer of training, 14
infant-adult, 109f
interocular, in pigeons, 45
Transmission, cortical, and A/S ratio,
124f
as problem of thought, xvi
Tryon, R. C., 117, 129
"Vacuum activity," 166
Valentine, C. W., 233, 241
Variability, in motor response, 153-
157
of learning, 140, 224f
Visual concepts in the blind, 129,
293
Vitalism, xiii
See also Animism
Volition, 144ff
motor control in, 153-157
Wade, M., 27
"Wakefulness of choice," 218
Waking center, 217, 219
Waking organization, 122, 141-146,
219ff, 234
mode of drift to intrinsic organ-
ization, 221f
Walker, A. E., 82, 86
Wallach, H., 23, 54ff
Walter, A. A., 26, 111
Watson, J. B., 40f, 241
Watts, J. W., 270, 272f
Weaver, T. A., 82, 86
Wechsler, D., 290
Weddel, G., 183
Weisenburg, T., 277, 281, 285, 289f
Weiss, P., 8, 63, 124, 184
Wellman, B. L., 295, 302
Werner, H., 291
Index 335
Wholes, integration of, 95-98, 103
Wies, D., 272, 282, 288
Williams, K., 93, 135, 137, 168, 297f
Williams, R. D., 262
Willson, E., 263f
Wilson, G., 260
Wolf, A., 285
Wolf, E., 168
Wolf, G. A., 266f
Wolff, H. G., 185, 266f
Woodrow, H., 110
Woodworth, R. S., 36, 46, 111, 131f,
135, 180, 233, 240
Wortis, H., 183
Wulf, F., 13
Yerkes, R. M., 90, 158f
Yerkes Laboratories of Primate Bi-
ology, 114, 242f, 266
Yoshioka, J. G., 246
Young, R T., 190, 193, 204, 240
Zangwill, O. L., 26, 111
Zener, K., 175
Zerrahn-Wolf, G., 168
Zollinger, R., 285