

\..S!; 


BRAIN MECHANISMS AND LEARNING 


BRAIN MECHANISMS 

AND 

LEARNING 


A SYMPOSIUM 

organized by 

THE COUNCIL FOR INTERNATIONAL 

ORGANIZATIONS OF MEDICAL SCIENCES 

Established under the joint auspices of 
UNESCO and WHO 


Coiisiihiug Editors 
A. Fessard R. W. Gerard J. Konorski 

Editor for tlic Council 
J. F. Delafresnaye 

C.LO.M.S. 

Paris, France 


BLACKWELL 

SCIENTIFIC PUBLICATIONS 

OXFORD 


(jC) Blackwell Scientific Publications Ltd., ig6i 

Tins I'ook i.< copyright. It niny not he reproduced by any means in whole or in part without permission. 
Application with regard to copyright should be addressed to the publishers. 

Published simultaneously in the United States of America by Charles C Thomas, Publisher, ^01-327 
East Lawrence Avenue, Springfield, Illinois. 

Published siimiltaneously in Canada by the Ryersou Press, Queen Street West, Toronto 2. 


FIRST PUBLISHED igfil 


PRINTED IN GREAT BRITAIN IN THE CITY OF OXFORD 

AT THE ALDEN PRESS 
AND BOUND BY THE KEMP HALL BINDERY, OXFORD 


CONTENTS 

Page 

List of those participating in the symposium vii 

Foreword by J. F. Dclafresnayc, Executive Secretary, CIOMS ix 

Prefacio by A. Establicr, Director, LASCO xi 

Introduction by A. Fcssard, ori^mnzer of the symposium xiii 

Darwin and concepts of brain function, by H. W. Magoun i 

The fixation of experience, by R. W. Gerard 21 
Distinctive features of learning in the higher animal, by D. O. 

Hcbb 37 
The interactions of unlearned behaviour patterns and 

learning in mammals, by I. Eibl-Eibcstcldt S3 
Some characteristics of the early learning period in birds, /))' 

W. H. Thorpe 75 
Some aspects of the elaboration of conditioned connections 

AND FORMATION OF THEIR PROPERTIES, by E. A. Asratyaii 95 

The PHYSIOLOGICAL APPROACH TO THE PROBLEM OF RECENT MEMORY, 

by ]. Konorski 1 15 

Conditioned reflexes established by coupling electrical 

EXCITATIONS OF TWO CORTICAL AREAS, by R. W. Doty and C. 

Giurgca 133 

Interference and learning in palaeocortical systems, by 

J. Olds and M. E. Olds 153 

A NEW CONCEPTION OF THE PHYSIOLOGICAL ARCHITECTURE OF CONDI- 
TIONED REFLEX, by P. K. Anokhin i8y 

Changing concepts of the learning mechanism, /))' R. Galanibos 331 

The significance of the earliest manifestations of conditioning 

in the mechanism of learning, by E. Grastyan 243 

Behavioural and EEG effects of tones 'reinforced' by cessation 
OF PAINFUL stimuli, /)/ J. P. Segundo, C. Galcano, J. A. 
Sommer-Smith and J. A. Roig 265 

Neurohumoral factors in the control of animal behaviour, 

by K. Lissak and E. Endroczi 293 

Considerations on the histological bases of neurophysiology, 

by C. Establc 309 

The effects of use and disuse on synaptic function, by]. C. Ecclcs 335 

La facilitation de post-activation comme facteur de plasticite 
dans l'etablissement des liaisons temporaires, par A. 
Fessard et Th. Szabo y^JolCik 4 r^ i 9 353 


^ 


U8RA«Y ,^ 


vi contents 

Lasting changes in synaptic organization produced by con- 
tinuous NEURONAL BOMBARDMENT, by F. Morrcll 375 

Functional role of subcortical structures in habituation 
AND CONDITIONING, by R. Hernandcz-Peoii and H. Brust- 
Carmona 393 

Functions of bulbo-pontine reticular formation and plastic 

PHENOMENA IN THE CENTRAL NERVOUS SYSTEM, by M. PalcStilli 

and W. Lifschitz 413 

Changes in cortical evoked potentials by previous reticular 

STIMULATION, /)/ M. R. Covian, C. Timo-Iaria and R. F. 

Marscillan 43:5 

Recherches sur les mecanismes neurophysiologiques du som- 

MEiL ET DE l'apprentissage negatif, par M. Jouvct 445 

Corpus callosum and visual gnosis, by R. E. Myers 481 

Anatomical and electrographical analysis of temporal 

neocortex in relation to visual discrimination learning 

IN monkeys, by K. L. Chow 507 

Observations sur le conditionnement instrumental alimen- 

taire CHEZ LE CHAT, par P. Busci" et A. Rougcnl 527 

Non-sensory effects of frontal lesions on discrimination 

learning and performance, /)}' rt. E. Rosvold and M. Mishkin 555 
Studies of hippocampal electrical activity during approach 

LEARNING, by W. R. Adcy 577 

Role of the cerebral cortex in the learning of an instru- 
mental conditional response, by T. Pinto Hamuy 589 
Significance of the photic stimulus on the evoked responses 

in man, by E. Garcia-Austt, J. Bogacz and A. Vanzulli 603 

Conditionnement de decharges hypersynchrones epileptiques 

CHEZ l'homme et l'animal, par R. Naquct 625 

General discussion 641 

Bibliography 665 

Index • 699 


LIST OF PARTICIPANTS 
W. Ross Adey 

University of California, Los Angeles (U.S.A.) 

P. K. Anokhin 

Academy of Medical Sciences, Moscow (U.S.S.R.) 

E. ASRATYAN 

Academy of Sciences, Moscow (U.S.S.R.) 

P. BUSER 

Universite de Paris (France) 

K. L. Chow 

University of Chicago (U.S.A.) 

M. COVIAN 

Faculdadc de Medicina, Sao Paulo (Brazil) 

R. W. Doty 

University of Michigan, Ann Arbor (U.S.A.) 

J. C. ECCLES 
The Australian National University, Canberra (Australia) 

I. ElBL-ElBESFELDT 
Max-Planck-Institut tucr Verhaltensphysiologie, Seewiesen (Germany) 

C. ESTABLE 
Instituto de Investigacion de Ciencias Biologicas, Montevideo (Uruguay) 

A. Fessard 

College de France, Paris (France) 

R. Galambos 

Walter Reed Army Medical Center, Washmgton (U.S.A.) 

E. Garci'a-Austt 

Instkuto de Neurologia, Montevideo (Uruguay) 

R. W. Gerard 

University of Michigan, Ann Arbor (U.S.A.) 

E. Grastyan 

University of Pecs (Hungary) 

D. O. Hebb 

McGill University, Montreal (Canada) 

R. Hernandez-Peon 

Centro Medico del Distrito Federal, Mexico 

M. JOUVET 

Faculte de Medecine de Lyon (France) 


VIU LIST OF PARTICIPANTS 

J. KONORSKI 

Institute of Experimental Biology, Warsaw (Poland) 

K. LiSSAK 
University of Pecs (Hungary) 

J. Luco 

Universidad Catolica, Santiago (Chile) 

H. W. Magoun 

University of California, Los Angeles (U.S.A.) 

F. MORRELL 

University of Minnesota, Minneapolis (U.S.A.) 

R. E. Myers 

Walter Reed Army Institute of Research, Washmgton (U.S.A.) 

R. Naquet 

Facultc de Mcdccine de Marseille (France) 

J. Olds 

The University of Michigan, Ann Arbor (U.S.A.) 

M. Palestini 

Facultad de Mcdicina, Santigo (Chile) 

T. Pinto Hamuy 

Universidad de Chile, Santiago (Chile) 

H. E. RosvoLD 

National Institute of Mental Health, Bethesda (U.S.A.) 

J. P. Segundo 

Instituto dc Investigacion de Cicncias Biologicas, Montevideo (Uruguay) 

W. H. Thorpe 

University of Cambridge (U.K.) 


FOREWORD 

Six years after the Laurentian symposium on 'Brain Mechanisms and 
Consciousness', the Council for International Organizations of Medical 
Sciences (C.I.O.M.S.) convened a meeting to explore the neurophysio- 
logical basis of learning. The meeting was held in Montevideo, from 
August 2nd to 8th, 1959, and was jointly sponsored by and organized with 
the Science Co-operation Office of Unesco for Latin America. 

Dr A. Fessard was responsible for the scientific planning of the meeting 
and shared with Dr Ralph Gerard the responsibility of presiding over it. 
May they both find here an expression of the Council's gratitude for the 
magnificent way in which they discharged their duties. 

Looking back over the years, it is impossible not to draw a parallel 
between the meeting held in Canada and the one held in Uruguay. 

In each case, the meeting was co-ordinated with the International 
Physiological Congress. Both were 'firsts': the first CIOMS symposium 
in North America and the tirst in Latin America. And then, of course, a 
number of scientists took part in both meetings. But here the parallel 
must end. 

In Montevideo, we were able to cast our net wider and we were pleased 
to welcome a number of distinguished investigators from Australia, 
Brazil, Chile, Hungary, Mexico, Poland, Uruguay, and the U.S.S.R., 
who had not attended the meeting in Canada. An organizing committee 
composed of Drs R. Arana, D. Bennati, C. Estable, and of Drs E. Garcia- 
Austt and J. P. Segundo, who acted as joint secretaries, took charge ot all 
technical arrangements and provided much appreciated entertainment for 
all who took part in the meeting. Credit must go to this committee for 
having created the right conditions for the development of an atmosphere 
of understanding and friendliness. May this committee find here an 
expression of the Council's gratitude. 

It is also a pleasant task to put on record the Council's indebtedness to 
the various agencies which supported directly or indirectly the meeting, 
and in particular to the European Office, A.R.D.C.-^ whose grant was 
used to cover the travelling expenses of a number of participants. 

I am also pleased to thank Miss Henderson and Dr Christie, who 
transcribed the discussion, and the secretarial staff, who worked very hard 
to get everything ready in time. 

^ Contract No. AF61 (052)-207. 

ix 


X FOREWORD 

This monograph is the record of what happened in Montevideo and 
has been put together with the help of the consuking editors. May I thank 
all three but especially Dr Fessard who, being on the spot, has advised me 
on many points. 

It is the hope of all who contributed to this book that it will be of value 
to all those who study the fundamental processes of learning, wherever 
they may be. 

J. F. Delafresnaye 


PREFACIO 

El Centro de Cooperacion Cicntifica de la UNESCO para America 
Latina ha contribuido a organizar la reunion que sobre 'Brain Mecha- 
nisms and Learning' decidio celebrar en Montevideo el Consejo Intcr- 
nacional de Organizaciones dc Cicncias Medicas (C.I.O.M.S.). 

Es ya de tradicion la organizacion de Symposia por los Centros de 
Cooperacion Cientifica de la UNESCO y comienza a ser habitual el 
que algunos de ellos se monten con la colaboracion de la C.I.O.M.S., 
organizacion que tan activamente se ocupa de los problemas basicos de la 
Medicina. 

La iniciativa de esta reunion vino dc un latinoamericano, el Dr. Raul 
Hcrnandcz-Peon, y era logico pensar que el Centro de Montevideo 
aportase su plena colaboracion a esta manifestacion. 

Hace varios anos, igualmente en Montevideo, organize el Centro de 
Cooperacion Cientifica un Symposium scerca dc otro tenia de fisiologia, 
nos referimos al titulado Symposium sobre 'Problemas fundamentales de 
cstructura y fisiologia celular' que reunio a invcstigadorcs latino- 
amcricanos, norteamcricanos y europcos. 

Esta insistencia en organizar este tipo dc reuniones sobre tcmas de 
Biologia en Montevideo se justifica por la prescncia en el Uruguay del 
Institute de hivestigaciones de Ciencias Biologicas dirigidopor elconocido 
Profesor Clcmente Establc. 

El Symposium sobre 'Brain Mechanisms and Learning' es un ejemplo de 
la cooperacion entre organismos internacionales y comites nacionales. La 
C.LO.M.S. orientada cientificamente en esta reunion por el Dr. A. 
Fessard, con la incansable tenacidad de su Secretario, el Dr. J. F. Dela- 
fresnaye, encontro el apoyo de un Comite local uruguayo que le brindo su 
plena colaboracion. El Centro de Cooperacion Cientifica de la UNESCO 
para America Latina aporto toda la ayuda que podia prestar. Ha sido una 
reunion de un alto nivel cientifico y que ha demostrado plenamente hasta 
que punto puede llegarse en la colaboracion cientifica cuando se consigue 
una pcrfecta coordinacion de esfuerzos. 

A. Establier 
Director del Centro de Cooperacion 

Cientifica para America Latina de la 

UNESCO 
xi 


'lu' LIBRARY ,3= 
INTROD UCTION \ J'f--. ^/ass-^^ \ 

Lcs mccanismes du ccrvcau oftrcnt uiic niaticrc incpuisablc pour des 
themes de symposiums. Apres le Brain Mccluviisiiis and Coiiscioiisiiess de 
Ste Marguerite (Canada), en aoiit 1953, bien d'autres reunions scienti- 
hqucs de meme espece eurent lieu, ou I'interet se porta tantot sur lcs 
proprietcs de la formation rcticulee, tantot sur lcs rapports des reflexes 
conditionncs avcc relectroenccphalogramme, ou encore sur les bases 
nervcuses du comportement, pour ne citer que trois exemples. Le sym- 
posium dont il s'agit ici iut consacre aux problcmes de I'Apprentissage, 
ou Learning, I'accent etant mis sur lcs mccanismes ncurophysiologiques 
dont CCS phenomenes dependent. 

L'initiativc du choix de cc sujet, ainsi que la conception generale du 
symposium, reviennent au Dr Raul Hernandez-Peon dont on connait les 
importantcs contributions cxperimcntales dans cc domaine. Son projet 
re^ut un excellent accucil. Il cut aussi la bonne fortune de pouvoir etre 
parrainc par le CIOMS, cc qui signifiait que son execution scrait remise 
aux mains du Dr Dclafrcsnayc, dont lcs qualites d'organisateur eurent 
ainsi I'occasion de se manitester une fois de plus. Il vicnt dc nous rappelcr 
tout cc que cette reunion a du a rextreme obligeancc et a I'activitc dili- 
gcnte de nos botes uruguayiens et aux institutions qui aidcrent matcricllc- 
ment le projet a se realiscr. Il n'appartient pas aux organisateurs de porter 
un jugement sur la valcur du resultat: comme il est de regie, Icurs buts nc 
turent que partiellemcnt attcints et ils ne sc croient pas a I'abri de toutc 
critique. 

Tout d'abord, un theme connnc cclui-ci pourrait ctre traite de points dc 
vue bien diflerents. On nous reprochcra probablement ccrtaines lacuncs, 
(HI d'avc^ir donne trop peu dc place a certains aspects. Il etait impossible en 
iait dc couvrir entierement un champ aussi vaste que cclui qui nous etait 
impose. On notera cependant que nous nous sommes efforccs de fairc 
appcl a des disciplines differcntcs. Si lcs neurophysiologistes formcrent, 
comme il se devait, notrc majorite, les points dc vue du psychologuc et 
de I'ethologiste purent s'exprimcr, ct du cote des donnecs de base, une 
place fut rcservee a I'anatomie et aux aspects physicochimiqucs. Les 
neurophysiologistes eux-mcmcs representaient des tendances diverses, 
depuis les electrophysiologistcs du cerveau jusqu'aux specialistes du 
rcflexc conditionne classique. 

Le choix des participants a etc — est-il besoin de le dire ? — le probleme 

xiii 


XIV INTRODUCTION 

le plus cmbarrassant pour les organisateurs. Il y a partout dans Ic mondc 
d'excellents representants de ce genre de recherches ct bcaucoup d'entre 
eux nous ont manque. Quelques-uns, dont le noni s'iniposait, prcssentis, 
declinerent I'invitation. Lorsqu'il y eut necessite de choix, ce furent 
generalement des raisons contingcntes qui dccidcrent. L'idce a prevalu 
qu'il etait souhaitable de niettre en presence sur ce theme de I'Apprentis- 
sage, des representants de pays que les circonstances avaient longtemps 
empeches de confronter leurs theses. Que ccux qui sont venus de I'Europe 
de I'Est aient eu ainsi la possibilite de discutcr, sur un sol sud-americain 
avec les representants de ce qu'on appelle I'Occident, voila un signe 
encourageant pour I'avenir des relations scientifiqucs internationales. 

Le Symposium de Montevideo n'a ete comme tout autre-, qu'un echan- 
tillonnage, dans I'ordre des personnes, comme dans celui des questions 
traitees, et une grande part de hasard a preside a cet echantillonnage. Nous 
espcrons cependant que le livre qui en est le reflet apportera a peu prcs ce 
qu'ils cherchent a ceux qui desirent prendre une vue d'ensemble sur une 
question qui represente un aspect majeur de la connaissance de THomme 
et des lois de la nature. 

A. Fessard 


DARWIN AND CONCEPTS OF BRAIN FUNCTION 
H. W. Magoun 

It is appropriate in the Darwin Centennial Year (1959) to recall the 
impressions made by a visit to South America upon a young graduate of 
Cambridge whose studies, begun in theology, had turned instead to the 
sciences. Darwin arrived in South America as a naturalist on the voyage of 
H.M.S. Beagle, circumnavigating the globe. Reaching Buenos Aires, in 
October of 1832, he found a violent revolution had broken out and 
wrote, 'I was glad to escape on board a packet bound tor Monte Video, 
the second town of importance on the banks of the Plata' (Darwin, 1S39). 

His young man's eye was critical, and his opinions mixed: 'The Plata', 
he wrote, 'looks hke a noble estuary on the map; but it is in truth a poor 
affair. A wide expanse of muddy water, it has neither grandeur nor 
beauty. At one time of the day the two shores, both of which are extremely 
low, could just be distinguished from the deck. The land, with the one 
exception of the Green Mount, 450 feet higli, from which it takes its name, 
is level. Very little of the undulating grassy plain is enclosed, but near the 
town, there are a few hedgebanks. There is something very delightful in 
the free expanse, where nothing guides or bounds your walk, yet I am 
disappointed as regards scenery.' 

The short surveying trips of the Beagle up anci down the coast left 
Darwin with time ashore. 'In the sporting line, I never enjoyed anything 
so much as ostrich hunting with the wild soldiers. They catch the birds in 
a fine, animated chase by throwing two balls which are attached to the 
ends of a thong so as to entangle their legs.' (One was a new species, later 
named Rhea Darwinii.) The young man took an interest in gaining dex- 
terity with the bolas: 'One day as I was amusing myself by galloping and 
whirling the balls around my head, by accident the free one struck a 
bush and, like magic, caught one hind leg of my horse; the other ball was 
then jerked out of my hand, and the horse was fairly secured. The gauchos 
roared with laughter ; they cried out that they had seen every sort of an 
animal caught, but had never before seen a man caught by himself 

On a trip into the interior, Darwin's party stayed at an estancia belong- 
ing to one of the greatest land owners of the country, with whom was a 
captain in the Army. 'Considering their station, their conversation was 

I 


BRAIN MECHANISMS AND LEARNING 


rather amusing. Upon fniding out wc did not catch our horses and cattle 
in England with the lazo, they cried out, 'Ah, then, you use nothing but 
the bolas.' The idea of an enclosed country was quite new to them. The 
Captain at last said he had one question to ask me. I trembled to think how 
deeply scientific it might be. It was 'whether the ladies of Buenos Ayres 
were not the handsomest in the world'. I replied like a renegade, 'Charm- 
ingly so.' He added, 'Do ladies in the other part of the world wear such 
large combs?' I solemnly assured him that they did not. 

'They were absolutely delighted. The Captain exclaimed, 'Look there! 
A man who has seen half the world says it is the case; we always thought so 
but now we know it.' My excellent judgment in combs and beauty 
procured me a most hospitable reception.' Darwin's remarks may have 
expressed more than disinterested diplomacy, however, for at this time a 
letter to his sister read: 'On shore, our chief amusement was riding about 
and admiring the Spanish ladies. After watching one of these angels 
gliding down the street, involimtarily we groaned, "How foolish English 
women are. They can neither walk nor dress." And then, how ugly Miss 
sounds after Signorita.' 

Leaving now the attractive ladies of Buenos Aires, the experiences of 
Darwin's South American visit led gradually to his formulation of the 
theory of evolution by natural selection, certainly one of the most out- 
standing contributions to biology and one with many broad implications. 
The ScaJa naturae, in which living beings were arranged in a spectrum of 
increasing complexity, was familiar to earlier naturalists and to the bio- 
logists of the eighteenth century, by whom its order was generally con- 
ceived as the immutable product of divine creation. Darwin's revolu- 
tionary conception, published a century ago, as the Origin of Species 
(1859), proposed instead that natural selection, working on the range of 
normal variations, led to survival of the fittest and so accounted, in a 
materiahstic way, both for evolution and for the adaptation of existing 
forms to their environments. 

Darwin's later writings, on the Descent of Man (1871) and the Expression 
of Emotion in Man and A)iinials (1872), called more particular attention to 
the phylogenetic development of the brain. In keeping with his contribu- 
tions, and related to the interest in evolution created by them, views were 
subsequently developed by Hughlings Jackson in neurology, by Pavlov in 
physiology and by Freud in psychiatry, each of whom accounted for the 
phylogenetic elaboration of the central nervous system in terms of a series 
of superimposed levels, added successively as the evolutionary scale was 
ascended. 


H. W. MAGOUN 3 

In each of these conceptual systems (Fig. i), the management of pnmi- 
tive, innate, stereotyped behaviour, having to do with the preservation of 
the individual and the race, was attributed to older, subcortical, neuraxial 
portions of the central nervous system, which formed Jackson's lowest 
level and subserved the Pavlovian unconditioned reflex and the 
Freudian id. 


ENGLISH 


RUSSIAN 


COMPARATIVE 


PSYCHOANALYTIC 


NEUROLOGY 


NEUROPHYSIOLOGY 


NEUROANATOMY 


PSYCHIATRY 


SYNTHESIS 


Hughlings Jockson 


Ivan P Pavlov 


Edinger, Koppers, Herrick 


Sigmund Freud 


HIGHEST 
LEVEL 


SECOND 
SIGNAL 
SYSTEM 


^ 


SUPER 
EGO 


ABSTRACTION 
DISCRIMINATION 
SYMBOLIZATION 
COMMUNICATION 


MIDDLE 


CONDITIONED 


r/^ 


ACQUIRED 


LEVEL 


REFLEX 


-^ 


EGO 


ADAPTIVE 
BEHAVIOR 


LOWEST 


UNCONDITIONED 


'Ti 


INNATE 


LEVEL 


REFLEXES 


^ 


10 


STEREOTYPED 
PERFORMANCE 


Fig. I 
Chart coinpanng the evolutionary concepts of the organization and function ot the brain 
whiclr developed after Darwin and Spencer. 

Modified from a chart by Stanley Cobb, Human nature and the understandintj; of disease, 
in: Faxon, N.W. The Hospilnl in Coiiiciiipordry Lift'. Harvard Univ. Press, Cambridge, Mass., 
1949- 


Next, the more mutable, adaptive, learned behaviour of Pavlov's 
conditioned reflex, together with the capacity of the Freudian ego for 
perception and the initiation of movement were ascribed to higher neural 
structures, including the sensori-motor cortex of Jackson's middle level, 
which developed above or upon the older subjacent parts. 

Finally, in the brain of man, hypertrophy of the associational cortices of 
the frontal and parieto-occipito-temporal lobes, forming Jackson's highest 
level, was correlated with the capacities of Freud's superego and, in the 


4 BRAIN MECHANISMS AND LEARNING 

dominant hemisphere, with the capabihties of Pavlov's second signal 
system for symbolization and communication by means of spoken and 
written language. 

Further testimony for the evolution of neurological function in these 
terms was provided by Jackson's view of dissolution, or reversal of the 
phylogenetic process when clinical impairment proceeded from highest 
through middle to lowest levels during neurological disease in man 
(Jackson, 1958). Jackson specified that the resulting deficit was usually 
accompanied by some release of lower activity, normally subjugated to 
higher control. This latter feature was elaborated also in the Freudian 
system, in which conflicting interests of the different levels were emphasized 
as a source of psychic disturbance. 

In much the same way that increased complexity and specialization 
appeared as the ladder of nature was ascended by the earlier classification- 
ists, more and more elaborate functions came to view as one climbed 
cephalically up the successive levels of the central nervous system. In its 
progressive enccphalization, the brain came to resemble the earth itself, not 
simply in its globular form but in consisting as well of a series of strata laid 
down like those of geology, one upon the other, in evolutionary time. 
Each neural accretion was associated with a characteristic increment of 
function and, following Jackson, a dis-solutionary school of neurophysio- 
logy developed, in which enccphalization was reversed by operative 
transection and evolution traced backward by observing residual capaci- 
ties diminish in the increasingly truncated, decorticate, decerebrate and 
spinal preparations. 

Probably because such views arc still so contemporary, little attention 
has been given to exploring the role of Darwin, and the interest in evolu- 
tion excited by his work, in establishing these concepts of neural organiza- 
tion and function. The views of Hughlings Jackson (1958), which might be 
presumed to be the most directly Darwinian, were, on the contrary, 
derived chiefly from Thomas Laycock, with wht~)m Jackson began his 
career in York, and from Herbert Spencer, whom he admired greatly. 
Both Laycock and Spencer had applied evolutionary principles to con- 
cepts of the organization and function of the brain independently of and 
preceding Darwin. In his Mind and Brain, first published in 1859, Laycock 
wrote: 'As wc ascend the scale, the difterentiation of tissue takes place and 
instincts of plants or animals appear. As we ascend still higher in animal 
life, the instincts gradually lose their unknowing character and the mental 
facidties emerge with their appropriate organic basis in the encephalon. 
Finally, with the highest evolution, we find man evincing in art and 


H. W. MAGOUN 5 

science the results of the operation of mental powers which in the lower 
animals are purely nistinctive and in the lowest organisms simply vital 
processes.' 

There can also be found in Laycock an expression of the conflicting 
interests of the different levels with the higher holding the lower in check, 
to be elaborated later by Jackson and by Freud: 'This entire group of 
corporeal appetites and animal instincts is characterized by the quality of 
necessity. They are imperative on the individual; in lower organisms they 
are performed blindly. In man and higher vertebrates, in whom there is a 
development of cognitive faculties, they may be made to act as a check 
upon each other and thus states of consciousness, termed motives, will 
coincide with a knowing restraint exercised over them. But even with the 
highest and strongest of human motives, it is often found difficult to curb 
them effectually. The entire group constitutes "the Flesh" of St Paul. 
Those classed under the head of primordial instincts or corporeal appetites, 
most necessary to the well being and maintenance of the organism anci the 
species, are the farthest removed from the will and consciousness.' 

His clinical observations in neurology led Laycock to consider disease 
as 'retrocession', in which changes taking place were the inverse of evolu- 
tionary. He proposed a law of 'disvolution in certain kinds of brain 
disease, when there was a decay of the mental powers and return to an 
earlier, infantile status. Concepts of evolutionary levels of function, con- 
flicting in their interests and exhibiting dissolution in neurological disease, 
can thus be detected in a germinal stage in Laycock's views. 

Jaekson's psychological concepts were strongly influenced by Herbert 
Spencer, from whom Darwin borrowed the term 'survival of the fittest'. 
After having been an evolutionist for some time, in 1851 Spencer formu- 
lated the basic principles that were to be elaborated in most of his later 
work. He had been asked to write a notice of a new edition of Carpenter's 
Principles oj Physiology and 'in the course', he noted (1904), 'of such perusal 
as was needed to give an account of its contents', came across the theory of 
von Baer — that the development of all plants and animals was from homo- 
geneity to heterogeneity. This concept of progressive differentiation, 
added to that of Lamarckian adaptation, became his distinctive evolu- 
tionary principle. 

Having just turned forty, Spencer determined to devote the remainder 
of his life to the systematic apphcation of this concept to the whole field 
of knowledge. Flagging a dilatatory cerebral circulation, with which he 
was hypochondriacally preoccupied, with bouts of exercise preceding 
dictation to an amanuensis, Spencer embarked on the exposition of a 


6 BRAIN MECHANISMS AND LEARNING 

System of Syiitlictic Philosophy, the successive parts of which appeared at 
intervals through the balance of the nineteenth century. 

In the first edition of his Priiicipk'S of Psychohii^y, published in 1855, and 
thus four years before the Orij^iii of Species, Spencer (1899) pointed out 
that his arguments 'imply a tacit adhesion to the development hypothesis, 
that Life in its multitudinous and infinitely varied embodiments has risen 
out of the lowest and simplest beginnings, by steps as gradual as those 
which evolve a homogenous, microscopic germ into a complex organism, 
by progressive unbroken evolution, and through the instrumentality of 
what we call natural causes. Save for those who still adhere to the Hebrew 
myth or to the doctrine of special creation derived from it, there is no 
alternative but this hypothesis or no hypothesis'. 

Applying his 'development hypothesis' to psychology, Spencer 
reasoned: 'If the doctrine of Evolution is true, the inevitable implication 
is that Mind can be understood only by observing how Mind is evolved. 
If creatures of the most elevated kinds have reached those highly inte- 
grated, very definite and extremely heterogeneous organizations they 
possess, through modification upon modification accumulated during an 
immeasurable past, if the developed nervous systems of such creatures 
have gained their complex structure and functions little by little; then, 
necessarily, the involved forms of consciousness, which are the correlates 
of these complex structures and functions, must also have arisen by 
degrees.' 

In the study of Mind, 'in its ascending gradations through the varic^us 
types of sentient beings', Spencer conceived of 'a nascent Mind, possessed 
by low types in which nerve centres are not yet clearly differentiated from 
one another', and consisting of 'a confused sentiency formed of recurrent 
pulses of feeling having but little variety or combination. At a stage above 
this, while yet the organs of the higher senses are rudimentary. Mind is 
present probably under the form of a few sensations which, like those 
yielded by our own viscera, are simple, vague and incoherent. From this 
upwards, mental evolution exhibits a difterentiation of these simple 
feelings into the more numerous kinds which the special senses yield; an 
ever increasing integration of such more varied feelings, an ever increasing 
multiformity in the aggregates of feelings produced, and an ever increas- 
ing distinctness of structure in such aggregates; that is to say, there goes on 
subjectively a change from an indefinite, incoherent homogeneity to a 
definite, coherent heterogeneity. 

Support for his views was marshalled also from pharmacological 
observations, when Spencer presented, in remarkably vivid detail, the 


H. W. MAGOUN 7 

subjective impressions during induction of chloroform anaesthesia for 
dental extraction, and concluded : 'This degradation of consciousness by 
chloroform, abolishing fnst the higher faculties and descending gradually 
to the lowest, may be considered as rcversnig that ascending genesis of 
consciousness which has taken place in the course of evolution; and the 
stages of descent may be taken as showing in opposite order the stages of 
ascent.' 

"What is the implication oi this law as applying to different grades of 
men?' Spencer asked. And answered, 'It is that those having well developed 
nervous systems will display a relatively marked premeditation, a greater 
tendency to suspense of judgment and an earlier modification of judg- 
ments that have been formed. Those having nervous systems less developed 
will be prone to premature conclusions that arc difficult to change. 
Unlikeness of this kind appears when we contrast the larger with the 
smaller brained races, when, from the comparatively judicial intellect of 
the civilized man, we pass to that of the uncivilized man, sudden in its 
inferences, incapable of balancing evidence and adhering obstinately to 
first impressions.' Tliough Spencer's association with the female sex was 
limited, he nevertheless felt qualified 'to observe a difference similar in 
kind but smaller in degree between the modes of thought of men and 
women; for women are more quick to draw conclusions and retain more 
pertinaciously beliefs once formed'. 

Returning to the problem 'of how such higher co-ordinations are 
evolved out of lower ones and how the structure of the nervous system 
becomes progressively complicated', Spencer proposed the interpolation 
of new plexuses of fibres and cells between those originally existing. In 
diagrammatic sketches, apparently of an invertebrate ganglion, Spencer 
distinguished (Fig. 2,) 'a nervous centre to which afferent fibres bring 
all order of peripheral feelings, and from which efferent fibres carry 
to muscle the stimuli producing their appropriately combined contrac- 
tions'. If a part of the co-ordinating plexus (A) 'takes on a relatively greater 
development in answer to new adjustments which environing conditions 
furnish, we may expect one part of this region (a) to become protruberant, 
as at A". Because space within the plexus was already pre-empted, 'the 
interpolated plexus, which effects indirect co-ordination, must be super- 
imposed [Fig. 2, A' above; d, below], and the co-ordinating discharges 
must take roundabout courses as shown by the arrow. Little by little, there 
is an enlargement of the superior co-ordinating centre by the interpolation 
of new co-ordinating plexuses at its periphery' (Fig. 2, e, f, g, below). 

With the publication of Darwin's Origin of Species, Spencer gave some 


BRAIN MECHANISMS AND LEARNING 


consideration to the concept of natural selection, but continued to account 
for the more complex portions of neural evolution primarily in Lamarc- 
kian terms: 'Regarding as superimposed, each on the preceding, the 
structural effects produced generation after generation and species after 
species, we have formed a general conception of the manner in which the 
most complex nervous systems have arisen out of the simplest. This 


A'.... 


Fig. 2 
Diagrams of a ganglion prepared by Spencer (17), sliowing the 
development of superimposed levels of neural co-ordination^ 


general principle can be alleged only on the assumption that changes 
wrought in nervous structures by nervous functions are inheritable. 
Throughout the earlier stages of nervous evolution, a leading and perhaps 
most active cause has been the survival of individuals in which indirect 
influences have produced favourable variations of nervous structure but, 
throughout its later stages, the most active cause has been the direct 


H. W. MAGOUN 


production by functional changes of corresponding changes in nervous 
structure, and the transmission of these to posterity. Considering how 
involved are the nervous systems of superior creatures, there apply here 
with special force reasons for concluding that natural selection is an 
inadequate cause of evolution where many co-operative parts have to be 
simultaneously modified; and that in such cases, inheritance of functionally 
produced modifications becomes the leading agency — survival of the 
fittest serving as an aid.' 

Ii'ivi P. Parlor. Though Pavlov's work in the physiology of the central 
nervous system did not commence until his fifties, its conceptualization 
was influenced strongly by the ideas of Darwin and of Spencer, encoun- 
tered in liis youth through the writings of Pisarev and Sechenov. In his 
Aiirohio^irapliY (1955), Pavlov wrote: T was born in the town of Ryazan 
in 1849 and received my secondary education at the local theological 
seminary, hifluenced by tlie literature of the 'sixties, and particularly by 
Pisarev, our intellectual interests turned to natural science and many, 
myself included, decided to take the subject at the university.' 

Pisarev was a writer and critic whose articles in the Riisskoyc Sloro 
promoted revolutionary-democratic anti materialistic ideas among the 
intelligentsia of the 'sixties. There seems little doubt that Pavlov first 
became captivated by Darwin and the theory of evolution from reading 
Pisarev's lengthy, systematic, popular exposition of the Ori'^iii of Species 
entitled, Proi^ress in the Aiii)iuil (vui Ve^ietable li'orlds (1858). The ecstatic 
attitude towards Darwin, which Pavlov preserved to the end of his days, 
can easily be identified with Pisarev's lofty expression: 

'This brilliant thinker, whose knowledge is enormous,' Pisarev wrote of 
Darwin, 'took in all the life of nature with such a broad view and pene- 
trated so deeply into all its scattered phenomena that he discovered, not an 
isolated fact, but a whole series of laws according to which all organic life 
on our planet is governed and varies; and he told of them so simply, 
proves them so irrefutably and bases his arguments on such obvious facts, 
that you, a common human, uninitiated in natural science, are in a state of 
continual astonishment at not having thought out such conclusions your- 
self long ago. 

'For us ordinary and unenlightened people, Darwin's discoveries are 
precious and important just because they are so fascinating in their 
simplicity, so easy to understand; they not only enrich us with new 
knowledge, they give fresh life to all the system of our ideas and widen 
our mental horizon in all dimensions. In nearly all branches of natural 
science, Darwin's ideas bring about a complete revolution. Even 


TO BRAIN MECHANISMS AND LEARNING 

experimental psychology tmds in his discoveries the guiding principle 
that will link up the numerous observations already niade and put investi- 
gators on the way to new fruitful discoveries. 

'Every educated man', Pisarev continued, 'must make himself familiar 
with the ideas of this thinker and, therefore, I think it fitting and useful to 
give our readers a clear and fairly detailed exposition of the new theory. 
In it, readers will find the rigorous precision of an exact science, the 
boundless breadth of philosophical generalization and, finally, the 
superior and irreplaceable beauty which is the mark of vigorous and 
healthy human thought. Darwin, Lyell and thinkers like them are the 
philosophers, the poets, the aestheticians of our time. When human reason, 
in the person of its most brilliant representatives, has succeeded in rising 
to a height from which it can survey the basic laws of universal life, we 
ordinary people, unable to be creative in the realm of thought, owe it to 
our own human dignity to raise ourselves at least enough to be able to 
understand the leading brilliant minds, to appreciate their great achieve- 
ments, to love them as the ornament and pride of our race; to live in 
thought in the bright and boundless realm that they open for every 
thinking being. We are wealthy and powerful through the works of these 
great men.' 

A second early influence upon Pavldv was provided by the writings of 
Sechenov (1935). Later in his career, Pavlov referred (1928) to the begin- 
ning study of higher nervous activity with the objective techniques of 
conditional reflex physiology: 'The most important motive for my 
decision, even though an unconscious one, arose out of the impression 
made upon me during my youth by the monograph of I. M. Sechenov, 
the Father of Russian physiology, entitled Cerebral Reflexes and published 
in 1863. The influence of thoughts which are strong by virtue of their 
novelty and truth, especially when they act during youth, remains deep 
and permanent, even though concealed. In this book, a brilliant attempt 
was made, altogether extraordinary for that time, to represent our 
subjective world from the standpoint of neurophysiology.' 

In a report on objective study of higher nervous activity in 191 3, 
Pavlov (1928) began: 'With full justice, Charles Darwin must be counted 
as the founder and instigator of the contemporary comparative study of 
the higher vital phenomena of animals; for, as is known to every educated 
person, through his highly original support of the idea of evolution, he 
fertilized the whole mentahty of mankind, especially in the field of 
biology. The hypothesis of the origin of man from animals gave a great 
impetus to the study of the higher phenomena of animal life. The answer 


H. W. MAGOUN 


II 


to the question as to how this study should be carried out and the study 
itself have become the task of the period following Darwin.' 

Pavlov concluded: 'I have hnished my communication, but I should 
like to add what seems to me to be of great importance. Exactly half a 
century ago, ni 1863, was published in Russian the article Reflexes of the 
Brain, which presented in clear, precise and charming form the funda- 
mental idea which we have worked out at the present time [see Fig. 3]. 
What power of creative thought was necessary under the then existing 


Fig. 3 
Diagram of the central nervous system of the frog (left), from Sechenov (16). Stimulatiiin of 
the sites marked by crosses inhibited spinal reflexes, illustrating the hierarchy of neural levels 
and the domination of higher over louder. 

At the right is a diagram of the mechanism of the Pavlovian conditioned reflex (13). The 
animal makes adaptive adjustments to its environment by means of new links between the 
cortical analvsers and connections from them to subcortical, unconditioned reflex arcs. 


physiological knowledge of nervous activity to give rise to this idea! 
After the birth of this idea, it grew and ripened until, in our time, it has 
become an immense force for directing the contemporary investigation 
of the brain. Allow me at this fiftieth anniversary of the Reflexes of the 
Brain to invite your attention to the author, Ivan M. Sechenov, the pride 
of Russian thought and the Father of Russian physiology!' 

It is interesting to note that Sechenov, like Jackson a contemporary ot 
both Darwin and Spencer, was more directly influenced by Spencer's 


12 BRAIN MECHANISMS AND LEARNING 


writing than by that of Darwin. Though not in time to influence prepara- 
tion of the Reflexes of the Brain (1863), both Darwin's and Spencer's works 
were early translated into Russian: the Orii^iii of Species in 1864, by Pro- 
fessor S. A. Rashinsky of Moscow University, of whose efforts Pisarev 
was highly critical; and a year later in shorter exposition, by K. A. 
Tiniiryazev, the leading Russian Darwinist (Platonov, 1955). Spencer 
(1904) learned of a Russian translation of his First Principles in 1866 and, a 
decade later, heard with surprise, from Professor Sontchitzici, of the 
University of Kiev, that all of his works had then been translated into 
Russian, excepting the Sociology, which was soon to be added to the list. 

In his Elements of Thought, published in 1883, Sechenov (1935) wrote: 
'Darwin's great theory of the evolution of species has placed the idea of 
evolution on such a firm basis that it is at present accepted by the vast 
majority of naturalists. This logically necessitates the recognition of the 
principle of evolution of psychical activities. Spencer's hypothesis may 
actually be called the application of Darwinism to the sphere of psychical 
phenomena.' 

And later: 'Another and no less important success in the study of the 
mental development of man in general we owe to the famous English 
scientist, Herbert Spencer. It is only on the ground of Spencer's hypothesis, 
concerning the sequence of stages of 'neuropsychical development from 
age to age that we can solve the ancient philosophical problem of the 
development of mature thought from initial infantile forms. To Spencer 
we owe the establishment, on the basis of very wide analogies, of the 
general type of mental development in man, as well as the proofs of the 
fact that the type of evolution of mental processes remains unchanged 
through all stages of the development of thought. The present essay is 
based on the theories of Spencer; therefore, our first task will be to 
expound the main principles of his theory. It even appeared at the same 
time as Darwin's theory and is practically a part of the general theory of 
organic evolution.' 

Signiund Freiid. Passing now to Freud, his autobiography refers to the 
influences leading him to medicine as a career: 'At the time the theories 
of Darwin, which were then of topical interest, strongly attracted me, for 
they held out hopes of an extraordinary advance in our understanding of 
the world; and it was hearing Goethe's beautiful essay on Nature read 
aloud at a popular lecture, just before I left school, that decided me to 
become a medical student.' 

There are singularly few other allusions to Darwin in Freud's writings, 
and the factors responsible for his visualization of the psychic apparatus 


H. W. MAGOUN 13 

as spatially stratified were, doubtless, unconscious ones. It seems an 
exaggeration to propose that a continuum can be detected, in any literal 
sense, in Freud's anatomical, neurological and psychoanalytical works. 
Instances of a recurring effort to interpret neural organization and function 
in evolutionary terms can, however, be noted. In his monograph on 
Aphasia, published in 1891, Freud wrote (1953): 'In assessing the functions 
of the speech apparatus under pathological conditions, we are adopting as 
a guiding principle Hughhngs Jackson's doctrine that all these modes of 
reaction represent instances of functional retrogression (disin volution) of a 
highly organized apparatus, and therefore correspond to earlier stages of 
its functional development. This means that under all circumstances an 
arrangement of associations which, having been acquired later, belongs to 
a higher level of functioning, will be lost, while an earlier and simpler one 
will be preserved. From this point of view, a great number of aphasic 
phenomena can be explained.' 


Pcpt-C5 


Fig. 4 
Two diagrams by Freud (8, 8b), presenting the mental apparatus as 
though spatially stratified. 


In a letter to Fleiss in 1896, Freud (1954) discussed a revision of his 
Project for a Scientific Psychology and referred to his 'latest bit of specula- 
tion, the assumption that our psychical mechanism has come about by a 
process of stratification'. A quarter of a century later, Freud made two 
attempts to diagram these ideas, with interesting differences in the form of 
the figures. The first (Fig. 4, left), prepared in 1923, resembled an inverted 
brain, although reference was made to it as an ovum. The second (Fig. 4, 
right), prepared a decade later, was on the other hand really egg-shaped. 
In his lecture on 'The Anatomy of the Mental Personality', Freud (1933) 
elaborated upon the contents of these figures: 'Superego, ego and id are 


14 BRAIN MECHANISMS AND LEARNINC; 

the throe reahns, regions or provinces into which we divide the mental 
apparatus of the individual, and it is their mutual relations with which we 
shall be concerned. 

'The id is the obscure, inaccessible part of our personality and can only 
be described as being all that the ego is not. We can come nearer to the id 
with images, and call it a chaos, a cauldron of seething excitement. We 
suppose that it is somewhere in direct contact with somatic processes and 
takes over from them instinctual needs. These instincts till it with energy, 
but it has no organization and no unified will, only an impulsion to obtain 
satisfaction for the instinctual needs in accordance with the pleasure 
principle. Contradictory impulses exist side by side in it, without neutral- 
izing each other or drawing apart; at most they combine in comprtMnise 
formations under the overpowering pressure towards discharging their 
energy. In the id, there is nothing corresponding to the idea of time. 
Conative impulses which have never got beyond the id, and even impres- 
sions which have been pushed down into it by repression, are virtually 
immortal and are preserved for whole decades, as though they had only 
recently occurred. They can only be recognized as belonging to the past, 
deprived of their significance, and robbed of their charge of energy, after 
they have been made conscious by the work of analysis, and no small part 
of the therapeutic effect of analytic treatment rests upon this fact. Natur- 
ally the id knows no values, no good and evil, no morality. There is 
nothing in the id which can be compared to negation. Instinctual cathexes 
seeking discharge — that, in our view, is all that the id contains. 

'The (\'o is directed onto the external world; it mediates perceptions of 
it and in it are generated, while it is functioning, the phenomena of 
consciousness. The ego has taken over the task of representing the external 
world for the id. In the fulfilment of this function, it has to observe the 
external world and preserve a true picture of it in the memory traces left 
by its perception. The ego also controls the path cif access to motility, but 
it interpolates between desire and action the procrastinating factor of 
thought, during which it makes use of the residues of experience stored up 
in memory. In this way, it dethrones the pleasure principle, which exerts 
undisputed sway over the processes in the id, and substitutes for it the 
reality principle, which promises greater security and success. The 
relation to time, too, is contributed to the ego by the perceptual systems; 
indeed, it can hardly be doubted that the mode in which this system works 
is the source of the idea of time. What, however, especially marks the ego 
out in contradistinction to the id is a tendency to synthesize its contents, to 
bring together and unify its mental pr(^cesses, which is entirely absent from 


H. W. MAGOUN 15 

the id. In popular language, we may say that the ego stands for reason and 
circumspection, while the id stands for the untamed passions. One might 
compare the relation of the ego to the id with that between a rider and his 
horse: the horse provides the locomotive energy, and the rider has the 
prerogative of determining the goal and of guiding the movements of his 
powerful mount. 

'The role which the supcrc<^o undertakes later in life is at hrst played by 
an external power, by parental authority. It can be traced back to the 
influence of parents, teachers and so on, and is based upon an over- 
whelmingly important biological fact, namely, the lengthy dependence 
of the human child on his parents. We have allocated to the superego the 
activities of self-observation, conscience and the holding up of ideals. It 
is the representative of all moral restrictions, the advocate of the impulse 
towards perfection. In short, it is as much as we have been able to appre- 
hend psychologically of what people call the "higher things in human 
life". It becomes the vehicle of tradition and of all the age-long values 
which have been handed down from generation to generation. The 
ideologies of the superego perpetuate the past, the traditions of the race 
and the people, which yield but slowly to the influence of the present and 
to new developments.' 

In discussing the interrelations ot these parts, Freud, like Spencer, 
appeared to invoke Lamarckian views : 'The ego has the task of bringing 
the influence of the external world to bear upon the id. In the ego, percep- 
tion plays the part which, in the id, develops upon instinct. The experiences 
undergone by the ego seem at tnst to be lost to posterity ; but, when they 
have been repeated often enough and with sufficient intensity in the 
successive individuals of many generations, they transform tlicmselves so 
to say into experiences of the id, the impress of which is preserved by 
inheritance. Thus in the id, which is capable of being inherited, are stored 
up vestiges of the existences led by countless former egos; and, when the 
ego forms its superego out of the id, it may perhaps only be reviving 
images of egos that have passed away and be securing them a resurrection. 

'The poor ego has, then, to serve three harsh masters and to do its best 
to reconcile their claims and demands. These demands are always diver- 
gent and often seem quite incompatible; no wonder the ego so frequently 
gives way under its task. The three tyrants arc the external world, the 
superego and the id. It feels itself hemmed in on three sides and threatened 
by three kinds of danger, towards which it reacts by developing anxiety 
when too hard pressed. 

'Having originated in the experiences of the perceptual system, it is 


l6 BRAIN MECHANISMS AND LEARNING 

designed to represent the demands of the external world, but it also wishes 
to be a loyal servant to the id, and to draw the id's hbido onto itself. In its 
attempt to mediate between the id and reality, it is often forced to clothe 
the unconscious commands of the id with its own preconscious rationaliza- 
tions, to gloss over the conflicts between the id and reality. 

'On the other hand, its every movement is watched by the severe 
superego, which holds up certain norms of behaviour without regard to 
any difficulties coming from the id and the external world; and if these 
norms are not acted up to, it punishes the ego with feelings of tension 
which manifest themselves as a sense of inferiority or guilt. In this way, 
goaded on by the id, hemmed in by the superego, and rebuffed by 
reality, the ego struggles to cope with its task of reducing the forces and 
influences which work in it and upon it to some kind of harmony. When 
the ego is forced to acknowledge its weakness, it breaks out into anxiety; 
reality anxiety in the face of the external world, normal anxiety in the 
face of the superego, and neurotic anxiety in the face of the strength of 
the passions of the id. 

'It can easily be imagined that certain practices of mystics may succeed 
in upsetting the normal relations between the different regions of the mind 
so that, for example, the perceptual system becomes able to grasp relations 
in the deeper layers of the ego and in the id which would otherwise be 
inaccessible to it. Whether such a procedure can put one in possession of 
ultimate truths, from which all good will flow, may be safely doubted. 
All the same, we must admit that the therapeutic efforts of psychoanalysis 
have chosen much the same method of approach; for their object is to 
strengthen the ego, to make it more independent of the superego, to 
widen its field of vision, and so to extend its organization that it can take 
over new portions of the id. Where id was, there shall ego be. It is re- 
clamation work, like the draining of the Zuyder Zee.' 

Spciiar and Darwin. From what has been presented it seems clear that, to 
their contemporaries, Spencer's ideas of the evolution of the brain and 
its functions were fully as influential as Darwin's, if not more so. This may 
be attributable in part to the fact that Spencer applied evolutionary prin- 
ciples to an understanding of the brain earlier than Darwin and, indeed, 
before the lattcr's ideas were published at all. Additionally, appeal doubt- 
less attached to the broad sweep of Spencer's interests and to his efforts to 
account for the whole range of neural function, from instincts to the most 
complex features of the mind, in keeping with his propensity for global 
syntheses. 

Darwin, by contrast, stuck closer to the observational data then 


H. W. MAGOUN 17 

available with emphasis mainly upon instinctive behaviour, supporting 
the survival of the individual, in feeding, aggression or defence, as well as 
survival of the species, in behaviour relating to sex. Even in Darwin's 
Descent of Man (1871), for example, greater emphasis was placed upon an 
exposition of sexual selection than upon development of the associational 
and communicative functions of the brain, which are easily the most 
strikingly distinctive features of human evolution. 

Darwin and Spencer might be rated cquivalcntly in the impressions 
they made upon Pisarev and Sechcnov and, through these latter, upon 
Pavlov and Russian neurophysiology. There is no question, however, of 
the predominant influence of Spencer upon Hughlings Jackson and, 
through him upon the formation of evolutionary concepts of the organiza- 
tion and function of the brain in Western neurological thought. 

Contrasting features of their personalities and outlooks appeared to have 
led Darwin and Spencer to develop reservations about one another. An 
early phrenological characterization of Spencer (Spencer, 1904) con- 
cluded: 'Such a head as this ought to be in the Church. The self-esteem is 
very large.' Darwin's tendency to personal deprecation seemed, on the 
other hand, to have amounted to a real sense of inferiority when compar- 
ing himself with Spencer. Each seemed also to have cultivated a possibly 
wilful ditiiculty in understanding the other's views. In a letter to Hooker 
in 1868, Darwin (1925) wrote: 'I feel Painyeiiesis is stillborn. H. Spencer 
says the view is quite difl:erent from his (and this is a great relief to me, as I 
feared to be accused of plagiarism but utterly failed to be sure what he 
meant, so thought it safest to give my view as almost the same as his), and 
he says he is not sure he understands it.' 

In other letters, Darwin blew hot and cold. He characteristically 
acknowledged Spencer's brilliance, but usually expressed some question 
of the soundness or reliability of his views. In a note thanking Spencer for 
a present of his Essays in 1858, Darwin wrote: 'Your remarks on the 
general argument of the so-called development theory seem to me admira- 
ble. I am at present preparing an Abstract of a larger work on the changes 
of species; but I treat the subject simply as a naturalist, and not from a 
general point of view; otherwise, in my opinion, your argument could 
not have been improved on, and might have been quoted by me with 
great advantage.' 

In a letter to Hooker in 1866, Darwin wrote: 'I have now read the last 
No. of H. Spencer {Principles of Biology). It is wonderfully clever and I 
daresay mostly true. I feel rather mean when I read him: I could bear and 
rather enjoy feeling that he was twice as ingenious and clever as myself; but 


l8 BRAIN MECHANISMS AND LEARNING 

when I feel he is ahnost a dozen times my superior, even in the master art 
of wrigghng, I feel aggrieved. If he had trained himself to observe more, 
even at the expense, by the law of balancement, of some loss of thinking 
power, he would have been a wonderful man.' 

In a letter to E. Ray Lankester in 1870, Darwni's regard reached a high 
point: 'It has pleased me to see how thoroughly you appreciate (and I do 
not think this is general with men of science) H. Spencer; I suspect that 
hereafter he will be looked at as by far the greatest living philosopher in 
England; perhaps equal to any that have lived.' Darwin's regard was also 
expressed in a note to Spencer himself at this time (1872): 'Dear Spencer 
— I daresay you will think me a foolish fellow but I cannot resist the wish 
to express my unbounded admiration for your article. Everyone with eyes 
to see and ears to hear (the number, I fear, are not many) ought to bow 
their knee to you, and I for one do.' 

In a letter to Fiske, in 1874, reaction had set in: 'With the exception of 
special points, I did not even understand H. Spencer's general doctrine; 
for his style is too hard work for me. This may be very narrow minded; 
but the result is that such parts of H. Spencer as I have read with care 
impressed my mind with the idea of his inexhaustible wealth of suggestion, 
but never convinced me.' 

In a fmal judgment, in his Aiitohi(\^rapliy (1958), Darwin commented: 
'Herbert Spencer's conversation seemed to mc very interesting but I did 
not like him particularly and did not feel that I could easily become 
intimate with him. I think he was extremely egotistical. After reading any 
of Spencer's books, I generally feel enthusiastic admiration for his trans- 
cendent talents and have often wondered whether in the distant future he 
would rank with such great men as Descartes, Leibniz, etc., about whom, 
however, I know very little. Nevertheless, I am not conscious of having 
profited in my own work by Spencer's writings. His deductive manner of 
treating every subject is wholly opposed to my frame of mind. His 
conclusions never convince me: and over and over again I have said to 
myself after reading one of his discussions, "Here would be a fine subject 
for half a dozen year's work." His fundamental generalizations (which 
have been compared in importance by some persons with Newton s 
Laws !) — which I daresay may be very valuable under a philosophical 
point of view — are of such a nature that they do not seem to me to be of 
any strictly scientific use. They partake more of the nature of definitions 
than of laws of nature. They do not aid one in predicting what will 
happen in any particular case. Anyhow, they have not been of any use to 
me.' 


H. W. MAGOUN 19 

Reciprocal comments on Darwin and his work, by Spencer, were made 
primarily from the point of view of their relations to Spencer's own ideas 
and niterests. With publication of the Orioiii of Species, Spencer wrote 
(1904) : 'That reading it gave me great satisfaction may be safely inferred. 
Whether there was any set-oft to this, I cannot now say; for I have quite 
forgotten the ideas and feelings I had. Up to that time, I held that the sole 
cause of organic evolution is the inheritance of functionally-produced 
moditications. The Ori^Ui of Species made it clear to me that I was wrong; 
and that the larger part of the facts cannot be due to any such cause. 
Whether proof that what I had supposed to be the sole cause, could be at 
best but a part cause, gave me any annoyance, I cannot remember; nor 
can I remember whether I was vexed by the thought that, in 1852, I had 
fiiled to carry further the idea then expressed that, among human beings, 
the survival of those who are the select of their generation is a cause of 
development. But I doubt not that any such feelings, if they arose, were 
overwhelmed in the gratification I felt at seeing the theory of organic 
evolution justified. 

'To have the theory of organic evolution justified was, of course, to get 
further support for that theory of evolution at large, with which, as we 
have seen, all my conceptions were bound up. Believing as I did, too, that 
right guidance, individual and social, depends upon acceptance of 
evolutionary views of mind and of society, I was hopeful that its effects 
wouki presently be seen on educational methods, political opinions and 
men's ideas about human life. Obviously, these hopes that beneficial 
results would presently be wrought, were too sanguine. My confidence in 
the rationality of mankind was much greater then than it is now.' 

On the revision of his Principles of Psycholooy, in 1870, Spencer wrote, in 
a similar vein: 'Several feelings united in making me enjoy the resumption 
of this topic, which I dealt with in 1854-55. At that date, an evolutionary 
view of Mind was foreign to the ideas of the time, and voted absurd: the 
result of setting it forth brought pecuniary loss and a good cieal of reproba- 
tion. Naturally, therefore, after the publication of the Ori(^iii of Species 
had caused the current oi public opinion to set the other way, a more 
sympathetic reception was to be counted upon for the doctrine of mental 
evolution in its elaborated form.' 

In 1872, Spencer acknowledged a copy of Darwin's work on The 
Expression of the Emotions as follows: "Dear Darwin: I have delayed some- 
what longer than I intended acknowledging the copy of your new 
volume which you have been kind enough to send me. I delayed partly in 
the hope of being able to read more of it before writing to you; but my 

c 


20 BRAIN MECHANISMS AND LEARNING 

reading powers arc so small, and they arc at present so much eniployed 
in getting up materials for work in hand, that I have been unable to get on 
far with it. I have, however, read quite enough to see what an immense 
mass of evidence you have brought to bear in proof of your propositions. 

'I will comment only on one point, on which I see you tiifter from 
me ... ' 

Differing interests in the presentation of observational data and in the 
derivation from it of speculative syntheses, so apparent in the attitudes of 
Darwin and Spencer, will doubtless appear in the programme of this 
present conference as well. To find a topic upon which all may initially 
agree, let us return to Darwin's South American visit of a century ago. 
His narrative (1839), under December 6th, 1832, notes: 

'The Bcai^lc sailed from the Rio Plata never again to enter its muddy 
stream . . . When speaking of these countries, the manner in which they 
have been brought up by their unnatural parent, Spain, should always be 
borne in mind. On the whole, perhaps, more credit is due for what has 
been done, than blame for what may be deficient. It is impossible to 
doubt but that the extreme liberalism of these countries must ultimately 
lead to good results. The very general toleration of foreign religions, the 
regard paid to the means of education, the freedom of the press, the 
facilities offered to all foreigners, and especially, as I am bound to add, to 
everyone professing the humblest pretensions to science, should be re- 
collected with gratitude by all those who have visited Spanish South 
America.' 

The most cordial reception which has been providec^ the present visitors 
to Montevideo, in 1959, will, I know, make each of us wish to echo and 
approve emphatically of Darwin's concluding remarks. 


THE FIXATION OF EXPERIENCE^ 
R. W. Gerard 

The tixation of experience is a wider topic than is learning, which is a 
subhead under it. It inckides changes in an nidividual system, at all levels 
from molecule to taxa or society, that have become irreversible under 
single or repeated experiences and so have left some material record of a 
past activity; and it includes racial changes that have cumulated over 
generations of a self-reproducing system. It miderlies one of the three 
universal attributes of all systems at all levels; 'becoming'. The architec- 
ture oi any system, its inhomogenities at a given cross section of time 
which remain reasonably constant over succeeding cross sections, its 
'being', is the base of the behaviour or functioning of the system along 
time. 'Behaving' represents the transient or functional responses of the 
system to stimuli or stresses imposed by the environment and are rever- 
sible, so that the system essentially reverts after the perturbations have 
passed. When, however, the stimuli are sufficiently intense or repetitive or 
meaningful to the system as to leave an irreversible change, and therefore 
a material residue, the system undergoes a secular change along the 
longitudinal time axis and has tixed experience. 'Becoming' thus encom- 
passes individual development, racial evolution, cultural history, and 
many other basic phenomena of orgs in general (niaterial systems) and of 
animorgs ni particular (living material systems). 

Irreversible changes of individual units at the molecular level include 
gene mutations and adaptive enzymes and antibodies. At the cellular level, 
comes the whole process of cellular differentiation, including the fornia- 
tion or lack of formation of particular organelles and particulates. At the 
organ level, inductions and gradients and mechanical forces mold the 
particular organs during cievelopment, just as their use through life leaves 
behind hypertrophied soles or muscles, wrinkled and weathered skin, 
bone shapes to meet functional strains, and the like. Engrams within the 
nervous system are entirely comparable residues of experience. At the 
individual level, come the collective processes of ageing, perceptual and 
motor habits, conscious memories and the like. And at the group or social 
level, individual cultures create customs, laws, languages, artifacts, libraries 
and many other concrete entities or functional roles occupied by concrete 
entities. 

1 Aided by a grant from the National Institute of Mental Health, U.S.P.H. 

21 


BRAIN MECHANISMS AND LEARNING 


At the racial dimension, or successive generations in time, there are the 
typical character changes which come to identify a taxa — morphological, 
physiological, chemical and, progressively, behavioural attributes that 
have accumulatively changed, in a more or less directional manner, over 
many generations. More important, and less universally recognized, has 
been the evolution, not of adaptations, but of adaptability. Selection 
pressures from an environment will prociuce evolutionary changes only 
when the stock is malleable and can respond to the pressures. Systems must 
be able to respond to experience and to tix it in some way if they arc to be 
changed by it; and since heredity must supply the initial plasticity which 
enables the system to respond adaptively to selection pressures — and there 
is much current evidence that natural selection does not operate quite so 
blindly as was at one time believed (e.g. Waddington, 1950, i960; 
Gerard, 1960b) — the sharp line between Darwin and Lamarck is begin- 
ning to blur. One can inherit not only mutated genes, but genes that are 
more mutable, even genes that induce mutations in others. Adaptive 
enzymes come into being only when both the genetic potentiality and 
also the environmental substate are present. An organism not only can 
learn, it can learn to learn. Learning set, attention level, motivation 
intensity, past experience, present physiological state, type of stimulus 
presentation and many other factors can influence the speed and effective- 
ness of learning. And the contributions of heredity, of individual experi- 
ence, and of current situation to many, if not all, of these factors have not 
remotely been disentangled. 

It remains true, none the less, that learning to learn, accelerating adapta- 
tion, speeding auto- and hetero- catalysis is the great invention of life 
stuff". This is the epigcnetic mode. It allows living organisms to respond 
ever more rapidly and adaptively to the environmental challenge; it 
similarly enables mindful organisms to meet their environmental problems 
with greater skill and speed; and it has brought about that accelerating 
cultural change in civilizations which seems almost to have reached an 
explosive point. Epigenesis was enhanced by increased gene mutability, by 
the development of chromosomes and sex assortment, by adaptive changes 
in individual characteristics (in themselves or as a richer array for the 
action of simple natural selection), by the invention of a nervous system 
and of highly differentiated or coded responses. 

Environment operates upon a system at all levels, differentially selecting 
for survival particular genes or gene arrays, cells and cell aggregates, 
organs and organ systems, and individuals and groups of various sizes. 
The environment operates not so much on the finished product as on the 


R. W. GERARD 23 

formative process. It supplies the physico-chemical milieu determining 
molecular changes, the electro-chcmico-mechanical held guiding cellular 
changes, the neuro-chemico-mechanical influences modulating organ 
development, the material and biotic stimuli that guide the maturation oi 
the individual, and the coded and meaning-laden signs and symbols which 
arc added to these in the course ot enculturating an individual into his 
group. The stresses applied by the environment determine the direction of 
development of the individual and the selection of the individual in the 
group. It may deternnne the adaptations of the body, the behaviour of 
the individual, the nornis of the culture, and, in general, the goals or values 
that guide the course of future change of a system. 

Living thmgs are engaged in a continuous tracking operation, attempt- 
ing to bring their existing state or their anticipated state into congruence 
with a desu-ed state. At any instant in time, the system faces its universe, 
from which it is separated by some kind of boundary, with a certain 
patterned inhomogenity which is by then its enduring property. In the 
course of interacting with its environment, u-reversible changes are 
produced in the system and it reaches a new state, the new architecture 
which it then offers for interaction with further environmental stimuli or 
stresses. Any individual system, and all its sub-systems, thus brings a new 
self to succeeding chunks of time. What the zygote brings we ordinarily 
call heredity, the end-product of ancestral learning and selection. What 
happens to the zygote we call individual experience, often divided further 
into experiences /// iitero, which we call congenital, and those affecting the 
separate organism, tirst the infant and then the adult. 

At each stage and at each level, the system or sub-system presents to the 
environment a structure which has at least some aspects of a template, and 
so can lead to the production of more or itself; and at least some aspects of 
a programme or set of operation rules, so that the kinds of responses it 
will make to certain situations are roughly indicated. The outcomes are 
never identical and never foreseeably deterministic, because the tine 
details of the particular template and programme, even in identical twins, 
are not absolutely identical and, even more, because the environmental 
conditions to which these are exposed are never even roughly identical. 
Despite relative constancy in 'beings', therefore, outcomes are always 
more variable, the exact one in each case depending on the particular, often 
chance, details of the individual — environment interaction. Clearly, the 
line between heredity and individual experience becomes vague indeed. A 
gene array is a template and a programme; so is an engram. 

It is a duty of neurobiology to discover the action rules. (If the nervous 


24 BRAIN MECHANISMS AND LEARNING 

system knows the equation for tt, it can grind out an unlimited number of 
digits, and yet carry none of this as bits of information.) It is its function, 
also, to decipher how existing engrams form the scaffolding for new ones. 
Here is the crux of the problem of learning. It depends upon the evolved 
structures — the improved units, patterns, and numbers in the more 
advanced nervous systems, but it depends no less on the improved 
physiology — lowered thresholds, faster conduction, greater spontaneity, 
easier fixation, and a host of other attributes, which arc just coming to be 
recognized as important characteristics of the 'higher' animals as compared 
with the 'lower' ones. But before coming to grips with the problem of 
the neural mechanisms of learning, a few more general considerations 
involving the fixation of expierience will be helpful. 

Several questions must be faced for all systems that fix experience. The 
first, of course, is: How does a system at a given time, with an enduring 
architecture, contributed by the past experience of the race or itself, 
interact with the environment to give a new enduring architecture? As 
alreaciy indicated, this applies at each level and over all time spans. 
Second, when does a reversible change — a homeostatic response to stress, 
or a behavioural change in response to a stimulus — become irreversible? 
What is the limit of homeostatic tolerance, the Rubicon that is crossed, 
when the transient response becomes either an adaptive change or a non- 
adaptive (pathic) breakdown? When does a dynamic memory become a 
structural one, much as the spoken word becomes fixed in writing? At 
what stage does the totipotent embryonic cell become irreversibly differen- 
tiateci and specialized for a particular function? When does the individual 
growing up in his society acquire the set of values, customs, skills that 
characterize it? When can he no longer learn a new language without an 
accent, or face a different culture with no sense of xenophobia ?Third, 
what is the mechanism, or the carrier, of the operation? How is the 
change in the system given an adaptive (or a non-adaptive) direction? 
Fourth, turning from the individual to the race, how is the adapted 
individual selected or, to the extent that individual change is passed on, 
how is this achieved? Last, what is the mechanism of cumulative racial 
change? 

Such questions are not merely disembodied abstractions, even as regards 
the nervous system and behaviour. Learning must utimately be at the 
molecular level, as well as at cellular, organ, and individual levels. The 
material record of experience must be found in some change in the 
number, kind, or position of particles; in the pattern of ions or molecules 
in neurones or at their junctions. As acquired racial information is passed 


R. W. GERARD 2$ 

on by the molecular array m the nucleus, the gene number, kind and 
position, so the acquired individual information must be carried somati- 
cally, perhaps at synapses. But that experience fixation may go beyond the 
comfortable level of ordinary neural activity is nidicated by such pheno- 
mena as result from operative manipulation of organisms. The classical 
experiments of Weiss and of Sperry, for example, demonstrated that the 
discharge of impulses along motor nerves depends on the peripheral 
comiections ; if a supernumerary sartorius muscle is implanted in the back 
of a frog and neurotized by any nerve from the back or legs, the muscle 
w^ill come to contract simultaneously with the normal muscle of the same 
name. Some kind of micro-specification of the centres connected to the 
new muscle 'teaches' them to respond to the same central activity. More 
recently, the work of Thompson (1957, 1958) and McConnell ct al. 
(1955, 1959) has shown that learning occurs in the essentially non-neural 
tail of a planarian at least as well as in the ganglionated head. If an indivi- 
dual planarian is taught a conditioned response, cut in two, and the 
pieces allowed to regenerate, the new worm formed from the tail end 
performs almost perfectly as soon as put to the test. 

Turning at last specifically to the nervous system, the same interaction 
of heredity and environment is seen in full operation. Heredity gives the 
embryonic cells which will form neurones under their normal environ- 
ment, but become skin or lens under a different one; which will continue 
to multiply, or biturcate, their extensions sufficiently to satisfy fully the 
physiological field or need (Weiss, 1955); which will grow fibres in a 
direction guided by micellar structure and by chemical concentration 
(Cohen cr nl., 1954), to reach an appropriate end organ (as shown by 
various regeneration experiments), or the appropriate central neurone, 
despite operative mixing (Detwiler, 1936). But all these capacities are 
present only in the very young embryo; at each successive stage of 
development the potency shrinks further. Totipotent cells can later be- 
come only neurones, distorted patterns can no longer be corrected and, in 
general, new growth and chemical metabolism decrease m rate. Rate falls 
off' with life time, a basic ageing process, roughly as a decaying exponen- 
tial curve; growth and its special manifestation, learning, shrink in speed 
and scope with advancing years until plasticity is essentially lost. Then no 
new material trace is formed and experience is no longer fixed by the 
individual. 

Specifically for the nervous system, the following questions are impor- 
tant : what experience is retained ; under what conditions ; where does the 
change occur, is it local or diffuse; what is the nature of the change, 


26 BRAIN MECHANISMS AND LEARNING 

chemical, electrical, structural; and how does fixation occur, at the 
molecular and at the cellular levels? Some attention will be given to the 
nature and the locus of the material record, but most attention will be 
given to the mechanism of fixation. 

A neurone, at least in tissue culture, is a restless entity. It shows the usual 
swellings and churnings of other cells and its processes thrust out and 
retract pseudopodial branches and terminations unceasingly. The neurone 
has an unusually high rate of metabolism and, judging by the rate of 
regeneration and of peripheral flow in axons, a neurone may renew its 
entire mass of protoplasm three times daily. It is hard to see, therefore, 
how an enduring modification can be left at the cellular level. Experience 
must presumably alter some macromolecule, DNA or RNA or protein, 
which can continue to reproduce itself in the altered form, much as a 
mutated gene. But then other formidable problems arise. How docs 
neural activity, and the particular pattern of electric currents and attendant 
change in position and concentration of ions and polar molecules which 
activity engenders, lead to an altered array of nucleotides or amino-acid 
moieties in a macromolecule? How does such an altered complement of 
macromolecules in a neurone come, in turn, to modify its future physio- 
logical activity so as to give a new and appropriate pattern of discharges? 
How is the specificity of the molecular change related to the specific 
functional past and functional future in an adaptive fashion? Is a sort of 
natural selection process in a neurone population involved; if not, we 
again face the sort of problem raised by Lamarck. 

Whatever the answer at the molecular level, there are certainly morpho- 
logical changes with neural activity at the levels of organelle and of cell, 
and some of these endure for a relatively long time. The chromatolysis of 
fatigue, with diminished Nissl substance, swollen and rounded cytoplasm, 
and eccentric nucleus, has long been known. More recently, a change in 
microsome number and locus arouiid the nucleus has been shown to 
accompany changes in activity of neurones in tissue culture (Geiger, 1957). 
The apical dendrite of pyramidal neurones becomes thicker and more 
twisted with continuing activity. Nerve fibres swell when active (Hill, 
1950; Tc^bias, 1952), sprout additional branches, as seen in the spinal cord 
(McCouch ct (iL, 1958), and presumably increase the size and number of 
their terminal knobs. New fibre branches and ccMincctions, at least, might 
endure long enough to constitute a morphological engram. 

It is highly doubtful, both from the total number of bits remembered 
and from the survival of memories despite extensive brain lesions, that 
each remembered item is located at a particular neurone or synapse. Some 


R. W. GERARD 27 

localization is, of course, present, as shown by the aphasic defects with 
regional lesions and by particularized recollections inducec^ by local 
temporal stimulations. But even these are hardly cell by cell; and it is far 
more probable that large numbers of neurones, in assemblies or masses, in 
different patterns and other orderings, are involved in each memory. 
Partial engrams — of percepts, images and acts — are built into larger ones 
— concepts, imaginings and skills — much as a small assortment of amino- 
acids is used to build a limitless variety of proteins. In the same way, learn- 
ing goes from letters to words to sentences, with plateaux of achievement 
at each larger unit; and bits of information become aggregated into larger 
'chunks' so that a greater quantity can be handled in a given time (Miller, 
1956). Not only spacial relations, but also temporal ones must be proper, 
witness the great disturbances to thought and speech produced by delayed 
auditory feed-back. 

Whatever the micro-locus of the memory trace, most learning involves 
the cortex. Besides the evidence of cortical localization by stimulation and 
lesion, there is the further evidence of a general parallel between learning 
and memory capacity on the one hand and general cortical size on the 
other, and also there are the recent psycho-physiological experiments 
initiated by Sperry (1959). With the optic chiasm cut, so that incoming 
messages from each eye reach only the ipsilatcral hemisphere, the two 
hemispheres remain connected primarily through the corpus callosum. 
If conditioning is carried on with cMie eye, a correct response can be elicited 
through either eye so long as the callosum is intact; but after this is also 
cut, only the eye used in training can elicit the learned response. The 
engram, while available to both hemispheres when these remain anatomi- 
cally connected, is as clearly localized in only one. Comparable tmdings 
have been made on learning set or learning to learn. Other evidence for a 
cortical engram, and one which acts as a template for further engram 
development, comes from the work of Meyer (1958). Removal of both 
occipital lobes, but with a two-week interval between ablations, leaves a 
rat with pattern vision essentially intact if ordinary visual experience is 
possible during the interval; but if the animal experiences no pattern 
vision between the removal of the hrst lobe and that of the second, a 
pernianent loss of pattern vision results. The endurance for months of 
figural after-effects in the Kohler's satiation-illusion (Wertheimer and 
Leventhal, 1958) further indicates a cortical locus of the engram. 

Not all fixation, however, is in the cortex. Although ordinary condi- 
tioning of the spinal animal remains highly dubious, there is solid evidence 
of card changes produced in the intact animal. If one cerebellar hemisphere 


28 BRAIN MECHANISMS AND LEARNING 

or peduncle is cut a few hours prior to a high spinal section, an enduring 
postural asymmetry remains in the hind quarters (de Giorgio, 1943). This 
is reminiscent of the continued unidirectional circling by a decorticate 
dog, the direction depending upon which hemisphere was first removed. 
A comparable postural asymmetry in the spinal cat is seen long after a 
severe inflammatory lesion is produced in one paw; the previously 
damaged leg is pulled into sharp flexion, with crossed extension of the 
other, when a spinal section is made (Sperensky, 1944)- Clinical experience 
with trigger points, and the demonstration that anginal and other pains 
can develop a permanent referral to another bociy region which has been 
irritated at the time the pain occurs, points in the same direction (see 
Gerard, 1951). 

The growing evidence of a relation of deep forebrain structures to 
learning and recall has not yet crystallized. The amygdala seems to exert 
an adverse influence on fixation, and the reticular formation and hypo- 
thalamus have also been found to be involved. Conversely, recent memory 
fails with damage to the mammilary body, the fornix, and the amygdala 
(Morrell, 1956; Bickford ct ah, i9.vS; Jasper and Rasmussen, 1958; 
Samuels, 1959; Doty, 1960; Gerard, 1960a). Recently, the stimulation of 
the midbrain tegmentum has been reported (Thompson, 1958; Glickman, 
1958) to cause forgetting; while stimulation of the caudate impairs fixa- 
tion. The whole situation is confused by difliculties in distinguishing 
between fixation on the one hand and retention and recall on the other. 
The use of hypnosis as a tool to maximize recall has long given unexpected 
results; and recent reports by responsible workers — such as the hypnotic 
recall of a conversation which occurred during surgical anaesthesia of the 
individual doing the recalling; or recall of experiences during a given year 
of childhood but only when an hypnotic suggestion had brought the 
individual back to that age (Sheerer and Reift, 1959) — demonstrates our 
small understanding of the complex phenomena of fixation and recall. 

The mechanism of fixation of experience is not known, but two sets of 
data give strong clues concerning its nature. First, is a group of well- 
known changes that attend continued activity of neurones: thresholds go 
through increased and decreased phases; after-potentials are increased 
enormously in magnitude and duration; post-tetantic potentiation is 
associated with greatly increased reflex responses ; and the like. The other 
phenomena relate to the existence c^f a considerable period, of minutes to 
hours, between having an experience and fixing it. If neural activity is 
interfered with during this fixation period, by electroshock, by cold, 
by heat-block, and the like, fixation is interfered with and permanent 


R. W. GERARD 29 

memories are feeble or abolished (Duncan, 194.S ; Gerard, [953 ; Ransmeier, 
1954; Leukcl, 1957; Otis and Cerf, 1959). Thus, hamsters or rats trained in 
grouped runs on some learning situation — a maze or an avoidance 
conditioning or the like — show a normal learning curve if a convulsive 
electroshock is delivered after each set of runs with an interval of 4 hours 
or more. It the shocks follow the experience by an hour there is some 
deterioration, at 1 5 minutes loss is very considerable, and at 5 minutes or 
less learning is in effect prevented. If a hamster is promptly cooled after 
the learning experience, a shock given an hour later can be as deleterious 
as one given a few minutes later at body temperature; the Q^^ of the 
fixation process has been thus determined at nearly three (Ransmeier and 
Gerard, 1954). 

Anoxia acts much like electroshock, and the two sum their effects 
(Ransmeier and Gerard, 1954; Thompson, 1957). A number of drugs has 
now been tested for influences on the electroshock effect. Reserpine, like 
anoxia, potentiates the ECS disruption of learning (Weyner and Reimonis, 
1959); ether protects against electroshock effects (Seigel ct al, 1949; 
Potter and Stone, 1947); and, in man, meprobamate decreases the con- 
tusion produced following electroshock (Thai, 1956). Miss Rabc and I 
(1959) have just completed studies of the action of phenobarbital and of 
meprobamate on ECS action in rats on an avoidance conditioning test. In 
effect, phenobarbital slows and prolongs the fixation time, as judged by 
the greater disruptive effects of a convulsive stimulus at i, 2, 5 and 15 
minutes in animals under the barbiturate as compared with undrugged 
ones; while meprobamate, contrary to expectation, seems to have the 
reverse effect. Interestingly enough, meprobamate, but not phenobarbital, 
slows the learning process, aside from any ECS. Strychnine, according to 
an informal communication from Dr Krech, shortens the fixation time; a 
convulsive shock at a given time interval is less disruptive in the strychnized 
rat than in one without the drug. 

The above facts fit well into a theory of continued activity in the nervous 
system, following the arrival of sensory impulses, in the course of which a 
dynamic memory is fixed as a structural one (Gerard, 1960a). Summation, 
irradiation and reverberation of messages would lead to repeatec^ activity 
of the same neurones, with progressively greater and greater residual 
changes from the continued activity. At 50 reverberations per second, 
100,000 actions could easily occur during the fixation time; presumably 
sufficient to leave an indelible trace. 

Any change that would enhance the extent or intensity of reverberation 
should hasten the fixation process; any agent acting in the converse 


30 BRAIN MECHANISMS AND LEARNING 

direction, should slow it. A fall in threshold of cortical neurones, or an in- 
crease in impulse bc:)mbardment, should hasten tixation. Since epinephrine 
lowers thresholds, and is released in vivid emotional experiences, such an in- 
tense adventure should be highly menaorablc. Perhaps during imprinting 
periods, the relevant neurones arc similarly in a low threshold or 'soft- 
shell' stage. Strychnine, by enhancing general activity level and lowering 
thresholds, should also speed up the fixation process. The reticular forma- 
tion, the amygdala and other components of the limbic system, the hypo- 
thalamus, etc., act through one another or directly on cortical neurones to 
alter their dendrite potentials, their thresholds, and their responsiveness to 
the discrete impulses which reach the cortex by non-diffuse afterents or by 
spread within it. By raising or lowering thresholds of cortical neurones, 
associated with lov/ered or increased attention and vigilance, these deep 
regions could easily modify the ease and completeness of experience 
fixation even if the nuclei were not themselves loci of engrams. Ether and 
meprobamate, by raising thresholds, should slow reverberation and 
therefore make electroshock effects more severe; unless they also 
decrease the effect of the shock itself due to the increased neurone thresholds. 
This last possibility is being further tested in the case of meprobamate; it 
does explain the synergy of rescrpine and shock, and of anoxia and shock, 
since both anoxia and rescrpine lower convulsive thresholds. 

Fixation would also be modified by changing impulse bombardment; 
in fact it is the interference with such continued bombardment, by shock 
or cold or concussion — which latter rather nicely parallels clinically the 
amnesic effects of electroshock m animals or man — that interrupts the 
fixatic^n process. Early in any learning experience, as the organism's 
actions fail to solve the problem and eliminate the disturbing input, there 
is great central irradiation; muscle tension is increased, there is generalized 
contraction of irrelevant as well as of the desired muscles, autonomic dis- 
charges occur, tension and attention are intense, the 'consciousness of 
necessity' is high, and many neurones m cortex and deep centres show 
electrical activity on mass or micro-electrode recording. Later, when 
learned responses have been established, general radiation disappears, 
muscles relax, there is little tension and maybe not even attention, 
habituation is evident in performance and experience, and electric activity 
has disappeared from all neurones except those specifically involved in the 
response. With errors or other kinds of emergency situations, the electrical 
activity and other signs of irradiation promptly return (see Gerard, 1960a 
for references). As an action is learned and certain paths through the 
nervous system become canalized, irradiation is eliminated. This requires 


R. W. GERARD 3 1 

cither feed-back inhibitory arcs or a decrease in the general level of 
facilitation, as interneurones become less active. This latter would follow 
the diminished irradiation consequent to effective responses. 

Although the effect of continuing or recurring activity has been 
described in terms of simple feed-back loops, the same results can be 
obtained by improved synchrony of beating neurones or, especially, by 
repeated waves of activity passing through a sheet of neurones. This last 
model, developed by Beurle (1957), depends only on more or less random 
connections of neurones in a mass, activated only fractionally by con- 
trolled waves passing through them. Such waves can cross and at the 
locus o( intersection will leave a group ot lowered-threshold neurones. 
From such a locus the original waves can reinitiate without external 
stimulation; such loci thus of^er engrams for memory, recall, planning of 
action and for the combination of smaller elements of perception and 
action into larger wholes of conception and of planning. The model and 
the physiological support for and predictions from it are more fully 
discussed elsewhere (Gerard, 1960a). 

Reverberation, or some form of continuing reactivation, is probably 
involved in a number of other mental processes, including perception, 
attention, repression, anxiety, and the like; but this is not the place to 
develop these points. Certainly attention is necessary ior learning and for 
discrimination. A dog, faced with an impossible oval or circle choice, 
develops a neurosis only after it has already learned to pay attention to this 
discrimination as a problem. Presumably the structures feeding diffuse 
system impulses to the cortex, discussed above in connection with the 
altering of fixation and recall, are involved in such influence of attention 
on the learning process. 

The nature of the material change in the brain, of its locus, and of the 
mechanism of fixation that brings it about are not established, but all these 
questions are clearly on the way to satisfactory solution. The really diffi- 
cult problem is none of these, however, and it remains as mysterious as 
ever. This has to do with spccificiry in the selection and discrimination of 
what is perceived, in the degree of attention given it, in the presence and 
firmness of fixation of a memory, in its retention and subsequent recall to 
consciousness. This qualitative aspect is, of course, not limited to the 
receiving and retention of experience, but is equally present on the 
behavioural side and is attached to plans, to values and to actual per- 
formance. How choices are made, how priorities are assigned, how shifts 
occur from one set of active neural processes to a different one, remains 
scill a complete mystery. 


32 BRAIN MECHANISMS AND LEARNING 

Certainly this is related to our subjective experience of free will in 
choosing what we do and what we attend to. But before accepting un- 
caused causes, let us recall that computers can also learn to develop a set of 
values and to choose between programmed activities, so that we need not 
yet despair of finding the neural mechanisms for this ultimate core of 
higher behaviour. Given a programme or plan, a computer can rapidly 
scan the existing situation and select that set of actions which will come 
closest to matching an 'ideal' that has been given it. Behaviour is similarly 
tracking; as individual or as race, organisms respond to the challenge of 
their environment. Behaviour, as remarked earlier, is such as to bring the 
expected future condition of the organism into congruence with the 
desired condition. This can be done blindly, starting from the existent 
state, or it can be done with greater or lesser foresight bv projecting the 
existing state into its probable future condition. 

The nervous system and learning endow the animals possessing them 
with this ability to extrapolate the curve of existence and so to act with 
foresight. The cat jumps not to where the mouse is, but to where it may 
be expected to be. Man not only can himself run away from the batter in 
order to catch a flyball; he can also build into computing tracking 
mechanisms, as anti-aircraft guns, the same ability. Man himself, with his 
ten billion or more cortical neurones', can out-perform all other projecient 
machines, so far built by nature or by him, in projecting further and more 
elaborately into the future. This is possible because of a great repertoire of 
past memories, or partially organized and interacting programmes, and of 
complex probability computers that take account of past and present 
factors and assign value or utile weights to them. The basic molecular, 
cellular and organismic mechanisms involved, however, are probably not 
much different from frog to physiologist. It is an intriguing question 
whether equally effective ones will one day be evolved for man's machines. 

GROUP DISCUSSION 

Hebb. I would like to raise two points. One is with respect to the work or 
McConnell on the flatworm which I found extremely puzzling. I do not know 
whether it can be compared to learning in mammals. If this type of mechanism 
applied in the mammal, presumably Sperrv's work should not have given the result 
it did by just cutting the corpus callosum. We must be dealing with a different 
phenomenon, when we talk of learning in the mammals, from the phenomenon 
that McConnell studied. Secondly, with respect to the effects of anaesthetics on the 
consolidation period of learning, I would like to cite some work by Muriel Stern, 
concerning the effect of barbiturates on learning. The effect you observed may be a 
general disturbance, rather than a specific interference with the retention period. 
The effect of barbiturates is that of depressing learning capacity for a period of some 


R. W. GERARD 33 

weeks 111 the laboratory rat. The effect disappears in 5 or 6 weeks. The effect is not 
found with other anaesthesia. 

Gerard. I am not prepared to go along with the extreme psychiatrists who say 
that learning is all over the body and not in the brain, nor with the slogan that 
'we think with our blood'. We think in the nervous system and in the upper part 
of it, but there is continuity. The same kind of mechanisms, which are specialized 
in the nervous system as a basis tor recording experience, probably evolved from 
basic cellular processes, which are seen at the chemical level, hi the case of the 
flatworm this is closer to the level ot total behaviour. As to the second point, the 
action of these drugs on the fixation process and on learning processes is very 
complicated and depends on many parameters. With increasing doses of mepro- 
bamate, one finds an increasing interference with the learning curve. These experi- 
ments were on rats with an avoidance conditioning, given massed trials and tested 
the next day. The drug, when given, was injected i hour for meprobamatc, or 
3 hours for the barbital (depending on maximum time of action), before the learn- 
ing experience. Rats were given electroconvulsive shocks at i, 2, 5, or 15 minutes 
after the mass runs, which were completed usually well within 15 minutes total 
time. Although meprobamate definitely interfered with learning, it did not inter- 
fere with the hxation; it even improved it. hi other words, an electric shock at one 
minute did not interfere so much with retention on second test of a rat with, as 
for one without, meprobamate. We are checking to be certain that a raised threshold 
is not a fictor, even though all shocks gave typical convulsions. The phenobarbital 
had a marked effect in prolonging fixation time; as shown by much deterioration 
with electroshock, and the greatest deterioration with the closest shock. It did not 
interfere with initial learning. Your findings seem contrary to what happened in 
our laboratory; we should examine differences in conditions. 

Chow. May I just make a point about this ffatworm work, because that work 
seems to me to be most striking. If it is true it will probably inffuence our thinking 
about learning at the cellular level. There are two points I wish to make about this 
particular piece of work, the results of which I think should be accepted with some 
caution. First, there are probably nerve cells in the nerve tract, therefore you have 
a second or caudal half which still remembers the problem, which may not be due 
to memory in the nerve fibre itself. Secondly, a very important point is that 
apparently an essential control was not made. This control is : they should have a 
group of animals which are given a series of shocks without learning, and cut them 
into two, wait until they regenerate and do the conditioning, and see if this group 
without previous training, also show a fast learning. 

Gerard. These controls have now been done. I talked to McConnell some 
months ago and, as I recall, results were going in the proper direction. Even with- 
out them, however, it seems hard to explain awa\' the fict that, even if there are a 
few neurones in the tail as compared to the many in the head, the tail did do as well 
as, or better than, the head. Maybe if one gets too much stuff"in the head, ideas get 
fixed in a certain way and performances are no longer malleable; a sort of ageing 
phenomenon. The work will certainly have to be examined carefully, because it is 
a very disturbing finding. However, the appearance of certain phenomena at this 
primitive level does not necessarily explain those at a more complicated level. It 
may well be true that this is a perfectly good type of learning, but it is certainly not 
the whole story of our learning. Let me give one example of this: Many years ago 


34 BRAIN MECHANISMS AND LEARNING 

Dr. Libet and I found that travelling slow waves in the hemisphere ot the frog, 
starting at the olfactorv bulb, would cross a complete section through the brain if 
this was made with a sliarp razor blade and if the two halves were accurately placed 
together. (I do not think neurohumoural agents were involved). Several investi- 
aators checking in mammals found the spread of epileptic activity was blocked by 
cutting. Aside from the fact that they were not able to get anything like an 
equivalent opposition, there was no necessity that this be the same phenomenon. 

AsRATYAN. I would like to ask a question about the same experiment. For how 
long is It possible to evoke these responses from the head part and the caudal part ? 
How long does this habit remain? Is it possible to elaborate reflexes on the head 
part and on the caudal part of the animal after cutting the animal in two and what 
is the difference m the rate of elaboration and of appearance of learning? 

Gerard. I do not know the answer to the second question. As to the first, it 
takes about 3 weeks for regeneration and, as I recall, the test was made a month 
after cutting. Uncut planaria also retained the learning for such a period. 

BusER. Did the author exclude the possibility of peripheral network as happens 
in many invertebrates, as being hypodermic and distinct from the C.N.S. It could 
be a diffuse system. 1 • j 

Gerard. I can only answer that work is now in progress with a zoologist and 
histological tests are being made more extensively. This is perhaps the widest loop- 
hole; whether there are as yet unrecognized neurones in the tail part of the animal. 

RosvOLD. I was interested in your statement that there is no necessary reason why 
what happened m planaria need concern us with respect to man's brain or monkey's 

brain ... , i 1 • 1 ■ 

Gerard. May I correct that? I would not say it does not concern us; 1 think it 
concerns us enormously or I would not have mentioned it. But I do not think it is 
the whole story. It is part of the story, the substrate on which other mechanisms 

are built. , 

RosvoLD. From a comparative point of view, however, where can one draw tiie 
line? How close can one make the analysis? If one takes your statement literally 
could not one always, so to speak, wriggle out of an inconsistency? Simply statmg 
that the comparison is too remote and, therefore, is hardly appropriate. 

Gerard. I am not really clear as to the question. Is this responsive : I have become 
a behavioural scientist lately and am m an Institute which devotes itself to be- 
havioural science. We are concerned with finding the constants in the properties of 
systems at all levels from molecules to societies. We are often attacked for dealing 
with meaningless analogies because the details are different in each. I suggest that it 
is extremely Important to be aware both of the particularities wliich differentiate 
the one from the other and of the general likeness. When I say, as I do, that an idea 
in society is in many ways like a gene in biology, and that this is the new disturbmg 
element which makes social evolution possible as a new gene makes biological 
evolution possible, I do not mean that male and female ideas mate and that there 
are mitoses and all the rest of it. This would be arrant nonsense. But it is still useful 
to see some of the likenesses. If one can in fact get something, assummg that this 
holds up, which, objectivelv, we would have to call learning, in non-neural cells, 
even though they have undergone division to form a new nervous system, this is 
something^vhich I should expect to apply in any comparable situation that could 
be found In man. Indeed, the reason I brought up these findings is the high prob- 


R. W. GERARD 35 

ability that the engram in neurones, which themselves undergo 'regeneration' by 
changing their material three times a da\', might also be something at a molecular 
level. Whatever the mechanism, in either case, a new kind ot template is formed, 
new molecules are reproduced according to the template, and this is the way 
fixation ot experience or learning is carried on in the nervous system. 

Thorpe. With regard to this matter ot the planaria there have been a number of 
examples of what have been called pre-imaginal conditioning in insects where the 
larva was exposed to an intluence which produced the corresponding modification 
of behaviour in the adult. This was first done with regard to exposure to chemical 
substances, but there was also a series of experiments reported by Borell du Vernay 
(1942) in which the larvae of the beetle Tciiehrio iiwliter were trained to take a 
certain path in a 'T' maze and the same effect was observed significantly in the adults. 

Gerard. To what extent is the nervous system discontinuous? 

Thorpe: A steady change takes place. Although the locomotory systems of the 
larva and the adult are very different there would certainly have been organized 
nervous tissue present throughout. So, though we are not here dealing with a 
complete replacement of nervous tissue, I think nevertheless that these experiments 
are interesting to recall in connection with the subject of our present discussion. 

Might I ask you about one other point? A very different one. You referred to 
this matter of fixation and recall and the distinction between it. I was very puzzled 
some years ago by some work that Dr McCulloch reported, on the ability of some 
patients to recall, when under hypnosis, extraordinary details of their own early 
actions and experiences. One particular case I remember concerned a bricklayer 
who was able to recall minute details of the bricks he had laid at a certain date of his 
life. I tried to get the exact particulars ot this trom Dr McCulloch at the time but 
he was not able to give anytliing like the detail which I hoped would be torth- 
coming. I mentioned it in the hope that other people here, such as Dr Hebb, 
might have some fresh data to give us on this particularly important topic. Only 
last year Wilder Penfield {Pwc. Nat. Acad. Sci., 44) stated his belief, based on 25 
years experience ot surgical treatment of brains of epileptics, that the temporal 
lobes possess a permanent record of the stream of consciousness in amazing detail 
— as if the brain cells behaved like a tape recorder which may be played back by a 
suitable electrical stimulus. 

Gerard. I have also been intrigued by that report. As I recall, a bricklayer in his 
sixties or seventies was asked to describe, say, the fifth brick in the seventh tier of a 
wall that he had laid in his twenties. The bumps on the top and the bottom of the 
brick he described under hypnosis could be checked and they were there. It is a 
dramatic statement and I have referred to it in a paper. I do hope McCulloch is 
correct, for he is also my source. 

Another example, given informallv, at a symposium in Kansas City along with 
the statement about recall of events under deep surgical hypnosis, frightened me 
even more, because it makes me think ot Bridie Murphy. An adult was asked to 
recall some details of a classroom, desks, etc., in which he sat when 6 years old. He 
was quite unable, even under hypnosis, to recall any of these things until the 
suggestion had been made: 'You are now 6 years old'. Then he described them. 
When he was told he was 7 years old, he couldn't do so. I can only say that a 
thoroughly reputable psychologist told us this with fear and trembling, but he 
could not find anything wrong with his experiments. 

D 


DISTINCTIVE FEATURES OF LEARNING IN THE 
HIGHER ANIMAL 

D. O. Hebb^ 

In the study of learning our problem is not only to understand the way 
in which synaptic function is modified; it also fundamentally involves 
control of the concurrent activity of the rest of the central nervous system. 
One question is more 'molecular' and physiological, essentially concerned 
with the relations between two individual neurones ; the other is 'molar' 
and psychological, involving the operations of the total system. The 
second part of my thesis is that we must develop means of dealing experi- 
mentally with the so-called autonomous central processes whose activity 
is at the heart of the learning process in the higher animal, and some 
research is reported which attempts to make a contribution of this kind. 

THE PROBLEM OF EXCESS ACTIVITY IN CNS 

Learning consists of a modified direction of transmission in the CNS so 
that, in the clearest example, a sensory excitation is now conducted to 
effectors to which it was not conducted before. A new S-R or stimulus- 
response connection has been established (the clearest example, but not the 
only form of learning). The term learning may be used to refer to other 
changes of behaviour in primitive animals, but at least in the case of the 
mammal's acquisition of prompt, efticicnt responses, dependent on the 
all-or-none action of neural cells conducting over considerable distances, 
the direction of transmission must be determined at the synapse (or close 
to it: Milner, 1959). It might therefore be thought that the problem of 
learning is simply to discover when and how synaptic function is modified. 

The question is indeed fundamental, and must be answered before the 
problem of learning will be solved; but it is by no means the whole 
problem, because of complexities introduced by the structure of the CNS 
in higher animals, and particularly in mammals and in primates. In the 
learning process we do not have a one-to-one relation between progres- 
sive changes of behaviour and changes at the synapse. The simplest 

^ The presentation of this paper, and some of the experimental work reported herein, was 
made possible by the National Research Council of Canada. The experiments were done by 
Jean Campbell, Joseph Deitcher and Eric Renncrt. 

37 


38 BRAIN MECHANISMS AND LEARNING 

mammalian behaviour involves an enormous number of synapses, the 
changes that occur simultaneously may be in opposite directions, and 
there may be little detectable relation between the course of events at any 
one synapse and changes of overt behaviour —just as there may be little 
relation between the activity of an individual neurone and the gross record 
of the EEC 

If we assume that many of the neurones in the brain of a waking 
mammal arc tning at any given time, and especially if we further assume 
that some of these are inhibitory, there are certain consequences which 
have sometimes been overlooked in physiological discussions of learning. 
The possession of a large brain capable of learning a great many different 
things inevitably nieans that there are far more neurones present than is 
necessary for learning some one specific task. Any random activity in 
these excess neurones (the ones not needed for the task being learned) is 
'noise', which must tend to interfere with the learning. (If the activity is 
organized instead of random it is not noise, technically speaking, but the 
effect may still tend to be adverse.) It seems obvious that the number of 
excess neurones must be very much greater — perhaps thousands of times 
greater in the brain of the higher animal — than those needed for the 
learning going on at the moment. It therefore seems that the rate of 
learning, as observed in the behaviour of the whole animal, may be not 
an index of capacity tor adding new synaptic connections so much as an 
index of the noise level, and that learning will be fast or slow according as 
one is successful in establishing an environmental control of the excess 
neural activity, in order to prevent or niinimize interference. 

Practically, the significance of this point of view is clear in the term 'lack 
of concentration', in the case of the student whose learning is inefficient 
even in a quiet environment, because other thoughts obtrude besides the 
ones he should be concerned with. Experimentally, the point is made by 
Ricci, Doane andjasper (1957), in reporting that a significant part of the 
conditioning process — perhaps the main part — is in the dropping out of 
irrelevant connections, rather than the acquisition of new ones. It is 
clearest of all in a classic experiment of Yerkes. 

Yerkes (19 12) trained an earthworm to choose one arm of a T-maze, 
using electric shock as punishment for error and the moist burrow as 
reward for correct choice. The habit was acquired in twenty trials, 2 days 
at ten trials per day, about what might be necessary for the laboratory rat. 
No errors were made on the third day, though behaviour was somewhat 
inconstant in the following week as between good days and bad days 
(even worms have them). Yerkes then removed the brain, or principal 


D. O. HEBB 39 

ganglia, by cutting oft the head — the anterior four and a halt segments. 
The animal continued to respond correctly, showing that there were 
suiTicient synaptic modifications in the remaining ganglia to mediate the 
response — until the new head regenerated, at which time the habit was 
lost. The noise generated by the new ganglia, the irrelevant neural activity 
of the uneducated brain, was sufficient to disrupt learning completely. 

In this case we are dealing with the effects of noise on an established 
habit, and in a lower animal. It seems clear that the potential disrupting 
effects must be as great or greater while learning is still going on, and in 
the large brain of the higher animal. If then the rate of learning reflects the 
noise level in the system, it becomes intelligible that the higher animal 
does not learn faster than the lower, when each is given a task to which it 
is constitutionally adapted. The rate of learning per sc is not an index of 
intelligence or level in the phylctic scale (Lashley, 1929); we may note, 
for example, the occurrence of one-trial learning, by inspection only, in 
the solitary wasp (Baercnds, and Tinbcrgen and Kruyt, cited by Tinber- 
gen, 195 1). There is of course no reason why modifications of the individual 
synapse should be made more quickly in a system with many synapses 
than in one with few. When receptors, effectors and intervening neural 
structures are adapted to a small number of acquired responses, and there 
is little or no irrelevant concurrent activity, the necessary synaptic 
modification may occur very rapidly. Much that is considered to be 
innate, because there is little evidence of practice, may in fact depend on 
immediate learning at first exposure to the stimulating conditions. 

The problem of excess activity or noise in the larger brain is character- 
istic of those regions of divergent (rather than parallel) conduction in 
which learning is supposed to occur. If the experimental stimulation is to 
result in a modification of function at the appropriate synapses, the 
excitation produced by the stimulus must be conducted to those particular 
synapses. From the sensory surface to the cortical projection area there is 
no difficulty; here conduction is in parallel and the same population of 
units can be reliably excited, time after time, since the units which begin 
together (and thus are simultaneously excited) end together, and reinforce 
one another's action at the next synaptic level (Hebb, 1958). But this 
condition does not hold for the further cortical transmission, especially 
where the mass ratio of association to sensory cortex is large (the A/S ratio: 
Hebb, 1949). A relatively small sensory projection area cannot dominate a 
large region of divergent conduction, maintaining on successive trials the 
same conditions of transmission with respect to single units. This applies 
to the large thalamo-cortical sectors which comprise the so-called 


40 BRAIN MECHANISMS AND LEARNING 

association areas, and also to other rhinencephalic and basal-gangliar regions 
which seem to be involved in higher mental processes (cf. the recent 
review by Rosvold, 1959). Whether a given synaptic junction will be 
activated in such regions of divergent conduction cannot be determined 
by control of the stimulating environment, the experimental situation in 
which the learning is to be established. It is fully dependent also on 
whether the units concerned are ready to be fired, and on the concurrent 
activity of other units, not controlled by the present sensory stimulation, 
which impinge on the synapse in question and which may produce either 
summation or inhibition. Cumulative learning, in which the second trial 
adds to a synaptic modification begun by the first, thus depends on what 
activity is already going on in these regions. 

Here is a point at which we tmd a full convergence of the physiological 
with the psychological evidence. Psychologically, the problem is that of 
'set', 'attention', 'motivation', 'attitude', or the like: all terms developed 
in an earlier day to refer to (i) the puzzling but unmistakable deviation of 
behaviour from control by environmental stimulation, and (2) deviations 
of learning from the simple S-R and CR conceptions, implying a direct, 
through connection which, as we have seen, cannot be expected to occur 
in regions of divergent conduction, though it was taken for granted by 
earlier physiologists as well as psychologists. 

Even today, the problems involved here have not always been faced, 
perhaps because of the complications which they entail. The histological 
structure of the tissue in question appears to mean that transmission is via a 
series of the closed pathways described by Lorente dc No (1943), or 
groups of them functioning as systems capable of self-maintained activity. 
Such systems are what I have called 'cell-assemblies' (Hebb, 1949), and 
Lashley (1958) 'trace systems'; the most adequate discussion of their pos- 
sible mode of development and internal function is that of Milner (1957). 
Such conceptions certainly introduce complexities into behavioural 
theory, but on the other hand the behaviour itself is even more complex, 
and means of experimental analysis arc becoming available. Burns (1958) 
for example has provided us with much information about the conditions 
in which self-maintained activity is possible in a slab ot isolated cortex, 
from a physiological approach ; and such studies of perception as those of 
Broadbcnt (1956) with respect to hearing, and Heron (1957), Kimura 
(1959) and Bryden (1958) with respect to vision, have begun to transform 
a very speculative class of theory into something more solidly grounded 
in fact and susceptible of direct experimental attack. The following experi- 
mental work attempts to carry this process further. 


D. O. HEBB 41 

THE NATURE OF THE TRACE IN SHORT-TERM MEMORY 

In principle, we may distinguish two ways in which a memory trace 
can be estabhshed (cf. e.g. Eccles, 1953). One is a kind of after-discharge, a 
reverberatory activity set up without necessarily depending on any 
change in the units involved, other than the discharge of impulses; and 
one consists of some change in the units which outlasts their period of 
activity. The first can be called for convenience an actii'iry trace, the second 
a stniaiiral trace. 

In discussing this point earlier (Hebb, 1949) I assumed that the repetition 
of digits in a test of memory span provides a pure example oi the activity 
trace. Here the experimenter presents verbally a series of digits, and the 
subject is asked to reproduce them in the same order. After the subject has 
attempted one series, the experimenter presents a second series and the 
subject seems to forget the preceding series completely. He docs not get 
the two mixed up just as, in a calculating machine, punching a second set 
of numbers wipes out the preceding set completely. But is this what 
happens? Is there no lasting after-effect: no structural change produced by 
hearing and repeating the digits? 

With this question in mind, and with the intention of learning more 
about the nature of the trace, in short-term memories for highly familiar 
material, the following experiments were carried out. 

Method: the subjects were college students, each tested individually. 
They were informed that the purpose of the experiment was to see 
whether the memory span for digits would improve with practice. 
Twenty-four series of digits were presented. On each of these trials, the 
experimenter read aloud a series of nine digits at the rate of about one per 
second. The subject was instructed to listen carefully and repeat the digits 
in exactly the same order. 

Each series consisted of the digits from i to 9, in varying order, each 
digit occurring once only. An example is 

591437826. 

There was, however, one special feature about which the subject was not 
informed. On every third trial (3rd, 6th, 9th ... 24th) the same series was 
repeated, and the object of the experiment was to discover what effect this 
had. Would the repetition result in learning? If immediate memory for 
digits in these circumstances is mediated solely by an activity trace, with 
no structural changes, each new stimulus-event would be expected to 
wipe the slate clean and set up a new pattern of activities. No cumulative 
learning would occur. If such learning occurs, however, we may conclude 


42 


BRAIN MECHANISMS AND LEARNING 


that there is some structural modification, in the sense defined above (in 
addition to ■whatever activity trace there may be). 


25 


^^ 


ZQ 


/ 


Repeat tcL Series 


^ 


ys 


/ 


\ 


\^ 


"^ 


10 
5 


1 


\ 
\ 

?■ 

— 1 — , 


/ 


/ ^ ^ 
— \ — . — . — 1 — 1 — , — 1— 


•- 


/ 

' — I — . 


— . — 1 — . — 1 — 1 — I — 1 — \ — 


Fig. I 


/5 


I? 


II 


Zf 


Number of subjects (out of forty) successfully repeating nine digits on each of twenty-four 
trials, when every third set of digits is the same ('repeated series') ; non-correction method. 

The results show clearly that cumulative learning occurs. Fig. i illustrates 
the data for one procedure. Here the subject's errors are not corrected, and 
the record is made in terms of the number of trials in wliich the nine digits 


3.5 


-I — I — I — I- 


IZ 

Fig. 2 


/5 


z/ 


Z¥ 


Mean number of digits correctly repeated on each of twenty-four trials, by twenty-five sub- 
jects, when every third set of digits is the same ('repeated series'); correction method. 

are repeated without error. Fig. 2 shows the similar result that is obtained 
when the subject is stopped as soon as he gives one digit out of the proper 


D. O. HEBB 43 

order (thus providing 'negative reinforcement', or 'punishment' for 
error) ; here the record is in terms of the number of digits correctly 
repeated on each trial. In both cases it is clear that learning occurs. 

This can be seen in another v^ay. The forty subjects v^hose results are 
diagrammed in Fig. i were each asked, following the nineteenth trial, 
whether they had 'noticed anything unusual' about the procedure. 
Twenty-two reported that there had been some repetition in the series 
presented to them, and three of these could give the crucial series without 
error. If the subject did not volunteer anything, he was asked explicitly 
about the repetition; three more subjects reported that they had observed 
it, and one of them could repeat the crucial series correctly. The remaining 
fifteen subjects had not observed the repetition. (The questioning was 
done between trials 19 and 20 in order not to direct attention to the 
crucial scries, occurring on trials 18 and 21; none the less, the questions 
may have affected the subsequent performance on trials 21 and 24, which 
can be seen in Fig. i to show a further sharp improvement.) 

The implications of this rather simple-minded experiment are more 
extensive than may be apparent at first. With such results, I can find no 
way of avoiding the conclusion that a single repetition of a set of digits, 
with or without the reinforcement of being told when an error has been 
made, produces a structural trace which can be cumulative. I assume that 
an activity trace may also be involved in the actual repetition, but it is the 
structural change which is of interest here. 

It is important to note that we are dealing with highly practised material. 
Associative connections already exist between any two digits, for the 
educated subject especially. In addition to the very highly practiced 
sequence 1-2-3-4 ..., the learning of historical dates, telephone numbers, 
street addresses, quantitative values such as the speed of light or the number 
of feet in a mile, and the batting averages of the Boston Red Sox in 1937 
— all these varied uses of the nine digits mean that the subject has already 
learned many sequences, in one or other of which any digit is followed by 
any other digit. When he is given a specific series to repeat, the memory 
for that scries must depend somehow on a further strengthening of the 
connections already established. This is diagrammed in Fig. 3, where for 
simplicity the trace systems or cell-assemblies of three numbers only are 
represented. 1 has a strong connection with 2, and 2 with j (because of the 
frequency with which the sequence 1-2-3 ■•• has been repeated in the 
past) ; but 2 has connections with / as well as with j, and j with _' and i as 
well as with 4 (not shown). 

Now let us suppose that the subject is given the sequence 3-2-1 ... to 


44 


BRAIN MECHANISMS AND LEARNING 


repeat. Some change occurs which means that when the experimenter 
stops speaking, the corresponding trace systems fire, and in the proper 
order. Undoubtedly there are complexities which this does not take 
account of; there is certainly not a mere chaining of the systems involved, 
because — for example — the subject may be able to tell you what the 
first and last of a group of digits were, though he has lost the intervening 
ones. Memory for the individual item in the scries is not entirely depen- 
dent on the preceding item. But the order of repetition seems to require a 
changed synaptic relation between the specific cell-assembly groups, or 
trace systems, so that in the example of Fig. 3, j tires 2, and not 4 or the 


Fig, 3 
Diagrammatic representation of the synaptic 
relations involved in the repetition of the 
digits 3-2-1. The heavier arrows 1-2 and 2-3 
represent more strongly established connec- 
tions ; the encircled .v indicates the temporary 
strengthening of less strongly established 
ones. 


central representative of some other number; and 2 fires 1 and not j. The 
synaptic 'strengthening' implied is shown by the encircled .v in the figure. 
What can this be? 

The difficulty here is that all the synapses concerned must be already 
asymptotic, in the development of their structural connections. Synapses 
connecting system 2 with other systems are highly developed; if now 
when 2 fires it is to fire 1 reliably, and not 4 or 6 or some other system, it 
must be sii^^iifiavitly more closely related to 1 for the moment, and it seems 
most unlikely that this closer relation can consist of a sudden further 
development of the size of the knobs of the systeiii-2 — system- 1 synapses. 
Some other mechanism besides growth of the knob, or its closer juxta- 
position with the cell body, must be involved. 

The mechanism could well be that recently proposed by Milner (i959)- 
He has pointed out that failure of conduction along some axon fibrils may 


D. O. HEBB 45 

readily occur where they begin to enlarge to form knobs — that is, short 
of the synapses proper. When there is already some depolarization of cell 
body or dendrites, the flow of current, it is argued, will make the bottle- 
neck traversable, and the subsequent depolarization will keep it open for 
an appreciable period of time. This will be particularly true if there is any 
considerable activity of the dendrites. 

Such a mechanism would clearly help to account for the short-term 
memories of the present experiment, where we are dealing with associative 
connections which are already highly practised (i.e. whatever growth 
process there may be at the synapse is near its maximum), and where a 
momentary experience is able to make one set of well-developed synaptic 
connections temporarily dominant over others. The conditions of the 
experiment demand, of course, that this dominance must be very brief, 
lasting only for the period in which one set of digits is being held and 
giving way when the next is presented. 

The explanation is of course speculative, but it accounts in principle for 
phenomena which cannot be plausibly dealt with solely in terms of (i) 
a reverberatory trace, and (2) growth at the synapse. What I am saying, in 
short, is that there may be three mechanisms of the memory trace and not 
two, as suggested earlier. It should be clear, of course, that these are not 
alternative mechanisms in the actual phenomena of brain function; they 
occur together, and reinforce one another's actions. 


WIDER IMPLICATIONS 

It has been urged above that we have no hope of understanding learning 
in the adult mammal until we know much more about the organized 
activity (in cell-assembly or trace system) of the regions of divergent 
conduction in the cerebrum. Lack of such knowledge is the main reason 
for the great gap between the theory of learning and the practical advice 
one can give to the student who wants to know how to study more 
efficiently. The great question always is how to 'concentrate', and how to 
'motivate oneself — that is, to keep on concentrating — and this, clearly, 
is the theoretical problem of the control of the excess activity, to prevent 
its interference with the task in hand. If such experiments as the one des- 
cribed can help us codify the ideas involved, and if further they provide 
some information about the interaction of cell-assembly groups in 
learning, we can also see them as a slight contribution to the ultimate 
understanding of the problem of serial order which, as Lashley (195 1) has 
shown, is the crux of the problem of behaviour in the higher animal. 


4.6 BRAIN MECHANISMS AND LEARNING 

A more specific implication is with respect to delayed response. If 
extrinsic reinforcement is not necessary for the establishment of a 
structural trace — if the subject who only hears and tries to repeat a series 
of digits none the less has some memory of that specific series which can 
last despite hearing and repeating other series afterwards — then some 
revision is called for in my interpretation of the phenomenon of the 
delayed response (Hebb, 1958). A monkey sees a piece of food being put 
under a cup at the left, none under a cup at the right; his only response at 
the time is to look at the left cup, since both are out of reach. Both cups 
are then hidden by lowering a screen, let us say for 20 seconds. When the 
monkey is permitted to choose he goes at once to the left cup and gets the 
food. It seemed to me that this could only be accounted for in terms of a 
perceptual activity held in reverbcratory circuits. In view of the present 
results, this does not follow. One look may be enough to establish what is, 
in effect, an S-R connection between the stimulation of the sight of the 
two cups, and an eye movement to the left. When the monkey is again 
exposed to the same visual stimulation, he looks left, his hand follows the 
direction of gaze, and he obtains the food. It is still clear that this 'S-R 
connection' in the brain of the higher animal can hardly be the kind of 
direct, one-way route envisaged by earlier theory, and that it must also 
involve some transmission by re-entrant circuits, but it is still a relatively 
simple mode of learning. In a lower animal, presumably, the closed-circuit 
element may not be present, and we can understand better the one-trial 
learning in the wasp already referred to (Tinbergcn, 195 1). 

GROUP DISCUSSION 

EcCLES. I am very grateful to Dr Hebb for putting up some clear ideas to shoot 
at. I want to shoot at the synaptic knob story he has given wliich is based on the 
outmoded theory of electrical synaptic transmission. However, I thought I might 
leave this synaptic story until tomorrow when I will be talking about the details of 
what goes on in the synaptic knob during repetitive stimulation and afterwards. 

I would like also to comment on one-trial learning. I was listening very carefully 
when Dr Hebb spoke his series of digits. As soon as he started off the series I repeated 
this mentally and I went over the first 3 digits quickly when he got to 3, and so on 
at the 4 and the 5, so as to develop my memory. It wasn't a single trial for me. 

Hebb. Did you get it right? That is the question. 

EccLES. I don't suppose I did. (Checking) The first ones were right, and also the 
last two ; there was some muddle in the middle. 

Hebb. I think you will find another method is better. Just to listen and then 
repeat. 

EccLES. While you are even listening to one digit there will be continued activity 
in the cortex. In 1949 both Dr Gerard and Dr Hebb published the concept that 
reverberating circuits played the key role at the early stages o( learning. I have 


D. O. HEBB 47 

always thought that it is a very good idea that rcvcrberatory circuit activity gives 
the basis for tormation of memory traces. 

Hebb. As Dr Eccles has picked me as a target, I would say that, in my opinion, 
the notion that transmission is solely chemical is pure dogma. That is to say, I 
think that Dr Eccles has provided convincing and powerful evidence that the 
primary transmission is chemical. However I can't see what evidence possibly 
could rule out an electrical current flow as an ancillary or supporting mechanism 
of synaptic transmission. 

The second point is that I would agree entirely, and I take it that I^r Gerard docs 
also, that one trial means one presentation of the stimulus, but it doesn't mean one 
burst of impulses. Of course I don't suppose that 'one trial learning' means that 
only one impulse passes the synapse, and does the trick. But further I would point 
out that in this experiment of course we are dealing with cumulative effects 
because we have no direct evidence of learning after one presentation oi the 
digits. 

Gerard. First, I think it is unfortunate that l^r Hebb has jumped three levels: 
from total behaviour, to organ, to cell, organelle or molecule. This is just too far a 
jump at any one time to be profitable theoretically. I think it unfortunate that he 
contaminated some very precise and provoking issues by bringing in the synaptic 
endings, which are irrelevant at this point. I don't think the argument that is 
brewing, as to whether the synaptic change is electrical or chemical, or whatever, is 
really relevant to the particular problem raised. It is relevant to all kinds of 
problems but too diffuse to matter for this one. 

Second, I raise the question whether you really are dealing in your experiments 
with a structural memory, in contrast to a dynamic one. It seems to me you have 
something very similar to what happens when I look up a telephone number in the 
phone-book, keep it in my mind for a bit, make the call and immediately forget 
the number so completely that if something goes wrong, and I must dial again, I 
must look it up again. This is perhaps related to the point that Dr Eccles was making 
about keeping the memory going; and the psychologists here are certainly aware 
ot Broadbent's work on the temporary storage of information before passing it 
into a permanent storage as a lasting memor\-, whatever the neurological mechan- 
isms there may be. 

Third, I completely agree with the point that Dr Hebb makes about what we 
might call 'imput overload', and the fact that the more elaborate nervous systems 
often tangle themselves up in their own sophistication. Let me remind you of two 
experiments. Some of the original work of Bavelas, following Kurt Lewin, 
involved five people in a certain communication network. Each started with part 
of the answer to a problem, such as what the colours of certain marbles are; and by 
communicating in the intervals each had to get the total information. This they did 
very easily when the marbles were solid colours; but, having worked with a pattern 
problem, they could not solve the solid colour problem because their attention 
focused on minute flaws in the colour. This kind of knowing too much was proved 
at another level, of disease dia2;nosis. Skilled clinicians diagnosed trom a history less 
correctly than did clerks given certain key items to look for. 

So I am not surprised that monkeys may learn certain simple problems with 
greater difficulty than rats, and perhaps man with even greater difficulty than 
monkeys. But it is going much too far from that to say that learning has nothing 


48 BRAIN MECHANISMS AND LEARNING 

to do with the mass of the brain or the number of neurones. The minute one goes 
from these simple things to complex ones, there arc correlations of performance 
with the total mass of the brain. Indeed I find myself in most exciting conflict with 
Dr Hebb on the role of large numbers ot neurones. I have used the term 'physio- 
logical neurone reserve', distinguishing the potentially activable neurones of the 
brain, in contradiction to the anatomical population. For various reasons of in- 
creased thresholds or previous activation, many neurones may not at a given time 
be functionally available. The functional neurone reserve does seem to parallel 
nicely the ability to 'master a situation', a broader and more behaviourally interest- 
ing term than simple learning, hi other words, while one may not do better on a 
prescribed simple stimulus response relation, when it is necessary to find the correct 
concatenation oi possible behaviours out of an almost infinite variety, then the 
larger neurone complement is certainly an advantage and allows greater richness. 
Such an endowed brain can learn things that the other cannot. Maybe in this last 
point I am violating what you meant to say, but that is the way it struck me. 

Hebb. I don't believe the synapse is irrelevant (what has happened is that I have 
been restricted to a 15-page document by Dr Delafresnaye and to a 30-minute 
presentation by Dr Gerard). Behind this statement of mine is an extremely long 
attempt to see if there was any other way of dealing with the problem. The fact of 
the matter is that the theory may be bad. I have no doubt that the position suggested 
by Dr Galambos, by Dr Olds and by others is logically sound. But the flict also 
is that people do think of a simple theoretical dichotomy — structural modification 
or a reverberation; 

Gerard. It need not be one or the other, but either or both — or neither. 

Hebb. I understand that. My only point is that I didn't mean to suggest that it is 
only structural, and not reverberatory. I had thought before, as you did with the 
telephone book, that I could look up a telephone number, dial it, and forget it 
completely. That I could listen to a set of digits, repeat them and forget them 
completely. I suggest to you that your statement about completely forgetting the 
telephone number is almost certainly wrong, on the basis of the experimental 
evidence that I have reported. I would have agreed with you 4 years ago. My idea 
now is that it is a little more complicated. 

One last point on rate of learning in different species. I assumed that we were 
dealing with simple learning in all cases. Man certainly learns some complex tasks 
quicker than a monkey does, but the complex task may be regarded as the equiva- 
lent of a number of simpler ones. But there is a point that I think needs to be made. 
It has happened that Baerend's and Tinbergen's data have not been believed because 
the wasp is too low in the scale to have as fast learning as we have. But I propose 
that learning for which an organism is suited is likely to be faster, rather than 
slower, at the lower levels. 

KoNORSKi. According to my opinion any neurophysiologist concerned with the 
problems of conditioned reflexes is entitled, and even obliged, to explain the facts 
he deals with in the terms of synaptic transmission. As a matter of fact our first- 
hand knowledge about the principles of this transmission comes out from the most 
simple two-neurone reflex arcs studied by the most refined electrophysiological 
techniques. We extrapolate these findings to interpret the more complicated spinal 
reflexes, and we have also a right to utilize them to understand the mechanisms of 
conditioned reflexes, although they are much more complicated. 


D. O. HEBB 49 

As far as L)r Hcbb's experimental data are concerned, I hnd them very interesting 
and possibly requiring further elucidaton. They show that the d}-namic memory 
trace, produced by presentation ot a given series of numbers, is not totally oblite- 
rated by the subsequent scries. It would be interesting to know what will be the 
rate of memorizing of the repeated series if it recurs less frequently than in Dr 
Hebb's experiments, and whether it is possible in this way to prevent memorizing 
of tliis series at all. 

I realize the difficulty connected with the understanding of transient memory of 
series of items in terms of reverberating circuits. May be that the place of a given 
item in the sequence is determined, among other things, by the strength of the 
retroactive inliibition produced b}' the following items. I also agree with Dr Hebb 
that all the series he used in his expermients are in a greater or lesser degree already 
included in our stable memory repertory and one has only to remember that this 
and not that sequence was presented at the given moment. 

Hebb. I should make clear that Milner did not develop liis conception to account 
for those results. On the contrary, I was still struggling to find some intelligibility 
in the data when I saw Milner's paper. 

If I have given the impression of thinking that Dr Eccles's position was pure 
dogma I should like to get that corrected right away. It seems to me that it is dogma 
to maintain the negative proposition that something like electrical fiicilitalion 
cannot happen. The positive side ot chemical transmission has been established well 
beyond anv questioning bv me, but there is still the possibility that you might have 
some ancillary mechanism. 

I can find no way, to come to Dr Konorski's second question, of accounting for 
the ordering in that series. I remind you that the same digits are used trial after 
trial, and that the essence of the successful response is in the proper ordering. 
Suppose that each one of these systems is firing and reverberating. What deter- 
mines the order in which the)' cause the motor sequence? I would only say that as 
nearly as I can determine there must be in addition to reverberation, some sort of 
connection between the separate reverberation systems, so that A fires its motor 
paths and then causes B to do so — and not vice versa. There is some short-term, 
essentially structural, modification which with repetition may turn into something 
more lasting. 

Gerard. Dr Konorski, I was not for one moment objecting to going down to 
the synapse; you will remember I went all the way down to the molecule in the 
first paper. I simply said that whatever specific mechanism one assumes at the 
synapse is irrelevant to the problem Dr Hebb is facing. The problem is whether or 
not a message gets to the synapse, not how it produces a change there. The questions 
are at different levels. 

Thorpe. With reference to the one trial learning experiments with the insects 
Philaiitliiis and Auiuiophila bv Tinbergen and Baerends I would like to say first that I 
don't think there is the slightest doubt about the facts. It is quite easy to observe the 
orientation flights of these insects. Secondly I would like to point out that the task 
learned on these flights is in itself a very complex one. Moreover it is a kind of 
serial learning since, because of its very nature, the insect's eye cannot be getting a 
unitary vision of the whole field; on the contrary it is flying around picking up the 
various landmarks one after the other — and they may be numerous. So it seems to 
me to be a decidedly complex piece of learning; not as complex a task perhaps as 


50 BRAIN MECHANISMS AND LEARNING 

learning these large numbers that Dr Hcbb was discussing, but very complex when 
one compares it with the kind ot learning task the insect normally has to achieve. 
1 think that there is a very real problem here, namely that of explaining the very 
rapid mastery by the insect brain ot a substantial task in serial learning. And then of 
course one must remember that there are no synaptic knobs, so far described, in 
the insect's nervous system which could be acting m the way that Milncr proposed. 

EccLES. If I understood Dr Hebb correctly, he said he couldn't see how he could 
explain some of the digital learning processes on the basis of synaptic trace changes. 

Hebb. Only when dealing with synapses already so highly developed as must be 
the case in repeating a sequence as much practised as 1-2-3. Here a single exposure, 
a single repetition, could hardly make a significant additional change. 

EccLES. Against that concept I would like to say that in the very simplest cerebral 
actions w^e are using millions of neurones in the most complicated imaginable 
patterns. In, say the 1-2-3 sequence, we don't have a group of neurones that are 
related to i, another group for 2, and another for 3. I think that in each digital 
association the most complicated neuronal network is in operation. With a 2-1-3 
sequence, we would have a quite different asseinblage of neurones in activity. It is 
not sutf iciently realized what an immense number of neurones we have to draw 
upon for the simplest memory. Lashley says that in the simplest engram there are 
millions of neurones involved. I would go much further than that myself There is 
evidence from EEG record. During mental arithmetic there may be disturbance of 
the EEG over a wide region of the cerebral cortex. Thus immense assemblage of 
neurones are used doing some unusual multiplication, say 23 X43. Therefore I 
don't subscribe to Dr Gerard's idea that there is a functional neurone reserve. I like 
to think that we are using all the neurones in some kind of pattern or other and 
using them thousands of times over in patterns, but that we can still use them many 
more thousand times over. It is the pattern that is important, not the neurones. 

Gerard. This is not contrary to my notion, except that the more neurones you 
have the more patterns you can deal with. 

Eccles. I agree entirely. One final point, and that is if there is electrical interac- 
tion, and we have seen from Dr Estable's work the complexity of connections, 
and we now know from the electronmicroscopists that there is no free space, only 
200 A. clefts, everywhere in the central nervous system, then everything should be 
electrically interacted with everything else. I think this is only electrical background 
noise and, that when we lift with specific chemical connections above that noise we 
get a significant operational system. I would say that there is electrical interaction 
but it is just a noise, a nuisance. 

Hebb. I think we are in agreement, at least in part, if I understood this last com- 
ment. When you say that the whole cortex is thrown into action when somebody 
multiplies 23 X 43 you are suggesting that most of the activity contributes nothing 
positive. The Jasper, Ricci and Doane experiments at least suggest strongly that the 
course of learning is throwing out the neurones that are irrelevant, keeping them 
out of the way wliile the others do the job. And this may mean that the course of 
learning involves many more neurones than would be desirable. It may be that 
though the operation is very complex, it could be done better if there were no other 
neurones present, for that operation. But the great characteristic of the human 
brain is the extraordinary variety and complexity of things it can deal wih. 

Gerard. I might add, since, the work of Jasper's group has been mentioned so 


D. O. HEBB 51 

many times in connection with this, that extensive studies have been done by Beck 
along the same hnes, showing the dropping out of neurones. 

Chow. I think that the problem Dr Hebb proposes is a very important one. 
What kind of mechanism can produce such effects that one exposure will redirect 
nerve impulses in a certain direction? 

I would like to point out Lashley's notion that maybe you are creating some 
kind of field effect by your repeating the digits. Therefore, in essence you may 
create a DC potential gradient there, as if you had applied an external DC current. 
This presumably will modify all your neuronal excitability and will make your 
impulse flow easily in one direction. I think this may be a possible mechanism. 

Hebb. It appears to me that we are dealing very often, not with the activity of all 
neurones in one topographical area as against the neurones in another topographical 
area but with selective fictors among neurones interlinked with one another. It is 
very diflicult to conceive a functional value from a DC field that is more extensive 
than a single neurone or 2 or 3 neurones. This is my problem. 


THE INTERACTIONS OF UNLEARNED BEHAVIOUR 
PATTERNS AND LEARNING IN MAMMALS 

Irenaus Eibl-Eibesfeldt 


I. INTRODUCTION AND OBJECTIVES 

The term 'innate' as applied to behaviour patterns has become con- 
troversial. Beach, 1954, Hebb, 1953, Lehrmann, 1953, 1956 and Schneirla, 
1956 have criticized the ethological approach, which they accuse of 
performing an artificial dichotomy into innate and acquired patterns. 'I 
strongly urge that there are not two kinds of factors determining behaviour 
and that the term "instinct" is completely misleading, as it implies a 
nervous process or mechanism which is independent of environmental 
factors and different from the processes into which learning enters '(Hebb, 
1953). Lehrmann (1953) accuses ethologists in general and Lorenz in 
particular of misrepresenting unanalysable part-function in such a way as 
to give them the specious appearance of natural units, caused by the same 
physiological mechanism. 

The main task of this paper is to show that the two oldest and most 
important ethological conceptions, that of the fixed motor pattern and 
that of the Innate Releasing Mechanism (IRM), correspond to very real 
functional units and that they prove their analytical value in the attempt 
to analyse the ontogenetic process by which unlearned and learned 
behaviour becomes integrated into an adaptive functional unit. 

2. EXPERIMENTAL EVIDENCE 

((?) The analysis of nest biiildiw^ and rctricvuK^ yotiin^ in the rat. As is well 
known, every rat raised in isolation is capable of building a nest. A number 
of scientists have, therefore, termed this behaviour instinctive. In Mumi's 
(1950) Handbook oj Psychological Research on the Rat, it is discussed under 
the heading 'unlearned behaviour', along with retrieving activity; every 
rat which has given birth retrieves nestlings which were put outside the 
nest. 

Riess (1947) has investigated the problem of the imiateness of this 
behaviour. He raised rats under conditions which gave them no oppor- 
tunity to manipulate solid objects. From the age of 21 days (some even as 

53 


54 BRAIN MECHANISMS AND LEARNING 

young as 14 days) they were kept isolated in wiremesh cages and fed only 
with powdered food. After they had mated he put them in a wooden box 
with strips of paper hanging from the walls. Nest building and retrieving 
were looked for in this testbox, but the animals failed to build nests and to 
retrieve their babies. They only carried the young around and scattered 
the paper strips all over the floor. Most of the nestlings died, as they were 
not sufficiently cared for by their mother. From this failure to build nests 
and to retrieve, Riess concluded that the nest-building behaviour of the 
rat must be learned during ontogeny through handling solid objects. One 
could not apply the term instinct, therefore, to this type of behaviour. 
The way in which the rat might learn nest building was pointed out by 
Lehrmami (1954). The rat might collect food and other objects at its 
sleeping place and thereby observe that some of these collected objects 
can be used to prevent loss of heat. It would in this way learn to build a 
nest as soon as it felt cold. No attempt was made to gain detailed observa- 
tional information concerning either recurring motor patterns or the 
stimulus situations indispensable to elicit them. 

From observing a small number of different rodents [Miis iiiiisaihis L., 
Rattiis uorve^iais L., Merioiies persiciis Blanf., Meriones shawi, GerbiUus 
gerhilhis L., Jaciihis jaciihis L., Cricctiis cricetus L., Mesocrketus aiiratiis 
Waterhouse, Mkrotus arvalis Pall., CitclUis citeUiis L., Glaiicoiuyys t'ohms 
Thomas, Sciiinis i'iiJ(^ciris L., Glis glis L., Muscardiinis avellaiiariiis L., 
Dasyprocta aguti, Oryctolagiis ciiiiiciihis L.) I was familiar with a number of 
motor patterns recurring in most or all of them (Eibl-Eibesfeldt, 1958). 

In all ground-dwelling rodents observed, two phases of nest building 
can be distinguished: (i) The digging of a burrow (2) the construction of a 
nest for sleeping and nursing. 

The nest-building Norway rat for example starts digging a tunnel, 
20 to 30 cm. under the surface, which ends after about i metre in a small 
nesting chamber. Later, additional tunnels are added (Steiniger, 1950). The 
forelegs scratch the earth with alternating movements from the front 
under the belly [scratchiiiii). From there the accumulating earth is pushed 
backwards by the hindlegs {kicking backwards). From time to time the 
animal turns around pushing the earth by alternating movements of its 
forelegs out of the tunnel (piisliiiig). Pushing with the snout is also observed. 
After digging the animal collects nesting material. This collection activity 
consists in grasping the nesting material with the teeth, piillitig it free and 
if necessary biting it loose from where it is attached, then carrying it to the 
nesting site and depositing it there. 

The nesting material is pushed into a heap by movements of the fore- 


IRENAUS EIBL-EIBESFELDT 55 

legs, identical with those performed when pushing earth out ot the tuimel 
and by shoving with the snout. In the centre of the heap the rat starts 
scratchiiiii, forming a bowl and, furthermore, turning on its axis and 
pyiishiiK^ the nest material towards the periphery, it forms a ring-shaped 
mound around itself. Intermittently the rat reaches over the mound, 
grasping scattered nesting material with its teeth, and deposits it on the 
rim of the mound or inside the nest. It reminds one of the way geese 
construct their nests, but whereas the latter are able only to lay the 
material backwards over their shoulder, the rat shows more plasticity, 
being capable of depositing nesting material from different positions in 
relation to the nest. Coarse nesting material, like straw, is split with the 
teeth longitudinally. The rat holds the material at both ends m its paws 
and bites along it with the lower incisors. By a sudden lifting of its head 
it splits the straw [splittiiKT). Straw, and even solid wood, are transformed 
into soft nesting material. Similar movements of nest building are ob- 
served in many other rodents and the patterns of splitting, scratching, or 
pushing, are common to all those species mentioned above. Sometimes 
additional movements are observed. Tree squirrels, for example, have 
special movements for bundling the nesting material before transport. 

In addition, I was aware of certain environmental situations which were 
obviously indispensable, for instance, previously explored environment 
containing known nest locality, or else good cover. ^ At once I suspected 
that the failure of Riess's rats to build was due to the fact that they did not 
have a definite nesting place in the unfamiliar testing situation with which 
they were confronted. 

When I took ten virgin rats experienced in nest building, duplicating 
Riess's test situation, none oi them started building. After they had over- 
come their shyness they first started exploring, in between retreating to 
one corner where they cleaned themselves and rested. Some pulled out 
paper strips and scattered them all over the Hoor, thus behaving like the 
rats of Riess. But none built a nest within the first hour and only three 
built within 5 hours. 

On the basis of my observations on different species of rodents, Riess's 
statement of the problem, asking whether 'nest building' as a whole was 
innate or not, seemed much too simple and consequently his conclusions 
too strongly generalized. Therefore, I attempted to clarify the following 
special problems, using experimental procedures as closely as possible 
similar to those employed by Riess. The question was: 

1 The rats have furthermore to be fainihar with the presence of an observer, otherwise they 
are irritated, showing curiosity or shyness. 


56 BRAIN MECHANISMS AND LEARNING 

(i) Do any motor patterns and taxis components exist which are 
developed in the individual independently of learning ? 

(2) Does the assumption made by ethologists hold true that motor 
patterns which, by the comparison of species can be shown to be phylo- 
genetically homologous, are independent of learning ? 

(3) What is the role of individual learning, (a) in changing or developing 
the pattern itself and {b) in intcgratnig such innate elements, it any, into a 
functional whole ? 

Albhio rats [Ratttis iion'c[^ictis Erxl.) were isolated from the age ot 11 to 
21 days and raised in cages with a grill floor and given powdered iood. 
Since, in the absence of nest material, rats often carry their own tails, 
these were amputated in the experimental animals. In contrast to Riess, I 
tested the rats in their living cages, snicc it has been shown that in rodents 
placed in a new room, escape and exploratory behaviour predominate 
over other activities. For the test, a rack holdnig 30 g. of crepe paper 
strips was fastened on one wall of the cage. The room temperature varied 
between 17' and 19 C. In this way, the ncst-building behaviour of ((7) 
eighty-two virgin rats (2-3 months old), (/)) three pregnant rats and the 
nest building and retrieving of {c) forty-two females, 4 months old, 
immediately after parturition, were tested. 

Of group ((?) (subgroup i) thirty-seven animals had nothing else ni their 
cage than the glass with powdered food, fastened on one wall, and the 
phial with water hanging from the roof Eight of these animals started 
nest building as soon as the paper was presented and four started within 
an hour. Thirteen of the animals ran around with paper strips in their 
mouths, and eventually let them fall, thus behaving as Riess described, 
and six gnawed or played with the strips. They all built nests within 
5 hours. The remaining six rats of group ((7) (subgroup i), however, did 
not build a nest. Observations in the previous days had shown that 
twelve of these rats had had definite sleeping places. All eight animals 
that built immediately belonged to this group. I, therefore, conjectured 
that the structural poverty of the experimental cage might have interfered 
with the establishment of a definite nesting place. Indeed, this may have 
hindered many experimental animals and made building in some cases 
impossible. That some of the experimental animals built in two places 
is evidence favouring this assumption. To facilitate the choice of a place, 
I divided one corner of the cages of the other experimental animals 
(subgroup 2) with a small vertical screen. Of forty-five virgin rats tested 
in such a cage, thirty-three immediately started nest building behind the 


IRENAUS ElBL-EIBESFELDT 57 

screen. Among these were animals who tended to sleep in another corner 
of the cage. An unlearned preference for the most covered place prevailed 
in these cases over the effects of previous experience. Three of the experi- 
mental animals started behind the screen within i hour. Nine decided, 
within 5 hours, to build in another corner. To sum up: of the eighty-two 
virgin females, seventy-six built a nest, forty-four of these within the fnst 
hour of the test, and only six rats did not build. 

In all animals that showed an interest in the nesting material, the above 
mentioned movements of nest building were observed. Most of the 
animals explored the rack and the paper at their first encounter by nibbling 
and sniffing. Then they tore one or a few strips out of the rack and carried 
them without much hesitation to the prospective nesting site. There they 
deposited the nesting material, and often nest-building movements like 
splitting, scratching and pushing appeared, although of no use at this 
stage. These movements usually lasted for only a few seconds before they 
started for more paper. From the behaviour of the animal one did not get 
the impression that they had any idea what the result of their behaviour 
would be. The behaviour released by the nesting material consisted simply 
of certain building movements, in disorderly sequence. But, in all cases, 
a nest was the result of the rat's activity. In five cases the experimental 
animals were given pieces of straw instead of paper. These, too, started to 
build, and all started to split the straw, in the characteristic way previously 
described, thus producing soft nesting material. Here, too, it was evident 
that the rat did not follow a certain plan by insight. It did not, for example, 
split one piece of straw after another, but oidy grasped one piece after 
another, making the splitting movements and dropping the straw after- 
wards without looking at the result. Often the blade was only cracked, or 
a small piece ripped off when it was dropped in order to grasp the next 
blade of straw. There was only one tendency evident, namely, to let 
certain movements run off on certain material. But by repetition of this 
behaviour, all the straw got split eventually. Experienced rats, by the way, 
seem to follow a plan or scheme, but that still has to be studied in detail. 

Although I was interested only in actual nest construction, I let ten of 
the experimental animals, after testing nest building, dig in earth. They 
did so with complete co-ordination of all digging movements. 

For the purpose of filming, three females (group (/;)) were tested when 
pregnant. The reason was that we needed light to film but this produced 
heat, and as is well known (Kinder, 1927) virgin females do not build a 
sleeping nest when it is too warm. But in pregnant females, temperature 
does not influence the behaviour as much. The urge to build a nest is then 


58 BRAIN MECHANISMS AND LEARNING 

very strong. As Kollcr (1955) has shown, this is due to the corpus-hiteum 
hormone, progesterone. All three females built immediately; their 
behaviour is shown in the film. 

Forty-two inexperienced females (group [c)) were tested for retrieving 
and nest building immediately after they had given birth in the experi- 
mental cage. Thirty-five of these females retrieved a nestling taken from 
the corner where they were suckling and deposited in another corner. 
Only seven of the females did not carry back their babies. In six of these 
seven cases the nestlings, which were not protected by nesting material, 
were so cold, that they seemed nearly lifeless, and did not squeak. Squeak- 
ing is, however, one of the sign stimuli for the females, releasing search 
and retrieving in the mother (Zippelius and Schleidt, 1956; Eibl-Eibesfeldt, 
1958). Those retrieving behaved like normally raised mothers, with the 
only difference that they showed some hesitation in grasping the nestlings 
and often lost them during transport and had to pick them up again. 
When the nestling squeaked while being grasped, the female changed the 
grip, a behaviour which is now being studied in more detail. 

All forty-two females, including those that did not retrieve, started 
immediately with the construction of a nest when, after the retrieving 
test, nesting material was offered. Furthermore they showed the covering 
of the nestlings typical for the breeding mother. This will be shown in the 
film. 

Our experiments have shown that the handling of solid objects during 
ontogeny is not a prerequisite for the development of nest-building and 
retrieving behaviour. Riess did not realize this and he furthermore over- 
looked that nest building is a complex behaviour, and therefore did not 
look for the elements which compose it. 

The several motor patterns which, on the basis of a comparative study 
of many species of rodents, had been assumed to be unlearned 'fixed 
patterns' appeared completely unchanged in the rats reared in the manner 
employed by Riess. One important difference was, however, found 
between experienced and inexperienced rats. The sequence in which the 
above-mentioned motor patterns are used was considerably better adapted 
to the function in experienced rats, inexperienced rats often employing 
motor patterns which can develop their function only at an advanced 
stage of the newly built nest, at a very early stage of building at which the 
patterns in question did not yet perform their function. The question 
whether 'nest building' must be learned or not cannot be given a simple 
answer. Certain essential motor patterns and taxis components are 
completely independent of learning. The proper sequence in which 


IRENAUS EIBL-EIBESFELDT 59 

several patterns are best employed to perform an integrated function is 
indubitably learned. But no ethologist had ever doubted — as critics of 
ethology often imply — that learning processes are of the greatest impor- 
tance in behaviour, Lorenz (1937) has often emphasized that learned and 
innate elements of behaviour are closely interwoven. His ravens for 
example had innate nest-building movements, but had to learn which 
material to use for nest building. Eibl-Eibesfeldt (1956b) has recently 
shown that red squirrels develop individually different techniques of nut 
opening on the basis of a few innate patterns, such as gnawing and a 
certain splitting movement. 

In retrieving, learning plays a lesser role than in nest building. The 
inhibition of biting the nestling seemed even stronger in inexperienced 
rats. It would seem that the rat has to learn that a baby is not so vulnerable 
after all. 

(h) The killing tecluiiqtic of the pok'cat (Putorius putorius L.). The follow- 
ing deals with the technique of prey killing of inexperienced and exper- 
ienced polecats. Kuo (1930) has studied the behaviour of cats raised under 
different conditions towards albino rats, grey Norway rats and nnce. 
Twenty cats were raised in isolation from weaning, twenty-one cats 
remained with their mother and were allowed to observe how she killed 
prey, and eighteen cats isolated from other cats were given a rat as com- 
panion from an early age on. Of those raised in isolation, nine killed prey 
(= 25 per cent); of the second group eighteen killed prey (85 per cent); 
but of those raised with rats, only three killed prey and then only of a 
type to which they were not accustomed. Towards their rat-companions 
the latter individuals showed peaceful and positive reactions. They licked 
and defended them and searched for them persistently if they were taken 
away. Kuo thus clearly showed that experience influences the behaviour 
of the cat towards prey, even though the difference in the percentage of 
killing in the first two groups might be explained as a result of a different 
state of health due to deprivation. Kuo expresses the opinion that the 
concept of instinct has been proved useless by his experiments. The bodily 
structures alone explain why a cat behaves like a cat and not like an ape : 
one does not need to have recourse to instincts, based on special structures 
of the central nervous system. What the animal actually does, within the 
potentiality given by its body, is learned during ontogeny. 'The behaviour 
of an organism is a passive affair. How an animal or a man will behave in a 
given situation depends on how it has been brought up and how it is 
stimulated' (Kuo, p. 37). 

It seems, nevertheless, that Kuo did not realize where the problem 


60 BRAIN MECHANISMS AND LEARNING 

actually lies. Such a complex behaviour pattern as prey hunting is, of 
course, not fully innate or learned as a whole, and we agree with Kuo that 
to pose such an alternative is a mistake. But, like Riess, Kuo did not see 
that the complex functional unit in any animal's behaviour is formed by a 
number of smaller behavioural patterns. In his paper, no single, particulate 
behaviour pattern is described. We investigated the prey-hunting be- 
haviour of the polecat, asking it there were any fixed motor patterns or 
innate orientation movements involved in the pattern, and if so where and 
how experience enters. 

Every adult polecat kills rodents which are able to defend themselves, by 
biting them in the neck or occipital region. Prey in flight is often grasped 
{{^raspiii'^) on another part of the body, but after the polecat has thus 
stopped the prey's flight, it immediately releases its preliminary hold to 
grasp the neck. Often it turns the prey on its back and shakes it torcetully 
{shakiiii^), in the way employed by many other carnivorous animals. The 
polecat furthermore loosens and fastens its grip, in rapid succession, sinking 
its teeth repeatedly in exactly the same spot, thus gradually perforating 
the skull {hillino hitc) (Eibl-Eibesfeldt, 1956a). 

Studying the ontogeny of this technique in non-deprived animals, one 
finds that they show little skill in their first attempts to kill a prey, prob- 
ably partly due to the lack of bodily strength. At the age of 3 months, 
however, a polecat is a skilful killer. 

Twenty polecats were raised under deprivation. They were not con- 
fronted with living prey until the test, but were fed with meat and even 
dead rats to keep them in a good state of health. Eight of the polecats 
remained with their mother and litter mates. Twelve were isolated at the 
age of 21 days from members of their own species. Six of the polecats 
raised in isolation and eight of those raised with their litter mates, but also 
never being confronted with living prey, were given an adult rat when 
they were 5 months old. One of those raised in isolation was tested at 10 
months, and five raised in isolation were 2 years old when first given a 
prey. The latter were given chicks. 

Those tested at the age of 5 months behaved as follows. It the rat 
remained motionless on the spot, the polecat approached slowly, sniflmg 
with curiosity at the prey and touching it with the paws. Some licked or 
carefully tried to bite, but they did not attack. If the rat ran towards the 
polecat, the latter retreated. But as soon as the rat showed flight reactions 
by running away, the polecat attacked it vigorously, trying to grasp and 
bite it. It did not direct its attack towards a special part of the rat's body, as 
an experienced polecat does, but just bit into what it grasped, the tail, the 


IRENAUS EIBL-EIBESFELDT 6 1 

shoulder or a leg. Then the rat immediately turned in defence, and the 
polecat, evidently surprised, released its grip, normally attacking again and 
again. The more anterior the hold on its prey, the more difficult was it for 
the latter to defend itself. If the polecat was successful in grasping the 
rat's neck, then it could kill the rat easily. The polecats learned the right 
grip quite rapidly. One killed its first prey within 20 seconds, with only 
three attacks. Others needed 1-15 minutes, depending on the behaviour of 
the rat. After having killed four to six rats, one each day, a polecat was a 
skilful hunter and its killing bite was always directed towards the neck of 
the prey. One female that was bitten by the rat showed fear and on the 
following 3 days, avoided the rat which had been left with her. The rat 
slept in the polecat's nest, which the polecat avoided. When awake the 
polecat restlessly ran up and down in the cage. On the fourth day it killed 
the rat. It became a good hunter from then on. 

The one polecat tested at an age of 10 months got bitten too, and it 
avoided the rat even i month later. Unfortunately this polecat escaped. 

There were no differences between the behaviour of those seven raised 
in isolation and that of those left with their litter mates. In both groups, 
the first attack was released by the fleeing prey, and both had to learn the 
correct orientation of the killing bite. There arc, however, indications 
that those raised with litter-mates learned this orientation faster, probably 
as the result of experience while playing. Four succeeded within i minute. 

The movements described above, such as shaking, killing bite and 
turning the prey on its back, were observed at the first encounter. 

Five polecats, at 2 years of age, were confronted with young chicks of 
the domestic fowl. They were previously fed with dead chicks. Neverthe- 
less, none of the experimental animals attacked the chicken as long as it did 
not move. They sniffed at it and only when it ran away did three of the 
polecats follow and grasp it. Two of them killed it within a few seconds 
by bites in its back, one that had not got a very good hold released the 
grip when the chicken started struggling, but killed it with the next bite. 
All three killed thereafter without hesitation, but they did not aim their 
bites towards the neck of the prey, very probably because these animals, 
not capable of self-defence, did not demand the development of a special 
killing technique. They were equally easily killed by bites m the back, side, 
chest, or neck. 

The other two polecats also showed great interest in the chicken. They 
sniffed and licked it, meanwhile uttering the low sounds (muttering) that 
normally express readiness for social contact (Eibl-Eibesfeldt, 1955). 
After some minutes they turned away, but returned after a while to 


62 BRAIN MECHANISMS AND LEARNING 

explore the chick agahi. When it tried to escape, both chased and caught 
it, but they showed a very strong inhibition to bite. They just seized the 
chicks, without causing injury, and carried them into their nesting box. 
After 5 hours the chicks were still alive there. I took them away. On each 
of the following 7 days I offered one chick to each of these polecats. 
During that time they were fed only with very little meat every third day. 
Until the fourth day, the behaviour of the polecats towards the chick 
remained nearly the same. After a short investigation they grasped the 
chick and carried it home, uttering the call of social contact. In one of the 
polecats a little hostility was observed on the third day; it hissed when it 
met the chick, but soon started muttering. On the fourth day I left the 
chicks with the polecats over night. Next morning polecat {b) had killed 
and partly eaten its chick, while the other chick was sitting in the warm 
nest without showing the slightest damage. A newly offered chick was 
attacked by the polecat, grasped in the neck but with inhibition to bite, 
and carried alive to the nest, where it was killed and eaten half an hour 
later. Polecat ((7) however, hissed a little, then carried the new chick to its 
nest, where I found it still ahve next morning. After I killed it, the hungry 
polecat immediately started eating. On the same afternoon, both were 
given chicks again; both carried them in alive. After 3 hours the chicks 
were found to be still ahve. This time polecat {a) killed the chick during 
the night, whereas polecat (/>) left it undamaged. The experiments are 
not fmished yet. 

The experiments have shown that learning plays an important role in 
the development of prey-hunting behaviour, but also that there are some 
important inborn reactions. What is learned is the orientation of the 
killing bite towards the neck of the prey, a learning process which takes 
place only if the prey proves difficult to kill otherwise. The polecat has, 
furthermore, to learn that a quietly sitting rat or chick is prey, even when 
it has eaten dead rats or chicks before. The first attack is released by a 
fleeing object, this reaction of chasing and biting fleeing prey is iimate, 
as well as those behaviour patterns, like shaking or killing bite, described 
above. The inhibition of the two-year-old polecats to kill chicks is 
secondary and probably due to social frustration. I got the impression that 
the chicks became substitutes for a social companion. 

The different evaluation of the results of the experiments by Kuo, 
Lehrmami, Riess and the author might first be due to the fact that the 
former were interested only in the final outcome of the experiments. 
They noted only: 'prey killed or not killed', 'retrieving or no retrieving', 
'nest building or no nest building' but did not care very much about the 


IRENAUS EIBL-EIBESFELDT 63 

behaviour of the animal in the test situation. It the author had followed the 
same line when studying the prey-killing behaviour of the polecat, he, too, 
would have come to the conclusion that this specific technique is learned, 
since the animals need several trials before killing efficiently. But the 
crucial point is that it does not need to learn the shaking movements and 
it does not need to learn to attack and grasp a fleeing object. If one observes 
the behaviour one finds that a number of movements are there from the 
moment the first reaction towards prey is released. Kuo's (1930) experi- 
ments on cats do not contradict our experiments, but simply fail to give 
information on certain important questions. The results of Riess, on the 
other hand, clearly contradict our observations, due to the fact that Riess 
made the methodological mistake of testing his animals in a surrounding 
strange to them. The immediate nest building of our inexperienced animals 
can hardly be explained by the learning hypothesis. How should our rats 
have learned that nesting material keeps the animal warm if collected 
and formed into a nest? And what taught the mother that newborn nest- 
lings need to be covered with nesting material, and that straw has to be 
split? 

3. DISCUSSION OF THE CONCEPTIONS 

According to Lorenz (1952) fixed motor patterns are sequences of 
movements which can appear in an animal without previous exercise and 
experience. Presumably they are based on specific central nervous 
mechanisms, which are inherited in the same way as other morphological 
structures. The characteristics of fixed patterns are: 

(i) They are constant in form. 

(2) They are characteristic for the species. 

(3) They appear in an animal reared in isolation from members of its 
own species. 

(4) They develop even if the animal is prevented from exercising the 
behaviour pattern in question. 

(5) They are hardly ever found in one species alone at least if closely 
related species are in existence, but are characteristic of taxonomic groups 
just as morphological characters are. Thus can they be used for taxonomic 
purposes, as has been shown by Heinroth (191 1), Whitman (1919), 
Lorenz (1941) and others. 

Our experiments have shown that in every case in which, on the basis 
of criteria gained by comparative observation alone, a sequence of move- 
ments was suspected of being a fixed pattern, the co-ordination in question 


64 BRAIN MECHANISMS AND LEARNING 

appeared unchanged in the animal reared in a deprivation situation, used 
hke a ready-made tool by the inexperienced animal. Conditioning deter- 
mined the way of their application only (see also Eibl-Eibesfeldt, 1956b 
and Eibl-Eibesteldt and Kramer, 1958). 

Fixed patterns or instinctive movements as defined by Hcinroth and 
Lorenz are therefore not a fiction as implied by the 'anti-instinctivists', but 
something real, and the distinction between innate and acquired is of 
analytical value. The learning psychologist who docs not know what is 
genetically determined is in a position equivalent to a geneticist perform- 
ing modification experiments on genetically unanalyscd material. We 
know, of course, that we abbreviate, when we say a behaviour pattern is 
inherited or innate instead of saying that its anatomical and physiological 
conditions are. But there is no reason for not using the term this way, as 
long as we remain aware of the fact that characters always develop on the 
basis of an inherited range of variation, all the more if the range of 
modifiability is practically nil. 

Although we do not know yet what a fixed pattern is, a number of 
investigations have made it highly probable that they share a common 
physiological basis. Besides the peculiarities already mentioned, the fixed 
pattern is characterized by a specific spontaneity (Lorenz, 1937), which 
may be hypothetically explained by, the assumption of neural centres 
creating impulses. The spontaneous activity of the central nervous 
system has been demonstrated by von Hoist (1935, 1936, 1937), first in the 
earthworm in which he discovered in the isolated ventral nerve cord salvos 
of rhythmical impulses, corresponding exactly to the contraction of 
segments in the normal creeping movements. Later von Hoist found that 
even such highly complicated inborn motor patterns like swimming in 
fishes are based on a central nervous system automatism. Von Hoist 
explains this spontaneity by the hypothesis that there are groups of cells in 
the CNS creating impulses. These groups of cells influence each other, 
leading to certain forms of co-ordinated movements. 

Very little is known about the neuroanatomical basis for fixed patterns 
but the experiments of Hess (1948), Hess and Brugger (1943) and the 
recent investigations of von Hoist and von St. Paul (1959) strongly 
indicate that fixed patterns are based on an inherited neurophysiological 
mechanism. 

Experiments have shown that fixed patterns are released by certain key 
stimuli which characterize simply but unmistakably the biologically 
adequate situation, and release the behaviour pattern even in an in- 
experienced animal. In other words, besides the innate ability to perform 


IRENAUS EIBL-EIBESFELDT 65 

certain movements, there is an innate ability to recognize (Lorenz, 1942). 
Although our investigation concentrated on the motor patterns we 
found that a certain stimuli situation must be given to release nest building, 
retrieving and prey hunting. Ignorance oi this caused all Riess's errors. 
Wild brown rats failed to build, not because they had less unlearned 
responses, but because they reacted only to more specific stimuli. They 
needed a more covered nesting site and when I offered them little tin 
huts they started to build under this cover. 


4. THE VALUE OF THE DEPRIVATION EXPERIMENT 

Lehrmann (1953) is right in his critique, when he says that in every 
deprivation experiment one has to answer the question: Of exactly what 
has the animal been deprived? But we cannot follow him to his conclusion, 
that all actual behaviour of an organism is determined by the animal's 
experience, the only inherited basis being the bodily structures deter- 
mining potential behaviour. He, as well as Schneirla and Hebb, insists that 
it is practically impossible to prove that experiences during ontogeny 
are not at work. An animal might even learn /// iifcro or in the egg. To 
elaborate this line of thought, Schneirla (1956) and Lehrmami (1956) 
define as experience any influence of external stimuli on development 
and behaviour. According to this definition even biochemical changes or 
stimuli arising from growth processes, as well as any kind of external 
stimulation, must be termed experience. Learning is only one form of it. 

The way in which such experiences may enter development is explained 
by Schneirla and Lehrmama on the basis of Kuo's (1932) experiment with 
chicks. Kuo examined the ontogenesis of the food-pecking reaction in the 
domestic fowl. These animals are able to peck for small objects immed- 
iately after hatching, a behaviour which we would look at as innate. But 
Kuo observed that in the three-day old embryo the neck is bent passively 
as a result of the heartbeat, which lifts and lowers the animal's head 
resting on the chest. At the same time the head is stimulated tactually 
by the yolk sac, which is moved by contractions of the amnion that are 
synchronous with the heart beat. One day later, the chick reacts to 
external stimulation by bending its head actively, and at the same time it 
opens and closes its beak during nodding. According to Kuo, this is due 
to irradiation of the neural excitation caused by the nodding. At the age 
of 8 to ID days, liquid, forced into the mouth by the head and beak move- 
ments, is swallowed. The movements get more and more stereotyped and 
the previously more independent head, bill and throat movements become 


66 BRAIN MECHANISMS AND LEARNING 

integrated into one functional pattern. According to Kuo, this is due to 
learning. The first arousal of the animal by visual stimuli is explained by 
a diffusion of impulses from the optical region in the midbrain centres 
which previously released the discussed behaviour to tactile stimulation 
(Maier and Schneirla, 1935). 

Lorenz (195X) objects that this hypothesis completely leaves out of 
consideration one fact ot the greatest importance: 'All these structures of 
behaviour, whether they function on the receptor or on the motor side, 
are adapted to innumerable environmental data. Unless one assumes a 
mystical "prcstabilizcd harmony" between the organism and its environ- 
ment — which would be "preformationism" indeed, information con- 
cerning all these environmental data must, at some time, have been "fed 
into" the organism to make this adaptation possible. This acquisition of 
information can have occurred only in the interaction between the 
organism and its environment, either during the evolution of the species 
or during the ontogeny of the individual.' 

To uphold the hypothesis that pecking is learned in the egg in the way 
suggested by Kuo one would have to make at least one of three assump- 
tions: 

'The tirst ot these is that the heartbeat and the primarily independent 
swallowing movements just happen, by pure chance, to teach the chick 
motor co-ordinations that can later on be used in pecking up food. The 
second, equally impossible assumption is that there is a preformed harmony 
between what the chick learns inside the egg and the environment with 
which it is to be confronted later on. The third, improbable but not 
impossible, is that the heartbeat and the other conditions inside the egg, 
have, in the evolutionary interaction between the species and its environ- 
ment, been developed into a highly specialized apparatus whose function 
it is to teach the chick just those well adapted motor co-ordinations. In 
other words, the attempt to avoid the assumption of genetically fixed 
movements saddles us with that of an equally genetically fixated teacher!' 
(Lorenz, 195S). 

Once we have realized the adaptive character of a behaviour pattern, 
we can deprive the animal experimentally of the specific information 
concerning the data to which the behaviour is adapted. If the animal 
nevertheless shows the adaptive behaviour pattern in the test situation, 
then we call it innate, despite the fact that learning in the widest sense, e.g. 
exercise of single muscle units, might have played a role during the 
ontogeny of the individual. This does not explain why the behaviour 
pattern as a whole later fits the environment. 


IRENAUS EIBL-EIBESFELDT 67 

The distinction between innate and acquired behaviour patterns, there- 
fore, is not at all an artificial one. If a male salticid spider needed to learn 
by trial and error his complicated courtship display, which enables him 
to approach the cannibalistic female, he woulci get eaten before copula- 
tion. A newborn baby, unable to suckle, would starve before having 
learned the complicated pattern of movements. On the Galapagos 
Islands no carnivorous land mammal exists and therefore many of the 
endemic birds have lost their escape reaction. They are tame, with just a 
few exceptions. One such exception is the little endemic Galapagos Dove 
[Ncsopclia (^alapai^ociisis) which shows 'injury feigning'. As soon as one 
approaches a nest with nestlings, the adult bird flutters from the nest to 
the ground. There it runs away from the nest, fluttering its widespread 
wings. This "injury feigning' is well known in birds of other countries, 
where its function is to attract the attention of a predator and to lead him 
from the nest. But on Galapagos there arc no carnivorous mammals, 
and the behaviour pattern is normally not used. It seems rather difficult, 
therefore, to explain this behaviour on the basis of the learning hypothesis. 
From where might the animal have gathered the information about 
predators, to which its behaviour is adapted ? Since the information could 
not have been acquired during the ontogeny of the individual, the only 
remaining possibility is that this information was acquired by the species 
during the course of evolution, at a time when it was confronted with 
predators; this 'experience' must have been preserved in the genoma of the 
species. It appears in the Galapagos Dove as a behavioural relict. 

In the normal behaviour of an animal, innate and acquired patterns are 
closely linked together into functional units, so that it is very often rather 
difficult to distinguish the two components from each other. Nevertheless, 
we can deprive an organism of special information and thereby prove that 
certain behaviour patterns are innate. Doves (Grohmann, 1939) and 
swallows (Spalding, 1873) for example have been reared under conditions 
that made it impossible for them to use their wings and practise flying 
movements. Upon being released, however, these animals were able to fly 
as well as their siblings, who had not been deprived of the opportunity to 
learn flying. 

How independently from learning processes certain behaviour patterns 
develop, can also be seen in mallards reared in isolation. With the onset of 
sexual maturity they will show such highly differentiated movements as 
grunt whistle and head-up-tail-up, without ever having seen another 
mallard. If we raised a tufted drake or a mandarin drake under identical 
conditions we should observe them performing their species specific 

F 


68 BRAIN MECHANISMS AND LEARNING 

courting. In short, those movements arc inherited: they mature as part of 
normal development in each individual of the same species in an 
almost identical way. While this paper was under press I continued my 
investigation of the ontogeny of behaviour patterns in mammals. These 
will be described in detail in a later publication, but I should like to 
mention here one more example illustrating very well a behaviour 
pattern in a mammal which is extremely little influenced by learning. 
As is well known, squirrels {Sciuriis vtt\(^aris L.) hide nuts in holes which 
they dig in the ground. They deposit the nut, stamp it into the ground by 
repeated blows with the front of the upper incisors and afterwards cover 
the hole with earth by means of the forepaws. Squirrels raised in grill 
cages with powder food, by a procedure similar to that described in 
connection with the rat experiment, were given open nuts. After they had 
eaten some nuts these animals immediately began to search for a hiding 
place. With the nut between their teeth they moved over the ground, 
and wherever the nut met an obstacle they started to scratch as if digging 
a hole. In many instances they deposited the nut, banged it down with 
their teeth and finally made covering movements with the forepaws, as if 
covering the nut with earth, although in fact there was no earth. Only 
rarely is such a comparatively long chain of reactions innate in mammals. 
This behaviour of the squirrels was filmed and a detailed analysis will be 
presented in the near future. 

Some drawbacks of the deprivation experiments have, nevertheless, to 
be considered. First of all, we must always keep in mind that an animal 
raised artificially may not show its normal behaviour. How easily an 
animal is disturbed under the conditions of captivity, is well known to 
every animal keeper. If certain behaviour patterns characteristic for the 
species under observation, do not appear in the deprivation experiment, 
this might be a result of disturbance and not of lack of information. Strictly 
speaking, the deprivation experiment is conclusive only in case of a 
positive result. If a specific behaviour pattern comes out despite the 
specific deprivation situation, we can say that it is a fixed pattern. But if it 
does not appear, it does not necessarily indicate that the animal needs to 
learn it. If we keep the animal in good health we may be able to avoid 
such disturbances. That a certain behaviour pattern does not appear in the 
experimental situation might, furthermore, be due to lack of the stimuli 
normally releasing it. To avoid this error a normally raised control animal 
must be tested under the same conditions as the experimental animal and 
the experimental animal must be confronted with the situation in which 
the behaviour pattern normally appears. 


IRENAUS EIBL-EIBESFELDT 69 


GROUP DISCUSSION 


EiBL-EiBESFELDT. Wc saw that learning plays an important role in the integration 
of innate behavioural elements into a functional whole. If we study such learning 
processes, we find that many animals are capable of learning certain tasks very 
rapidly. These arc supposed to be of biological importance to them. Is tliis rapid 
learning a specific talent based on central nervous structures adapted to that special 
task during evolution or is it due to a special high motivational pressure? 

Galambos. Dr Eibl-Eibcsfcldt asks if this is the result of prime motivation or of a 
special adaptive mechanism. Would you please elaborated 

EiBL-EiBESFELDT. When I studied the prey-hunting behaviour in the common 
toad, I found that this animal, which normally does not learn very fast, learns 
rapidly to distinguish between palatable and unpalatable foods. If it gets stung by a 
wasp or if it receives unpalatable food it needs only very few experiences to avoid 
such prey in the future. Is this rapid learning based on specially adapted structures 
of the C.N.S. or is it just due to an especially high motivational pressure? Would 
they learn other tasks just as fast if a similar pressure was put on them? Are there 
experiments where animals learned the same tasks under different motivation? 

Olds. We have experiments where certain animals learned complicated and 
unusual behaviour patterns. We worked with rodents too. Provided the animal by 
its behaviour could turn on an electrical stimulus to the hypothalamus — the 
animal learned with great speed, provided the stimulus was in the right part of the 
hypothalamus. The same animal looked very stupid if we turned the stimulus down 
so that it was barely above threshold. This will be clarified later when I give my 
paper. 

Hebb. I thought Dr Eibl-Eibestcldt was disagreeing with the idea that there is any 
role of learning in the instinctive patterns ot behaviour. The experiments he has 
described clearly support the idea that learning and the constitutional structure of 
the animal collaborate closely, and that we are not dealing with two different and 
totally unrelated sorts of processes. We must not distinguish instinctive processes 
from processes influenced by experience. 

It is my impression that we have agreement in principle on the question. I have 
not talked to Riess, but I am sure that Lehrmann would not disagree with any of 
your propositions today. It would be ridiculous to suppose today that there are no 
innate mechanisms, no laid-down patterns of response. What I add, reinforced by 
your paper, is that those laid-down patterns are what are modified by the effects 
of experience. In some animals little modification, in others much. The lower 
animal particularly is built to learn a few things and he learns them quickly. If this 
is the essence of the notion of instinct, this and the tendency to act in certain direc- 
tions, I suggest that man's behaviour also is essentially instinctive. We are so 
conscious of individual differences that we do not see the extraordinary uniformities 
that occur through all human cultures. 

Eibl-Eibesfeldt. If you read Lehrmann's paper you will sec that he does not 
agree with all these concepts. There are certain behaviour patterns which are innate 
in the sense that they develop in the animal without certain types of learning being 
involved. We do not imply that experience plays no role at all. Lehrmann has 
defined experience as every impact of external stimuli on the development of 
behaviour, formative processes and so on. We agree that all these can influence 


70 BRAIN MECHANISMS AND LEARNING 

behaviour, but behaviour is adapted to certain environmental situations. There arc 
two possibihties to explain such an adaptation: during evolution of a species, or bv 
learning during the ontogeny of the individual. And wherever we find that such 
highly adapted behaviour comes out in the animal in spite of complete lack of 
previous knowledge of the specific stimulus situation to which its behaviour is 
adapted, we speak of innate behaviour. If a male spider would need to learn its 
complicated courtship postures it would get eaten by the female. When a male 
mallard raised in isolation still shows its highly complicated courtship postures at 
sexual maturity it demonstrates innate behaviour, in spite of the fact that the 
animal might have learned something in addition to it. How our opponents argue 
is demonstrated in Kuo's paper. He observed that in the chick embryo the head 
rests on the heart and is lifted with every heartbeat. Thus nodding movements, a 
part of the later pecking behaviour — get induced. Later nodding becomes inte- 
grated with the previously independent swallowing movements into food-pecking 
behaviour which every newly hatched chick shows. Our opponents express the 
opinion that this behaviour was learned within the egg. (A detailed discussion is 
given in my paper.) 

Hebb. No. 

EiBL-EiBESFELDT. Yes they do. Concluding from this and similar experiments 
they say there is no innate motor pattern : everything is learned. 

Hebb. There is no support for that extreme view. My disagreement is with the 
dichotomy you made earlier, between learned behaviour and instinctive behaviour. 

Thorpe. I agree with Dr Eibl-Eibesfeldt on this. Dr Lehrmann does not now 
take this view, but he certainly did appear to do so at one time. I am glad that he 
seems now to have come round to what I believe is the only reasonable position to 
take. With birds you get the same sort of thing, in regard to this interlacement of 
learned and innate behaviour patterns. It depends on the group as to how big these 
innate chunks of behaviour are. In the mammal they are small and as you go down 
the zoological scale they become larger, and they may be very large in insects or 
spiders. One other point I would like to mention is that these experiments on 
birds give clear evidence that the learning is too selective to be 'motivational' in 
this sense. A theory which relies mainly on a supposed change in the general 
motivational level would not carry us very far in explaining this type of learning 
in birds. 

EiBL-EiBESFELDT (Added later): I can confirm Dr Thorpe's statement that Dr 
Lehrmann does not take this extreme view now. I discussed the matter with him 
in Cambridge at the Ethological Conference. 

Grastyan. I would like to know in more exact terms what is meant when you 
say that the animal learns quickly — How many trials ? 

Eibl-Eibesfeldt. It has to kill four or five rats to learn how to kill. 

Grastyan. If it cannot kill the first time, what happens in the next trial? 

Eibl-Eibesfeldt. He tries again. If he receives a punishment stimulus, he learns 
faster. 

Grastyan. What happens if you work with a satiated animal? 

Eibl-Eibesfeldt. It kills anyway. It was shown by Kuo that in cats the satiation 
has no influence on hunting behaviour. I still would like to ask why Dr Hebb teels 
that it is not justified to speak of instincts as innate behaviour in contrast to a 
learned response? 


IRENAUS EIBL-EIBESFELDT 7I 

Hebb. My objection to 'instinct' is that it refers to a mechanism which is ciistinct 
from what enters into normal learning processes. 

EiBL-EiBESFELDT. But thcrc are differences, for example, m the physiology of 
these innate patterns. I would like to mention their spontaneity. Furthermore they 
are often released by certain key stimuli — innately. All this shows that we are 
justified in distinguishing innate and accjuired behaviour patterns. 

Hebb. The argument against that is exactly the evidence that you have in your 
paper. You show how closely the two collaborate. The implication of a separate 
mechanism is that some behaviour is controlled by instinct, some by intelligence 
and learning. It seems to me that it is all intelligence or all instinct. The higher 
species has a wider range of situation to which it can adapt, innately. The operation 
of intelligence and learning is within the scope of the innate aptitude, hi certain 
cases we see the deviation from the species pattern clearly, and call that learning. 
But learning also operates in the things which are characteristic of the species. My 
objection to instinct as a term is the suggestion that we have two different kinds of 
mechanism controlling behaviour. 

Gerard. There is a continuity from growth, through heredity, prebirth experi- 
ence, to youth and ageing experience. Only in the fertilized egg have you heredity 
— from then on, the chromosome pattern interacts with its surroundings. If we 
find it useful to separate innate reflexes from conditioned reflexes, it is not mean- 
ingless to try and distinguish other different levels; some higher level reflexes 
common to the species we may call instinctive. What reallv matters in any given 
situation is whether the variance normally introduced by the environmental 
experience of the individual is relatively important compared with the variance 
introduced by the past experience of the race and put in the genes. 

The most interesting aspect is the adaptive one, for example, the rats building 
nests to keep their young warm. The pattern seems innate, but the execution of the 
details of the pattern is clearly learned. 

As an illustration of the innate character, the behaviour pattern should cease if a 
non-adaptive situation is brought about. For example, it you put the rat in a hot 
room, where the young might become overheated would it still make a nest? 

EiBL-EiBESFELDT. Kinder (1927) has studied the influence of temperature on the 
nest-building activity of the rat. If it gets too warm, the animal stops nest-building. 
But that this behaviour is controlled by temperature and thus variable to a certain 
extent does not prove that it is not innate in the way pointed out in my paper. 
Pregnant rats, by the way, are less easily influenced by temperature, because of the 
secretion of progesterone. 

Gerard. A negative result does not disprove it but if you have an instance when 
an animal does a complex act despite the harmful effects, it would be a strong case. 

Eibl-Eibesfeldt. May I mention two examples. Turkey cocks have a highly 
ritualized way of fighting with rivals. They do not jump in the air and hit their 
opponent with the spurs, as other gallid birds do, but wrestle with their necks, 
grasping each other at the bill. As soon as one wants to give up it assumes a sub- 
missive posture, by lying down on the floor. In captivity it can happen that a 
turkey cock and a peacock get involved in a fight. Both closely related species have 
a similar display and therefore understand the expression movements of the 
preliminary display. But their techniques of fighting do not fit together. While the 
turkey cock expects wrestling the peacock hits him with his spurs. This leads to 


72 BRAIN MECHANISMS AND LEARNING 

submissive posture in the turkey cock but the peacock unable to 'understand' 
continues fighting and the more he beats the turkey cock the more the latter gets 
latched in the mechanism of the submissive posture. He does not run away and 
often a fatal end results. Another example of innate behaviour misleading in 
artificial circumstances normally not occurring in nature is the light orientation of 
mail)- insects which leads them to circle around lamps and burn to death. 

Concerning the previous remark of Dr Hcbb, I want to add that he himself 
speaks of how closely 'the two' collaborate, referring to innate and acquired 
patterns. Evidently he makes a distinction and only avoids the terms innate and 
acquired. We use them as we feel a need to distinguish those behaviour patterns 
which show adaptcdness to specific environmental situations without the animal 
having experienced this situation before, from those behaviour patterns which are 
learned by the animal by active communication with the environment during 
ontogeny. We do not know if they are based on two different kinds of mechanisms 
and our term does not imply it. It is probable but still needs to be analysed. The 
separation of both by experiment is a first step towards such an analysis. 

LissAK. The other side of the question is the humoral or hormonal background. 
May I mention that well-known experiment of Richter and the mouse-killing or 
frog-killing test of the lactating animal. The maternal aggressivitv can be avoided 
by a simple injection of ocstrone or by destroying the amygdalate. I suppose that 
the hormonal afterentation must have a very important role in such aggressiveness 
or killing experiments. 

EiBL-EiBESFELDT. There exists a paper published by Roller some years ago, dealing 
with the hormonal control of nest-building behaviour. 

LissAK. His method is very convenient for studying the humoral mechanism too. 

Magoun. Are physiologists making an effort to identify activity of the C.N.S. 
associated with such behaviour? 

EiBL-EiBESFELDT. Dr vou Holst Studied brain mechanism by stimulating certain 
areas in the brain of the domestic fowl — and he released several highly complicated 
behaviour patterns. He will present a paper at Cambridge in September. 

Thorpe. I would like to comment on one aspect which seems important in this 
question of learned and instinctive: that is the complexitv of the stimulation to 
which the animal is exposed. If a bird is raised in a sound-proofroom and it sings 
a normally elaborate song we are justified in saying that this is innate because there 
is no corresponding complexity in the environmental situation to wliich the bird 
is exposed which could have called forth the behaviour. One must assume that any 
observed complexity must come from somewhere; in this case from the animal 
itself If on the other hand you can find a sufficient complexity in the sensory input 
to which the animal has been exposed you are inclined to put it down to the 
account of experience. This is a very important fundamental distinction between 
the two points ot view. The essential point is the origin of complexity in behaviour. 
In so many examples of animal behaviour environmental stimuli may be acting as 
triggers to set off^ complex behaviour patterns but they are not such that they can 
be regarded as the full 'cause of the behaviour; the complex adjustment of it must 
be sought elsewhere and in such circumstances we are fully justified in regarding it 
as internal and in labelling the behaviour as innate. 

Hebb. To add to the question asked by Dr Magoun, I think that it is relevant to 
mention the extensive neurolosiical investitrations b\' Beach made on instinctive 


IRENAUS EIBL-EIBESFELDT 73 

patterns in the rodent. Instinct in behaviour depends on the cortex, and there is a 
close correlation between problem-solving capacity and sex behaviour in the male, 
and problem-solving capacity and maternal behaviour in the female. 

REFERENCE 
Loreiiz, K. (1937) Uber die Bildung dcs Iiistinktbcgriftcs. Die Natuni'iss. 25, 289-300, 307-18, 
3:14-3 1- 


SOME CHARACTERISTICS OF THE EARLY LEARNING 
PERIOD IN BIRDSi 

W. H. Thorpe 

In the last 3 or 4 years much new work on the early learning period in 
birds has been accomplished so I propose in today's talk to attempt to 
survey some of these recent advances and to consider what lines of 
investigation now appear to be the most promising. I also wish to discuss 
the relation of modern developments in the study of bird learning to the 
general body of knowledge and theory concerning the flexible and 
individually adaptable behaviour of animals. 

The acquisition of new actions and the development ot skills by the 
process of trial and error learning has recently been the subject of a good 
deal of investigation. A type of behaviour which is particularly con- 
venient for detailed analysis of this kind is the feeding technique shown by 
a number of species of birds, particularly tits [Paridav), when presented 
with a piece of food suspended by a thread. Individuals of many species of 
birds will sooner or later learn to pull up the food, the pulled-in loop being 
held by the foot whilst the bird reaches with its beak for the next pull. In 
goldfnichcs {Carduelis carducUs) at least this behaviour has been known 
from tunc immemorial. Goldfinches are so adept at the trick that they 
have for centuries been kept in special cages so designed that the bird can 
subsist only by pulling up and holding tight two strings, that on one side 
being attached to a little cart containing food and resting on an incline, 
and that on the other a thimble containing water. The keeping of birds in 
this type of cage was so widespread in the sixteenth century that it gave 
rise to the name 'draw-water', or its equivalent, in two or three European 
languages. Nor was this type of aviculture restricted to Europe. Dr Dillon 
Ripley tells me that Parus variiis in Japan is kept as a cage bird and is 
taught many fancy tricks, such as the solving of little puzzles, running a 
betting bank, and pulling up strings. No doubt all of these depend upon 
trial and error learning of the kind described. 

Those of us who have seen this kind of performance by wild birds at a 
bird table, as can so easily be done with the great and blue tits (P. major 

^ Part of this work has been pubhshcd in Nature, 182, 554-7, 1958, and part in Ibis, lOi: 
337-353. 1959- Full references will be found in these two papers and in "Current Prohleim oj 
Auiiihil BclhU'iour" cd. W. H. Tiiorpc and O. L. Zangwill, Canib. Univ. Press, i960. 

75 


76 BRAIN MECHANISMS AND LEARNING 

and p. caeriileiis) in Britain, can hardly have failed to have been impressed 
by the smooth easy certainty with which the complete act is accomplished. 
Under these conditions one seldom sees anything so suggestive of trial 
and error learning. On the contrary, the act appears at first sight to be a 
real and sudden solution of the problem from the start, as if the bird, 
before responding to the string at all, had seen the answer to the problem, 
and as if its behaviour was the planned result of this insight. But when one 
comes to investigate the development of such behaviour in the nidividual, 
one gets a very different picture. Far from it being a smooth and easy 
response, one fmds instances of birds which learn with extreme slowness 
and difficulty, acquire some part of the response only to lose the abihty 
again, or get to a certain point in the acquisition of the trick and arc unable 
to complete the learning. 

One point that emerges clearly, however, is that it is much easier — 
other things being equal — for a bird to take food from a short string 
than a long one ; and that an essential stage in the training of birds to feed 
by pulhng up strings is to get them to take food from a short string, say 
2 inches long, where the seed is reached without the necessity of actually 
pulling in the string. Once this has been achieved, the problem posed by 
an increase of length to 2| inches, which must then be pulled in before the 
seed can be eaten, is a relatively easy one; although it of course imme- 
diately brings in a new movement, that of holding the string. Once the 
string has been held and the bait secured in this manner, a gradual increase 
in the length of string can very often be made without causing the bird 
any great further difficulties or setbacks. 

Prehminary observations at the Madinglcy Field Station for the study 
of Animal Behaviour, Cambridge University, with captive wild-caught 
and hand-reared tits and finches, gave puzzling irregularities, uncertainties 
and differences in individual and specific behaviour. It was clear that 
many more factors were coming into the process of acquiring this rather 
striking new feeding technique on the part of the birds and that a careful 
experimental study of it might be expected to give results not merely 
of interest to the ornithologist, but of wide application to problems of 
animal learning in general; and possibly even to bear on problems of 
human learning. Miss M. A. Vince of the Cambridge Department of 
Experimental Psychology and of the Madingley Field Station accordingly 
took up this study which, although far from complete, has already 
yielded results of remarkable interest (Vince, 1956-59 and in press). First 
experiments with adult wild-caught great tits showed that the bird's 
first response to the experimental set-up was to keep away from it, but 


W. H. THORPE 77 

that with further trials the bird's positive response to bait and or string 
tended to increase over a period of time. Subsequently the bird's response 
to both might tend to decrease and even birds which were learning well, 
or in which the string-pulling habit had been completely mastered, might 
be set back, or the habit lost, when the experimental situation was changed 
or the bird was moved to a new cage. 

Further tests, not only with wild and hand-reared blue and great tits 
but also with greenfmches {ChIori<; chloris), chaffmchcs [Friti'^iUa coelebs) 
and canaries [Serimis c. canarius), have thrown a good deal of light on 
what is evidently a very complex situation. It seems clear that juvenile 
birds tend to be superior to adults in the string-pulling situation and that 
this success can be correlated with the amount of time spent responding 
to the string aiid/or the bait. This amount of time responding could be 
due to a higher level of activity, and in tasks requiring a high level of 
activity (requiring, in fact, large persistence and many trials) juveniles — 
which are at their maximum of exploratory activity at about 13 weeks — 
are likely to be more efficient than adults. But positive response is not the 
whole of learning, and in such a task as string-pulling it is necessary also 
for the bird to learn to refrain from responding to situations in which 
reinforcement is absent or withdrawn. This ability to refrain is, as Vincc 
shows, dependent on something similar to what Pavlov called internal 
inhibition.^ It is weak in very young birds, develops as a result of age, and 
as a result of experience during the juvenile stage, is still being strengthened 
at an age when positive responsiveness is well past its peak and later 
weakens sHghtly again. Consequently tasks which depend less on the level 
of activity, and more on the ability to control behaviour by precisely 
timed inhibition are more likely to be mastered quickly and efficiently by 
older birds than by younger ones. Other experiments in which great tits 
were required to obtain food by opening dishes covered with white lids 
and refrain from opening dishes covered with black lids gave further 
confirmation of this view. These studies also suggest that internal inhibi- 
tion is unstable in young birds and that the process of development from 
the unstable type of inhibition found in younger birds to the more stable 
type found in older birds, depends not only upon age but also on experi- 
ence. 

In another experiment, great tits were first trained to feed from a blue 
dish. When they had become habituated to the blue dish, a white 

^ A full discussion of the relationship between Habituation and Internal Inhibition would 
be out of place here; those interested will find it dealt with at some length in my book 
(Thorpe, 1956, 57-62). 


78 BRAIN MECHANISMS AND LEARNING 

cardboard lid was placed over it. The number of trials (i.e. the reinforced or 
training trials) needed for the consistent removal of the white lid was then 
recorded. Unreinforced trials consisted of empty dishes with black lids. 
When hand-reared birds were compared in this test with wild-reared ones 
caught asjuveniles over a period of 8-20 weeks after fledging, both groups 
showed internal inhibition (indicated by the ability to refrain from remov- 
ing the black lid) rising as a function of age and later falling slightly. But 
there are indications that a richer and more varied experience may 
change the slope of the curve, giving a sharper rise to a higher level. It 
seems as if, at any rate in this experimental situation, richness of early 
experience is an important factor in contributing to the perfect mastery of 
a task. Thus in a task primarily requiring superabundant activity (as, for 
instance, does maze learning, where a large number of different cues have 
to be investigated), younger animals arc more likely to excel. On the 
other hand, a task which requires activity directed to a particular feature 
of the environment may well be learned better by older animals. Both 
these features arc shown in the studies of bird learning which I have been 
describing. 

To sum up, there are four aspects of development which have emerged 
from such experiments. Firstly, there is the question of responsiveness; 
this appears, in the hand-reared great tit, to increase with age and then 
decrease. Secondly, there are changes in a different type of responsiveness, 
namely habituation or internal inhibition. Here again there appears to be a 
rise and also probably a slight subsequent fall with age, but the rates of 
the first and second changes may be quite different, and so the variation in 
these two factors couJd well give rise to the puzzling differences in learning 
ability which the earlier experiments revealed. Thirdly, there is the 
question of the hunger drive. The birds in this work were kept under 
conditions such that approximately the same measure of food deprivation 
was experienced by all. Fourthly, the effects of early experience or 
environment on behaviour are clearly shown and it seems that internal 
inhibition may develop more rapidly and more completely in a more 
varied environment, presumably as a result of greater activity, greater 
stimulation and perhaps greater opportunity for adaptation and so on. It is 
plausibly assumed that aviary rearing is a richer experience than hand- 
rearing and that rearing in the wild is a richer experience still. 

hiipriiitiin^. The phenomenon of imprinting is now so well known to the 
student of bird behaviour that a definition is hardly necessary. Recent 
experiments have confirmed that the period during which young birds 
can tirst learn to follow moving objects is strictly limited. But once a 


W. H. THORPE 79 

bird has learnt to follow models, the ability to transfer from one type to 
another can be demonstrated in birds as young as 3 days and as old as 
60 days; although when first presented with a new model they will show 
a slight initial diminution in response, indicating that they perceive a 
difference and that the new model is not at first quite as acceptable as 
the old. With coot and moorhen, Hinde, Thorpe and Vince (1956) found 
that in the early days of life the following-response is stronger than the 
tendency (which is also present) to flee from strange objects. The result is 
that, in the course of the following experiments, the younger bird gets 
used to a number of strange objects and so its fleeing tendency is weakened 
by habituation. But the bird's response is always ambivalent in some 
degree and later the fleeing drive becomes stronger, and more difficult to 
habituate; and consequently, with age, a strange object becomes less 
likely to elicit following. In the coot, practically all the waning of a 
following response with age can be attributed to this growth in the 
fleeing drive and in these species no evidence was forthcoming for any 
permanent effect able to influence instinctive behaviour patterns which 
are to mature later. In some species, however, including some duck such 
as the mallard (Weidmann, 1955, 1958) and also geese, there seems little 
doubt that the tcrnunation of the imprinting period cannot be due solely 
to the development of a competing fleeing drive but there that is an 
internally controlled waning of the tendency to follow, independent of 
ihe development of fear responses (jaynes, 1957). 

It is clear that, in the kind of experiment which I have just described, the 
foUowing-reponse can be said in some way to be 'its own reward'. All 
that the bird is doing in the first instance is giving rein to a need to 
maintain itself by its own efforts in constant spatial relationship with a 
moving object. This moving object can be almost anything which is not 
too large or too small, and which does not move too fast or too slowly. 

To sum up this work on imprinting to visual stimuli, we can say 
tentatively that the imprinting period is initiated primarily by the matura- 
tion of an internally motivated, appetitive behaviour system which is in 
readiness, at the time of hatching or very soon after, to express itself in the 
following-response. The time during which this internally controlled 
tendency can find its first expression in action is limited, to a matter of a 
few hours or at most days, by internal factors; but during this time the 
response is ready to appear as soon as the circumstances permit it to do so. 
The termination of the ability to make that first response is, as we have 
seen, very largely influenced by the development of competing fleeing 
responses. Nevertheless, the evidence remains very strong that, in some 


80 BRAIN MECHANISMS AND LEARNING 

species at least, an internal change directly reduces the appetitive following- 
drive with age. Conversely, once the following-behaviour has had a 
chance to establish itself, it will continue towards those models which 
have become associated with the achievement of the consummatory 
situation resulting from successful following, and may be yet further 
generalized to others. 

It is clear that the broad nature of the stimulus to which following 
animals respond represents a difference only of degree from other instinc- 
tive responses, especially those of young animals. Not only does the 
following-tendency itself become primed through exercise and so more 
firmly established, but the following gives opportunitv for the animal to 
become conditioned in various ways to the characteristics of the object 
which has elicited the response in the first place. 

Imprinting can occur in response not only to visual stimuli but to 
auditory ones also. Thus an animal with an initially weak tendency to 
respond to both objects and sounds within a wide range may learn to pay 
attention to, and ultimately to follow, certain sounds which have been 
associated with the visual stimuli first releasing the following response. 
This may work both ways. Auditory stimuli may be conditioned to 
visual characteristics of the object, and vice versa. The recent studies of Dr 
Klopfer at Madinglcy and in the United States have given particularly 
good examples of this. In the first place certain surface-nesting species of 
waterfowl, chiefly mallard and redhead [Aytliya aimricana), were examined 
and it was found that they were able tc^ learn appropriate sound-signals 
involved in the maintenance of the brood-parent relationship as a con- 
sequence of visual imprinting. No unlearned preferences for any particular 
auditory signals seem to exist; they tended to approach most rhythmic 
repetitive signals without distinguishing between them. Nor could 
auditory imprinting to a particular one occur in the absence of visual 
stimulation. When Dr Klopfer came to investigate the wood duck (^4/.v 
sponsa) which is a hole-nesting species, the situation was found to be 
reversed: they did not initially tend to approach the repetitive signals, but 
exposure to a sound signal alone could produce a subsequent preference 
for that sound, thus demonstrating auditory imprinting. Further, the 
imprinted response was not reinforced or altered in any way by following 
or by a visual stimulus. This fact can be explained by the nesting habit of 
wood ducks, which require that the young duckling's first response to the 
mother be based on auditory stinauli. In surface-nesting birds, visual cues 
can play a larger role from the start, and thus assume paramount import- 
ance. The quackless muscovy duck [Cuiriiia niosclmra) was shown to be 


W. H. THORPE 8l 

incapable of auditory learning under the conditions oi the experiment, and 
this striking failure in a species whose wild ancestors arc closely related to 
the wood duck is supposed to be a consequence of domestication. Klopfer 
(1959) then studied the following-responses of European shelduck 
[Tadorua tadoriia). Young oi^ the species hatched and reared in sound- 
proof rooms showed no tendency to approach repetitive sound signals 
when tested at 18-26 hours of age. In this they were similar to wood ciucks 
and domestic muscovies, and unlike any species of surface-nesting water- 
fowl. However, with the shelduck, exposure to repetitive sound signals in 
early life produced no change in the response pattern, whereas in the wood 
duck it did so. Nevertheless, associating the sound signals with an object or 
person which the ducklings were allowed to follow for short periods 
enabled a highly specific preference for sound to be developed, and in this 
shelducks resembled several species of surface-nesting waterfowl, particu- 
larly of the genus Auas. Of course this work does not prove that there are 
no sounds to which these birds would respond in the absence of visual and 
motor experience, but it does appear that a preference can be established 
for sounds which are linked to a visual model. As in the surface-nesters, 
one can suppose that the following-response serves as a necessary rein- 
forcement in the learning of particular sound signals. Thus the behaviour 
of the young shelduck does not show the pattern which would seem to be 
most appropriate for hole-nesting species, for whom auditory stimuli 
should be of much greater importance than visual ones. Moreover, hole- 
nesting species should either be endowed with response tendencies to 
specific auditory stimuli at the time of hatching, or else highly susceptible 
to auditory imprinting. That, for instance, seems to be true of the wood 
duck. It would thus seem likely that newly hatched shelducks emerge 
from their burrows in response to visual or perhaps tactile stimuli, with 
auditory cues assuming a secondary importance. The fact that shelduck 
will nest in thickets above ground when burrows arc not readily available 
also suggests a possible explanation tor this difference in behaviour. Here 
is an interesting opportunity for a held investigation which will be 
essential before the whole matter can be completely understood. 

In nidicolous birds this particular type of following-response cannot, of 
course, operate. Nevertheless there is in certain song birds good evidence 
for something very suggestive of imprinting in the process of learning 
characteristic song, and indeed the vocalizations of birds can be regarded 
as offering particularly crucial problems concerning the acquisition of 
complex behaviour patterns as a result of individual learning. 

The normal song of the chaffinch is an elaborate integration of inborn 


82 BRAIN MECHANISMS AND LEARNING 

and learned components, the former constituting the basis for the latter. 

A good example of a normal song of Friin^illa coclchs qciiojcri is given in 
Fig. I. The inborn component of the chaffinch song can be revealed by 
hand-rearing the young birds from early nestling life either in acoustic 
isolation or at least out of contact with all chaffinch song. Six birds thus 
individually isolated produced songs of an extremely simple type (Figs. 
2-4) consisting of a song-burst of approximately the correct length (2-2.5 
seconds) made up of about the right number of notes. The pitch, or funda- 
mental frequency, of these notes was somewhat lower than normal and 
the songs produced by these isolates lacked the division into the three 
phrases so characteristic of the normal chaffinch song. Moreover, the final 
flourish and all the other tme details by which the chaffinch song is 
normally recognized as such were also absent. Of all the hundreds of wild 
and aviary-kept birds which have passed through our hands during the 
course of these experiments, none has produced songs of such extreme 
simplicity as these six birds, although such undeveloped songs are known 
to occur in the wild at times. 

In contrast to the simple, restricted song produced by the isolated birds, 
we find that if, after babyhood, two or more such birds are put together 
in a room but still without the opportunity of hearing experienced 
chafhnches, they will develop more complex songs. It seems that the 
attempt to sing in company produces mutual stimulation which en- 
courages the development of complexity. The members of each group of 
hand-reared birds thus kept together will, by mutual stimulation, build 
up a distinctive community pattern. The birds conform so closely to this 
pattern that it is sometimes barely possible to distinguish the songs one 
from another even by electronic analysis. The song of such birds may be 
quite as complex as that of a normal wild chaffinch, but its complexity 
tends to be of a different kind and a song thus produced may bear little 
resemblance to the characteristic utterances of the species. From further 
experiment it is clear that, in the wild, young chaffinches learn some 
features of the song from their male parents or from other adults during 
the first few weeks of life. But most of the finer details of the song are 
learned by the young bird when, in its first breeding season, it first comes 
to sing in competition with neighbouring territory-holders. There is little 
doubt that this is the way in which local song dialects are built up and 
perpetuated. In addition to this true song, the chaffinch, like many other 
species, has what has been called a sub-song, which is a much quiet and 
less aggressive affair than is the full song. It is, however, not merely a song 
of low intensity: it is of an entirely different pattern from the true song and 


W. H. THORPE 


83 


/. n 


m\ 


Fic;. I 
'Normal' song of British chaffinch {Friiii;illa coclchs (.'t'/zsj/rn') Vertical 
scale: frequency in kc./s. Horizontal scale: time m seconds (Similarly 
Figs. 2-1 1). 


'i k '1 I 


WWWs 


10 15 2 2-5 


^^WM'^iw««\iWrti\ 


05 


l-O 15 20 2 5 


MhVVVUsv 


Figs. 2-4 
Song of three hand-reared isolates. GW, lune 24th, 1954; B/BkY, 
May 25th, 1957; and BkW/W, June 24th, 1955. 


84 BRAIN MECHANISMS AND LEARNING 

may be described cis a long succession of chirps and rattles (Thorpe, 1955, 
1958; Thorpe and Pilcher, 1958). It seems to have no communicatory 
function and is most frequently heard in the early spring, when it is 
produced, so far as we know, much more by tirst-year birds than by older 
ones — the latter seeming to come into full song with much less of this 
preliminary sub-song. There seems little doubt, however, that the sub- 
song provides in some degree the raw material out of which, by practice 
and by the elimination of unwanted extremes of frequency, the full song 
is 'crystallized'. The chaffinch has, of course, like other species, a number 
of call notes which arc in the main signals for co-ordinating the behaviour 
of the flock, mate and family; some of these may be used as components 
of both sub-song and full song. It is interesting that call notes are much 
more in evidence as components of the songs of isolated birds than they 
are in the songs of normal ones. This is presumably because the isolated 
birds have had a greatly restricted auditory experience to draw upon as 
compared with normal wild individuals. 

Chaffinches are not imitative birds, in that they do not normally copy 
anything but sounds of chaffinch origin. Once a chaffinch has heard a 
chaftinch song as a young bird in the wild, it appears to have learned 
enough about it to refuse to copy any sound pattern which departs far 
from the normal chaffinch song; that is, it will learn only the finer indivi- 
dual variations of the song of other chaftmches. So, in the wild, chaffinches 
practically never, in their full songs, imitate anything but other chaffinches. 
Hand-reared isolated birds will learn songs of far greater abnormality, 
however, provided always that the tonal quality is not too far different 
from that of a chaffinch song. Voices as 'abnormal' as that of a domesti- 
cated canary may be learned by hand-reared birds (Thorpe, 1955), and very 
occasionally by wild birds; but when this happens the alien notes are kept 
as components of the non-communicative sub-song only; the full song is 
not contaminated with them. 

If one catches wild chaffinches in their first autumn and keeps them 
until the following spring with other chaffinches, the song of which is 
already fixed, one gets clear evidence that the young birds have modelled 
their songs, in some respects at least, on those of their associates, which 
have probably come into song first as the spring season comes on, and 
which they thus hear before they themselves have got very far along the 
path of song production. Similarly, if, instead of exposing such wild- 
caught first-year birds to the songs of other chaffinches, one plays such 
songs to them by means of a repeating tape machine which we may call a 
'song tutor', a clear positive effect is manifest. An experiment carried out 


W. H. THORPE 


85 


ill 1954-55 with iour wild-caught birds given such tuition in the winter 
(January 6th-20th) and with three birds similarly treated in February- 
March showed clearly the effect of the 'song tutor', there being correct 
articulation of the phrases and a good approximation in pitch and quality, 
although there were abnormalities in length and emphasis. If, however, 
under these conditions one exposes the young birds to highly abnormal 
songs, no result is obtained. Thus in experiments in 1955-56 such birds 
were given tuition with a reversed chaffinch song during March. The fact 
that no definite effect was obtained suggested that enough of the song had 


^/■//'/' 


25 


Figs. 5 and 6 
Two sons of a hand-reared auditory isolate after having been 
exposed to normal song on the 'song tutor' for lo days only during 
November of its first year. GY/BW, April 24th, 1956. 


already been learned in the autumn for the reversed song to be ineffective 
even with visually isolated birds. In 1956-57 a similar experiment was 
done using six wild-caught autumn males and exposing them in this case 
not to a reversed song, which was thought to be perhaps too abnormal, 
but instead to three different 're-articulated' songs (these are songs in 
which the position of the three phrases has been interchanged). In this 
experiment the birds were isolated in a sound-proof room. Two birds 
had exposures in September and October, repeated in January, two had 
exposures in mid- to late-October and late January and early February, 
and two were similarly exposed in November and February. On the 


86 


BRAIN MECHANISMS AND LEARNING 


whole, this experiment also had little effect, conhrming the tests with the 
reversed song and suggesting even more strongly that the natural training 
which the bird had already received by the time it has reached the fnst 
winter moult has equally been sufficient to prevent it from learning after- 


FiG. 7 
Re-articulatcd model with 'end in middle'. 

wards the rather large abnormalities characteristic of these re-articulated 
songs. 

When we come to similar experiments with hand-reared birds, very 
different results are obtained. It was found that a hand-reared visual isolate 
can be taught features of a normal song from the tape by being exposed to 
it during November only of its tu'st winter for a period of lo days. Figs. 5 


Fig. 8 
Song of hand-reared auditory isolate B/BkP after having been exposed 
to this model for 12 days January-February and 12 days February-March. 
March 25th, 1958. 

and 6 give the two songs developed in 1956 by such a bird treated in 1955. 
The first song shows clearly the influence of the 'tutor' in that it has three 
phrases otherwise unknown in the hand-reared isolate, having a clear 
stepwise descent but no end-phrase. The second song is of the isolate type 
plus two end-notes. These experiments thus produce evidence that tutor- 


W. H. THORPE 87 

ing, even for a short period in November, can effect the same kind of 
result as the normal song experience of a wild bird in the field during the 
autumn. Similar positive results were obtained in similar experiments 
with re-articulated songs carried out in 1956-57 and 1957-58. Figs. 7 and 
8 show such a re-articulated model and its effect on the song of a hand- 
reared isolate. Hand-reared birds show similar differences from the wild- 
caught ones, in that if given a reversed song a considerable amount of 
copying is achieved. Similar experiments with other birds provide 
further evidence for a previous conclusion that although a hand-reared 
isolate is incapable by itself of producing anythhig approximating to a 
normal ending, yet if once it hears a normal end-flourish on the 'song 
tutor' (even though in the rearticulated song on the tape it occurs at the 
beginning or in the middle of the song) it will recognize it as appropriate 
for an ending and will attempt to place it properly in its own song. 

In experiments in 1954-55 ^-i^ attempt was made to indoctrinate hand- 
reared isolate chaffinches with the song of the tree pipit {Ainliiis trivialis) 
(Fig. 9), this being chosen because the spectrograph reveals that the tonal 
quality of the notes is similar to that of the chaffinch. The i:irst experiment 
gave a doubtful result, but a repeat in 1956 resulted (Fig. 10) in the 
achievement of a remarkably good copy by such a chaffincli of the song 
of the tree pipit, producing a chaffinch song entirely unlike any other 
uttered in my experience by a wild or hand-reared bird. It is of interest to 
note that the rather long song of the model has been condensed to conform 
to the standard length of chaffinch song by condensing the middle phrase 
of the tree pipit song to two notes instead of tour. This song became still 
further 'tightened up' and shortened by the beginning of May (Fig. i i). 

The reason why chaffinches do not imitate and acquire songs from other 
species in the wild now seems fairly evident. There is little doubt that they 
restrict their imitativeness to the right models as a result of being respon- 
sive only to notes of approximately the right tonal quality. It is interesting 
that the imitation of the tree pipit song just referred to is the best copy of an 
alien song so far obtained. Here the tonal quality of the model was right, 
although the song itself was much too long and the phrasing of the notes 
of the model abnormal from a chaffinch point of view. 

To summarize this work with the 'song tutor', it can be said that in 
eleven experiments carried out over the years 1955-57 a-i^d involving 
thirty-four wild autumn-caught first-year males, the only positive effect 
obtained (in the sense of a significant similarity between model and 
mimic) occurred in two experiments (seven birds) only, and in these two 
the model was a normal chaffinch song. Of the remaining ones (nine 


BRAIN MECHANISMS AND LEARNING 


experiments, twenty-seven birds) the models were always abnormal — 
the songs being artificial, re-articulated or reversed — and the results were 
always negative. The contrast when hand-reared isolated birds are used is 


AAA A 


\ 


3-0 


Fig. 9 
Song of Continental tree pipit {Aiithtis trividlii) used for tutoring 


Fig. 10 

'Copy' of tree pipit song produced by chaffinch R/R after tuition 

on 'song tutor'. Apnl 21st, 1956. 


mxnt 


Fig. II 
'Improved' tree pipit copy. R/R, May 3rd, 1956. 


Striking. Thus during 1954-59, thirteen experiments involving sixteen 
hand-reared birds were carried out using similar song models. Of these, 
ten experiments (eleven birds) were positive, in the sense of yielding a 


W. H. THORPE 89 

iignihcant effect. The remaining three experiments (tive birds) gave a 
negative result. 

The counter-singing that occurs between birds in adjacent territories is 
an important factor in stimulating and restricting the imitative abilities of 
chaffinches. When a chaffinch has acquired more than one song-type, each 
song-burst consists of a sequence of one song-type followed by a sequence 
of another. When songs are played back to a chatiinch, we find that those 
songs which it uses most frequently itself arc the most effective in evoking 
song (Hinde, 1958). A chaffinch in the wild will thus tend to reply to a 
neighbour with that song of its own repertoire which most nearly 
resembles the song of its rival. 

It is suggested that, with the more 'imitative' finches, such as the bull- 
finch [Pynhuhi pyrrliiila), hawfinch [Coccothratistcs coccotliriiiistes) and to 
some extent greenfinch [Cliloris cliloris), the song, while functional in co- 
ordinating the breeding cycle and behaviour of the mated pair, is of less 
importance as a territorial proclamation. Thus in some respects it re- 
sembles the sub-song rather than the full song of the chafimch; similarly, 
contamination with alien notes can be tolerated to an extent which might 
be very disadvantageous in a strongly territorial song that must maintain 
its character as a reliable specific recognition mark. A preliminary study of 
a number of species of buntings shows that many of the species, for 
example reed and corn buntings [Embcriza schociiichis and E. calaiidrn), 
have songs which arc highly stereotyped and completely innate. The song 
of the yellow bunting (R citriiiclla) appears to consist of an integration of 
innate and learned components much as in the chaffinch. The buntings as 
a whole appear to have strongly territorial songs corresponding to that of 
the chaffinch in function. 

Apart from a very few and partial exceptions, chatiinches can learn song 
patterns only during the first 1 3 months of life, and towards the end of 
this time there is a peak period of learning activity of a few weeks during 
which a young chaffinch may learn, as a result of singing in a territory, the 
fine details of as many as six different songs. This special period of high 
learning ability is brought to an abrupt close by internal factors. It matters 
not whether a chaffinch has learned one or six songs by the time it is 
13 months old, it can afterwards learn no more, and so remains with its 
one song or its six for the rest of its life. This restriction of learning ability 
to a particular type of object and to a sharply defined sensitive period 
recalls the phenomenon of imprinting. 

The inborn recognition and performance of its specific song by the 
chaffinch involves (a) duration of approximately 2^ seconds, (h) interval 


90 BRAIN MECHANISMS AND LEARNING 

between songs of between 6 and 20 seconds, (c) tonal quality (though this 
last may have been learned by the experience of the bird of the qualities 
of its own voice). The readiness with which the bird learns to divide its 
song into three sections and learns to attach a simple flourish at the end as 
an appropriate termniation shows that there must be an imperfectly 
inherited tendency to respond to, and perform, tlie features of the normal 
song as soon as the singer is stimulated by hearing another bird. Such an 
inherited tendency would explain the basic similarity of the songs of the 
species throughout its range. 

In conclusion, I believe that recent studies of bird behaviour have added 
a considerable body of evidence in support of the view, now being 
advanced from many quarters (sec Thorpe & Zangwill, i960) that in 
regard to both learned and innate behaviour patterns the organism is to be 
regarded as 'searching' largely for cues and stimuli, consummatory 
situations in fxct, rather than for the release of consummatory acts. This 
is to say that the 'goal' of the behaviour is not solely or even primarily the 
discharge of a specific motor-nicchanism but the achievement of a 
specific sensory stimulation. Moreover, in this connection we must 
remember that the animal perceives its own consummatory acts primarily 
through its own interoceptors and proprioceptors, and that in the majority 
of cases at least it is again a special pattern of sensory stimulation, a re- 
afference, which is the effective reward and not simply the general level of 
well-being or activity resulting from the concomitant adjustment of 
nutritional or endocrine balance. 


GROUP DISCUSSION 

Magoun. 1 am trying to relate this behaviour to mechanisms ot the brain. It is 
possible in the cat and in the inonkey to introduce electrodes into the peri-aqucductal 
grey or the tegmentum of the midbrain and by repetitive stimulation to evoke 
flicio vocal responses which are identical with the normal expression of emotion by 
these animals. These can be elicited when the brain stem is transected just ahead of 
this region, so evidently one is not exciting an ascending pathway to some higher 
level. After this region of the cephalic brain stem is destroyed elcctrolytically, the 
cat no longer vocalizes in response to an adequate environmental stimulus. This 
suggests that there is built into this part of the brain a mechanism for faciovocal 
performance in the expression of emotions, to be differentiated from the speech- 
mechanisms that have developed in the association cortex in the dominant hemi- 
sphere of man. This had led me to wonder whether it would be possible, by intro- 
ducing electrodes into the midbrain, to evoke those patterns of vocal performance 
which you point out as being so specific for individual birds. Have experiments 
along those lines been tried? Do you think they might be feasible; 

Thorpe. Yes, stinnilation experiments are being performed, but not so tar with 


W. H. THORPE 91 

song birds which arc, ot course, very active and mostly small. We have started 
some ablation experiments with song birds, but these have not got very far yet. It 
is as yet very difficult to stimulate the brain of a very small bird so that it will behave 
in a fairly normal fashion and still sing. I think it will be some rime before the 
technique is developed to the point at which signihcant results will begin to accrue. 

Fessard. What about parrots f 

Thorpe. Parrots should certainly be easier to investigate but, although they have 
of course an astonishing facility for vocal mimicry, would not be much good tor 
the study of the innate sounds. In this matter of vocal imitation the Indian hill 
mynah excels even the parrots. But perhaps the most puzzling feature is that 
neither parrots nor mynahs appear to use their power of mimicry in the wild at all 
(Thorpe, 1959). They have stereotyped cries and calls which seem to serve tor 
co-ordinating the movements oi the tiock but show no sign of mimicry. From the 
point of view of the study of vocal learning these species would be admirable; but 
not for studying mechanisms tor innate sound production. 

BusER. It might be of interest to mention here observations made in Paris by 
Dr Rougeul and Dr Assenmacher, studying the electrical activity of deep brain 
structures in unanacsthetized ducks by stereotaxic methods: each time when simply 
introducing or moving up and down the Horsley-Clarke electrode, the animal 
would start quacking whenever a certain level was reached. King in the tegmental 
midbrain. Electrical stimuli were not systematically tried and the mechanical 
excitation produced h\ the electrode appeared actualK' surticient to produce the 
described effect. 

KoNORSKl. I would like to make two comments. The tirst one is this: As far as I 
know, only birds, and among mammals only man, have the ability ot reproducing 
sounds, i.e. they possess what mav be called acoustic-vocal reflexes of a specitic 
kind. This ability seems to indicate that in these animals direct connections exist 
between the acoustic area and the part of motor area concerned with vocalization. 
We know that in man such connections in fict exist between anterior temporal 
area and frontal opercular region. If the}- are destroyed the peculiar torm of 
asphasia ensues. The patient understands what is said to him and is even able to 
verbalize his thoughts, but he is not able to repeat exactly the words he hears. It is 
much easier for him to say something spontaneously than to repeat it. It would be 
very interesting to know whether such an acoustic vocal system exists in birds, and 
if so, to study it both from anatomical and physiological points ot view. 

The second point concerns the problems of imprinting. We know trom exten- 
sive experimental data (the literature is given in a paper by Konorski and SzweJ- 
kowska, 1952) that classical conditioning shares one important property with 
imprinting, namely its partial irreversibility. By the way, this is why I do not like 
the term 'temporary connections' because these connections, once established, are 
not temporary but stable, and, as I shall discuss later, they are not obliterated by 
'disuse'. If you elaborate an alimentary conditioned reflex to some stimulus, it is 
much more difficult to transform it afterwards into, say, a detensive conditioned 
stimulus b}' changing the reinforcement trom tood to shock, than to elaborate the 
defensive conditioned reflex to quite a new stimulus. Even after a long defensive 
training of the alimentary conditioned stimulus, its previous alimentary role may 
be detected. We explain these facts by saying that the old conditioned connections 
established between the central representation of the conditioned stimulus and 


92 BRAIN MECHANISMS AND LEARNING 

feeding centre are by no means removed by the new training but the new connec- 
tions are formed independently of them. 

This suggests a form of continuity between the inborn reflexes, imprinting and 
normal conditioning and shows that there arc perhaps no sharp limits between 
these phenomena. Experiments made long ago in Pavlov's laboratory by Citovitch 
are also relevant here. This author found that the alimentary reflex to the smell of 
tood is not inborn but conditioned because it appears only after a few reinforce- 
ments of the stimulus, but this reflex is established very rapidly and is then very 
stable. I think that we could label this phenomenon either as conditioning or as 
imprinting. 

Thorpe. Some experiments have been done with deafened birds by Schwartakoft 
and Messmer. The production of the essential elements in an innate song is not 
fundamentally affected by being deafened but there may be a change in the overall 
'pitch' of the song. This seems to provide a sharp distinction from conditioning. 

EsTABLE. In certain species of birds, males sing more and more and differently 
when exliilarated by the presence of the female or when after being in the neigh- 
bourhood of the female the latter is removed from his sight. Hormonal influences 
have been mentioned as possible explanation of this, but though they are probably 
necessary, they cannot be suflicient to explain the production ot such a variety oi 
songs. Besides, an exclusively endocrine mfluence cannot explain the flict that the 
male sings better than the female; the nervous system and tlie sound-producing 
apparatus must necessarily participate. 

When the male bird is in the presence of the female, it sings ; if the latter is removed, 
it sings even more. When the female is absent it stops singing. What happens in 
the brain of the male bird would be interesting to know. Is there some endocrino- 
logical influence in the mechanism of the brain which conditions this behaviour ; 
For example, the problem of the gestalt theory if the conditioned reflex is the 
cause, if it is first or second. I have talked on this subject with Kohler, who was a 
pioneer in this field. He believes that the gestalt structures precede conditioning. 

Segundo. To complement the observation of Dr Baser I would like to note that 
we have implanted electrodes in the mesencephalon of the Tero bird [Beleuopterus 
chilciisis). Electrical stimulation with threshold voltage produced investigation or 
orientation movements; with higher voltages, the bird stretched its wings, started 
to fly and in addition, squawked loudly. (Silva, Estable and Segundo, unpublished 
observations.) 

Grastyan. a marked following reaction could be observed in some oi our 
experiments by stimulating the hypothalamus in cats. This behaviour seemed to be 
also in close correlation with the function of the rhinencephalon. You say that 
there is a limited period of the animal's life when this reaction can occur naturally. 
In which stage of maturation of the central nervous system does this reaction occur? 
Which structures are responsible for its integration in the birds? 

Doty. I have seen this following reaction consistently in cats deprived of all 
visual cortex in infancy. They will approach and follow moving acoustic stimuli 
and persist in this activity for minutes. 

Grastyan. l^r Adey has made similar observations, after the destruction ot the 
cntorhinal cortex in the marsupial phalanaen. 

Adey. After a limited reaction of the hippocampus cortex which forms the 
entorhinal area. The animal was tamed and at the same time would follow anv 


W. H. THORPE 93 

moving object. This was not accompanied by any aggressive behaviour towards 
the object they were following. They normally attack with jaws or teeth objects 
that come within the visual field. This ceases after ablation of the entorhinal cortex. 

Magoun. I am glad to hear reference to the rhinencephalic or limbic brain in 
this connection because I think attention should be called to the generalizations of 
Paul Maclean that this part of the C.N.S. is concerned with these types of behaviour 
which preserve the individual on feeding or aggression or defence, and which 
preserve the race by managing sexual behaviour. 

Hernandez-Peon. The curves that Dr Thorpe showed us concerning the rate 
ot development of internal inhibition may explain the great individual variation 
observed during experimentally induced habituation. I wonder if he could elabo- 
rate the matter further and tell us what is the presently available evidence obtained 
by using a systematic approach. Wliat kind of responses have been tested and 
which animals have been studied f 

Thorpe. The subjects were all birds ot tlie following species: Canary {Scriiiiis c. 
caiiariiis). Greenfinch {Cliloris chloris) and Great Tit [Pants itiajor). These curves refer 
to string-pulling behaviour and to learning to take food from two types of dishes, 
one with a black lid and one with a white one. In the string pulling the wild bird, 
it it learns the trick at all, often behaves with extreme precision and the whole 
movement appears highly organized. In experimental studies, however, the birds 
present very varied behaviour. This, we hope, will be explained in the future as a 
result of this discovery of the diiferent rates of change of these factors, namely the 
level of responsiveness and of internal inhibition, which arc themselves partly an 
expression of the previous experience of the bird. 


SOME ASPECTS OF THE ELABORATION OF 

CONDITIONED CONNECTIONS AND FORMATION OF 

THEIR PROPERTIES 

E. A. ASRATYAN 

Much in the field of conditioned reflexes was achieved by Ivan Pavlov 
and his pupils and by his other followers ni many countries of the world. 
Nevertheless, many aspects of this problem have not yet been satisfactorily 
investigated and require further experimental study and theoretical 
elaboration. At present, many scientists in the U.S.S.R. and elsewhere 
are studying, as Pavlov's scientific heritage, the elaboration of conditioned 
reflexes, the formation of their properties, their preservation, anc^ the 
significance of various factors acting favourably on these processes. My 
collaborators and myself are among them. 

At present the main trend of our experimental and theoretical investiga- 
tions in this field is the study of the role of the strength of stimuh and 
their sequence in the elaboration of conditioned reflexes and the formation 
of their properties. The purpose of this paper is to report our findings and 
present our interpretation of them. 

The important role played by the relative strength of the excitatory 
processes which take place in nervous structures and especially in the cor- 
tex, is generally known. Pavlov has stated: 'Prime importance must be 
ascribed, of course, to strength relations in the activity of the cortex.'^ He 
also repeatedly emphasized the great importance of the pattern of com- 
binations of stimuli in time for the elaboration of conditioned reflexes. He 
stated that 'since conditioned stimuli play the role of signals, they must 
become effective only wlien they precede the signalized physiological 
activity. '- 

Our collaborators investigated these problems in dogs in the usual 
semi-soundproof chambers designed for the study of conditioned reflexes. 
In order to elaborate conditioncc^ reflexes with definite functional pro- 
perties, we combined in various ways either two so-called 'indifferent' 
stimuli, or two different unconditioned stimuli, or one 'indifferent' and 
one unconditioned stimulus. The characteristic feature of our experiments 

^ I. P. Pavlov, Complete Works, vol. Ill, 1949, p. 382. 
- 1. P. Pavlov, Complete Works, vol. Ill, 1949, p. 381. 

95 


96 BRAIN MECHANISMS AND LEARNING 

was the attempt to select and apply only those 'indifferent' stimuli which 
provoke in experimental animals reflex reactions which can be objectively 
observed and graphically recorded. 

I. In 1958, our collaborators Varga and Pressman pertormed a scries 
of experiments in dogs in which they combined two so-called 'indifferent' 
stimuli, namely, a sound and a passive lifting of the animal's leg. In some 
of these experiments stimuli were always applied in the same stereotyped 
sequence, m others in a variable sequence. The advantage of a passive 
movement of the leg as one of the 'indifferent' stimuli is that it gives 
direct and immediate indication of a connection between the combined 
stimuli when the movement of the leg or the electromyogram of its 
muscles is observed. In addition, with these indicators the investigators 
were able to study directly the whole process of formation of conditioned 
connections when these stimuli are combined from the very beginning of 
the connections to any subsequent stage of their evolution. 

((?) In experiments by previous investigators of this problem (Podko- 
payev and Narbutovich (1936), ZeHony (1928), Oreshuk (1950), Roko- 
tova (1952), Karmanova (1955), Bregadzc (1956), Sergeyev (1957) and 
others, this was not possible. The reason was that they used only the 
method of intermediated and, besides, sporadic indication ot the process 
of elaboration of connections between 'indifferent' stimuli. Data obtained 
by Varga and Pressman clearly show that between the brain points of 
these stimuli a double (or two-way) 'associative connection', i.e. actually 
a conditioned reflex connection is established. This occurs not only when 
sound and passive lifting of the animal's leg are combined in a variable 
sequence, but also when they are combined in a stereotyped one. (Fig. i) 
Further, in the course of their experiments, Varga and Pressman also 
obtained data which on the one hand corroborate facts previously estab- 
lished by other workers, and on the other are new. For example, they have 
shown, in conformity with the findings of other authors, that the connec- 
tions between 'indifferent' stimuli are characterized by extreme fragility 
and instability. After being established, they rapidly begin to weaken and 
soon disappear after a period of more or less regular function. They 
recover again for a very short time but only after a certain interval, 
during the experiment, or when an alimentary conditioned reflex has been 
elaborated in response to one of the combined stimuli. 

(/;) At the same time the experiments of Varga and Pressman established 
other facts which illustrate the great signiticancc of a detnntc sequence in 
combining stinndi. Their experimental data show that although the 


E. A. ASRATYAN 97 

above-mentioned properties — fragility and instability — are characteristic 
for both conditioned connections between the brain points of two stimuli, 
they are not inherent in these reflexes in an equal measure. When sound 
and passive lifting of the animal's leg are combmed in a variable sequence, 
the conditioned connections arising between them are almost equivalent 
in their rate of formation, consolidation, stability, resistance to extinction, 
etc. But when these stimuli are combined in a stereotyped sequence, the 
resulting conditioned connections essentially difi^er from each other in all 
of the above-mentioned criteria. A direct or forward conditioned connec- 
tion, i.e. a connection leading from the preceding stimulus (P) to the 


/ iii»i w >h>i ) iiit " 'i ^.i^ M >^^ *>iiit->41ii4»>iHi 

Z ZZZ^ZZI 


I 

Fig. I 
Conditioned connections resulting from various combinations of a sound and a passive lilting 
of the dog's leg. (a) Active lifting of the leg (indicated by the rarefiction or disappearance of 
electrical potentials in the extensor of the leg, i.e. in the gastrocnemius muscle) in response to 
the application of a tone after its thirty-seven combinations with a passive lifting of the leg in 
the stereotyped sequence: 'tone — passive lifting of the leg' ; (fo) reaction to a tone after twenty- 
eight combinations of a passive lifting of the leg with this tone in experiments with a reverse 
sequence, i.e. 'passive lifting of the leg — tone'. 

I. — clectromyogram of the gastrocnemius muscle; 2. — mark of conditioned stimulation. 

subsequent one (S) becomes elaborateci more easily and is more powerful 
and resistant to extinction and other altering factors, than a reverse or 
backward conditioned ct^nnection, i.e. a connection leading from S to P. 
Further, a direct connection recovers more rapidly than a reverse one 
after an acute extinction or when, following their natural disappearance, 
a break is made in the experiment, or when one of the combined stimuli 
turns into an alimentary conditioned signal. Essentially similar results 
were obtained by Varga and Pressman in a series of experiments on other 
dogs in which they combined passive lifting of the animal's leg with the 
blowing of air into the animal's eye which evoked the blink reflex. The 
advantage of the combination of these two stimuli over those applied in 
the previously described experiments was that both stimuli (and not only 


BRAIN MECHANISMS AND LEARNING 


one of them) called forth effects which could be recorded objectively. 
Moreover, as shown in the course of the experiments, the new stimulus, a 
puff of air in the animal's eye, applied in the given combination, proved 
to be stronger than the stimulus (sound) applied in the previous combina- 
tion. With this technique, it proved possible to demonstrate the elabora- 
tion of direct and reverse connections in a more distinct, graphic and what 
is particularly important, direct form. This was possible both when one 
of the combined stimuli was applied first in stereotyped sequence, and 


/ mmm 


■^ 


.S^^^v^. ^ .^ , ,.. || ^ | 


28. U 3. 5 9 


m 


ff4'**.y^'y)l» ^i|ii^ p Mi»^^l § ii^ 


Fig. 2 
A two-way connection resulting from the combination of a passive lifting of the leg with the 
blowing of air into the eye. {a) A conditioned motor reflex arising when the combination is 
applied in the stereotyped sequence — 'blowing of air into the eye — passive lifting of the leg'. 
In response to the blowing of air into the eye [d) the dog blinks and lifts the leg which is 
testified to by the disappearance of action potentials in the gastrocnemius muscle. The lifting 
of the leg (p.p.) is again accompanied by blinking which testifies to a reverse connection; 
(/)) A conditioned eyelid reflex arising when the combination is applied in the stereotype 
sequence 'passive lifting of the leg — blowing of air into the eye'. In response to the passive 
lifting of the leg (p.p) the dog blinks. After the blowing of air into the eye ((/) there is observed, 
along with the blinking, a disappearance of action potentials in the gastrocnemius muscle, 
which testihes to a reverse connection. 

I. — action potentials of the gastrocnemius muscle; i. — movement of the eyelids; 3. — 
marks of stimulation (long line — passive lifting of the leg; short line — blowing of air into the 
eye). 

when the other stimulus was applied first (Fig. 2). As a puff of air in the 
animal's eye proved to be a stronger stimulus than sound, the direct 
conditioned connections were much more stable than when sound was 
combined with a passive lifting ot the animal's leg. To some extent this 
also applies to reverse conditioned connections. 

II. New facts related to this problem were established by Lyan-Chi-an, 
who continued the investigation of a very pressing problem which 
Varga (1953) had previously studied in our laboratory. 

Using dogs, Varga combined in a strictly alternating sequence two 


E. A. ASRATYAN 99 

classical unconditioned stimuli of moderate intensity, namely, presentation 
of food and electrical stimulation of the leg. She established the possibility 
of elaborating a stable double (or two-way) conditioned connection be- 
tween cerebral nervous structures corresponding to these two stimuli. It 
should be noted that under standard experimental conditions (strictly 
alternating sequence of combination of stimuh, relative equality of their 
intensity, observance of the routine conditions of maintenance of the 
experimental animal, etc.) these conditioned connections are practically 
equivalent as regards sign, strength, stability and so on. 

On the contrary, in the experiments of Lyan-Chi-an, two unconditioned 
stimuli were combined in a stereotyped sequence. In other words, he 
combined in a stereotyped sequence a defensive motor reflex of a fore-leg, 
evoked by applying an electric shock of moderate intensity, with an 
alimentary reflex to the presentation of a moderate portion of powdered 
meat and bread. In certain series of experiments, performed on one group 
of dogs, application of the electric shock preceded presentation of food, 
while in other series, performed on another group, presentation of food 
preceded application of the electric shock. Both series of experiments 
yielded essentially similar results which agree in some respects with the 
data of Varga and of Pressman and Varga, but differ in some essential 
points. In the experiments of Lyan-Chi-an, double (or two-way) con- 
ditioned reflex connections between the brain points of the combined 
unconditioned stimuli were also formed. This is shown in Figs. 3 and 4. 
However, here also, as in the similar Varga and Pressman experiments, 
direct and reverse conditioned comiections between these points were far 
from being equivalent. In comparison with Varga and Pressman's 
experiments those of Lyan-Chi-an showed a considerably more pro- 
nounced difference in the rate of elaboration of conditioned connections 
leading from the preceding (P) to the subsequent (S) stimulus, than from 
(S) to (P). In both scries of Lyan-Chi-an's experiments, the direct condi- 
tioned reflex connection between the combined unconditioned stimuli is 
formed sHghtly more rapidly than the reverse conditioned reflex connec- 
tion. This connection is also of much greater strength and stability and is 
characterized by a considerably greater regularity than the reverse connec- 
tion. In contradistinction to a direct connection between two indifferent, 
i.e. physiologically weak stimuli, a direct connection between two un- 
conditioned, i.e. physiologically strong stimuli, forms easily, consolidates 
increasingly from day to day and persists if the conditions which have 
given rise to it are maintained; namely, when these stimuli are combined 
in the given sequence. 

H 


100 


BRAIN MECHANISMS AND LEARNING 


-J i i 


JLJil.7wt^vvlc/--VL. 


~13 V 


Fig. 3 

An alimentary conditioned reflex to an electric current. A dog to which an electric current 
and the presentation of food were applied in the stereotyped sequence 'electric current — 
food'. (<i) A conditioned salivary alimentary reflex to the application of electric current 
(direct connection) ; (/)) movements of the leg during the presentation of the food receptacle 
(reverse connection). 

From top to bottom: i. — mcchanogram of movement of the left foreleg; 2. — mechanogram 
of masticating movements; 3. — salivation in drops; 4. — mark of electro-cutaneous stimula- 
tion; 5. — mark of presentation of the food receptacle; 6. — time in seconds. 


/LAjL_4L-/iLJL 


i 


Fig. 4 

Electro-defensive motor reflex to presentation of food. A dog to which presentation of food 
and electrical current were applied in the stereotyped sequence 'food — electric current'. 
((7) A conditioned motor defensive reflex to the end of the act of eating (direct connection); 
(/)) a conditioned alimentary reaction to the electric current (reverse connection). 

From top to bottom: i. — mechanogram of movements of the left foreleg; 2. — mechanogram 
of movements of the right foreleg; 3. — mechanogram of masticating movements; 4. — saliva- 
tion in drops; 5. — mark of electro-cutaneous stimulation; 6. — mark of presentation of the 
food receptacle; 7. — time in seconds. 


E. A. ASRATYAN lOI 

The matter is entirely different with a reverse conditioned reflex connec- 
tion between unconditioned stiniuh. They greatly resemble, especially 
in their fragility and irregularity, conditioned connections between in- 
different stimuli. It must be pointed out that a reverse connection, is so 
feeble, so unstable, multiphasic and diverse in its manifestations and 
depends on so many factors for its functioning that initially we even 
doubted its conditioned reflex nature. 

Some of the peculiar features revealed by the experiments of Lyan- 
Chi-an in the functioning of reverse conditioned reflex connections are 
of special interest. For instance, (S) evokes more easily the reflex corres- 
ponding to (P) if the strength of (P) has been increased or if the excitability 
of the central nervous structures is raised. By strengthening (P) or by 
an increase in the excitability of the relevant nervous apparatus it becomes 
possible to revive and activate the reverse conditioned connections even 
when under usual experimental conditions they cease to manifest them- 
selves. A similar result is obtained if (S) is weakened, or the excitability 
of the corresponding nervous structures is lowered. For example, in an 
experiment in which stimuli are applied in the sequences 'electrical shock 
— food', the presentation of food evokes a conditioned reflex lifting of 
the animal's leg more frequently and more distinctly if the leg was 
previously stimulated by a stronger shock than usual, or if, prior to 
the experiment, the dog was deliberately fed to satiety. The feature 
common to these two different preliminary experimental manipulations 
is that they both lead to the same picture of relative excitability or 
excitation of the central nervous structures corresponding to these two 
stimuli. 

In contrast to this, a prehminary strengthening of (S) or increase in 
excitability of the nervous structures corresponding to its reflex leads to 
inactivation of the nervous connection even when under usual experi- 
mental conditions they function regularly and satisfactorily. Similar 
results are obtained by a preliminary weakening of (P) or by a preliminary 
decrease of the excitability of the relevant nervous structures. Let us, for 
example, take again the type of experiments thoroughly studied by Lyan 
Chi-an in which stimuli were applied in the sequence: 'electric shock — 
food'. Here the presentation of food evokes no conditioned lifting of the 
leg if the animal is in a state of fasting before the experiment, or if it is 
given appetizing food at the beginning of the experiment, or if its leg is 
first stimulated by a weak electric shock. It may be noted that although 
these preliminary experimental manipulations are different, they have 
something in common: in different ways they lead to the creation of the 


102 BRAIN MECHANISMS AND LEARNING 

same picture of relative excitability or excitation of the central nervous 
structures of these two stimuli. 

III. New and interesting data obtained by our collaborator Struchkov 
(1958) also relate to the problem under discussion. While elaborating 
conditioned reflexes in dogs, he combined 'indifferent' stimuli with an 
unconditioned stimulus. This special method is known in the literature as 
the method of 'covering'. In this method the 'indifferent' stimuli are 
applied against a background of action of the unconditioned stimulus. 
Struchkov used food as an unconditioned stimulus and chose only those 
'indifferent' stimuli which by themselves evoke in the organism specific 
reactions that can be objectively observed and recorded. For example, in 
one series of experiments he applied a passive lifting of one of the dog's 
hind legs and in another a momentary cooling of a limited area of the skin 
on one side of the animal's body, which evokes a local vasomotor reaction. 
This sequence of stimuli was applied systematically in all experiments of 
each scries. The following very interesting facts were brought to light. 
After repeated combination of an alimentary reflex with one of the 
'indifferent' stimuli, presentation of food alone began to evoke reactions 
peculiar to these stimuli. In one series of experiments the reaction was an 
active lifting of the leg, and in another a local vasomotor reflex. This is 
shown in Fig. 5. In other words, in Struckhov's experiments the classical 
unconditioned stimulus — food — played the role of a conditioned or 
signalling stimulus, while the so-called 'indifferent' stimuli — passive 
movement of the leg or local cooling of the skin — appeared as reinforcing 
stimuli. These remarkable experiments show that it is precisely the 
position in time of the preceding stimulus (P) which determines the 
acquisition by the food of a signalling significance in relation to other 
reflexes, that is, the formation in one case of a somato-motor conditioned 
reflex to the presentation of food, and in the other case a vasomotor 
conditioned reflex. It may be assumed therefore that this is the consequence 
of the formation of a direct conditioned reflex connection between central 
nervous structures of the alimentary reflex and corresponding structures of 
the somato-motor or vasomotor reflexes. It was subsequently established 
that in these experiments there also arises a double (or two-way) con- 
ditioned connection between the unconditioned and indifferent stimuli 
which are combined in a definite, stereotyped sequence. This means that 
with the direct conditioned reflex connection there exists a reverse connec- 
tion which conducts excitation from the nervous structures corresponding 
to (S) (i.e. to the movement of the leg or local coohiig of the skin) to the 


E. A. ASRATYAN 


103 


nervous structures corresponding to (P) (i.e. the alimentary stimulus). 
This is clear from the fact that passive lifting of the leg or local cooling 
of the skin also acquire a signalling significance (i.e. the faculty of evoking 
an alimentary reaction in a conditioned reflex way when repeatedly 
combined with food, according to the method of 'covering' applied in 


^ -nrr 


l u iiiiiiiii in i um i ini i mm i nm i nii i u i mi iiiii m i 


197(20) 


nj\JiJi_AJi_n_n_n-JuriJv-n_n_a_rLJLruvn_n_n_n_rij^^ 


a 


6 


Fig. 5 

{a) A motor conditioned reflex evoked in the dog by the act of eating (direct connection); 
(/)) an ahmentary reflex evoked by the passive lifting of the leg (reverse connection). 

I. — Mark of masticating movements; 2. — mark of placing the leg on a bench; 3. — mark 
of movements of the same leg; 4. — mark of salivation ; 5. — mark of passive movement of the 
leg ; 6. — mark of presentation of food (figure indicates the number of combinations) ; 
7. — mark of time — in seconds. 


the experiments discussed above. This is demonstrated in Fig. 6. The 
data obtained from these experiments show that here too there is a 
considerable difference between the specific features of direct and reverse 
conditioned connections, and that, furthermore, it is the same kind of 
difference as is observed when two 'indifferent' or two unconditioned- 
reflexes are combined. Thus, in a given case the direct conditioned reflex 
connection is formed more rapidly than the reverse conditioned reflex 
connection and is much more powerful, stable and effective. 

The fact demonstrated by Struchkov that only weak and unstable 
conditioned reflexes become indifferent stimuli when 'covered' by un- 
conditioned reflexes, has been known since the work of Pimenov (1908), 
Krepa (1933), Pavlova (1933), Vinogradov (1933), Petrova (1933) and 
others; but these authors regarded it as an indication of the elaboration of 
an ordinary one-way conditioned reflex connection possessing certain 
specific features. In Struchkov's experiments this fact appears in quite a 


104 BRAIN MECHANISMS AND LEARNING 

different light, namely, as a particular case of the general phenomenon of 
elaboration of a reverse conditioned reflex connection (along with the 
direct connection) when two stimuli are combined in a stereotyped 
sequence. The transitory character of this connection, which was pointed 
out long ago by many workers and was regarded by Pavlov as an enig- 
matic phenomenon in conditioned reflex activity, may, to a considerable 
degree, be interpreted as a result of its systematic non-reinforcement 
during the periodic application of a stimulus for the purpose of testing a 
conditioned reflex to it. 


// 


Fig. 6 
Conditioned and unconditioned refle.x local changes in the skin temperature. 

A. Decrease of skin temperature in the dog under the action of food during 60 seconds and 
subsequent application of cold. I. Record of skin temperature. II. Time — 12 seconds; (a) back- 
ground; (/>) decrease of temperature during the process of eating and application of cold; 
(f) recovery of skin temperature after the process of eating and action of cold. i. — mark of 
presentation of food and beginning of the act of eating; 2. — mark of application of cold; 
3. — mark of the end of eating and application of cold. 

B. Conditioned reflex decrease of skin temperature during the process of eating, (ci) — back- 
ground ; (/)) — decrease of skin temperature during the process of eating; (c) — recovery of skin 
temperature after the process of eating, i. — mark of presentation of food and beginning of 
the process of eating; 2. — mark of the end of eating. 


IV. The significance of a definite sequence in the combination of stimuli 
for elaboration of new conditioned reflex connections and for preservation 
of existing ones is graphically shown in the experimental data obtained by 
our collaborator Pakovich. 

From the earlier investigations of American psychologists, Schlosberg 
(1928), Hilgard (1937), Wolflc (1930), Bernstein (1934) and others, which 
were carried out on human beings, it was known that when the indifferent 
stimulus precedes the unconditioned stimulus by o.i second and less, the 
elaboration of a conditioned reflex is impeded, and sometimes even be- 
comes impossible. In recent years, Pakovich has made a thorough investi- 
gation of this question in dogs, on motoi reflexes which can be objectively 


E. A. ASRATYAN IO5 

recorded with a sensitive mechanographic device, and in some experi- 
ments also by electromyography. His data show that when a combination 
of a sound of moderate intensity with an electrical stimulation of the 
animal's legs is used, a motor conditioned reflex is not elaborated if both 
stimuli begin and cease to act strictly simultaneously within a period of 
I to 5 seconds. According to Pakovich's fmdings, hundreds of combina- 
tions of these stimuli do not lead to the formation of the conditioned 
reflex even when sound precedes the electric current within a limit of 
about 100 msec, and when they cease to act simultaneously. The picture 
does not change if the intensity of the stimulating current is considerably 
increased. A conditioned reflex becomes elaborated only when the 
interval between the onset of the sound and that of the electrical stimula- 
tion of the skin exceeds this time-limit. This is illustrated in Fig. 6. Other 
data obtained by Pakovich show that the time fictor expressed in such 
micro-values makes itself felt also at a certain distance from the above- 
mentioned peculiar demarcation line. Pakovich established in particular 
that the latent period of the newly formed conditioned reflex, as well as 
its strength, duration, stability and other properties, depends to a con- 
siderable degree on the interval between the onset of action of the two 
stimuli, the indifferent and the unconditioned. 

Pakovich established another very interesting fact: a firmly established 
and regularly evoked clcctro-detensive motor reflex disappears in the dog 
if the conditioned stimulus is applied synchronously with the uncondi- 
tioned one in the course of a series of experiments. The same occurs also 
when the interval between onset of the conditioned and of the un- 
conditioned stimulus is reduced to less than lOO msec. The conditioned 
reflex reappears only when these conditions of combined action of both 
stimuli arc changed and when the interval between the onset of the two 
stimuli again exceeds lOO msec. The restoration of the conditioned reflex 
proceeds more rapidly and is more complete when this interval is some- 
what higher than its threshold value. 

Recently, Pakovich carried out an experimental analysis of the dis- 
appearance of the established conditioned reflex when conditioned and 
unconditioned stimuli were applied synchronously. 

He found that this phenomenon is due to the sporadic application in the 
course of his experiment of the conditioned stimulus alone for the purpose 
of testing its conditioned effect. It seems, therefore, that simultaneous 
application of the conditioned and unconditioned stimuli leads to such a 
weakening of the conditioned reflex that it disappears even after a few 
extinction trials. 


io6 


BRAIN MECHANISMS AND LEARNING 


The following fact is noteworthy. Although a strictly simultaneous 
action of both stimuli or a certain minimal precedence of action of one of 
them creates such conditions that the cerebral nervous structures cannot 
elaborate new conditioned connections, or even abolishes all previously 
conditioned reflexes, this does not preclude the possibility of a negative 
anci even positive interaction between the cerebral structures which 
correspond to these stimuli. Pakovich demonstrated that even in conditions 
of a synchronous action of sound and electrical current there may take 


a 


6 


Fig. 7 
Motor reflexes arising from different combinations in time of two stimuli — a tone and an 
electric current. (17) — a motor reaction resulting from a strictly simultaneous action of the 
stimuli; (fo) — absence of a conditioned reflex to the tone after 582 strictly simultaneous 
applications of it in combination with an electric current; (c) — emergence of a conditioned 
reflex to the same stimulus after fourteen applications of it in combination with an electrical 
current delayed for 2 seconds. 

I. — record of movements of the leg; 2. — mark of conditioned stimulation; 3. — mark of 
unconditioned stimulation; 4. — time in seconds. 


place both a reciprocal inhibition and a summation of excitation provoked 
by each of these stimuli in corresponding central nervous structures. The 
reciprocal inhibition is observed predominantly in the first period of the 
combined action of the stimuli, when the orienting reaction to sound is 
still strong. As to summation of excitation, it is observed predominantly 
following the combined action of the stimuli. During these periods a 
subthreshold excitation evoked by an electro-cutaneous stimulus applied 
separately turns into an above-threshold excitation when an electric current 
of similar intensity is applied in combination with a sound. In the same 


E. A. ASRATYAN IO7 

conditions of combined action ot both stimuli a weak above-threshold 
excitation evoked by a weak electric current becomes considerably intensi- 
fied, etc. Finally, it must be noted that according to these findings, the 
phenomena described above and the determining factors observed in the 
formation of conditioned reflexes are not connected to any appreciable 
degree with typology, age or other features of the experimental animals; 
it follows that they belong to the category of basic phenomena and 
determining factors which are characteristic of the conditioned reflex 
activity of higher parts ot the central nervous system. 

How are all these fmdings to be interpreted and generalized? 

Experimental psychologists, neurophysiologists and animal trainers have 
for long known a number of facts which prove the formation of a tw^o- 
way nervous connection and have provided a basis for the formulation of 
corresponding theories on this question (Goltz, 1884; Bekhtercv, 1887; 
Ebbinghaus, 1911-13; Kahsher, 1912-14, Savich, 1918; Zeliony, 1929; 
Beritov, 1932; Konorsky and Miller, 1936; Narbutovich and Podkopayev, 
1936; Rosenthal, 1936; Petrova, 1941; Kupalov, 1948; Fiodorov, 1952; 
Voronin, 1952, and others). In this connection special mention must be 
made of the theories of Pavlov (1949) concerning double (or two-way) 
conditioned connections, advanced by him in recent years and based on 
original facts brought to light in his laboratories. Pavlov stated that 'when 
two nervous points are connected, or associated, the nervous processes 
developing between them proceed in both directions'.^ At that time, 
however the founder of the theory of conditioned reflexes was of the 
opinion that the important question of two-way conditioned connection 
was still not sufficiently elaborated, experimentally or theoretically, and 
that it required further investigation. It appears to us that the facts recently 
obtained in our laboratory increase to some degree our knowledge of this 
problem. They reveal new aspects of direct and reverse connections, and, 
in particular, new aspects of the role played in this process by the physio- 
logical strength of the stimuli and by the sequence of their combination. 

All these facts, in addition to the previously known data of other investi- 
gators, lead to the assumption that the formation of a double (or two-way) 
conditioned comiection is not a limited phenomenon peculiar only to a 
certain group of conditioned reflexes, but is widespread, if not universal 
for this class of reflexes. This creates, so to speak, a structural basis tor a 
two-way tunctional connection and for a circular conditioned reflex 
interaction between the comiected nervous points, approximately in the 

1 1. P. Pavlov, Complete Works, 1949, vol. Ill, p. 452. 


io8 


BRAIN MECHANISMS AND LEARNING 


same manner as was assumed by Lorente de No (1934, 1938) and his 
followers for the activity of the central nervous system in general. In any 
case, there are sufficient grounds for this assumption during the stage of 
formation of conditioned reflexes and for the initial period of their func- 
tioning. Subsequently, depending on a number of circumstances, evolu- 
tionary development leads to different results: to the inhibition of both 
conditioned connections, to the inhibition of one of them, and some- 
times to the preservation of both connections in a state of stability and 
efficiency. In this case a usual one-way conditioned coiuiection can be 
regarded as a derivative of a double conditioned connection, as resulting 
from the subsequent inhibition of one of the paired connections. 


Fig. 8 


Schematic representation of a conditioned reflex arc. 


I should like also to point out that all the factual material described in 
this communication, as well as the appraisal of its scientific significance, 
may also be regarded as corroborating the proposition we advanced many 
years ago, that the primary conditioned reflex results from the synthesis 
of combined inborn reflexes evoked by both unconditioned and 'in- 
different' stimuli (Fig. 8). 

Of course, at present, we arc far from a satisfactory interpretation ot 
these facts. Nevertheless, some purely hypothetical statements can be made. 
In particular, we are inclined to believe that the elaboration of a double 
conditioned connection may be accounted for by the relative equality of 
intensity of the excitations which arise in the corresponding central 
nervous structures under the influence of the combined stimuli. It goes 


E. A. ASRATYAN 


109 


without saying that this basic condition can be observed best when two 
stimuh of ahnost equal physiological strength are combined (for example, 
two 'nidifterent' stimuli, two unconditioned stimuli, or one 'indifferent' 
and one unconditioned stimulus). This is also possible when two paired 
stimuli are combined in a variable sequence. Such a combination of these 
two factors can secure the formation and existence of double conditioned 


a 


p' 


F[G. 9 

I. Diagrammatic representation of a double conditioned connection 
with equivalent components. 

II. Diagrammatic representation alter development of non-equivalence. 


connection with absolutely equivalent components (Fig. 9 I). But when 
these paired stimuli are combined in a stereotyped sequence, the basic 
condition is not fulhlled, which, in our opinion, leads to the development 
of non-equivalence in the direct and reverse conditioned connections 
(Fig. 9 II). We explain it in the following way. When the (P) and (S) 
stimuli are applied in a stereotyped sequence, the central nervous structures 


no BRAIN MECHANISMS AND LEARNING 


corresponding to (S) become more excited after the emergence of condi- 
tioned connections than the central nervous structures corresponding to 
(P). This is due to the fact that the structures corresponding to (S) are in 
this case excited by two sources — by (P) in a conditioned reflex way and 
by (S) in an unconditioned reflex way. There does not exist the initial 
relative equality in intensity of excitation of the central nervous structures 
which corresponds to the combined stimuli. The excitation of the nervous 
structures of (S) prevails to a great extent over the excitation of the nervous 
structures corresponding to (P). As a result the direct conditioned connec- 
tion begins to prevail over the reverse connection in all characteristics. 
This explanation was corroborated by Varga in a series of experiments 
(1958) in which she demonstrated that the transition from a variable 
sequence of combined stimuli to a stereotyped sequence entails a con- 
siderable increase in the excitability of the central nervous structures 
which correspond to (S) and prevalence in this respect over the one 
corresponding to (P). 

We are inclined to offer a similar explanation for Struchkov's findings 
concerning the conversion of food to a signalling stimulus for somato- 
motor and vasomotor reflexes. In the first phase of combining food 
(which is the preceding stimulus (P)) with a passive lifting of the paw or a 
local cooling of the skin (which is the .subsequent stimulus (S) ) the central 
nervous structures corresponding to the latter stimuli become excited to 
a relatively high degree on account of the novelty of the stimuli for the 
experimental animals. Owing to this, the level of excitation of the central 
nervous structures corresponding to these stimuli approximates the level 
of excitation of the central nervous structures of the alimentary reflex 
which is the preceding stimulus. This creates favourable conditions for 
the elaboration of the double conditioned connection between these 
structures. Subsequently, after elaboration of this double connection, the 
above-mentioned factor comes into effect, namely, a more intense 
excitation of the central nervous structures of (S) coming from two 
sources by way of summation — at first, from the food in a conditioned 
reflex way, and then, against this background, from adequate stimuli of 
the nervous structures (a lifting of the paw or cooling of the skin) in an 
unconditioned reflex way. This maintains the previously created relative 
equality in levels of excitation of the nervous structures corresponding to 
the combined stimuli, i.e. the precondition necessary for maintaining the 
double conditioned connection is preserved. 

Facts concerning relatively different duration of efficiency for the condi- 
tioned comiections arising after combination of various stimuli of different 


E. A. ASRATYAN III 


biological significance (two 'indifferent' stimuli, two unconditioned 
stimuli, or one 'indifferent' and one unconditioned stimulus) can be best 
understood when considering the important role of the concrete physio- 
logical strength of the corresponding stimuli. There are grounds for 
assuming that this factor plays an important role in the further fate of 
the conditioned connections, whereas the relative strength of excitation of 
the central nervous structures corresponding to these stimuli and the 
sequence of application play an outstanding role in elaborating these 
connections and forming some of their properties. If we assume that the 
level of excitation of the central nervous structures originating conditioned 
reflex comiections expresses approximately the number ot functional 
units which come into activity, and that the strength of the conditioned 
reflex connection is its derivative, the following conclusion may be 
drawn : for maintenance and functioning of conditioned connections it is 
necessary that they should possess a certain threshold strength, i.e. consist 
of a definite 'threshold' number of functional units. Weak conditioned 
connections, that is, connections consisting of a small number of functional 
units, become rapidly fatigued and exhausted even after moderate activity; 
a protective inhibition occurs which blocks them. At present we arc unable 
to explain satisfactorily why such inhibition develops in structures corres- 
ponding to weak conditioned connection. We can point out, however, 
with some justification, that there is a similarity between this phenomenon 
and the rapid development of inhibition during a protracted action of weak 
conditioned and even indifferent stimuli. This phenomenon was known a 
long time ago, trom numerous investigations carried out by Pavlov in his 
laboratories. Moreover, the development of inhibition in coimection with 
a protracted action of weak stimuli seems to belong to the category of 
phenomena which are common to all nervous structures. 

For the time being we cannot give a satisfactory explanation for the 
facts obtained by Pakovich, which show that a conditioned reflex cannot 
be formed when two combined stimuli act synchronously. Nor can we 
explain the disappearance of an existing conditioned reflex when a 
conditioned stimulus and an unconditioned one are combined in a similar 
way, although in this case the possibility of a positive and negative 
interaction of these stimuli remains. It is possible that a certain role in this 
respect is played by mutual inductive inhibitory influences of the simul- 
taneously excited cerebral structures upon each other. These influences, 
however, do not preclude the summation of excitations that have 
originated in them. 


112 BRAIN MECHANISMS AND LEARNING 


GROUP DISCUSSION 


KoNORSKi. The problem investigated by Professor Asratyan and his group is 
very important, and has not been solved as yet, although it has been studied for 
many years. This is the old problem whether the so-called backward conditioning 
exists, or not. Both in Russia and in America there were authors strongly support- 
ing the idea of backward conditioning, as well as strong opponents of this view. 
While the first group of authors claimed that any association between stimuli 
(both forward and backward) leads to the formation of connections between them, 
the second group argued that the phenomena of backward conditioning were only 
pseudo-conditioning due to sensitization. In fact, while the biological role of 
forward conditioning is obvious, the biological significance of backward condi- 
tioning could hardly be understood. 

Professor Asratyan has brought out more extensive material concerning this 
problem than any other author. According to his data backward conditioning does 
exist, although the connections formed in the direction from the subsequent 
stimulus to the preceding one are much weaker than those established in the 
opposite direction. This is also in agreement with the results obtained by Czech 
authors on chimpanzees, and recently by Mrs Budochovska in the Psychological 
Department of the Nencki Institute on man. 

Chow. I would like to ask Professor Asratyan how these very interesting 
experimental results are to be incorporated, if they are to be incorporated, in the 
concept of reinforcement during the formation of the conditioned reflex. 

Asratyan. These conditioned reflexes also need reinforcement. If one of the 
stimuli is applied without any other, this, leads to the extinction of the reflex in 
every case. But in cases where we have a weak stimulus, the so-called 'indifferent 
stimulus' — I say so-called because I think that these indifferent stimuli are only 
relatively indifferent — indeed if they evoke a reaction at all in the organism they 
cannot be indifferent. They are indifferent in relation to another reaction evoked by 
another stimulus but by themselves they also originate reflexes like unconditioned 
stimuli, althoush their histological sia;n and their strength are different. When we 
combine two indifferent stimuli, or weak stimuli, in these cases absence of reinforce- 
ment leads to a more rapid extinction than in the case of strong stimuli. Reinforce- 
ment is necessary. 

Naquet. How do you explain the limit oi loo msec? 

Asratyan. I am afraid we cannot answer this question in a satisfactory manner. 
It may be a question of mutual influences, facilitation and inhibition, not elaborated 
connections. I really don't know. But it should be pointed out that tacilitation alone 
is not sufficient for the elaboration of conditioned connections. 

Fessard. I believe that new connections within the central nervous system 
cannot be formed unless the associated afferent messages converge towards common 
neurones so that the limit of lOO msec, might be explained by the recovery period 
of these neurones, at least of those having discharged aiter the first of the afferent 
signals reaching them. 

Naquet. I think loo msec, represents the recovery time of the reticular excitability 
and you may have some facilitation which permits the establishment of condition- 
ing. 

Hebb. I have been very interested for a long time in the problem of the sequential 


E. A. ASRATYAN II3 

aspect of cerebral function. It may tnid its simplest, its most experimentally 
attackable form in the apparent ordering in time of the conditioned stimulus and 
unconditioned stimulus. I was going to ask Professor Asratyan what speculations 
he was making about possible neuronal action that might explain this. It seems to 
me that Protcssor Fessard's point is excellent but yet there is the turther problem 
that conditioning is not good unless there is some overlap. It is as it the line had to be 
busy to a certain extent, or in certain circumstances. 

I would like to say also that I found the paper very interesting in the study of 
relatively unimportant, biologically unimportant, stimulations. So much of human 
behaviour is composed of reactions to biologically neutral stimuli, or very nearly 
neutral. 

Asratyan. First of all I should like to remind you that we are quoting from our 
data. In the case of stereotyped sec]uences ot application ot the stimulus the signifi- 
cance of these two conditioned connections is different. Dr Konorski also men- 
tioned this in his remarks that direct conditioning in all cases is much stronger and 
steadier and more regular than backward conditioning. This illustrates the bio- 
logical differences between these connections and the significance of the sequence 
of combination. How does one explain thisf Originally both nervous points of 
both stimuli are equally stimulated, however once the connection is established 
and the stimuli inverted, one point, namely the second point, receives stimuli from 
two sources, whereas the other receives only its original stimulus. This is the reason 
for the reinforcement of forward conditioning and the difficulty of backward 
conditioning. 

In answer to vour second question, in cases ot biologically weak stimuli, as tor 
example in cases when we combined two so-called indifferent reflexes, giving rise 
to the elaboration of weak two-wav conditioned reffex, these reflexes disappear 
very rapidly. Occasionally in our cases we have been able to prolong their duration. 
Why these conditioned weak reflexes disappear we don't know. Our theory, our 
'fantasy', is that it is due to the lower threshold level of excitation, to the small 
number of cells activated in both coupled nervous points during the elaboration of 
reflexes. As a result of the activation of such reflexes, their weak connections easily 
become exhausted and fatigued and soon disappear. A strong stimulus would call 
into play a larger number ot cells and consequently a steady conditioned reflex is 
elaborated. 


THE PHYSIOLOGICAL APPROACH TO THE PROBLEM 
OF RECENT MEMORY 

Jerzy Konorski 

It is a striking fact that the investigations of higher nervous activity of 
animals carried out both in physiological laboratories by methods of 
conditioned reflexes, and in psychological laboratories by use of such 
other methods as maze learning, discrimination box, etc., have been 
almost exclusively concerned with the problem of stable memory. In 
fact, in all these investigations the animal is routinely trained to perform a 
particular task or a set of tasks and then various properties of the acquired 
reactions, or the very process of their acquisition, are studied. But it is 
easy to conceive that the performance of the more or less firmly estab- 
lished conditioned responses, in the broadest sense of the word, and, of 
course, unconditioned responses, does not exhaust the whole behaviour 
of the animal. We know from everyday observation of animals and man 
that a large part of this behaviour is often based on transient memory 
traces which persist for some time and then are partially or even totally 
abolished. Recently the clear realization of a probable difference between 
the mechanisms of stable and transient memory traces has turnec^ the 
attention of investigators to the latter category of phenomena and has 
given an impetus to their studies. 

I intend in the present paper to present a brief review of the methods 
used so far in the study of phenomena of recent memory, to propose some 
new methods which may be applied in this field, to examine the relations 
between stable and recent memory and to discuss the problems of a 
probable physiological mechanism of recent memory, versus that of 
stable memory. 

I. TRACE CONDITIONED REFLEXES 

As a matter of fact, the phenomena of recent memory have been 
implicated for a long time in some conditioned reflex (CR) studies, 
although their significance in this respect was not clearly understood. I 
have in mind the so-called trace CRs which were studied by several 
research- workers of the Pavlov school (Pimenov, 1907; Grossman, 1909; 
Dobrovolskij, 1911; Feokritova, 1912; Pavlova, 1914) in the first decade 

I 115 


Il6 BRAIN MECHANISMS AND LEARNING 

of the work in this field. These workers estabhshed that if a given 'in- 
different' stinaukis is reinforced not during its action but some time — of a 
range of seconds or minutes — after its cessation it is possible to establish in 
a dog the CR not to the stimulus itself but to a short period of time just 
preceding the moment of reinforcement. Thus, if for example a dog 
repeatedly receives food 3 minutes after the cessation of the conditioned 
stimulus (CS), he learns to salivate not earlier than about half a minute 
before the moment of feeding. In a variation of such experiments the 
animal receives food simply at constant intervals and learns to expect it at 
the proper moment — 'the CR to time'. The end of the act of eating 
plays in this experiment a role of a trace CS. 

Facts of this kind mean that (i) the CS produces some changes in the 
CNS which persist some time after its cessation, and (2) that the animal 
is able to 'measure' somehow the time elapsed after it. 

The process of elaboration of a trace CR is such that tu'st both the CS 
itself and the whole period between the stimulus and the reinforcement 
evokes salivation and then gradually the conditioned reaction is more 
and more postponed. This shows that first there is a generalization of the 
CR to all moments following the CS and then differentiation of these 
moments occurs. 

A peculiar property of trace CRs is that they arc very widely generalized : 
even the application of new stimuli, much different from the original CS, 
produces salivation in the appropriate moment after their cessation; for 
example, when the original trace CR was elaborated to a tactile stimulus 
applied to one spot of the skin, it is produced by tactile stimuli applied to 
quite different spots and also in response to auditory stimuli. This property 
of trace CRs seems to suggest that they are formed not to the traces of the 
given exteroceptive stimulus itself, but rather to some of its consequences 
which are common for various sorts of stimuli. As any external stimulus 
elicits an orientation reaction it may be that the proprioceptive stimuli 
generated by this reaction form the true basis for elaboration and occur- 
rence of trace CRs. We shall return to this question in a later section. 

2. DELAYED RESPONSES 

Nearly at the same time Hunter (19 13), according to the suggestion of 
Carr, introduced into behaviouristic psychology another method of 
investigation of recent memory based upon the so-called delayed responses. 
The general principles of this method are roughly these: the animal is 
taught to receive food in two or more different places (or run to the food 


lERZY KONORSKI 


117 


through several different ways). When the animal is restrained in the cage 
or attached on the starting platform, a signal heralding that the food is 
presented in a particular place is given; in many experiments such a 
signal is provided simply by baiting one of the bowls in front of the 
animal. Then after a number of seconds or minutes the animal is released 
and if he remembers which signal was given, he will go to the correspond- 
ing place and be reinforced there. 

After the first paper by Hunter appeared, a number oi authors studied 
the delayed responses in various species, attempting to tnid the maximal 
delay periods which can be achieved, the cues which are used by the 


^^-^B^ L,\^ 


c^-^ 


F2 


^^^^ 


'.'-y- 


^ 


c 


\ 


^^C 


^ 


N m ^ 


^ 


Fic. I 
Experimental setting used in our study. 
Fi- ^2. F3' ''•"ff- iniddle, and right food tray. The bowls are auto- 
matically moved into position by the experimenter using an electro- 
magnetic device. E, table and seat for experimenter. S, starting 
platform. Bj, B.,, B3, buzzers, Lj, L,, L3, lamps situated on the respective 
food trays. (After Konorski, J. and Lawicka, W., 1959. Acu\ B'\o\og\ae 
E.xperiiiieiiriilis, 19, 175-198). 


animals to solve the task, and the way in which the delay period is 
'bridged' by the animal (Walton, 1915; Yarborough, 1917; Cowan, 
1923; Yerkes and Yerkcs, 1928; Tinklepaugh, 1928; McAllister, 1932). 
This form of experimentation grew in importance in the 'thirties of this 
century when Jacobsen (1936) showed that delayed responses are dramati- 
cally impaired or abolished by prefrontal lesions in monkeys and apes. 
The results of these studies will be discussed in a later section. 

The delayed responses have been recently studied in our laboratory by 
Lawicka and Konorski (Lawicka, 1957, 1959; Konorski and Lawicka, 
1959; Lawicka and Konorski, 1959) in dogs and cats by using the experi- 
mental setting presented in Fig. i . The animal was on the leash or in the 
cage on the starting platform and a signal, visual or auditory, which we 


Il8 BRAIN MECHANISMS AND LEARNING 

shall call a 'preparatory stimulus', from one of three food trays was given. 
When after being released the animal ran to the proper food tray, the 
bowl with food was automatically presented. In certain series of experi- 
ments the source of preparatory stimuli was situated not on the food trays 
but at the starting platform, and the animals had to learn by trial and error 
which signal denoted food in which food tray. 

Here are some results of these experiments (Lawicka, 1959). 

1. In response to the preparatory stimulus the animal as a rule displays 
an orienting reaction (of the whole body, of the head or only of the eyes) 
towards the corresponding Jood tray. This is true even in those experiments in 
which the source of the preparatory stimulus is not situated in the direc- 
tion of the goal. 

2. As shown earlier by other experimenters, the preservation of the 
bodily orientation towards the given food tray during the delay period is, 
as far as normal animals are concerned, not at all necessary for the correct 
reaction after release. In fact, during this period the animals nearly always 
change their position many times and take different attitudes which does 
not in the least affect their post-delay reaction. 

3. During the delay period normal animals may be distracted in many 
different ways, by presentation of extra-stimuli from different places, by 
screening the food trays, by giving the animals food on the starting 
platform, or by taking them out of the experimental room, and in the 
majority of trials these measures will not prevent them from running to 
the correct food tray. 

4. If in the triple-choice delayed response method two preparatory 
stimuli arc given one after another in the same trial, the animal, after 
being released, is able to go to both food trays and ignore the third one. 

These data seem to suggest that the 'locating' of the food tray in space 
during the action of the preparatory stimulus constitutes the cue the 
animal uses in the post-delay run. Since the bodily orientation is not 
maintained during the delay period — the animal may even have been 
removed from the room — the memory of this cue is based purely on the 
intracentral nervous processes going on m the brain during the delay 
period, which processes are precisely responsible for those forms of 
phenomena which are called recent memory. 

3. RECENT MEMORY TESTS FOR VARIOUS SORTS OF STIMULI 

From what has just been said about the delayed response test it can 
easily be seen that this test concerns a particniar kind of recent memory. 


JERZY KONORSKI II9 

namely the recent memory of directions in space. Although we do not 
know exactly which sorts of stimuli are involved in determining these 
directions (labirinthine, proprioceptive, or a compound of them) we do 
know that these stimuli were acting when the preparatory signal was 
applied and the animal remembers them during the delay period. But 
obviously not only these kinds of stimuli but also exteroceptive stimuli 
and their various modalities can leave their transient memory traces 
which may be easily detected in human beings, by use of introspection. 
However, as far as animals are concerned, the methods for examination of 
recent memory of these stimuli are not so obvious because we must be 
sure that the given test really concerns the stimuli in question and not 
their proprioceptive effects. We have seen for example in Section i that 
even in trace CRs we cannot be sure that the animal remembers the 
exteroceptive stimulus itself and we have some reason to believe that 
rather its proprioceptive counterpart constitutes the basis of this form of 
reflexes. In order to study the recent memory of various modalities of 
exteroceptive stimuli the following test has been devised (Konorski, 

1959). 

We choose a certain group of stimuli S^, Sg . . . Sn whose recent memory 
we wish to examine — e.g. tones of various pitch, lights of various inten- 
sity, tactile stimuli applied to various places of the body, etc. — and apply 
them according to the following schedule: the compound composed of 
the same stimulus, whatever it is, repeated twice (SxSx) is reinforced, while 
the compound composed of two different stimuli, whichever they are 
(SxSy), is not reinforced, or vice versa. And so when the first component 
of the compound is applied, the animal does not 'know' whether he will 
get reinforcement or not because this depends on comparison with the 
second component which is presented several seconds after the first one. 
Consequently, the animal has no possibility of preparing himself before- 
hand for a particular kind of reaction, and thus to make use of propriocep- 
tive cues, as is the case in delayed responses or in some other tests in which 
the first clement of the compound determines by itself the character of 
the conditioned reaction.^ 

The test described was applied by Chorcjzyna (1959) in dogs for the 
study of recent memory of tones, and by St^pieii and Cordeau (personal 
communication) in monkeys for the study of recent memory of rhythms 
of acoustic and visual stimuli. 

^ For instance, in conditioned inhibition the CS alone is reinforced, while the same stimulus 
preceded by another stimulus (conditioned inhibitor) is not reinforced. In this case already 
during the action of the conditioned inhibitor the animal takes the negative attitude towards 
the food tray and preserves it — or remembers it — during the action of the CS. 


T20 BRAIN MECHANISMS AND LEARNING 

4. ON THE PHYSIOLOGICAL MECHANISM OF RECENT MEMORY 

It is now almost generally accepted that recent memory is based on the 
activity of reverberating circuits of neurones which are connected with a 
group of neurones excited by the actual operation of a given stimulus. 
Such activity causes this group to continue to be excited for some period 
after the cessation of the stimulus itself (Hcbb, 1949), and either dies out 
spontaneously within a lapse of time or is knocked out by some inhibitory 
influence arising from other foci of antagonistic excitation (the so-called 
external inhibition). 

In order to develop this hypothesis a little further and make it more 
precise, the following well-known facts from the field of CR studies 
should be taken into account. 

1. If a CS is suddenly discontinued before the moment of its usual 
reinforcement, the conditioned response proceeds uninterruptedly almost 
with the same intensity as if the CS were still acting (Kupalov and Lukov, 
1932). On the contrary, if the reinforcement is usually presented several 
seconds after the CS is discontinued, the conditioned response appears 
already during its operation. This shows that a high degree of generaliza- 
tion exists between the actual action of a stimulus and its traces. 

2. If an actual stimulus is not reinforced, but the 'trace stimulus' is, then 
dift'erentiation of the two stimuli is 'gradually established. As a result the 
animal always displays the conditioned response only after the cessation 
of the stimulus. 

3. The animal is also able to differentiate early traces from late traces of 
the stinuilus (cf section i). This differentiation is, however, much more 
difficult than that between the stimulus itself and its traces. 

If we hold the view developed in detail elsewhere (Konorski, 1948) 
that generalization, or similarity, of stimuli is due to the overlapping of 
their central representations, while differentiation of them is possible when 
in at least one of these representations there are elements which do not 
belong to the other one, then the above facts can be understood as 
follows : 

According to the vast evidence of facts concerning the responses of 
various nerve-cells to the incoming impulses on all levels of the nervous 
axis, we may admit that the central representation of a given stimulus 
consists of the following types of neurones: Neurones which are activated 
only at the beginning of the operation of the stimulus (pure on-clements, 
group la); neurones which are activated during the whole action of the 
stimulus but not after its cessation (group ib); neurones which are 


JERZY KONORSKI 121 

activated not only during the operation of the stimuhis but also, by virtue 
of reverberating circuits connected with them, some period after their 
cessation (group 2) ; neurones which are activated only after the cessation 
of the stimulus (pure oft-elements (group 3)). This is illustrated in Fig. 2. 


Fig. 2 

Diagrammatic representation of the physiological structure of a trace stimulus. 

The X axis represents time, t,, is the beginning of the operation of the particular stimulus, 
tj, its termination. 

Along the y axis arc represented three groups of neurones: (i) a group of neurones activated 
during, and only during the operation of the stinuilus; (2) a group of neurones activated both 
during the operation of the stimulus and, owing to the reverberating circuits of neurones, for 
some time after its termination; (3) a group of neurones activated niter the cessation of the 
stimulus (off-neurones). All off-neurones are represented as acting for some time after the 
cessation of the stimulus by way of reverberating circuits. On the left the respective type of 
neurones of each group is indicated. The horizontal lines represent the periods of excitation of 
each particular neurone. The whole group I is activated only during the operation of the 
stimulus, some quickly adapting on-elemcnts being also shown. Group 2 is activated during 
the operation of the stimulus and after its cessation, gradually becoming less active in the 
course of time. The v.'hole group 3 is activated by the cessation of stimulus and then becomes 
gradually less active as in the case of group 2. Further explanations in text. (After Konorski 
and Lawicka, 1959. Acta Biologiae Experitnentalis, 19, 175-198.) 

As we sec the neurones of group i, being activated only during the 
operation of the stimulus, but not after its cessation, are responsible only 
for its actual ei^ccts; group la — i.e. the pure on-neurones group — 
causes the beginning of the stimulus to have a greater reflexogenic 
strength than its continuation — a well-known fact from CR experiments. 
On the other hand neurones of group 2 continue to be activated after the 
cessation of the stimulus and thus form a basis of the recent memory traces 
of that stimulus. As far as neurones of group 3 are concerned — pure off- 
elements — they account for the active role played by the cessation of a 


122 BRAIN MECHANISMS AND LEARNING 


Stimulus (again well known from the CR experiments) and also for the 
recent memory traces of this cessation. And so while neurones of group 2 
account for comiiioii features of a stimulus and its traces and provide a 
basis for their mutual generalization, neurones of group 3 account for 
diversity of the stimulus and its trace and form a basis for their possible 
differentiation. Various moments of the trace of the stimulus differ among 
themselves in that with a lapse of time fewer and fewer elements of 
groups 2 and 3 arc activated. And so the more remote the trace of the 
stimulus, the less its refiexogenic strength, a fact which is again supported 
by much experimental evidence. 

We have a strong inclination to believe that the 'sense of time' of men 
and other animals, i.e. the sense of the varying durations of time which 
have elapsed since a defmite event, is based on nothing else than the 
strength of traces left by this event at various moments after its cessation. 
The weaker these traces the more remote in time the given event seems 
to be. 


5. THE PROBLEM OF LOCALIZATION OF REVERBERATING CIRCUITS 

If the above hypothesis is correct then the problem arises as to where the 
reverberating circuits responsible for recent memory are situated. The 
simplest assumption is that they are localized in the associative areas of the 
cerebral cortex — as decorticated animals possess hardly any recent 
memory — in close vicinity to the respective sensory projection areas. In 
consequence we may expect that the destruction of these associative areas 
should lead to the abolition of recent memory for stimuli in the given 
sensory modality. 

Let us analyse from this point of view the functions of the prefrontal 
area, i.e. that associative area which was first recognized as having to do 
with recent memory in the classical experiments by Jacobsen (1936). As is 
well known this author demonstrated that after bilateral ablations of the 
prefrontal area in monkeys delayed responses are abolished and he 
attributed this defect to the loss of 'immediate niemory', as contrasted 
with the full preservation of 'stable memory'. The results of Jacobsen were 
afterwards confirmed by many authors, but his interpretation was sub- 
jected to much criticism. It was argued that the impairment of delayed 
responses is due not to the lack of 'immediate memory' but to the enhanced 
distractability of prefrontal animals (Malmo, 1942; Wade, 1947; Harlow 
et al., 1952), to their hypermotility (Wade, 1947) or to the impairment of 
associative functions (Nissen et al, 1938; Finan, 1942). 


JERZY KONORSKI 


123 


In our experiments on delayed responses with dogs (Lawicka and 
Konorski, 1959) we used an experimental setting (Fig. i) which allowed 
us to observe and analyse the changes in animal's behaviour after operation 
more clearly than was possible in the Wisconsin apparatus. We have 
found that after ablations of the frontal poles rostral to the presylvian 
sulcus (Fig. 3) the animals, completely or almost completely, lose their 


Fig. 3 
The cerebral cortex of dog represented in two dimensions (it the sheet ot 
paper is bent along the longitudinal axis, the three-dimensional picture of the 
cortex is obtained). The prefrontal associative areas and temporal associative 
areas removed in corresponding experiments are stippled. 


capacity to remember in which direction they have to go in the delayed 
response test. Being released after the delay period they 'follow their 
nose', i.e. go to that food tray to which they are just turned. If the delay 
period is short and no distracting stimulus has mtervencd, the dogs are 
able to preserve their bodily orientation towards that food tray which was 
signalled and then they are able to react correctly; we have called this type 
of reaction a 'pseudo-delayed response' since it is due not to the recent 
memory of the cue but to the actual direction of the body and head. But 
if the animals' attention is diverted for a moment, so that they change their 


124 BRAIN MECHANISMS AND LEARNING 

bodily orientation, then they will go either in the other direction or will 
not go anywhere. Of course they are also not able to go to the proper 
food tray after receiving food on the starting platform, or after being 
taken out of the room. When two signals are given, one after the other, 
the dog after release, if not distracted, goes correctly to the last signalled 
food tray and does not go to the other one. In other respects the general 
behaviour of our prefrontal dogs does not differ from that before opera- 
tion, in particular they do not display any hyperactivity or exaggerated 
reactions to new stimuli. 

Our experimental data seem to support the original idea of Jacobsen 
that the impairment of delayed responses in prefrontal animals is due to 
the loss of recent memory. However, this statement requires one substan- 
tial qualification, namely, that not all kinds of recent memory are impaired 
after prefrontal lesions but only recent memory of directional cues. In fact 
we have so far no evidence to show that other sorts of recent memory are 
also affected by these lesions and we have already some evidence that it is 
not so. One bit of it is provided by Mishkin and Pribram (1956) who 
have found that delayed responses in simple go-no-go tests were not 
impaired in prefrontal monkeys. Another one will be presented later. 

The important problem arises as to why it is that only the recent 
memory of directional cues is destroyed after prefrontal lesions. We think 
that a tentative answer to this question can be given. 

The extensive study made recently by Soltysik in our laboratory (in 
preparation) concerning the effects of caudate lesions on delayed responses 
revealed that these lesions produce striking disorciers of orienting reactions. 
The animals either do not visibly pay any attention to the auditory stimuli, 
or are not able to locate correctly the source of the stimulus even during 
its operation. When after some time this deficit is compensated the 
delayed responses are as much destroyed as in prefrontal animals. These 
animals arc also severely impaired in all locomotor CRs, not being able to 
find their way to the familiar places they ran to hundreds of times before 
operation. 

Although premotor cortex was not specially studied from this point of 
view, nevertheless as observed by I. Stg pien et al. (in preparation) premotor 
lesions, and even more so premotor-prefrontal lesions, also produce 
striking defects in the animal's orientation in space and orienting reactions. 
The connections between caudate nucleus and pericruciate region were 
emphasized by several authors (cf Purpura et al.). 

All these data show that both premotor area and the rostral part of 
caudate nucleus play an important part in general orientation of animals in 


JERZY KONORSKI I25 

space, i.e. in reacting correctly to directional cues and in formation and 
retention of locomotor CRs. Therefore, it is quite reasonable to believe 
that the prefrontal area, or rather some yet undehned part of it, is closely 
functionally connected with these regions supplynig the reverberating 
circuits responsible for recent memory of those cues which subserve this 
orientation. 

But It seems that much more precise analysis of the relations between 
projection areas and adjacent associative areas can be carried out with 
respect to the recent memory of exteroceptive stimuli on the basis of the 
test described in section 3. The corresponding experiments were performed 
by Chonjzyna and L. St^pieh in our laboratory (unpublished). After the 
dogs were trained to differentiate between pairs of identical tones (SxSx) 
and different tones (SxS>), the areas situated ventral to the auditory projec- 
tive area, namely gyrus sylviacus anterior and posterior (Fig. 3), were 
bilaterally removed. After this operation the dogs lost completely and 
irreversibly the ability to differentiate such pairs, although not only simple 
differentiations, but even conditioned inhibition (see footnote on page 6) 
was fully preserved (Fig. 4). In other words, the animal was able to differ- 
entiate between the auditory stimulus S — positive CS — and the auditory 
compound SnS — inhibitory CS — because stimulus So elicited a negative 
attitudiual reaction which was retained or remembered during the action 
of S. 

On the other hand ablation of the prefrontal area did not impair the 
performance of our test.^ This shows that the prefrontal area has probably 
nothing to do with recent memory of auditory stimuli. It is of interest to 
note that partial bilateral ablations of the auditory projection area also did 
not affect this test. 

Similar results were obtained by Goldberg ct al. (1957) in cats. After 
bilateral removal of ventral parts of the temporal region discrimination 
between groups of tones which differed only in temporal patterning was 
lost, although simple tonal discriminations were preserved. 

Even more convincing experiments were recently performed by 
St(t^pieh ct al. (personal communication). These authors, as mentioned 
before, used our test for investigating visual and auditory recent memory 
in monkeys. After the animals were trained to differentiate between 
pairs of identical and different rhythmical stimuli, both acoustic (clicks) 
and visual (flashes), different parts of the temporal lobes were bilaterally 

1 After prefrontal ablation the animal was disinhibited for several weeks (Brutkowski et al., 
1956) and therefore displayed a positive reaction to both excitatory and inhibiting stimuli, 
but this defect was soon totally compensated. 


126 


BRAIN MECHANISMS AND LEARNING 


% 


c 


anis 


100 


- 


A 


j |\ /v / 


80 


r"^ 


\ \ \l\ 


60 


- / 


\ \ v 


■<0 


■ 


\ 


20 


■L 


1 1 1 


\1 

1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 


IT 


^■^ 


,, I /\ . / 


20 ^0 60 80 100 120 140 160 180 200 10 30 50 10 30 10 30 50 


E Ibrus 


/ I 

/ 


-V^ : . 


20 40 60 80 100 120 140 160 180 10 30 50 10 30 10 30 


% 
100 

80 

60 

20 


20 ^O 60 80 100 120 UO 160 180 200 220 2<0 260 10 20 10 20 

Fig. 4 {see capHioii on opposite pnge) 


JERZY KONORSKI I2J 

removed. After ablation of the anterior parts of the first and second 
temporal gyri the differentiation of the pairs of auditory stimuli was lost, 
while the performance of the same test with visual stimuli was preserved. 
The opposite was found after the ablation of the posterior parts of the 
second and third temporal gyri and gyrus fusiformis. 

6. RELATIONS BETWEEN RECENT AND STABLE MEMORY 

It is now generally accepted that recent memory is based on the activity 
of reverberating chains of neurones, whereas stable, or permanent memory 
is due to some structural changes produced in the brain. The kind of 
changes with which wc have to do here has been a matter of much specula- 
tion. One of the possible hypotheses is that put forward long ago by 
Tanzi (1893), Ramon y Cajal (191 1), Ariens Kappers (1917), Child (1921), 
Coghill (1929) and others, and recently adjusted to the CR experimental 
evidence by Konorski (1948). According to this hypothesis when an 
indifferent stimulus is reinforced by an unconditioned stimulus, the 
'potential connections' established in ontogenesis between the correspond- 
ing groups of neurones are transformed into 'actual connections'. This is 
accomplished by growing and multiplication of synaptic contacts between 
the axons of the neurones representing the CS and bodies and/or dendrites 
of neurones representing the UCS. 

Without going more deeply into that problem one must notice that 
all structural theories of conditioning have so far encountered one major 
difficulty : it is well known that quite often a CR is formed and proves to 
be stable even after a single reinforcement, the so-called 'one trial learning'. 
This means that a few seconds' simultaneous operation of the two stimuli 
would be sufficient for the formation of stable connections between the 
respective groups of neurones. 

Fig. 4 
The effects of cortical lesions on acoustic recent memory in dogs. 

Each graph represents the percentages of negative (correct) responses, i.e. lack of movement 
for each dog to the inhibitory CS-i in successive blocks often inhibitory trials. 

IT, infratemporal ablation, EM, ablation of gyrus ectosylvius mcdius, F, prefrontal ablation. 
Continuous lines, responses to inhibitory compound stimuli, S^Sy (acoustic recent memory 
test). Dashed lines, responses to inhibitory stimulus in simple acoustic differentiation. Dashed- 
dottcd lines, reactions to inhibitory compound of conditioned inhibition, SoS. 

Arrows in the course of tone-relation differentiation (SxSx vs. SxSy) denote the end of 
preliminary training in which only particular pairs of tones were applied. 

In the first three dogs tone-relation differentiation was totally abolished after infratemporal 
lesion. Both differentiation and conditioned inhibition were easily established after operation. 
In the fourth dog bilateral removal of middle ectosylvian gyrus did not affect the tone- 
relation differentiation; the removal of prefrontal areas caused the general syndrome of 
disinhibition which was soon compensated. 


128 BRAIN MECHANISMS AND LEARNING 

This difficulty can now be overcome if we take into account the fact 
that each of the stimuh taking part in conditioning leaves transient traces 
in the nervous system such that the respective central representations of 
the stimuli are excited for a much longer time than the duration of the 
stimuli themselves. 

If the proposed mechanism of the transformation of recent memory 
traces into stable memory traces is correct, then one should predict that 
cutting the first ones short in some way or other, for instance by electro- 
convulsive shock, would lead to slowing down or even preventing the 
process of conditioning. 

Such experiments were in fact performed by a number ot authors and 
gave clear positive results. Duncan (1949) was the first to apply ECS after 
each conditioning trial in rats and he showed that the closer the application 
of the shock is to the trial, the stronger is its disrupting effect upon learn- 
ing. Later the co-workers of Gerard (1955) and Thompson and Dean 
(1955) applied ECS at various times after a single session of discrimination 
learning and again proved that the shorter the interval between the 
learning session and the shock, the poorer is the animal's retention after 
24 hours. 

All these data show that after the termination of the learning trial, or a 
massed series of trials, the consolidation of the CR continues and that this 
phenomenon depends on some on-going process in the nervous system. 
It is quite reasonable to believe that this is exactly the same process which is 
involved in the recent memory phenomena described in earlier sections. 

The important role played in the formation of CRs by recent memory 
may explain some other facts connected with learning which would 
otherwise be difficult to understand. It has been shown by Chow (195 1), 
Mishkin and Pribram (1954) and others that the ablation of posterior 
ventral parts of the temporal lobes in monkeys abolishes or greatly 
impairs visual discriminations established before operation. However, 
Orbach and Faiitz (1958) found recently that if before operation the 
animals were given prolonged, post-criterional overtraining, then the 
discrimination habit suffered little or no decrement after inferotemporal 
lesions. If we assume that the 'learning to criterion' of a visual discrimina- 
tion is largely based on recent memory, i.e. that the animal remembers 
from trial to trial and from day to day which figure is positive and which 
negative, then this would explain why after a partial destruction of the 
visual association area the habit is greatly impaired. On the other hand if 
with long training the habit becomes based on stable memory, then no 
post-operative deficit would ensue. It is worth while to stress that the 


JERZY KONORSKI 129 

ablation of the same area in monkeys produced a total abolition of recent 
memory of visual stimuli tested by our method in recent experiments by 
Stfpieii ct ill. 

To sum up, we believe that the inferotemporal area in monkeys contains 
reverberating chains of neurones connected with the visual projection area 
and therefore lesions in this area produce the deficit of recent memory 
observed either directly by using our test, or indirectly by using not very 
firmly established visual discrimination habits. 

To end this section it is necessary to draw attention to the striking 
deficits of recent memory found in recent years in humans after hippo- 
campal lesions (Milner and Penticld, 1955; Scoville and Milner, 1957; and 
others). Similar results were recently obtained after hippocampal lesions in 
monkeys by St^pieii et al. with respect to both visual and auditory stimuli. 
The physiological mechanism of these deficits seems to us so far not clear 
and they require more detailed investigation. 

SUMMARY 

In this paper the general review of the existing experimental material 
concerning recent memory in animals is presented and possible mechan- 
isms of this phenomenon are discussed. 

It has been shown that recent memory is involved in trace CRs (section i ) , 
in delayed responses (section 2) and in those forms of CRs in w^hich, in 
order to display a correct response the animal has to compare two succes- 
sive stimuli (section 3). It has been assumed that the mechanism of recent 
memory depends on throwing into activity reverberating circuits of 
neurones, connected with neurones engaged in perception of a given 
stimulus, and probably situated in the so-called association areas surround- 
ing the given projection area (section 4). As delayed responses appear to be 
based on the recent memory of directional cues, it is understandable that 
they are destroyed after lesions in prefrontal cortex situated in the vicinity 
of premotor area and caudate nucleus, structures directly concerned with 
the animal's orientation in space. Similarly acoustic recent memory is 
destroyed by lesion in gyri sylviaci in dogs and cats and anterior parts of 
temporal gyri in monkeys; visual recent memory is destroyed after 
inferotemporal lesions in monkeys (section 5). 

Recent memory plays a prominent role in the consolidation processes of 
conditioning, since it causes a much more prolonged activation of groups 
of neurones concerned in a given conditioning than is provided by the 
actual operation of the corresponding stimuli (section 6). Therefore, the 


130 BRAIN MECHANISMS AND LEARNING 

more powerful is the reverberating system of neurones within a given 
analyser in a given species, the more perfect and long lasting is the recent 
memory of the corresponding stimuli and the more rapid is the process of 
consolidation of the respective CRs. 


GROUP DISCUSSION 

RosvoLD. I would like to ask Dr Konorski, from the point of view of the position 
put forth in his paper, how he would account for the fact that in the literature there 
are many papers that deal with the impairment of sensory discrimination following 
frontal lobe lesions. 

Konorski. As we have established in our laboratory, prefrontal ablations produce 
two different sorts of impairment: one is that discussed in this paper, namely the 
loss of recent memory of directions involved in delayed response tests ; the second 
is the impairment of inhibitory conditioned reflexes. While the first one is, 
according to our data, irreversible, the second one is, on the contrary, compensated 
with further training. It is clear that the symptom described in Rosvold and 
Mislikin's paper belongs to the second category. We suppose that both these 
symptoms are of different origin and mechanism, and we try now to check this 
supposition by experiments. 

Rosvold. With reference to the permanence of these deficits, I would say that 
contrary to Jacobsen's earlier statements, in the chimpanzee the deficit is not 
permanent. Instead with training the animals gradually recover their ability to 
perform very well on tests of recent memory. The chimpanzee tells us another 
thing: a subjective thing to be sure, but we can see in watching the animals no 
evidence that orientation is a critical factor for those tests. In fact it is not unusual 
for the animal to be standing on his head when looking at the stimulus, and then 
turn completely over and respond correctly. Thus, even though this notion of 
proprioceptive recent memory being the unique function of the frontal lobes is 
intrigunig, I am not entirely convinced that it is the answer to delayed response 
deficit. As Dr Konorski stated, we simply do not know what the stimulus factors 
in the delayed response problem really are. We have assiduously tried to determine 
what it is in this problem that the animal is responding to or by what means he is 
responding. We have not been able to demonstrate definitely what it is in the test 
situation that the animal uses as the basis for his solution of the problem. This is 
why there is so much speculation about the probable function of the frontal lobes. 
Probably the only resolution is the development of a test method in which the 
stimulus and response variables will be clearly delineated. 

Hebb. I would like to add to this that bilateral total removal in another species 
produces no sign of defect: even better than in the chimpanzee. 

Konorski. In man, yes. 

EsTABLE. I am very much interested in Professor Konorski's communication 
that gives rise to so many problems. Obviously the only objection, if it can be 
called that, is that we have on one side techniques to study the central nervous 
system and on the other side techniques to study psychological problems, and the 
bridge is always there, but it is hard to correlate and the passage from one to the 
other is hard to interpret. 


JERZY KONORSKI I3I 

One must be precise with terms and when one says centre, for instance, one 
must not think that the centre of the spoken word means that the spoken word is 
venerated there. If a cortical area is destroyed and the spoken word is perturbed or 
disappears, all that one can say is that the cortical area is necessary but not surticient 
for the reproduction of the spoken word. 

The fact that neurones are incapable ot reproducing themselves is perhaps the 
price we have to pay to have memories and habits and other learned capacities 
which persist throughout life. If neurones died and were substituted by others, it 
is hard to conceive how these functions could be preserved. 

Reverberation circuits may be the basis of learning but perhaps they are not 
enough. Do vou, Dr Konorski, conceive as diiferent phenomena: habits, organic 
memory and memory itself? 

Konorski. As far as I know in dogs and in monkeys the loss of recent memory of 
directions after prefrontal ablations is irreversible. JVIoreover, in some of our 
annuals, there was only a partial deficit of recent memor)', due perhaps to smaller 
lesions. Why in Dr Rosvold's chimpanzee the deficit of recent memory is rever- 
sible, should be explained by further experimentation. As far as patients with pre- 
frontal lesions are concerned, they are able to solve the delayed response test very 
well. We believe that it is so because their memory of directions is supported, or 
even based, on visual cues and also on verbal recent memory which in man, is very 
powerful. 

I agree that we are not able so tar to state definitely which sorts ot cues are 
responsible in delayed response tests, although it seems that this problem may be 
solved by further experimentation. 

To answer Professor Estable, I would like to stress once more that, according to 
our view, conditioned reflexes, including habits as a particular form, are chiefly 
based on stable or static, or organic memory, while such forms of behaviour which 
are involved in delayed responses are based on transient or dynamic memory. 

Olds. I wish to address myself to the experiments where the so-called auditory 
association areas were removed, and where Dr Konorski says his experiment was 
on recent memor\-, for an auditory stimulus. I wish to suggest the possibility that 
these were not experiments on recent memory at all, but possibly on the animal's 
ability to compare two stimuli. It seems to me that the way this could be tested is 
to have the stimulus and the comparison stimulus (another S^ or Sy) presented 
somehow simultaneously. If the deficit is hi the animal's ability to compare, he 
will fail this test. If it is in recent memory, he will pass. One might also ask the 
question of the delayed response technique again here and ask if it is really fair to 
reject the delayed response as a test for recent memory of a particular (e.g. auditory) 
stimulus. One might develop some text in which a given tone Sx would not signify 
a direction but an abstract contingency and find whether excision o( the auditory 
association areas prevented an auditory memory which involved no comparison. 

Konorski. As to the first question of Dr Olds I agree that the test proposed b)' us 
is based on comparison of two stimuli, wliich is rendered impossible by the loss of 
recent memor)-. Whether or not there is any special function which may be called 
'comparison' — I do not know. 

As to the second question I would remind that, as shown in my paper, the 
removal of pre-frontal areas did not impair the recent memory of particular 
auditory stimuli, as proved by our recent memory test. 

E 


132 BRAIN MECHANISMS AND LEARNING 

Anokhin. I should like to make two comments on Dr Konorski's very interest- 
ing report. The brain, as a specific substance has two possibilities for 'remember- 
ing', each of which can be related to what Dr Konorski relates to 'recent memory'. 
There is first of all the ability of a nervous substance to associate any successive 
stimuli coming trom the inner or outer world ot the organism. On account of its 
ability to conduct stimuli rapidly and multilateralK' and oi its ability to retain in 
the synaptic systems the molecular changes which have taken place, the nervous 
tissue 'remembers' any succession of stimuli brought upon it. That is the most 
direct and universal memory which fits under the name of 'recent memory'. 

Tliis memory, however, represents a specific advance effected by the nervous 
system in relation to the acting agents of the outside world. 

For a given association to become stable, it has to end by a strong emotional 
discharge, i.e. has to end by some event which is meaniiigtul for the life of the 
organism. 

Thus in my view, to study the physiological mechanisms of 'recent memory' 
consists basically in discovering those concrete processes which occur as a result of 
emotional discharges and spread in the direction of the cerebral cortex, retaining 
there fleeting temporary associations. 

As is shown by the unavoidable extinction of desynchronization when training 
for the association 'sound-light', a final consolidation of such ephemeral associa- 
tions is indispensable. 

AsRATYAN. Food boxcs are very important factors in recent memory. Do you 
think the rule of the tonic conditional reflex, as we name it, plays a role in this 
recent memory? 

BusER. I wish to ask Dr Konorski if he has some information on the thalamic 
connections of the cortical areas which were ablated in his experiments. 

Konorski. I tliink that Asratyan's 'tonic conditioned reflexes' are based on the 
same principles as ordinary conditioned reflexes, i.e. when firmly established, they 
are due to stable memory traces. 

Our lesion producing the loss of delayed responses comprised gyrus proreus, 
subproreus and the anterior part of gyrus orbitalis. The vessels in presylvian sulcus 
were usually spared. These lesions produce degeneration in dorso-medial nucleus 
of the thalamus. Other desenerations were not so far studied. 


CONDITIONED REFLEXES ESTABLISHED BY COUPLING 
ELECTRICAL EXCITATION OF TWO CORTICAL AREAS^ 

R. W. Doty and C. Giurgea 

Is the mere coincidence of action of two stimuli sutiicient to form a 
learned association between them, or is some motivational factor also 
required? This has been a persistent question in psychology (and psy- 
chiatry) and is of major importance in any attempt to solve the riddle of 
the neural mechanisms subserving learning. Human experience is too 
complex to provide an answer despite the long history of 'associationism' 
as a science. In animal experimentation, on the other hand, it has been 
difficult to demonstrate learning or establish conditioned reflexes when 
motivation or 'drive reduction' have been unequivocally absent. 

The pertinent evidence has been reviewed elsewhere (Giurgea, 1953a). 
The experiments of Loucks (1935), however, are of special interest since 
they are the direct antecedents of our own. In several dogs Loucks 
stimulated the 'motor' cortex through permanently implanted electrodes 
effecting a movement of one of the animal's limbs. With each animal on 
as many as 600 occasions over several days he preceded the motor stimula- 
tion with an auditory conditional stimulus (CS). The CS never came to 
evoke the movement with which it was so thoroughly paired, nor any 
other movements. A food reward was then introduced to follow the CS 
and the induced movement. Within a few trials the animal began moving 
to the CS. Loucks drew the justifiable conclusion from these experiments 
that the motivational element was essential to the formation of conditioned 
reflexes and these results have had a wide and well-deserved influence on 
psychological theory since that time. 

Louck's position was apparently conhrmed by Masserman (1943) who 
failed to obtain any signs of conditioning when stimulation of the 
hypothalamus was used as US. However, Brogden and Gantt (1942) were 
able to produce movements by presenting a CS alone after repeated pair- 
ing of CS and stimulation of the cerebellum. In some animals the 'condi- 
tioned response' (CR) so elicited was very similar to the movement 
induced by cerebellar stimulation. Motivation seemed to be absent here. 

1 New research reported here was supported by grants to R. W. Doty from the Foundations' 
Fund for Research in Psychiatry and the National histitutcs of Neurological Diseases and Blind- 
ness (B-1068), and by a travel grant to C. Giurgea from the Academy of the Rumanian 
Peoples Republic. 


134 BRAIN MECHANISMS AND LEARNING 

In 195 r, while working in the laboratories of P. S. Kupalov seeking further 
confirmation of the hypothesis of 'shortened conditioned reflexes', it was 
discovered that conditioned reflexes could readily be formed by pairing 
stimuli at two cortical points (Giurgea, 1953 a, b). Stimulation of occipital 
cortex w^hich is initially without apparent effect ultimately produces 
movement highly similar to that elicited by the stimulation of the sigmoid 
gyrus with which it is paired. This direct contradiction of the results of 
Loucks seems best explained by differences in the timing of presentation 
of stimuli. In all his experiments Loucks (1935) used intertrial intervals of 
2 minutes or less, usually 30-60 seconds, whereas an intertrial interval 
of 3-5 minutes was used in our experiments. If the intertrial interval is 
reduced to 2 minutes even after such CRs have been established, the CRs 
disappear in the majority of dogs tested so far (Giurgea, 1953 a, b), 
although the unconditioned response (UR) is unaffected. 

Continuing these studies, it has been shown that formation of CRs by 
cortical stimulation is not dependent upon sensory endings in the meninges 
since the CR may be readily established after destruction of the Gasserian 
ganglion (Raiciulescu, Giurgea and Savescu, 1956). A CR established to 
stimulation of parietal-occipital cortex as CS was also elicited by a tonal or 
photic CS (Giurgea and Raiciulescu, 1957). In two animals with total, 
histologically confirmed section of the corpus callosum CRs were 
established even though CS and US were applied to different hemispheres 
(Raiciulescu and Giurgea, 1957; Giurgea, Raiciulescu and Marco vici, 
1957). The electrical activity recorded from the US area does not appear 
to be changed by this conditioning procedure and is within normal limits 
of low voltage, fast activity very shortly after the US is applied (Giurgea 
and Raiciulescu, 1959). In none of these experiments does the behaviour of 
the dogs indicate even the slightest element of motivation. This impres- 
sion has now been conhrmed objectively in the experiments reported 
below. 

TECHNIQUE 

The experiments at Michigan have so fir been performed on four dogs, 
two cats and two cynomolgous monkeys. The dogs are restrained easily 
by placing their legs through plastic loops (e.g. Fig. i). Cats are held by 
placing their heads through a heavy plastic stock leaving their limbs free. 
Monkeys are kept permanently seated in a Lilly-type chair (Mason, 1958). 
All animals are adapted to restraint prior to electrode implantation. 
During the experiment the animals are isolated and observed through a 
'one-way' glass. 


R. W. DOTY AND C. GIURGEA I35 

The cortex is stimulated through platinum electrodes, usually resting 
on the pial surface or just beneath it although in some dogs the thickness 
of the skull made such adjustment difficult. Two of four electrodes are 
carried in 7 mm. diameter plastic buttons which are held in trephine holes 
by means of screws (Doty, Rutledge and Larsen, 1956). Flexible 0.5 mm. 
diameter polyethylene insulated wires connect the electrodes to an 1 8 or 
34 contact receptacle permanently secured to the skull by stainless steel 
posts (Doty, 1959). 

Stimulation consists of i msec, rectangular current pulses at a frequency 
of 50/sec. and is monitored on a cathode-ray oscilloscope with a long- 
persistence screen. Great care is taken to keep the stimulating circuits and 
the animals isolated from ground and to avoid any other possibility of 
stimulation outside that intended. 

Prior to pairing CS and US the effects of stimulation are observed for 
each electrode pair. It is advantageous to have several pairs of electrodes 
in 'motor' cortical areas so that the chances are increased for procuring a 
relatively simple movement to serve as an UR. By means of automatic and 
silent control the CS is presented for 3-4 seconds and is slightly overlapped 
by the US of 1-1.5 seconds duration. Six to ten combinations of CS and 
US are made daily. 

After study of CS-US coupling is complete, the animals are trained in 
the same experimental chamber to press a lever to obtain food. The lever 
is then connected to administer cortical stimulation with each press. 

RESULTS 

Doi^ Alplhi. All stimulatit)n in the 'motor' cortical regions produced 
stiff, complex and unnatural movements. That finally chosen as an UR 
was a lifting and extension of the right hind leg, a slight lifting and curling 
of the tail, and a rotation of the head to the midline and down (Fig. 1). 
The US was 1.8 mA. applied just posterior to the left postcrucial sulcus. 
The CS of r.i niA. was applied to the left posterior suprasylvian gyrus. It 
elicited no response for the first forty-two CS-US pairings. The CS 
current was then increased to 2.2 mA. and elicited an opening of the eyes 
and turning of the head to the right, a response judged to be inherent to 
stimulation of this area. It was still obtained, when, later in the experiment, 
the CS was again reduced to i.o mA. The first distinct CR was seen on the 
thirtieth post-operative day after 108 CS-US pairings. It was a turning of 
the head to the midline and down, a movement similar to the head move- 
ment seen to the US and in opposition to that inherently evoked by the 


136 


BRAIN MECHANISMS AND LEARNING 


CS. Movement of the leg or tail was never elicited by the CS. This CR 
subsequently occurred up to lOO per cent of the time in some sessions and 
had a threshold of about 0.3 mA. It could also be elicited by 0.4 niA. 
applied to a second pair of electrodes about 2 mm. distant from the 
original CS pair. 


Fig. I 
Unconditioned response o( Do^ Alpha to stimulation just posterior to left postcrucial sulcus. 


The same CR was evoked by a CS of 3 seconds clicks after one session 
(eight trials) combining this CS with the UR. A second UR was coupled 
with 9/seconds clicks as CS. This at first produced the previous CR which 
gradually became modified to a sidewise oscillation of the head with nose 
pointing down. This new 'CR', however, had nothing in common with 
the second UR. 

It was very difficult to teach this dog to eat in the experimental situation. 
Once trained, however, the animal pressed the lever repeatedly despite 
accompanying CS or US stimulation which in the case of the US produced 
violent movements. In contrast, if the side of the cage was tapped gently 
each time the lever was pressed, two or three taps abolished all pressing 
for the rest of the session. 


R. W. DOTY AND C. GIURGEA 1 37 

Dog Beta. A us of 2.0 niA. at the right postcruciatc gyrus produced a 
brisk, well-integrated flexion of the left hind leg as its only apparent 
effect. The CS at the right marginal gyrus gave no overt sign at intensities 
up to 2.2 mA. The hrst sign of movement to this CS occurred during the 
sixth session, forty-fifth pairing. The slight tossing of the head and 
indefniite movements such as stepping or shifting posture seen then sub- 
sequently became very common. On the sixty-sixth piairing two lo-cm. 
flexions of the left hind leg, held for about i second each, were seen as the 
only movement to the CS. This was the first CR. This type of CR occurred 
seventy-four times in 171 subsequent CS presentations (including extinc- 
tion) and had a threshold of 0.95 mA. It was not extinguished by eighty- 
four presentations of the CS alone at 2-3 minute intervals for eleven 
sessions. These CRs were also obtained to stimulation ofthe right posterior 
cctosylvian gyrus indicating some generalization had occurred. 

Technical difficulties prevented testing this dog in the lever-pressing 
situation. However, the animal was extraordinarily sensitive and yelped 
violently even when grasped gently by the scruff" of the neck. Yet there 
was no evidence of pain or emotion during the training sessions, and the 
animal ran each day to the experimental room and jumped into the 
enclosure to be harnessed. 

Dog Gallium. Stimulation at 0.4 mA. 1.5 mm. above the right pyramid 
in the field H^ of Forel was used as US. It produced a forceful extension of 
the neck and rotation ofthe head over the right shoulder, wider opening 
ofthe eyes, flaring ofthe nostrils and occasionally a lilting ot the right lip 
(Fig. 2). This response was elicited sixty-eight times in ten sessions with no 
particular alteration in the animal's behaviour. Pairing ofthe US with a 
CS of 3/second clicks was then begun. A CR of turning the head up and 
130° right was elicited by the CS on the sixteenth trial and similar CRs 
occurred on forty-eight of ninety-six subsequent presentations. The CS 
also evoked great agitation, whining and yelping. Both the CRs and this 
agitation to the CS were extinguished in three sessions totalling twenty- 
two presentations of the CS alone. 

Lever-pressing behaviour was little altered by coupling with stimulation 
ofthe postcruciate gyrus producing a hind-leg left. It was slowed greatly 
by the click CS or other sounds, and abolished immediately by the US. 

Dog Epsiloii. Stimulation at right or left postcruciate gyri produced well 
co-ordinated, maximal flexions of left or right forelegs respectively. 
Stimulation of left posterior ectosylvian gyrus with 1.8 mA. or right mid- 
marginal gyrus with 1.6 mA. produced no response when tested initially. 
Stimulation at the latter points was then used as CS and at the former as 


138 


BRAIN MECHANISMS AND LEARNING 


US. For convenience the prefix 'R' or 'L' will be used to designate to 
which hemisphere the stimulus was applied, e.g. L-CS with R-US means 
CS applied to left posterior ectosylvian gyrus, US to right post cruciate. 
For the first forty pairings of R-CS with R-US there was no response to 
the R-CS. During the next few sessions the R-CS frequently elicited a 
lowering of the head and flexion of the right foreleg as shown in Fig. 3, 
i.e. the limb opposite that in which the UR was induced. The first left 


Fig. 2 

Unconditioned response of Dog Gainuui to stnnulation in right 

field Hi of Forel. 

foreleg CR occurred on the eighty-first presentation of R-CS and in the 
next 113 trials occurred hfty-cight times. These CRs were discrete, 4-20 
cm. elevations of the left foreleg held for several seconds and frequently 
returned to the floor prior to the UR. The threshold for CR elicitation was 
0.5-0.7 niA. Other movements, not so specific, occurred regularly, so that 
movement now occurred 95 per cent of the time to this CS. 

L-CS was now paired four times with R-US, and except for the first 
presentation produced flexions of the right foreleg (again opposite to the 
UR) each time. Two sessions later differentiation was begun and continued 


R. W. DOTY AND C. GIURGEA 
Table I 

INITIAL FAILURE OF DIFFERENTIATION IN DOG EPSILON 


139 


Stiiuiiliis location Total differentiation trials 


LFL CRs RFL 'CRs' 


R-CS, R-US 
L-CS, no US 


50 
44 


14 


for twelve sessions with results as shown in Table I. Obviously no differen- 
tiation occurred under these conditions; CRs were maintained to stimula- 
tion of cither area though the percentage of occurrence to R-CS may have 
declined. 


Fig. 3 
One of the first conditioned responses of Dog Epsilon to stimulation of right marginal gyrus. 
At this time the CR is in the 'wrong' leg, but in subsequent trials the left foreleg, the one lifted 
to the US, lifted to this CS (Doty, 1959). 


It is characteristic of these animals (Giurgea, 1953 a, b) that just prior to 
the appearance of the first CRs and frequently thereafter, movements of 
the affected limb occur sporadically during the intertrial period. This 
phenomenon was observed in Dogs Beta and Epsilon. These movements 


140 BRAIN MECHANISMS AND LEARNING 

might possibly arise from some 'irritative' process consequent to the 
repeated presentation of the US, but the following experiments show this 
is not the case. Using the usual timing and parameters, the L-US alone 
was presented lOO times in twelve sessions producing a forceful right fore- 
leg flexion each time. No spontaneous lifts of this limb occurred in this 
period. Another 150 presentations were then given in which L-US 
preceded L-CS by irregular intervals. Initially L-CS still gave some left 
foreleg CRs or other movements, but these reactions were extinguished by 
this procedure since for the last seventy presentations there was no response 
whatever to L-CS. Orthodox temporal relations for L-CS and L-US were 
then used. At first this restored the nonspecific movements, then left 
foreleg CRs and ultimately right foreleg CRs occurred about 50 per cent 
of the time to L-CS at a threshold of 0.7 mA. 

The R-CS was then presented for the first time in 79 days. The left 
foreleg was lifted 15 cm., flexing at the wrist, and was held so for about 
5 seconds. Ultimately it was possible to obtain right leg flexions to L-CS 
and left leg flexions to R-CS. 

Coupling L-CS, L-US or R-US with the animal's lever-pressing had 
no effect, whereas auditory stimuli or R-CS completely abolished pressing. 
The effect of R-CS had been predictable on the basis of behaviour changes 
seen as soon as the use of R-CS was resumed (after the hiatus described 
above). Since more than 5 months had then elapsed from time of electrode 
implantation it seems likely trigeminal fibres had grown into this medial 
electrode location. 

Monkey 1. The CS of i.o mA. applied to the left occipital pole con- 
sistently produced movement of the eyes, and sometimes the head down 
and to the right. At any time during this stimulation, however, the eyes 
might be moved elsewhere if attention was so directed. The 0.2 mA. US 
(300/second) in the left precentral cortex produced a smooth, vigorous 
flexion of the right forearm and contraction of the muscles on the right side 
of the neck and face causing the mouth to open (Fig. 4). Coupling of the 
CS and US was begun 2 weeks after surgery and continued for 4 weeks, 
200 trials in twenty-four sessions. In these 200 trials the right arm made 
random movements during the CS eleven times. Obviously there was no 
conditioning. During the next 25 weeks the animal was used sporadically 
for testing various lever-pressing procedures with fruit juice rewards. 

Coupling of CS and US was then resumed. The effects of these stimuli 
were still exactly as they had been 7 months earlier. LJsing a two-minute 
intcrtrial interval for the first fifty-one trials in five sessions there was no 
sign of conditioning. A four-minute interval was then used for trials 


R. W. DOTY AND C. GIURGEA 


14-1 


52 to 84. The first CR was seen on the sixty-seventh pairing. The right 
arm was flexed to the level of the restraining collar, then extended along 
its lower surface with hngcrs fluttering as though seeking an object. In 
the next forty-seven trials a movement similar to this occurred to the CS 
thirty-one times. At the threshold current of 0.55 mA. or during the later 
phases of subsequent extinction sessions the movement was more likely to 
be a simple flexion very similar to that evoked by the US (Fig. 4). Even 
when the movement was vigorous, it often terminated prior to the US. 


3 .? 


^^*SK 


Fig. 4 
Responses o( Monkey 1, enlarged from a 16 mm. colour film which was taken through a 
one-way mirror. Left and centre: conditioned flexions of right forearm to CS at left occipital 
cortex. The slight inclination of the head to the right, which also imitates a movement of the 
UR, was sometimes seen with high intensities of the CS before conditioning was begun and 
hence cannot accurately be considered a part of the CR, although its consistency and tiireshold 
of elicitation may have been altered by the conditioning procedure. The CS and US signal 
lights were used only during photography. Ri^ilit: unconditioned response to stimulation of 
left precentral cortex. Note similarity of end point of this response to that obtained by Dclgado 
from stimulation of the rhinal fissure (Delgado, 1959, Fig. i top). 


In ninety-one presentations of the CS given in five sessions without the 
US this CR was elicited fifty-six times. One-minute intervals were 
frequently employed. In two sessions after repeated presentations of the 
CS without the US, the CR was absent for five or more consecutive 
trials. The head and eye movements evoked by the CS were still present 
during this period of extinction. On several occasions it was noted, how- 
ever, that the eyes closed at the onset of the CS and the animal appeared to 
drowse even though the eyes could be seen moving beneath the lids. 

Stimulation with an electrode pair within 1-} mm. of those used as CS 
gave no eye movements if the polarity of the stimulus was negative for the 
electrode separated from the others by a small sulcus. None the less the 


142 BRAIN MECHANISMS AND LEARNING 

CR was evoked from this pair with this pohirity three times in six stimula- 
tions, never using the US. Stimulation in the right posterior parietal lobe 
elicited eye and head movements which were almost the exact counterpart 
of those elicited by the CS save they were to the left. No CRs were elicited 
by this stimulation in nine attempts during two sessions. 

With four-minute intervals and the same US there was no CR to an 
auditory CS despite no pairings in nine sessions. After this, giving the 
auditory CS simultaneously with the CS to the occipital pole often pro- 
duced 'external inhibition' of the CR to the latter stimulus. Seventeen 
tests with the auditory CS alone (plus the US) elicited no CRs even though 
the tests were given randomly throughout twenty-four presentations in 
which simultaneous auditory and cortical CS (plus US) was used. 

The effect of coupling various stimuli with lever-pressing was then 
studied. On July nth the animal pressed the lever forty-one times in a 
fivc-minute period and with each press received a US of 0.4 niA. (double 
the intensity used routinely) which produced violent movements, and 
often convulsive after-discharges, with each press. There was no hesitation 
whatever in lever-pressing behaviour beyond that attributable to the 
physical handicap consequent to the induced movement (e.g. Fig. 5). 
Coupling with cortical CS was similarly without effect. On the following 
day, as seen in Fig. 5, a novel clicking stimulus completely disrupted the 
behaviour when coupled with each lever-press. 

The animal was then taught in sessions of twenty trials per day to press 
a lever to avoid a shock to its tail. The first response to a tonal CS occurred 
on the thirty-third trial and a criterion of twelve avoidances in twenty CS 
presentations was reached in 128 trials. Generalization to 20/sec. and 5/sec. 
clicks as CS was immediate. The cortical CS, however, produced the 
former CR, a flexion of the right hand towards the chin rather than the 
extension to press the lever at the level of the abdomen. After twenty-five 
combinations of this cortical CS with the tail-shock, the animal began 
making lever presses to avoid this US. It took more than 100 trials, how- 
ever, before the former CR was fully extinguished under these conditions. 
The threshold for the shock-avoidance response to cortical CS was 0.2 
mA. which is significantly less than the 0.55 niA. threshold at the same 
electrodes for the flexion CR established without motivational context. 

Monkey 2. The CS and US elicited responses very similar to those seen 
in Monkey 1 save that the seizure threshold was much lower in this animal 
and the UR was frequently followed by a few clonic movements. The 
animal struggled almost continuously in the experimental situation (even 
when being given fruit juice) and no CRs were ever observed. Using 


R. W. DOTY AND C. GIURGEA 


143 


MONKEY # 
12 JULY 59 


r CLICKS T 

14 JULY 59 


.1- '. 


— L 


^-;-^— ; 


j^ 


-:^-|t::_ , 


; 


z— i .- 


F" 


1 


^: - 


- ii tfi~ 


— T 


E' : - : 


l_ . , - — 


1^- 


- - 


■ 1 -- 
1 


^ 


if i 


L£l-. 


■-■' ' . 


1- .A 


r 


Er 


aJV-^ 


^W 


t — '^!!^ 


T^:-.[- 


=^n^ 


|=.j==;i : 


\ 


_#^ 


: jj^y-. 


.=^^1 


^^jL-S*t 


:^^- ! ^;^- . 


^ 


-^ : . 


k^ — 


"-/■n-l' 


Y"' 


^:-^->-Z^~ - 


B 


-"-;-■ : -r;: : 


r - ■ 1 


t 


, T „ T 

MOTOR CORTEX 
0.4nnA 


15 JULY 59 


CLICKS TONE 


Fig. 5 
Records of lever pressing for grape juice. The animal is rewarded for each 
press and this is recorded by an upward excursion of the trace. On July 12th 
for the fivc-minute period labelled 'clicks' between the arrows each lever- 
press turned on a clicking sound for 3 seconds in addition to operating the 
solenoid dispensing the juice. No tests were made on July 13th. On July 14th 
coupling a cortical US with each Icvcr-press produced no disturbance in pressing 
behaviour beyond that attributable to the violence of the UR. On July 15th the 
same clicking sound used on July 12th now no longer disturbs pressing 
behaviour, but introduction of a 2000 cycle tone with each press brought this 
behaviour to a halt. Started again pressing without the tone by administration 
of several 'free' millilitres of juice, the monkey continues to press with only a 
minor slowing of rate when the tone is presented during a second period. 


144 BRAIN MECHANISMS AND LEARNING 

threc-niiiiute intervals nincty-onc couplings were administered in twelve 
sessions. Then after a hiatus of 25 weeks another 116 combinations were 
given with two-minute intervals in three sessions and eighty-five com- 
binations at four-minute intervals in four sessions. Finally, 265 couplings 
were given automatically at four-minute intervals in two overnight 
sessions, but still without indication of conditioning. 

The animal was then conciitioned to avoid a shock to the tail by pressing 
a lever. The first CR to a tonal CS required 105 trials and 238 trials to 
attain twelve avoidances in twenty CS presentations. Generalization to 
other auditory conditional stimuli was immediate and complete. The 
cortical CS was then employed. For 300 conabinations of this CS with the 
tail-shock US there was, judging by the animal's general behaviour and 
visually obscrveci respiration, no anticipation whatever of the impending 
shock during a four-second CS. Yet on every presentation the CS was 
patently effective since the eyes moved persistently down and to the right 
during each stimulation. The first CRs occurred after the CS was in- 
creased to 1.2 niA. The threshold was ultimately determined to be 0.6 
mA. for the elicitation of this CR. 

Cat ^2^. Conditioning with cortical stimulation was likewise a failure 
in this animal; but so too was conditioning using an auditory CS and 
foot-shock as US. The animal was also seizure-prone. A total of 345 
pairings of a right middle ectosylvian gyrus CS with right ansate gyrus 
US produced no CRs, and 420 pairings of a tonal CS with foot-shock US 
was almost equally ineffective although a few CRs were seen. 

Cat 48g. This animal has been thoroughly studied by Allan Minster. 
The CS applied to right middle ectosylvian gyrus usually elicited no overt 
response initially at the currents used. At higher currents the animal 
looked up and to the left. The US in the right ansate area produced an 
abrupt turn of the head to the left and flexion of the left foreleg. The first 
CR occurred on tlie thirty-ninth pairing, fifth session. These CRs, how- 
ever, never became consistent from one session to the next. In 100 trials 
after the first CR there were a total of only thirty-two and fifteen of these 
were made in three of the fifteen sessions. At first, the CR was a vigorous 
lifting and extension of the left foreleg, but after the seventy-second pair- 
ing the right foreleg executed this same movement to the CS more often 
than did the left. The threshold for CR elicitation by stimulation of middle 
ectosylvian gyrus was probably about 0.8 mA. Stimulation of the right 
posterior ectosylvian gyrus at 0.35 mA. yielded seven right and one left 
foreleg CRs in eleven presentations. To date, ninety-seven pairings of the 
right ansate US with a CS applied to left middle suprasylvian gyrus 


R. W. DOTY AND C. GIURGEA I45 

produced only four leg lifts that could equivocally be called CRs, yet the 
current for this suprasylvian CS has been kept high enough that head 
movements initially seen to this stimulation still occurred frequently. 

Ten-channel EEG records taken for each trial in this cat have not 
yielded much new information. Conhrming Giurgea and Raiciulcscu 
(1959) there is rarely any electrical abnormality even in the immediate 
vicinity of the electrodes following either CS or US. No changes char- 
acteristic of the conditioned state could be detected and it could not be 
predicted from the electrical record whether a CR would or would not 
occur. The CS in middle or posterior ectosylvian gyrus usually produced 
immediate electrocortical arousal whereas slower patterns frequently 
persisted through most of the middle suprasylvian CS. The arousal 
produced by CS and US, however, was minimal since 8-i2/sec. rhythms 
returned often within 1-2 seconds after stimulus cessation. 


INTERPRETATION OF RESULTS 

There can be no question that conditioned reflexes can be established 
with cortical stimulation as US. To the thirty dogs successfully contiitioned 
in this manner in the laboratory in Bucharest and the cats observed by 
Nikolayeva (1957) can be added the animals of this study extending the 
phenomenon to the monkey and to a wide variety of cortical electrode 
placements. The vagaries in the appearance of this phenomenon un- 
doubtedly arise not only from its complexity but from ignorance of its 
nature and the procedures most favourable for its induction. Such 
problems are not unknown in the study of more usual types of condi- 
tioning. 

With equal assurance one can state that, in the usual sense of the word, 
there is no motivation involved in the formation of CRs by coupling 
cortical stimulations. The animals tested so fir have remained entirely 
insouciant to self-administered cortical stimulation of near convulsive 
strength inducing violent movement, yet were profoundly inhibited by 
moderate and innocuous auditory stimuli. There is little reason to expect 
the cortical stimulation might be rewarding, and there was nothing in our 
observations to suggest this. On the other hand where motivational factors 
were expected as with the possible meningeal involvement {Doi^ Epsiloii) 
or in the held of Forel {Doi^ Gamma), the method readily conhrmed this 
impression. It is of some interest that in Dcn^ Gamma the presence of this 
aversive factor did not preclude some form of motor conditioning. 

Besides the direct evidence gained from coupling the cortical CS and 


146 BRAIN MECHANISMS AND LEARNING 

US with lever-prcssiug there is equally convincing circumstantial evidence 
for the absence of motivation in these conditioning procedures. The 
extreme brevity of electrocortical arousal follow^ing the CS-US com- 
bination in Cat 48g, or Monkey 1 closing its eyes to doze with onset of the 
CS, is scarcely expected from motivational stimuli. Nor are these CR 
movements purposeful; rather they are physiological absurdities. They 
seldom prepare the animal posturally for the ensuing UR and more often 
than not the CR is terminated before the UR begins. Watching Monkey 1 
day after day raise its arm to stimulation of its 'visual' cortex one could not 
escape the feeling that this nonsensical yet persistent movement was some- 
how analogous to the compulsive movements of neurotic humans. 

The work of Segundo and his colleagues, reported at this meeting and 
elsewhere (Segundo, Roig and Sommer-Smith, 1959), in which a tonal 
CS evokes movements similar to those elicited by electrical stimulation of 
centre median or the mesencephalic reticular formation as US, provide 
excellent confirmation of our observations — movements produced by 
central stimulation can be conditioned. Masserman's failure to obtain such 
results from hypothalamic stimulation (Masserman, 1943) stands in sharp 
contrast to the results with Doi^ Gaiiiiiia and the work of Segundo et al. 
Obviously the hypothalamus must be carefully re-examined in this 
regard. The US in the experiments 9f Segundo et al. probably produced 
motivational effects in some of their animals, but in others it appears less 
clear. In any event it would be difficult to ascribe some 'biological 
significance' or appropriateness to the movements induced by the CS. It 
is equally difficult to find a motivational basis for the 'backwards' type of 
conditioning linking the US with the CS as analysed by Asratyan in this 
volume. 

Since CRs can be established without a motivational factor, there is 
more hope that the basic phenomena of learning can be sought in neural 
systems sufficiently simple for meaningful analysis, perhaps with purely 
electrophysiological techniques. The temporal factor appears to be critical 
and the experiments with Doi^ Gaiiiiiia and especially Doo Epsilon show 
the conditioned state is not established by the repeated, randomized 
excitation of US and CS systems. 

Several observations support Sperry's hypothesis (1955) that the signi- 
ficant alteration is to be sought in the effector system. In motivated condi- 
tioning an alteration in excitability specific to the limb being conditioned 
can be observed some time before any CRs appear (Doty and Rutledge, 
1959)- In the present and earlier experiments (Giurgca, 1953 a, b) move- 
ments similar to the CR, never seen previously in the animal's behaviour, 


R. W. DOTY AND C. GIURGEA 147 

often appeared spontaneously at about the same tune that CRs were first 
noted. The threshold of the neural heirarchy controlling the complex CR 
thus seems to have been lowered. The frequent generalization of these 
CRs to other stimuli supports this view and m Doo Epsiloii and Cat 48g 
it was observed that convulsions induced accidentally by high-current 
CS began with remarkably prolonged CRs. 

Some animals seem to have inherently low thresholds for particular 
movement complexes so that stimulation at widely separated points with- 
in the nervous system will evoke the same response. For instance in Cat 
^23 stimulation in sensorimotor cortex, posterior ectosylvian and middle 
suprasylvian gyri, and the caudate nucleus produced an abrupt turning of 
the head which was Inghlv similar for the different stimulus points. 
Stimulation in middle ectosylvian gyrus, ventral anterior and ventral 
posterolateral nuclei did not produce this effect but a convulsion elicited 
from middle ectosylvian gyrus began with prolonged turning of the head 
r8o^ to the rear. In Cat j,C\^ flexion and, at higher currents, attack move- 
ments of the foreleg could be elicited by stimulation in the anterior portion 
of the caudate nucleus, septal region, ventral hippocampus, and periaque- 
ductal grey. This motor response was not correlated with the motivational 
effects of this stimulation since caudate and septal stimulation increased 
lever-pressing, periaqueductal stimulation was avoided and hippocampal 
stimulation was 'neutral'. Stimulation of the median forebrain bundle, 
pyriform area and habenula which produced aversive effects in this 
animal did not elicit the foreleg attack movement. 

The data are too limited to know whether electrodes in these areas in 
any cat would give these responses. The impression is gained, however, 
that they would not and that somehow a particular type of movement has 
in a given individual come to be 'prepotent' over others. The stimulating 
electrodes may thus be revealing the existence of 'individually acquired 
reflexes' (Beritoff", 1924) established through the animal's own activities. 
At least they are not different from the individually acquired reflexes 
deliberately given the animal by the stimulating electrodes during the 
course of our experiments. However, the relation, if any, between these 
phenomena must, as so many questions raised by these experiments, 
await further experimental analysis. 


GROUP DISCUSSION 

Segundo. I shall first comment on the finding that motor cortical stimulation is 
relatively 'indifferent'. It agrees with other observations, for excitation ot tlus 
region in the sleeping monkey produced no arousal unless a generalized seizure 

L 


I4o BRAIN MECHANISMS AND LEARNING 

occurred and, under the latter conditions, one should doubt whether wakefulness 
derived from stimulation itself or from proprioceptive 'feed back' resulting from 
clonic movements. The same 'indifference' was shown by the visual area: when the 
animal was awake, excitation ^vith few volts produced investigation movements 
directed towards the contralateral field; when the animal was asleep, stimulation of 
up to 80 volts produced no effect. A different result occurred when excitation was 
applied to temporal pole, cingular cortex or hippocampus: with low voltages, 
animals were immediately aroused, both behaviourally and electroencephalo- 
graphically (Scgundo, Arana and French, 1955). Therefore, and as far as we can 
infer irom the influence of animal brain stimulation upon sleeping state or tendency 
to excite itself, certain areas are 'indifferent' and others are not. 

Dr Doty mentioned that stimulation ot subcortical structures may be condi- 
tioned, hi our laboratory, brief tones have been reinforced by direct excitation of 
mesencephalic reticular formation, centre median, basolateral amygdala or head of 
caudate nucleus and learned responses to tones have eventually appeared, hi some 
cases (centre median, mesencephalic reticular formation, certain amygdaloid or 
caudate placements) conditioned effects were practically identical to the absolute 
responses. In other cases, responses from different pomts in caudate or amygdala, 
could not be conditioned //; Mo; some of their components, however, could and 
coiiset]uently conditioned responses were similar to a part of the absolute response, 
hi a third group o( animals (also caudate or amygdala) the conditioned reaction, 
though consistent, was completely different to the asbolutc effect. This variable 
relationship has been found in other types of conditioning and, therefore, though 
difficult to interpret here, is not altogether surprising (Hilgard and Marqtiis, 1940). 

To summarize we can say that stimulation effects of certain areas of the brain can 
be conditioned. The latter term seems justified in the sense that the process has 
many features (technique, effects, inhibitions) of classical conditioning. These 
studies may help to understand the physiology of learning and that of tested nuclei 
(Roig, Segundo, Sommer-Smith and Galeano, 1959; Seguiido, Roig and Somnier- 
Smith, 1959). 

Chow, hi your monkey work, do you worrv about whether vour stimulus 
would activate the pain sensation in your animal f 

Doty. We are very much aware that the factor of meningeal stimulation must 
be considered. Dr Rutledge and I have shown that stimulation of the dura mater of 
the saggital sinus in the cat can serve as a conditional stimulus and is undoubtedly 
also a nocuous stimulation. However, I feel satisfied that, as outlined in the text, 
this factor can be detected by the self-stmiulation procedure employed and was 
present only as stated, hi addition, Dr Girugea has established these responses in 
dogs after destruction of the Gasseriaii ganglion. 

Hernandez-Peon. The conditioned response obtained by Dr Doty using cortical 
stimuli as conditioned and unconditioned stimuli provides a method for testing 
defniitely whether there might be some cortico-cortical connections in some cases 
of conditioning. And I wonder whether he has done transcortical cutting isolating 
the respective cortical areas and testing whether these conditioned responses persist 
after the transection or not. 

Doty. I think the experiments of Girugea and his colleagues showing these 
conditional reffexes to be established after callosal section when the conditional 
stimulus and unconditioned stimulus are in different hemispheres, indicate that a 


R. W. DOTY AND C. GIURGEA 149 

subcortical pathway is likely to be involved m the production ot this phenomenon. 
Perhaps pertinent also are experiments which Dr Rutledge and I have been doing 
using electrical stimulation of marginal gyrus in the cat as conditional stimulus and 
shock to the foreleg as unconditioned stimulus. If the stimulated cortical zone is 
circumsccted so that most of the pathways available for intracortical elaboration of 
the excitation are severed, conditioned reflexes still occur to the cortical conditional 
stimulus. If, on the other hand, the stimulated cortical zone is undercut for a total 
length of more than about 8 mm. the conditioned reflexes are lost. They often 
return, however, after about a month ot training. The critical factors arc not fully 
determined in the reappearance of conditional reflexes after this undercutting of 
the stimulated cortex, but it may be that 'U' fibres are necessary. It is also not 
certain whether that mere passage of time is suflkient or whether it is the retraining 
which is critical. 

KoNORSKi. Did you try to extinguish these reflexes that you have established 
and how did you obtain the extinction? 

Doty. Dr Giurgea taught me a lot about extinction. Apparently it is extremely 
dirticult to bring about a total extinction of a salivary conditioned reflex. 
Those 'temporary connections' are surprisingly permanent. Hence a technique 
of 'acute' extinction is used wherein the conditional stimulus is presented 
as one might expect without the unconditioned stimulus, but also at much shorter 
time intervals than employed during the establishment of the conditioned state. I 
objected that this alteration of procedure would not yield a proper comparison so 
we tried extinguishing by rather long intervals between conditioned stimulus 
presentations. In Dog Beta we got no extinction in eighty-five presentations. 
Hence in Monkey i sliown in the film, we used shorter intervals of 1-2 minutes 
and on the two occasions produced 'acute' extension in which the conditioned 
reflexes were totally absent to the conditional stimulus and in five or more con- 
secutive presentations of the conditioned stimulus, although the animal remained 
alert. Giurgea has also published extinction data on some of his dogs. Our Dog 
Gamma in which an aversive factor, was present in the unconditioned stimulus 
showed rapid extinction. 

Anokhin. Much of the recently gathered data points to the possibility of obtain- 
ing 'conditioned responses' by an association between the most distinct points of 
the brain. In fact, this trend of thought covers also some of the better-known 
conditioned reflexes, as the one induced by training with sound and light. 

Yet a question comes reasonably to our mind: what aspect of the activity of the 
brain, taken as a whole, do such facts reveal ? Dr Doty's interesting experiments may 
be taken as instances of the brain tissue's ability to act as a specialized substratum, 
and to establish instant links between any two stimuli which aftect it 
simultaneously. 

Our team considers that such aptitude proves the capacity of the nervous tissue 
to unite any separate elements during stimulation. This capacity is the basic 
physiological ground of any spontaneous conditioned response. But if we assume 
at the start that conditioned responses are a physiological function of the animal, we 
must consider them as the outcome of a complex physiological system, leading 
necessarily to an adaptation process which involves the organism as a whole. 

Dr Doty tells us that he finds it diflicult to draw objectively a difference between 
the leg-lifting response which he has induced, and the one obtained in the classical 


ISO BRAIN MECHANISMS AND LEARNING 

Pavlovian test by direct application of an electric shock to the animal's paw. Yet 
to us, the difference seems considerable, hi Pavlov's test, the subject draws away 
the paws from the source of stimulation, thus displaying an adaptative behaviour, 
which terminates in a reverse afferent drive, in our sense of the word, indicating 
that the animal has avoided the impact of the electric current. Besides, Dr Doty in 
his experiment shows the movement of limbs as the mechanical effect of a stimula- 
tion wliich, after having been subject to association, is now conveyed outwards 
onto the motor system. This proves admirabK' the capacity of the brain tissue to 
register any sequence of induced stimuli. 

I wish, in concluding, to draw your attention to one miportant factor, which is 
of particular relevance, to Dr Hernandez-Peon's remark on the possibility of 
showing pure cortico-cortical connections. Our laboratory has shown that the 
most insignificant stimuli, as well as lesions of the cortical tissue will instantly 
involve subcortical formations, and possibly, reticular ones. Tliis is why, by the 
very nature of the process, no subsequent effects of stimulation can be only cortical. 

Doty. In reply to Dr Anokhin — I think we must simply adopt the mechanistic, 
objective approach which Pavlov used so successfully. If I form a conditioned reflex 
by using an avoidable, painful, 'biologically significant' stimulus to a forepaw so 
that the animal lifts it when a tonal conditional stimulus is sounded, and for the 
other paw proceed as Giurgea and I have done so that the other paw is hfted when 
the 'auditory anahser' is stimulated directly at the cortex, you would be unable to 
tell me simply by looking at the animal's behaviour, which conditioned reflex was 
which. True, by careful analysis I suspect you would find great differences in the 
autonomic nervous system responses to the stimuli employed, at least during some 
stages of the training. But their absence in the latter case only proves them to be an 
unnecessary complication in the process of establishing the neural alteration 
responsible for the change in effect of the conditional stimulus. Both tone and 
direct electrical stimulation of the 'auditory analyser' are initially without effect on 
somatic musculature save possibly for an 'orientation reflex' which is soon lost. 
Both become effective on the same motor apparatus in approximately the same 
number of trials. Both states are subject to the laws of conditioning, i.e. there is no 
apparent backward conditioning, there can be external inhibition, they can be 
extinguished. Why then call them different states or infer for them different 
mechanisms f 

We have shown that if the ventral roots are crushed and an animal immobilized 
with bulbocapnine, it can still be conditioned to make conditioned reflexes with its 
hind leg even though the leg has never moved during conditioning. Thus we can 
show proprioceptive feedback from the unconditioned reflex to be unnecessary for 
the formation of conditioned reflexes. This does not say, that such factors are not 
normally present; only that they need not be. Such experimental simplification ot 
the situation certainly does not alter significantly the process we desire to elucidate, 
nor change its name. In similar vein the elimination of motivational factors must 
not be construed to infer qualitatively different processes to be operative in 
'cortical-cortical' conditioning as compared to the more usual procedures which 
involve such unknowns as 'biological purpose' or unanalysable emotional and 
subjective factors. 


R. W. DOTY AND C. GIURGEA I5I 

REFERENCES 

Doty, R. W., Rutledge, L. T. and Larsen, R. M. (1956) Conditioned responses established to 

electrical stimulation of cat cerebral cortex. J. Neuropliysiol. ig, 401-15. 
Giurgea, C, Raiciulescu, N. and Marcovici, G. (1957) Reflex condition — at interheniisferic 

prin excitarea directa corticala dupa sectionarea corpului calos. Studiu anatonio-histologic. 

Ret'istii Fiziol. Ncnii. Potol. 4, 408-14. 
Doty, R. W. and Rutledge, L. T. (1959) Conditioned reflexes elicited by stmiulation of 

partially isolated cerebral cortex. Fed. Proc. 1$, 37. 
Beck, E. C. and Doty, R. W. (1957) Conditioned flexion reflexes acquired during combined 

catalepsy and de-efferentation. J. comp. Physiol. Psychol. 50. 211-16. 


INTERFERENCE AND LEARNING IN PALAEOCORTICAL 

SYSTEMS^ 

J. Olds and M. E. Olds 

Recent stimulation and single-unit studies in our laboratory have con- 
vinced us that the hypothalamus with its projections to palaeocortical and 
related structures bears some very special relation to mechanisms of 
instrumental learning, and possibly to associative mechanisms in general. 

introduction 

In one sense all animal-learning experiments involve response learning, 
for it is a changed response that signifies to the experimenter that some 
learning has occurred. Nevertheless, it is possible to distinguish between a 
type of learning that is mainly response learning, and another type that is 
mainly a learning of associations. The former is evidenced when an animal 
learns how to get food in a problem box; he has learned a new response, 
he has learned what to do. The latter is evidenced when an animal learns 
to salivate when a bell announces the present arrival of food; he does not 
salivate to get food but because the bell is associated with food, and the 
animal salivates as though the bell meant food. 

In the study of learning, two basic procedures have been used with 
respect to these two types of learning. 

In instrumental conditioning, response learning is the salient feature of 
the procedure. The experimenter has only one stimulus at his disposal (a 
reward or punishment). He awaits the response which interests him, and, 
when it occurs, proceeds to stimulate or to withdraw stimulation. If he 
stimulates with what is called a 'reward', the response in question has its 
repetition frequency augmented, and if the animal is rewarded again and 
again, the response may quickly come to dominate the animal's repertory. 

In Pavlovian conditioning, association receives the emphasis. The 
experimenter imposes first an 'irrelevant' stimulus and then a 'relevant 
one. Eventually, the irrelevant response makes the animal 'think' of the 
relevant one, or at least perform some response that we take as oblique 
evidence of such a thought. Pavlovian conditioning is thus conceptualized, 

1 The research reported here was supported by grants from the Foundations Fund for 
Research in Psychiatry, the National Institute of Mental Health, and the Ford Foundation, 


154 BRAIN MECHANISMS AND LEARNING 

usually at a hidden aud unexpressed level, in terms of the association of 
ideas. A new association in the environment is imprinted on the animal, and 
we observe the conditioned response not mainly as a measure of the 
animal's adaptive behaviour but rather as a measure of the degree to 
which we have impressed the association on the animal. 

Whether there is one basic mechanism with two aspects or two or more 
basic mechanisms is still almost entirely a matter to be decided by further 
experiments, although there is considerable argument pro or con associa- 
tion or response learning. At any rate, our experiments have taught us 
something rather defniite about the physiological processes underlying 
response learning. It appears to us that our studies also have relevance 
to the broader problem of associative learning, but that has not yet been 
established. Any argument that there is only one basic mechanism with 
these two different aspects in the learning process will, we believe, make 
it even more likely that these studies involve one very important basis of 
learning in general. 

I. SELF-STIMULATION EXPERIMENTS ON LEARNING 

Self-Stimulation tests (Fig. i) described in detail elsewhere (Olds, iQS-S) 
indicate that electric stimulation in medial-forebrain-bundle regions of 
the hypothalamus, or in related structures of basal ganglia, paleocortex 
and tegmentum (Fig. 2), causes positive reinforcement of learning to the 
degree that animals will spend Ic^ng times turning on the stimulus to the 
brain in preference to all other pursuits. Animals will learn to run in a 
runway (Fig. 3) with such stimulation as the only reinforcement; they 
will cross an electrified grid (Fig. 4) to reach a pedal and stimulate their 
brains; and they will learn rapidly to perform without errors in a multiple 
choice maze (Fig. 5) (Olds, 1956) with the electric stimulus to the brain as 
the only incentive. Both acquisition and extinction curves on all these 
problems compare favourably with those where the positive reinforce- 
ment (or unconditioned stimulus) is a food object. It is clear, therefore, 
that stimulation in these regions promotes repetition of a learned response. 

II. INTERFERENCE-STIMULATION AND LEARNING 

The question we went on to ask was a question that has troubled 
physiological psychology for some time: is there any sense in saying 
learning mechanisms are somehow localized in these heldsf Or before we 
ask that, is learning localized at alh. 


I. OLDS AND M. E. OLDS 


155 


3V 
TRANSFORMER 


I 1000 Jl 

'^ POTENTIOMETER 


C R 


Self-stimulation experiment. Animal's lever response causes i-second cut-off device to deliver 
60-cycle current as long as lever is depressed up to ^-second maximum, atter v^'hich the animal 
must release and press again for more current. Current causes cumulative recorder to register 
response (thereby converting response rate into the slope of a graph) and causes via transformer 
and potentiometer a stimulus to the rat's brain which passes on the way through a constant 
resistor across which it is measured by cathode-ray oscilloscope. Rats respond with rates up to 
12,000 responses per hour with electrodes in medial-forebrain-bundle regions. 


HPTH 

Fig. 2 
Map of self-stimulation and some escape points in rat's brain. The former arc 
indicated by cross hatching, the latter by stippling. Self-stimulation region 
extends from baso-medial areas of tegmentum through baso-lateral hypotha- 
lamic areas into telencephalon and then up through the basal ganglia to the 
whole palaeocortical system. Escape regions plotted here include the dorsal 
posterior hypothalamus and similarly placed points in the gementum. The 
escape system is now known to extend forward too. 


156 


BRAIN MECHANISMS AND LEARNING 


o — o— o— o S group 
(8rats ) 

»--*--»--* F" group 
( 7 ra1s ) 


TRIALS 

Fig. 3 
Runway experiment in which animals ran faster for electric reward 
(solid line) than a control group running for food (dotted line) (Olds, 
1956). 


Fig. 4 
Obstruction box in which animals were required to cross grid 
with shock to feet to reach pedal to stimulate brain. After three 
self-stimulations on one side, pedal became inoperable and 
animal crossed the grid again to get three more selt-stimulations 
on the other side. Animal continued to shuttle as shock increased 
until foot shock became so high as to inhibit further self- 
stimulation behaviour. Foot shock of 60 laamp. stopped most rats 
running for food; however, one hungry rat took almost 300 
namp. for food reward. Self-stimulating rats took over 400 
ijamp. for hypothalamic stimulus as reward. 


J. OLDS AND M. E. OLDS 


157 


In some earlier attempts to answer this question following the example 
of Lashley (1929), surgicallesions were used, the neocortex was the prnnary 
locus of the search, and each animal learned only once so he could not 
serve as his own control. Very often, surgical lesions appeared to leave 
parallel structures to do the job; the neocortex seems ipsoJiUto a narrow 
constraint upon the search; and individual as well as surgical differences 


FOUR DAYS 
FIRST RUNS 


EXTINCTION 


S group 
( 8 rots ) 


Fig. 5 
Multiple-choice maze in which animals improved performance from trial to trial and from 
day to day, the only reward being an electric stimulus below the septal area (solid line). 
They are compared with a control group running for food (dotted line). Insert shows day-to-day 
improvement m terms of first run of the day indicating that no primer-stimulation is necessary 
to get rats running for electric reward. 


from animal to animal make it desirable tor animals to serve as their own 
controls. 

Thus we turned to an electric stimulation (ci. Thompson, 1958; 
Mahut, 1957; Stein and Hearst, 1958) to jam' the network, to cause, that 
is, a temporary disarrangement of pattern which is believed to function as 
a temporary (reversible) lesion. We believe this has three advantages: 
(i) it is reversible; (2) it can be expected to project to parallel structures 


158 BRAIN MECHANISMS AND LEARNING 

and thus cause a diffuse interference; and (3) utilizing penetrating im- 
planted electrodes, we are in a better position to get beyond the neocortex. 
Furthermore, we use a discrimination reversal experiment, to be 
described below, so that after preliminary training the animal can learn 
and re-learn the same problem day after day, eliminating errors each day 
in roughly the same number of trials. This methodology allows us to test 
animals one day on rate of learning under the electric 'lesion' and another 
day to test the same animals on control ; and we may repeat the process as 
often as we choose. 


METHODS 

To date, 150 rats with implanted electrodes have been tested in a 
problem box where they had to learn over again each day which pathway 
and lever would activate a feeding mechanism. The problem box is shown 
in Fig. 6. A red plastic lever on the left activated the food magazine one 
day, and a white metal lever on the right activated it the next. Each day the 
rat had to learn anew which lever worked. This took a small number of 
runs, after which the animal would run only to the correct level. After 
each pellet was discharged, the animal had to go to the magazine and eat 
(thus breaking the photo-beam) befoi'e the magazine would work a second 
time. Thus animals could not stay at the pedal discharging several pellets 
and then go to the magazine to eat them all, but were forced to run the 
maze both ways for each pellet. Ten consecutive responses on the correct 
lever was the criterion of learning. All animals were put through extensive 
preliminary training (running every day for more than a month) before 
being used in experiments. By the end of this training period, animals 
would always begin each day by distributing runs randomly to right and 
left; then they would stabilize on the lever that activated the food maga- 
zine. Depending on the animal, this would take from three to twenty-five 
runs each day. Animals were not used in experiments until progressive 
improvement in this daily score had ceased. During training and tests, 
animals were maintained on a feeding schedule with 23 hours of starvation 
and I hour of feeding each day. They were tested after 22 hours of 
deprivation. 

The data were plotted in terms of cumulative response curves (see Fig. 7). 
Solid lines were used to indicate correct responses, dotted lines, to indicate 
errors. The slope of the line indicated the speed of responding (on the 
correct or error pedal). When the dotted line flattened out, this indicated 


J. OLDS AND M. E. OLDS 


159 


that the animal had stopped making errors. Every test consisted of 2 days, 
one with the left lever correct (indicated by graph on left), the other with 
the riglit lever correct (indicated by graph on right), to rule out position 


WIDE RED 
PLASTIC LEVER 


LONG WHITE 
METAL LEVER 


RED PLASTIC 
FLOOR MARKER 


PHOTO 
CELL 


11 


x:i. 


WHITE METAL 
FLOOR MARKER 


^ LIGHT 

'source 


FOOD TRAY 

Fig. 6 
Problem box. Response on correct lever causes food magazine 
to operate. Animal must retrace steps and break photo-beam in 
order for correct lever to work a second time. Correct lever 
changes from side to side from day to day. After about a month 
of preliminary training, each animal daily achieves criterion 
(ten correct runs in a row) in about three to twenty-five trials. 


preference. In the six tests shown in Fig. 7, errors level off in all cases, 
while correct responses continue to cumulate. 

In the first test, there was no stimulation. Then stimulation was intro- 
duced in an effort to interfere with the learning mechanism. For this 
purpose a series of i-second trains of sine wave current (60 o.p.s.) was 
introduced at a constant rate of one for every three seconcis. A pair of 


i6o 


BRAIN MECHANISMS AND LEARNING 


tests were given at each of five electric current levels, lo, 20, 30, 40 and 
50 iJanip. r.m.s. It is believed that the so-namp. stimulus sets up a supra- 
threshold electric field in the brain of about i-inm. diani. (Olds, i9.vS). 
The stimulation was divided into ^-second trains separated by ih 
seconds of no stimulation because it was found that such a discontinuous 


CONTROL 


SERIES 


CINGULATE CORTEX 


Fk;. 7 
Daily response curves. Slope of black line indicates rate of correct responding; slope of dotted 
line indicates rate of incorrect responding. Cumulative response totals are plotted along 
ordinate, minutes along abscissa. When dotted curve flattens, errors have been eliminated and 
only correct responding continues. Left curve in each pair indicates a day when left lever is 
correct; right curve indicates a day when right lever is correct. Large numbers above each 
pair indicate electric current setting in microamperes of the interference stimulus. In this scries 
with cortical stimulation, the stimulus does not interfere materially with animal's ability to 
learn. 


train caused few seizures and caused greater stimulus effects in other 
respects. 

Some animals were also given a series of special tests which will be 
described in more detail later. These were (i) a selt-stimulation test tor 
positive reinforcing effects; (2) a memory-performance test for confusion 
after criterion is reached each day; (3) a reinforcement-confusion test to 
find whether confusion persists when electric stimulation is programmed 


J. OLDS AND M. E. OLDS l6l 

SO as to reinforce learning; and (4) in a very small number of cases, an 
escape test to discover negative reinforcing effects of stimulation. 

After learning experiments were complete, each animal was sacrificed, 
and his brain was sectioned and stained to determine the precise location 
of stimulation. 


RESULTS 

(r) Differences. — The results indicate a dctmite division of the brain into 

(1) a larger set of regions where electric stimulation has very little effect 
on this learning behaviour, (2) a smaller (but very extensive) region where 
electric stimulation has very devastating effects and (3) a region where 
effects are ambiguous. 

First, when electrodes are implanted in the neocortex and in many 
parts of the adjacent cingulatc cortex, stimulation from 10 to 50 namp. 
does not cause any major deficit in behaviour (Fig. 7). Similar effects are 
often seen with electrodes in the thalamus (left of Fig. 9). In all these cases, 
there is learning, with, of course, some day-to-day differences. Stimu- 
lation may cause some increment in errors, but it does not prohibit 
perforniance or learning. Animals run well, eat well, and learn well 
whether the stimulus is going or not. 

Seconc^, when electrodes stimulate in other places, quite different results 
are obtained (Fig. 8). It appears totally impossible for animals to reach 
criterion with the stimulus going in these areas. Except for confusion of 
learning, the animals perform well. They run almost as fast as usual; that 
is, the total number of runs is about equal for the stimulated and control 
series. The animals eat food well when the correct response activates the 
food magazine. But they cannot learn to eliminate errors on days when 
the stimulus is going. Because the animals continue to run and to eat, it is 
assumed that there is no interference with behaviour or drives. The inter- 
ference has to do only with the ability to eliminate errors. The stimulus in 
these cases is very small, interfering usually at levels of 20 laamp. or less. 
Finally, it should be noted that the stimulus does not merely cause or 
enhance a position preference, for errors on right or left levers are both 
caused to increase greatly by the electric stimulus. 

Third, there are sometimes effects which make it impossible to make a 
satisfactory test: (i) animals stop running when the stimulus is introduced: 

(2) the stimulus produces seizures so that no test can be made; (3) the 
animal stops eating so that it is no longer possible to assume motivation 
for learning ; and (4) the stimulus causes some definite confusion which is 


l62 


BRAIN MECHANISMS AND LEARNING 


SO mild or unstable as to be ambiguous or a forced position habit renders 
results ambiguous (sec caudate placement in Fig. 9). 

(2) MappiiK^. — Mapping these effects, we tnid the surprising result that 


40n 


PERIAMYGDALOID 


ANT. THAL. 


20-1 / 17^v 


ANT. THAL 


U 


i n u t e s 

Fig. 8 
Interference effects. Control curves (in background) show errors eliminated (dotted lines) 
correct responses cumulating (solid lines). On days when animals run under ctTccts of stimula- 
tion (double lines in foreground), errors are not eliminated. 


the mcdial-forebrain-bundle regions of the hypothalamus, its extensions 
into the tegmentum and basal ganglia, and the relatcti structures of the 
paleocortex are the places most apt to yield confusion upon electric 
stimulation. Stimulation in the neocortex, sensory thalamus, and large 


40 1 


J. OLDS AND M. E. OLDS I63 

CINGULATE CAUDATE 


Minutes 

Fig. 9 
Stimulation with very little effect (cingulate and posterior thalamic electrode, left) and mild 
effects (two caudate points, right). With stimulation in cingulate or posterior thalamus, 
responding is sometimes slowed, and errors sometimes increase. But learning is not prevented. 
In two cases with electrodes in the caudate nucleus, results are ambiguous. At the top, the 
stimulus causes the number of errors to double (thus stimulation interferes with learning) ; 
but errors are eventually eliminated (thus learning is not prevented by stimulation). In the 
other animal with caudate electrodes, stimulation sometimes caused such a great decrement in 
correct responding that it was impossible to tell whether learning occurred or not. In this 
case, it may also be noted, the decrement caused by electric stimulation seemed to involve a 
position habit forced by stimulation, so that the animal failed when the left lever was correct, 
but succeeded when the right was correct. 

parts of the reticular activating system cause no detrimental effects at all. 
And moderate or ambiguous effects are yielded in the borderline structures 
such as the caudate (Fig. lo). And this is practically the same map as was 
obtained for self-stimulation effects (Fig. 2). 

Is it true, then, that the points of positive reinforcement, far from facili- 
tating learning, actually inhibit it? 

M 


1 64 


BRAIN MECHANISMS AND LEARNING 


Fig. 10 
Mapping of interference effects. Plus marks indicate complete interference with 
learning, i.e. more than three times normal number of trials without meeting 
criterion. Open circles indicate no clfect of stimulation. Closed circles indicate 
ambiguous effects as noted in text. Other nictations are: S — seizure, P — forced 
position preference, A — activation. 


J. OLDS AND M. E. OLDS I65 

(3) Correlation with Sclf-Stiiinihnioii. — In an effort to test for such a 
correlation, forty-six animals with electrodes placed diversely throughout 
the brain were tested first for interference and later for the self-stimulation 
effect. The result was a most remarkable correlation. Of eighteen animals 
completely confused by the electric stimulus, fifteen were self-stimulators. 
Of seventeen animals unaffected, fourteen were non-self-stimulators 
(Fig. 11). Of eleven animals moderately impaired by the stimulus, seven 
were self-stimulators. All told, there were only three cases of definite 
impairment that were non-sclf-stimulators; and there were four cases of 
ambiguous impairment. 

A first answer is, therefore, that points yielding the most interference in 
a learning experiment are the points which also cause self-stimulation. A 
question arises, however, whether the points which cause interference 
but no positive reinforcement occurred by chance, or whether they 
follow some pattern. Mapping the points (Fig. 12), we see that there is a 
design. Of the twenty-nine points yielding complete or ambiguous 
impairment, we have satisfactory histological localization on twenty-six. 
The points which yield both confusion and self-stimulation arc scattered 
throughout the hypothalamic-palaeocortical system, but the points which 
yield only confusion are clustered along the hippocampus proper (that is, 
as distinct from the dentate gyrus), plus one in the anterior thalamus. 
We may guess, then, that there are two kinds of confusion points: (1) the 
self-stimulation points, and (2) a second kind located mainly in the 
hippocampus proper. 

(4) Memory Test. — A further difference between these two kinds of 
points appears when we test for confusing effects of the stimulus on on- 
going performance, after criterion has been reached on a given day. We 
have called this, perhaps wrongly, a memory test because we introduce 
the jamming stimulus after learning has occurred, and we look to see if 
the animal can still 'remember' the correct way. We might equally well 
call it a performance test, inasmuch as the stimulus might interfere with 
correct performance even if some hypothetical 'memory' were intact. 

At any rate, the test shows up a striking difference between two kinds 
of placement. If electrodes are in placements which produce avid self- 
stimulation, the stimulus causes a complete relapse to errors (and errors 
continue indefinitely) when stimulation is introduced after learning. But 
if electrodes are in the interference points where this is dissociated with 
self-stimulation, the result is different. The animal may make a few errors 
upon initial introciuction of the stimulus, but correct performance is 
readily resumed (Fig. 13). 


1 66 


BRAIN MECHANISMS AND LEARNING 


X 


500 


" 


X 
X 

X 

5 


400 


X 


X 
X 

X 
X 


iOO 


- X 


X 


X 


X 


X 


X 


200 


- 


5 


X 


100 


' 


X 
X 


X 
X 


— XXXXXXXXXXXXXX 


X X X X 


XXX 


MODERATE IMPAIRMENT 


LEARNING DECREMENT 


COMPLETE IMPAIRMENT 


Fig. II 
Correlation of interference with self-stimulation. Each point stands for one electrode pair. 
The position on the ordinate indicates the self-stimulation rate for an 8-minutc test period 
(current set at optimal level in lo- to 50-pamp. range). Three categories on abscissa indicate no 
interference, ambiguous interference, and total interference. 

Self-Stimulotion Test of 26 Points Which Yield Confusion 

s = 5e/f stimulation 


Basal Ganglia 


*- No self stimulation 
Hpc 


Points where self-stiniulatic 


electrodes (S) and non-self-stimulation electrodes (#) 
yielded confusion. 


J. OLDS AND M. E. OLDS 


167 


Of seventeen self-stimulators given this test, fifteen showed total failure 
to regain criterion; the other tw^o were moderately impaired. Of six non- 
self-stimulators given the test, three were moderately impaired, and three 
were unaffected (Fig. 14). All animals were significantly impaired on the 
initial learning test. This suggests that the points where impairment is 
dissociated from self-stimulation are points involved in learning but not in 
later memory or performance. 


M EMORY 
PER I AMYGDALOID 


1/ .^ 


4 8 12 16 20 34 2B 32 3t 40 

H I PPOCAMPUS 


^--rr 

■! 20 30 


ANT. T HAL. 


/35//0 


.'-r 
-)—^ 


Fig. 13 
Memory test. Cumulative-error curves arc shown for days when stimulation started at 
onset of test (double lines) and days when stimulation did not start until after criterion 
was reached (single lines). Vertical dotted line indicates onset of stimulation for latter case. 
When stimulation starts at onset of test, errors are not eliminated in any of these cases. 
When stimulation is introduced after criterion, it causes complete relapse with electrodes 
in periamygdaloid cortex, but not complete relapse with hippocampal or anterior thala- 
mic electrodes. 


Now the question arises whether, in places where self-stimulation and 
confusion are associated, the confusion results from the extreme reward 
(which might render the food reward unimportant), or from some 
additional pattern disruption over and above this reinforcing effect. 

(5) Reuiforcement-Jutcrfcreuce Test. — To attempt an answer to this 
question, a test was devised in which the 'interfering' stimulus was so 
correlated with the correct response as to constitute to some degree an 


i68 


BRAIN MECHANISMS AND LEARNING 


added incentive. This was accomplished by having the wrons; lever 
terminate the series of stimulations, and the correct lever re-start the 
series. Thus the animal making all correct responses would be stimulated 
as before, i second on, 2^ seconds oft. But during learning, each day, the 
sequence would be interrupted every time an error occurred, and restarted 
only by the next correct response (Fig. 15). 


EFFECT OF STIMULUS AFTER LEARNING 


SELF STIMULATION POINTS 


NO SS 
PTS 


MAXIMUM IMPAIRMENT 


MODERATE IMPAIRMENT 


MINIMUM IMPAIRMENT 


Fig. 14 
Memory test for twenty-three points, all of which yielded impair- 
ment on learning test. Seventeen cases were self-stimulators. Si-x were 
non-self-stimulators. The figure shows that fifteen of the self-stimu- 
lation points yielded maximum impairment when introduced after 
learning; the other two self-stimulation points yielded moderate 
impairment. Of the non-self-stimulation points, three yielded moder- 
ate impairment, and 3 yielded almost no impairment. 


When this test was applied to the confused selt-stimulators, those with 
very rapid rates of self-stimulation were no longer confused. Several of 
the slower self-stimulators appeared, however, to be still confused 
(Table I). The possibility remained that those still appearing confused were 
in fact making errors to terminate negative reinforcement; i.e. they were 
receiving both positive and negative reinforcement from the same 
stimulus (Roberts, 195 8). To test for this in a very preliminary fashion 
three of them were given tests for escape responding yielded by the 
electric stimulus, and these three did give evidence of escape. Thus we 
cannot find from these tests any support for the notion that the self-stimula- 
tion points cause a bona-fide confusion. Rather, it appears that the large 
reinforcing effect of the electric stimulation somehow overrides the 
smaller positive reinforcing effect of the food. When food and electric 


J. OLDS AND M. E. OLDS 


169 


reinforcement pull the sanie way, the animal can achieve criterion, and 
can run perfectly. 

There is still room for doubt, however, because (i) the animals appear to 
be running and searching for food during stimulation trials in the learning 
and memory test, and they cat the food rapidly when they get it; (2) it is 


Tadle I 

REINFORCEMENT-INTERFERENCE TEST COMPARED 
WITH SELF-STIMULATION SCORE AND ESCAPE 
TEST. POINTS IN THE O COLUMN SHOWED NO 
IMPAIRMENT ON THE REINFORCEMENT-INTER- 
FERENCE TEST. POINTS IN THE PLUS COLUMN 
WERE COMPLETELY IMPAIRED. SELF-STIMULATION 
SCORES ARE RANK-ORDERED FROM TOP TO 
BOTTOM OF THE TABLE; THESE ARE RATES FOR 
AN 8-MINUTE TEST PERIOD. THE EVIDENCE 
SHOWS THAT POINTS YIELDING HIGH SELF- 
STIMULATION RATES CAUSED NO INTERFERENCE 
IN THIS TEST. THREE POINTS WHERE INTERFERENCE 
OCCURRED WERE TESTED FOR ESCAPE, AND THESE 
TESTS INDICATED THAT NEGATIVE REINFORCE- 
MENT ACCOMPANIED POSITIVE REINFORCEMENT 
AT THESE POINTS 


+ 


S. St. 


S. St. 


Esc. 


550 


soo 


468 


450 


370 


350 


+ 


3^.S 


300 


280 


270 


150 


100 


+ 


122 


+ 


not clear why animals which have reached criterion in the memory test 
proceed to make errors upon introduction of stimulation, when the same 
animals, having reached criterion under the reinforcement-interference 
conditions, can run and eat perfectly under the same electric stnnulation; 
and (3) in the reinforcement-interference test the added incentive of the 
goal might cause the animal to override the confusing effects of the 


1 70 


BRAIN MECHANISMS AND LEARNING 


stinuilus, rather than proving there were no confusing effects in the first 
place. We have often seen animals become 'brighter' as the 'reward' 
potentiometer is turned upwards in maze experiments. 


INCORRECT 


CORRECT 


STIM. EVERY 3 SEC. 

NO STIM. 

Fig.. 15 
Reinforcement-interference test. 
Interference stimulus starts when 
animal is placed in box and con- 
tinues unless error pedal is pressed. 
At that point it is terminated and 
does not start again until correct 
response occurs. 


DISCUSSION 


At the end of the first set of experiments we were well aware that the 
hypothalamic-rhinencephalic system was most likely the central core of 
the positive-reinforcement mechanism. And wc also thought in terms of 
positive reinforcement as involving (i) some close relationship to learning 
(a concept we still hold) and (2) some peculiar virtue in promoting learn- 
ing, sharpening the wits, oiling the course, or stamping in the right 
response. 

The second set of experiments surprised us at first, indicating as it did 
that the most efficacious way to interfere with the learning mechanism 
was to hyperstimulate the same system which we had formerly thought 
to promote learning. We should not have been surprised, of course, for 


J. OLDS AND M. E. OLDS I71 

this finding emphasized what we should have known at first: response 
learning involves, hrst, cessation of wrong behaviours, secondly, com- 
mencement of correct ones, and thirdly, repetition of correct ones. A 
stimulus causing repetition of behaviour would only promote the thu-d 
stage of the process, which in a sense is not learning at all but its antithesis. 

But again there was still room for doubt. Our stimulation of the large 
hypothalamic system causes repetition and inhibits adaptive changes in 
behaviour. In a sense, this is one and the same thing; by repeating, the 
animal cannot be adaptive, changing. This is consistent; but we must 
sometimes beware of consistency. 

Our animals could repeatedly get the brain stimulus (which is all the 
repetition they want) and yet adcipt the correct response to get food at the 
same time; and they appear to be trying to get food. Why does the 
response fail to become stereotyped down one lane? One might say there 
is a reinforcement ot the antecedent so per cent probability. But then why 
in the memory tests where the learning occurs first does the stimulus cause 
return to 50 per cent probability: Or even more strikingly, why, after the 
animal learns in reinforcement-interference tests, can he run successfully 
under stimulation, to food and back, with no errors even though the 
stimulation continues? The same animal running currently in the memory 
test starts and continues to make errors upon introduction ol the stimulus. 

The possibility we want to suggest is this: our stimulus might cause 
some confusion in addition to the output of positive motivation. In the 
reinforcement-interference test, the strong positive motivation emanating 
from hypothalamic stimulation keeps behaviour in line. But minor prob- 
lems cannot be solved because of confusing effect. It might be, then, that 
total activity in this hypothalamic-rhinencephalic system determines the 
positive-reinforcement function; but the pattern of activity in the same 
system is precisely the more or less lasting process that is altered by learning. 

Giving up, then, the idea that positive reinforcement promotes learning, 
we are left with alternative notions: either it is a simple inhibitor of 
learning or else it is somehow a function of the same neural aggregates 
which are chiefly involved in learning; but quantity is most important 
from a standpoint of reintorccmcnt, the patterning most important from 
the standpoint of learning. 

From the latter point of view, stimulation of the system could be said to 
augment the positive reinforcement and the repetitiousness of the whole 
system, but to disrupt the patterns formed by association of mild cues. 

The final experiment we have to report searched for ways of causing 
learned changes in neural patterns by operant-conditioning techniques. 


172 BRAIN MECHANISMS AND LEARNING 

III. THE OPERANT CONDITIONING OF SINGLE-UNIT RESPONSES 

A most important contribution of behavioural psychology to the 
brain and behaviour field is the conception that a response is not merely 
something to be elicited, observed, characterized, and recorded. In 
operant-behaviour analysis, a response is more than that: it is something 
to be conditioned. The single-unit response is, on the face of it, precisely 
the type of categorical and clearly defined event that should be amenable 
to operant-conditioning techniques. We report here initial successes in 
this endeavour. 


METHODS 

In these studies, rats were prepared first with self-stimulation electrodes 
in mcdial-forebrain-bundle regions. Preliminary tests established that 
very high self-stimulation rates were achieved, and no tendency to escape 
from stimulation was present. Rats which failed to meet these requirements 
were eliminated. Each rat was then placed in a stereotaxic instrument 
under barbiturate anaesthesia. A hole of 3-mm. diameter was drilled in the 
skull, usually i mm. behind, and i mm. lateral to the bregma. The dura 
was pierced repeatedly with a sharp instrument. Then tungsten micro- 
electrodes of i-ii diameter (Hubel, 1957) were lowered imni. into the 
cortex. 

As the animal recovered from the barbiturate anaesthesia, still in the 
stereotaxic instrument, he was given repeated doses of isopropyl meproba- 
mate (Soma, Wallace Laboratories). Each dose was 80 mg./kg. The dose 
was repeated whenever any tendency of the animal to try to escape from 
the instrument appeared. Previous tests had shown that an almost paralys- 
ing dose of isopropyl meprobamate (100 mg. kg.) fails to block self- 
stimulation (Olds and Travis). From this point on, the electrode was 
advanced downwards through the cortex, hippocampal formation, thala- 
mus and so forth, stopping whenever a clear spontaneous response 
appeared. Output from the microelectrode was led through a cathode 
follower into a Grass a-c amplifier and thence into a Dumont cathode-ray 
oscilloscope. Responses of single nerve cells appeared as 200- to 500-|jv. 
negative spikes, lasting about i msec. They were identified mainly by 
their duration, by being repeated responses of constant amplitude, and by 
their disappearance upon movement of the microelectrode by about 200 
microns. 

Unit responses were not considered satisfactory for these experiments if 
their resting frequency was more than about 2 per second ; and they were 


J. OLDS AND M. E. OLDS 


173 


preferred if this were something less than i per second. When such a 
response was observed, a three-step experiment was performed : first, after 
several minutes of waiting, a 30-second record was made on moving 
photographic paper of the resting rate of the unit response. Second, an 
elicitation test was made. A scries of twenty -j-second trains of stimulation 
(sine wave 60 c.p.s., 50 namp.) was introduccci via the medial-forebrain- 
bundle electrodes at a repetition rate of i every 2 seconds. These were not 
correlated with single-unit response. Instead, there was an explicit effort 


Fig. 16 

Schematic diagram of single-unit operant-conditioning 
apparatus. Single-unit response is amplified and fed into 
amplitude discriminator which disregards all lesser signals, 
but yields an output whenever the single-unit response 
occurs; the output then goes to a counter. The counter is 
preset to activate a stimulator and reset whenever it reaches 
some number from i to 9. At this point, a |^-second train of 
60-cycle stimulation (50 namp. r.m.s.) is delivered to the 
medial-forebrain-bundlc 'reward' point. During the stimu- 
lation, the counter is disengaged so that it will not count 
stimulus artifact. Immediately afterwards, it commences 
to count again until the preset ratio is again reached at 
which point it delivers another 'reward' stimulus. 


made to stimulate only in the absence of single-unit responses. Immediately 
after this test a second 30-second record of activity was made on film. In 
the event of elicited effects, each stimulation produced a series of responses 
from the unit, and no further tests were made. The microelectrode was 
then advanced until a new unit response appeared. In the event that 
elicited effects were not found, the experiment continued. Third, a 
reinforcement test was made. The experimenter awaited a single-unit 
response and, each time it appeared, immediately delivered a stimulus to 
the hypothalamus. When this was done by hand, reinforcement could be 


174 BRAIN MECHANISMS AND LEARNING 

applied after one appearance or after several appearances of the unit 
response. It was usually applied after each appearance of the response with 
a delay of about i second (this was the experimenter's response tinae). 

The design of the experiment, when the stimulus was presented auto- 
matically, is shown in Fig. 16. The output from the oscilloscope amplifiers 
was led into an amplitude discriminator circuit, and the output from the 
circuit activated an electronic counter which could be preset to deliver 
stmiulation after a preset number of responses from i to 9. In this case, the 
delay of reinforcement was about 3 msec. Each activation of the stimulator 
was recorded on a cumulative response recorder, so that the response rate 
was retained on a permanent record. 

In the case of a positive experiment, the single-unit response rate was 
greatly augmented by either the hand or the mechanical reinforcement 
procedure. The increased rate outlasted the procedure by a variable period 
of time. Immediately after this procedure, a third record of the unit's 
activity was made on film. It is the comparison of the three tilms that 
comprises our data. 

In the event of failure to reinforce, by manual techniques, the unit was 
often put on the automatic reinforcement procedure, and sometimes 
positive reinforcement bypassed by the former procedure was discovered 
after long runs with the latter method. Because of the close temporal 
relationship between the control picture and the augmented picture, when 
the manual technique was successful, it was always the more convincing. 

RESULTS 

The first point to emphasize is the difference between the neocortex and 
the palaeocortical structures studied. Palaeocortical units often appeared to 
'learn with ease'. Neocortical units never yielded any similar rapid modi- 
fication when subjected to reinforcement techniques. The details of this 
difference will be clarified below. Most often, stimulation caused neo- 
cortical unit respc^nses to disappear, never to return; but sometimes, rein- 
forcement caused slow augmentation of response. 

Where stimulation had any effect other than inhibition, it produced one 
of four kinds of changes in the pattern of single-unit response: (i) conver- 
sion of a sporadic grouped response to a continuous response by the 
reinforcement procedure; (2) augmentation of the response rate of a 
sporadic response by reinforcement; (3) ehcitation of activity imme- 
diately after the stimulus, but only when it was given as a reinforcement; 
and (4) ehcitation of activity by the stimulus irrespective of its use as a 
reinforcement. 


J. OLDS AND M. E. OLDS I75 

The most striking cases were of the hrst type (Figs. 17 and 20 I, II, IV). 
They were recorded mainly from areas such as the dentate gyrus, tmibria 
regions, and mammillo-thalamic tract regions, which appeared to be 
'seizure-prone' in our previous experiment (Fig. 10). In these cases, the 
unit was originally responding in a sporadic pattern with single responses 
or groups appearing at less than one per second. Stimulation when 

-M'3 1/3 )0°P Before Stimulation #1118 Dentate 

» |i 1 ,4 i i ji iiiiiii X" i< » r» i ii^i 4» ' ' ' *"'4 ''' ' 4* ***'**'''*'**^^ 

^ tf tt ^ ni i < i w i 4i III ! »■ ' iiWMw«4«*«wlN4>'i«^«"y ^ 

After 5' Stimulation No RemforcemenI 

-A- A ^ ' - - 


ii» i A y i 4 I t ■» n il 11 * 1 ■« |>| » I'i i iii f il i i |< '< > M Jii j i N 

'<| iii i^i 4 ii 4 mu iumm^immmmmmmmmmm m itt ifi ii » iiii|iiili > i>i w > rt» i M«ii iii 1 ■ i nwi^i u 

After 5 Reinforcoment 

mimitm4'»tm^mmmtm$»tmmm*mi»^^ ^ i :i i i i i i4 M ii«iiiii » ii i i» #i 

iii i iii i | iw i i ^ i i ' i j' i wii^ii I' ^m ' j iii i i iii i ii u I I Ai li \\ i \ n $\\ i «i i i 4 ii i 

Fig. 17 
E.xperimciit on single-unit response recorded from dentate gyrus. Small unit response 
(almost lost among larger slow waves) occurs at rate of approximately one per second 
tefore stimulation, and rate is not greatly increased by uncorrelated stimulation. 
When stimulation is correlated as reinforcement for unit response, however, continuous 
firing (at rate of about 30 per second) ensues. Time base as in Fig. 18. 

nitrociuced during silent peru)ds did not cause any elicited tiring. Such 
stimulation could be continued for periods of 5 minutes or more without 
materially augmenting the response rate. Then, if the stimulation was 
withheld and delivered only after the appearance of a single or grouped 
response, ten to twenty reinforcements would often suffice to cause a 
sudden burst of activity; the unit would respond continuously at rates as 
high as 30 per second. This burst would sometimes last for a period of only 


176 


BRAIN MECHANISMS AND LEARNING 


K)62 Anterior Rhirencepholo 


n Atl«r S No ftu -% %- 


% Ar 


•"V 


'\ %- 


Remloromeni Proceedu 


I II' III 


1062 Antsriot Rhinen<epholo 


\ 


(|M»itifl^l«>V««*^«*MlMfmM«a»WNMMHMPM«»«r^«Mipg>BMM^^ 


"^■1 ■ III nil !■•% 


■ »■ » II in iiii»>i 


MM^NMiMMMMlayPrilMiirfMiMMMMIMi 


IV 


J. OLDS AND M. E. OLDS 


177 


Slcwdcv.-. All 


1062 Anlenor Rhinencepka 


\X 


V 

Fig. 18 
Series of experiments on single-unit response recorded from anterior rhinencephalon. I. Before 
stimulation. II. After stimulation without reinforcement. III. After reinforcement. IV. After 
waiting for unit to slow down (about 5 minutes), reinforcement procedure is undertaken a 
second time, and a photographic record is made during reinforcement; large Litticc-type 
artifact indicates 60-cycle sine wave stimulus. V. Still later, after another wait, attempt is made 
to correlate stimulation with pauses, but this is unsuccessful. Later an electronic device was 
made to correlate stimulation with long silent periods, and the procedure caused some unit 
responses to cease altogether. 


several minutes, with response amplitude decreasing in an orderly fashion. 
Then the unit response would disappear for a period of some minutes, to 
return at the original amplitude. At other times, the repetitive activity 
would continue at a high level for longer periods. One might suspect that 
a hippocampal seizure had been started by the reinforcement procedure, 
except for the fact that movement of the electrode in these cases reveals no 
similar activity in neighbouring cells. 

The second type of changes (Figs. i8 and 20 III) were not greatly 
different from the first, but in these cases, the rapid responding caused by 
reinforcement seemed much more like the original sporadic responding, 
although its frequency was far greater. There was no tendency for the 
rapid responding to be accompanied by decrement in amplitude. It 


IjS BRAIN MECHANISMS AND LEARNING 

appeared to follow much more ordinary acquisition and extinction 
patterns, augmenting at each reinforcement, slowing upon cessation of 
the reinforcement procedure, often coming to rest at a higher firing rate 
than the original rate, but lower than during reinforcement. 

Changes of this type were graded along a dimension of rapid or slow 
conditioning. With units in the anterior rhincnccphalic cortex, condi- 


I Betoo S 


!092 ■ Hipptxompyi Proper 

miiimmimmm'imim'itmt m i nm » ■ n i ii i ii " 


in Mm Rl< 


fm0' 


V [X...r.g Si No Rft 


Fig. 19 
Expennieut on single-unit response recorded from hippocampus. It is noteworthy that stimula- 
tion which closely follows a unit discharge (see IV) causes a burst of activity which is not seen 
when stimulus occurs against a background of no-firing (see V). 

tioning was quite as fast as with type I units in the dentate gyrus. But the 
response was more moderate. Some 'learning' of a much slower order 
occurred with points estimated to be at the base of the neocortex. There 
was never any possibility of augmenting the rate by manual reinforcement 
with electrodes, here, but, left on automatic reinforcement for long 
periods of time, the unit response rate would sometimes increase 
gradually. This seemed quite like the conditioning of a skilled act in its 


J. OLDS AND M. E. OLDS 179 

time course, an idea possibly forced on us by the fact that the hnal condi- 
tioned unit activity was often accompanied by some minute movement of 
an extremity. One animal learned to move his tail to the right while 
remaining, otherwise, absolutely still. 

In contrast to the conditioning ot cortical units which sometimes seemed 
like conditioning the muscle output itself, the conditioning of subcortical 
units often led us to teel that we were observing the basic mechanism of 
instrumental conditioning itself. This was the case first in the conversion 
of dentate responses to seizure-like tempos as in type i. But also the type 
3 modification, observed in the hippocampus and diencephalon, appeared 
to involve some basic mechanism of learning. 

With microclectrodes here, the stimulus had quite different effects when 
given immediately after a unit response from those it might have against 
a background of silence (Figs. 19, and 20 IV). When it followed the 
unit, the stimulus would invoke either a further burst of response in the 
same unit, or a burst of very high potential activity not easily characterized. 
If the same stimulus was applied at some temporal distance from a single- 
unit response, there were no similar effects. 

It may be relevant that stimulation did sometimes elicit in hippocampal 
electrodes a high-voltage, repetitive, positive discharge of somewhat 
longer duration than the single-unit response. 

Finally, there is of course the type 4 modification: the stimulus causes 
an elicited effect so that reinforcement cannot be tested. It is hard to 
estimate how frequent this occurrence is because these effects have been 
quickly by-passed in our experiments in search of areas where elicited 
effects are absent. It is certainly clear that these cases increase in number 
as the microelectrode approaches the point of stimulation in the posterior 
hypothalamus (Fig. 20 V). 

SPECULATIONS 

It is certainly tempting to speculate at this point, so long as it is under- 
stood that no scientific factual import should be attached to these specula- 
tions. 

Ashby (1953) has suggested that perhaps each neurone is in itself a 
negative feedback system, at least a system that modifies its output 
depending on the feedback it gets from previous outputs. We have perhaps 
shown something of this kind here. 

There is of course nothing in our work to determine that the unit is the 
conditioned entity; perhaps it is simply our method for identifying a 

N 


i8o 


BRAIN MECHANISMS AND LEARNING 

gi2 — lil-'l^ ^iiJV III A iiiiri I I r I I II 


1118 

■^ — |t-V-Hl '"""""\ 


NEGATIVE 


_ll U I IL. 


J. OLDS AND M. E. OLDS 


I8l 


IV 


A;!!!!l "!" ' ' ! %ll l ll !! '! l l| l H ! IM I l i 


Fig. 20 
Localizations of microclectrode points yielding various effects. I. Point in or near maninnllo 
thalamic tract yielding reinforcement to contmiioiis firing (schematically symbolized by unit- 
sinewave pairings followed by repetitive vmit discharge). II. Point in fimbria yielding rein- 
forcement to continuous firing, plus three cortical points yielding negative results. III. Point 
in anterior rhinencephalon which augmented firing rate greatly during reintorccment 
procedure. IV. Point in hippocampus where reinforcement caused firing after the stimulus; 
and point in mammillo-thalamic tract region of thalamus where reinforcement yielded 
continuous firing. V. Point in dorsal posterior hypothalamus where stimulation elicited unit 
firings. 


larger integration which we proceed to condition. However, wc do have 
anecdotes which suggest that a unit discharge started by pressure (as our 
microclectrode enters an area) can be maintained active by reinforcement. 
If stimulation with the microclectrode can be performed systematically, it 
will at least be shown whether the unit response needs to be started by 
some afferent system to be conditioned. 


1 82 BRAIN MECHANISMS AND LEARNING 

Supposing for a moment that the unit itself is the conditionablc entity in 
these experiments, we would then be provided with some sort of a trace 
system. We could suggest that instrumental conditioning techniques 
leave a changed pattern of repetitive discharge in these units, a suggestion 
which someone more given to theorizing than we might inflate into a 
model for a brain. 

SUMMARY 

Three sets of experiments have been reviewed here, suggesting first 
obliquely and then more directly that the hypothalamic-rhinencephalic 
system is more intimately related to instrumental conditioning than the 
other parts of the brain. 

First, self-stimulation experiments which show that stimulation in this 
area can serve as a strong positive-reinforcement or unconditioned stimulus 
in Skinner box, runway, obstruction box, and maze experiments were 
cited. These suggested that somehow a good deal of activity emanating 
from this system might foster at least the repetition of preceding responses, 
if not the learning of responses themselves. 

Second, interference-stimulation experiments were conducted to find 
where electric stimulation might interfere with performance in a T-maze 
type of test. It was found that precisely the same system was implicated. 
Stimulation in positive-reinforcement points appeared to cause confusion 
because of the prepotency of the electric brain-stimulus 'reward' over the 
food reward to the hungry rats. But other points in the hippocampus 
proper caused confusion without any accompanying positive-reinforcing 
effects. These points were also distinct from positive-reinforcement 
points in that interference occurred only during learning but not during 
later performance. 

Third, instrumental-conditioning experiments were carried out on 
single-unit responses in cortical, palaeocortical and subcortical regions. 
Units in subcortical and palaeocortical structures were often readily 
conditioned. Only rarely and with great difficulty were cortical units 
conditioned. Conditioning in lower centres often appeared to suggest 
basic mechanisms, as stimulation after a unit discharge would have 
different effects from those it would have against a background of silence, 
or, several stimulations after a unit discharge, would cause seizure-like 
repetitive discharge. 

The data from all three experimental programmes might be summarized 
by the hypothesis that the quantitative measure of activity in the hypotha- 
lamic-palaeocortical system is a measure of the positive reinforcement, the 


J. OLDS AND M. E. OLDS 1 83 

tendency to repetitive behaviour. The pattern of repetitive firing in the 
units of this same system is the residual process which is modulated by 
instrumental-conditioning techniques. 

As to whether the same changing pattern of residual processes might 
serve as a model for associative conditioning, we will suspend judgment. 

GROUP DISCUSSION 

Gerard. In what manner are your experiments different in principle — ob- 
viously they are more elegant in degree — than ones using any other kind of 
effector response? Any response obviously must involve the discharge of some unit 
somewhere in the brain. 

Olds. Two answers. In the case of the cortical unit and even in the case of the 
anterior rhincncephalic unit that I showed, my impression is that there is no 
difference m principle at all. My feeling is, in both these cases, that the animal 
decided that doing something was getting him reinforced and so he did it. In the 
case, however, of the fimbria and dentate gyrus units and in the case of some of the 
liippocampals, I had a feeling that here is the mechanism, so to speak, so that you 
would say 'this is the stuff decisions are made out oC. That here we were playing 
with sornething that just automatically had its response rate augmented if we 
stimulated the hypothalamus right after it has discharged. 

Gerard. Did you say that stimulation of the fornix in the rat, under your condi- 
tions, interferes with learning f 

Olds. No. The answer is that in the hippocampus proper wc interfered. But in 
the fimbria and often in the fornix our stimulation usualK* produced seizures before 
we could get any experimental tests. 

Gerard" I raised the point only because of the recent report of an unfortunate 
operation on man, in which the fornix was cut bilaterally. Recent memory was 
completely abolished. 

Olds. I have a hunch that response Icarnnig and recent memory are somehow 
very closely related. And that if we said response learning and recent memory 
involve palaeocortical and hypothalamic systems we might be on to something, and 
the more structured memory might be, it seems to me, in the neocortex and the 
classical thalamic system. 

Grastyan. Bv stimulation of the hippocampus we got similar observations in 
somewhat different experiments. On the background of already established condi- 
tioned reflexes we obtained always an inhibition. The interpretation of these 
observations was, however, always very problematic. It is well knowii that the 
hippocampus is one of the structures that has the lowest threshold for eliciting after- 
discharges. I see that you disregard the experiments in which you elicited after- 
discharges. I am still not absolutely convmced, however, that by stimulating the 
liippocampus you did not get any after-discharge in the hippocampus itself 
Sometimes when recording in various parts of the hippocampus we got after 
discharges when we did not see any sign of after-discharge in other structures. I am 
not convmced that these effects represented a physiological inhibition. I have to 
admit at the same time that we had some observations m wliich we did not elicit 
any after-discharge in the hippocampus and we still obtained inlnbition. I would be 


l84 BRAIN MECHANISMS AND LEARNING 

very interested to know what might be the physiological meanijig of this inhibition. 
According to our supposition, the hippocampus inhibits the orientation reflex, 
which is a necessary step in the development of the conditioned reflex. 

Olds. Wc found that if we were on the surtace of the hippocampus itself, so that 
we were stimulating the hippocampal cells proper, we very rarely got grand mal 
seizure perceptible in the animal's behaviour. When we were in dentate we almost 
invariably did get grand mal at such low thresholds that it was impossible to do 
any research at all. Now as to whether our results in the hippocampus proper might 
be from hippocampal seizure which could be recorded although not observed I 
certainly would not be at all surprised if that were true. I have often conceived of 
these experiments in the light of trving to tind the place where localized seizures 
would have the same effect as a generalized electro-convulsive shock. We are trying 
to look for a place where we can, so to speak, jam the network by excessive activity. 

Adey. I was very interested in Dr Olds's observations of the role of the anterior 
hippocampus in this particular type of activity. I wonder whether he has also 
examined the ventral parts of the hippocampus arch and whether he has come to 
anv conclusions about the relative significance of the more ventral zones ot the 
hippocampus as opposed to the dorsal. My reason for asking this is primarily that 
we have studied quite extensively the course of the normal wave discharges in the 
hippocampus in the course of various learning procedures. I would mention that 
there does not appear to be a homogeneity of the whole aspect of the hippocampus 
in this regard, and that, briefly, the dorsal hippocampus and the entorhinal cortex 
are regions which appear to be particularly concerned in the type of behaviour in 
which we are interested, to the point where one might summarize by saying that 
the region between the hippocampal pyramidal cells and the dentate gyrus, where 
Dr Olds mentions this extreme sensibility to seizure, this region and the adjacent 
entorhinal cortex appear to be concerned in what we might term the execution of 
planned behaviour, basing that opinion on certain very regular slow-wave dis- 
charges. If I might raise one further point, that is the question of these curious 
polyphasic discharges which Dr Olds saw when he was recording from the vicinity 
of the dentate gyrus. I wonder whether he may not be seeing in fact a muscle 
discharge if his microelectrode recording set-up is the typical one with the mono- 
polar method of recording. The absence of the normal muscular paralysing agents 
will very often produce volume conducted muscular effects that may be extremely 
confusing in the interpretation of the micro-electrode recording. 

Olds. In all cases that we have reported, in which are found this special pheno- 
mena of peculiar effects immediately after the stimulus we have very careful 
control showing that we do not get these effects except when we apply the stimulus 
immediately after the unit response. In all other respects, I assure you this looks like 
artefact. And it is only because it has this physiological significance, namely that the 
proximal unit has to fire before we apply our stimulus in order for us to see this 
response, it is only for that reason that I even mention it. 

Let me go back to following the hippocampus around to its posterior arch. To 
me it looks like the hippocampus all the way around has the dentate gyrus facing it. 
I did show three sagittal sections in my large map and if you had looked carefully 
you would find that in all cases if the electrodes w^re placed just above the arch of 
the hippocampus proper, we got the rather total inhibitory effect without self- 
stimulation. Finally, as to your notion that the dentate might be involved particu- 


J. OLDS AND M. E. OLDS 185 

larly in planned behaviour, I had very much a similar notion after this work 
because we interfered completely with stimulation in hippocampus and I felt that 
the hand-in-glove arrangement between hippocampus and dentate cannot be for 
nothing, hi the single unit studies we tound the most remarkable mechanism-like 
effects in the outflow from the dentate itself where we could give two or three 
stimulations in a reinforcing position, that is following the response, and then 
multiple firing would suddenly ensue. 

Magoun. Stimulating ascending pathways to the rabbit's hippocampus evokes 
an undulating slow-wave discharge whose frequency is around 5 to 7 a second. 
Unit recordings from the hippocampus, by Dr John Green, often show firing in 
some relationship to this ryhthm. Did you elicit such a rhythm from stimulating 
reinforcing sites? if so, could you correlate unit tiring with the slow waves? 

Olds. We do not have fair data to answer that. Because wherever we got an 
elicited effect we went on very rapidly to another unit. Quite often units would 
respond on slow waves, this being rather hard to see when you have a lot of filters, 
but still it was rather obviously there and it made the unit an unlikely one for 
further recording. We did not spend much time with units that were associated 
clearly with slow waves. Thus in a sense we have worked out of our population 
a lot of material which you would like to have and we will certainly go back and 
look at it later. 

Segundo. It seems opportune to mention observations performed upon human 
subjects that illustrate contrasting features of archi- and palaeo-cortices. Firstly, 
excitation effects of hippocampus or fornix that included somnolence or even slight 
blurring of consciousness, a result has that not been encountered in extensive explora- 
tion of the human neocortical mantle (Pentield, and Rasmussen, 1950; Segundo, 
Arana, Migliaro, Villar, Garcia-Guelfi, and Garcia-Austt, 1955). Secondly, the 
report by Griinthal of dementia produced by hippocampal atrophy (Griinthal, 
1947); neocortical lesions of comparable size do not provoke dementia. This 
material stresses the manner in which interference with hippocampal mechanisms 
affects behaviour, thus supporting Dr Olds's observations. 

KoNORSKi. Did you try to use as reinforcing agent self-punishing points instead 
of self-rewarding ones? It is interesting to see whether in this case the units, whose 
activity you observe, will behave in the same way as in your experiments or, on 
the contrary whether they will stop firing, when their activity is negatively 
reinforced. 

Olds. The positively reinforcing points interfered with learning. Then we have 
anecdotes on the negative reinforcing points which show they do not interfere with 
learning in any similar fashion. But we do not have full evidence, just anecdotes, 
which make us sure that we will find that the negative reinforcing points do not 
interfere with learning. 

As for slowins; the units, I believe asain on the basis of anecdotes, the answer 
will be yes. In the early days of tliis experiment we had animals that were tested 
for self-stimulation but we did not find whether they were also escaping and I have 
had an animal that I worked on for a period of almost 24 hours and every unit that 
I ever tried to reinforce was slowed. Since then I think that when we do study the 
negative reinforcement we should find that it will work. 

BusER. I wish to mention here some observations made on rabbits, by Dr 
Cazard and myself, on the effect of repetitive stimulation of the dorsal 


1 86 BRAIN MECHANISMS AND LEARNING 

hippocampus on ncocortical evoked potentials. It appeared that a short tetanization 
of the hippocampus (subhminal for producing any hypersynchronic or paroxystic 
phenomenon), was toUowed by a net increase (lasting up to 30 minutes) of the 
amplitude of cortical sensory responses (primary as well as motor irradiated 
evoked potentials). These effects were obtained on acute (chloralose or curarized 
unanacsthcthized) as well as chronic preparations. 

Olds. I believe that that sounds extremely pertinent. Ot course under anaesthesia 
the evoked potentials are also augmented, is that truef 

BuSER. Yes, under anaesthesia and also in caudate rabbits. 

Olds. So it does look as if the same function could be involved in both of our 
situations. 

RosvOLD. I was wondering whetlier there is any difference in your results 
between the caudate and the hippocampus or whether you think they are very 
similar. 

Olds. There is a difference. Of course, because of the internal capsule it is very 
difficult to know when you are stimulating caudate and not internal capsule too. 
We found that all our caudate stimulators produced associated movements of one 
sort or another. Very often in the case of the ambiguous caudate results we were 
quite sure that the animal was trying to do one thing and being forced to do 
another by the stimulus. We found that also inevitably with caudate stimulation 
the running ceased, that is, the animal simply would not run at all. It looked more 
like total disruption. I might even be tempted to say that even long-term memory 
was disrupted in these cases. 

RosvoLD. Do you suppose that the stimulation of the hippocampus jams the 
function of the hippocampus or excited it? 

Olds. I think it is interfering. I find- it hard to think of a stimulation in a 
learning experiment of doing something to organize a pattern, though stimulation 
of certain places of the reticular formation has been shown to augment certain 
kinds of perceptual tasks, but I regard this as an exception. 

Myers. What proportion of your attempts to condition rhinencephalic neurones 
has resulted in failure; 

Olds. We have never had a failure of the sort we have had with cortical units 
with rhinencephalic units. We have had a failure of the experiment either because 
we reinforced too soon ; its activity remained high and we could not use it further, 
because we could not find whether we had elicited it or not. Similarly we have had 
cases where we thought we had something to work with but after trying to 
reinforce silence, the response did not come back. But this cannot be said to be a 
failure to reinforce. With cortical units, we have had the unit going for periods of 
5, 10, 15 minutes, reinforcing regularly and failed to augment its firing rate. This 
has happened many times with ncocortical units but never with palaeocortical 
units so far. 

Myers. Have you seen evidence that the conditioning of one unit has any 
influence on the conditioning of other units in the neighbourhood. 

Olds. Yes. This is a definite thing. Quite surprising and I don't know quite how 
to put sufficient constructions on the situation to explain it. We found some situa- 
tions where we had one unit in an area, and reinforced it. Eventually it went into 
continuous firing. Its amplitude decreased and it finally disappeared and it sub- 
sequently was replaced and we reinforced the other one, and those two then 


J. OLDS AND M. E. OLDS 187 

alternated. Another thing has happened, we seemed to be reinforcing a slow wave 
and a unit, and then we moved the electrode and then we found that now for 
perhaps a milhmetre we could not get rid of the slow wave, and other units firing 
on it; this should have been mentioned in connection with Dr Magoun's earlier 
question. 


A NEW CONCEPTION OF THE PHYSIOLOGICAL 
ARCHITECTURE OF CONDITIONED REFLEX 

P. K. Anokhin 

The present conception has taken shape over a period of more than 30 
years as the result of the work of the author and his pupils on the physio- 
logy of higher nervous activity. 

The conception proposed in this paper eliminates a number of contra- 
dictions that have accumulated in the physiology of the conditioned reflex 
in recent years; it opens new avenues of research in the mechanisms of the 
conditioned reflex discovered by Pavlov, who revolutionized the study of 
the behaviour of annuals and man. 

The generally accepted view of the mechanism of the conditioned 
reflex rests on Descartes's reflex theory expressed in the concept of the 
'reflex arc'. According to this view, the excitation evoked by a condi- 
tioned stimulus constitutes the afferent part of the conditioned reflex arc. 
In the central part of the arc the excitation is transferred from the analyser 
to the effector part of the reflex and, hnally, the excitation reaches the 
efferent part of the reflex arc where it stimulates some working organ or 
combination of organs to action. 

This classical concept has three characteristic features. 

1. In the reflex arc the excitation spreads according to the lincar- 
translational principle: at each successive moment it spreads to new- 
neural elements and never returns to the course already traversed. 

2. The reflex arc ends in an adaptive action which, from the point of 
view of these ideas, forms as an entirely new phenomenon in tlic padi of 
the linear-translational spread of the conditioned excitation. 

3. The formation of the reflex action in the peripheral working appara- 
tuses is conceived as a process complctiuo the reflex arc and, consequently, 
the very adaptive result of the reflex action is not the decisive factor for 
the dynamic alternation of the processes of excitation and inhibition in the 
reflex arc. 

The conception proposed below does not exclude the reflex as a prin- 
ciple of the organism's activity and its relation to the external environ- 
ment. The reflex invariably constitutes the nucleus of our new ideas. 

189 


ipO BRAIN MECHANISMS AND LEARNING 

However, this nucleus is considerably expanded and supplemented by new 
links physiologically conceived as components of an integral neurodyna- 
mic and not linear organization, which we have named the 'functional 
system'. 

The formulation of our concept became possible only when the basic 
classical method of investigating conditioned reflexes had been supple- 
mented by additional techniques which enabled us to reveal the other 
aspects of conditioned reactions and their physiological substrate. 

The 'secretomotor method', the method of studying conditioned 
reflexes proposed by us, has been of particular importance in the elabora- 
tion of the new concept. Owing to a special design of the stand this 
method made it possible to record simultaneously the secretory and motor 
components of the conditioned reflex, the motor component constituting 
not a mere manitestation of movement towards food but a movement of 
choice towards one of the two or four feeding troughs connected with the 
given conditioned stimulus (Fig. i). 

The special design of a two-sided experimental stand made it possible 
to connect various conditioned stimuli with the different sides of the 
stand and to compare the secretory and motor coniponents in the most 
diverse experimental situations. 

Since by the classical secretory method this had not been possible on a 
wide scale, the first experiments conducted in our laboratory with our 
method revealed new aspects in the physiological architecture of the 
conditioned reflex (Anokhin, 1932, 1933). All the results of our investiga- 
tions were published in detail in some of our generalizing papers (Anokhin, 

1949, 1955, 1958). 

To solve the problem of the physiological architecture of the condi- 
tioned reflex we have made extensive use, since 1937, of the clectro- 
enccphalographic method. A method of recording the EEG in a perfectly 
natural environment in the study of conditioned reflexes was first used in 
our laboratory (Laptev, 1941, 1949). Recently these aims have been con- 
siderably furthered by the numerous investigations of the physiological 
peculiarities of the brain stem reticular formation so brilliantly begun in 
the laboratories of Magoun, Forbes and Moruzzi and later applied to the 
elaboration of conditioned reflexes in the laboratories of Jasper, Hernan- 
dez-Peon, Morrell, Gastaut, and our own (Jasper, 1957; Morrell, 1958; 
Hernandez-Peon, 1958; Gastaut, 1957, loshii, 1956). 

Our conception of the general physiological architecture of the condi- 
tioned reflex is best considered fragmentally, so that at the end of this 
report it may appear before the reader in its totality. 


p. K. ANOKHIN 


191 


i£ef^ 


Riaht 


38.235 ' Bett 


' 


~c0j7S Ri-g^t 


Fig. I 

A. Two-sided stand with two feeding-troughs making it possible to record sinuiltaneously the 
secretory and motor components of the conditioned reaction ('Secretomotor method'). 
Pneumatic transmission. 

B. Sample of kymograph record showing movement to the left (i) and right (2) and the 
conditioned secretion of saliva in drops. 


I. CONCEPT AND MECHANISM OF AFFERENT SYNTHESIS 


According to the reflex theory the role of the aftercnt influences on the 
central nervous system consists in the fact that the external stimulus 
usually serves as the impetus for the beginning of some reflex action. This 


192 BRAIN MECHANISMS AND LEARNING 

conception ascribes the decisive role to an isolated stimulus, something 
that has found expression in the evaluation ot the conditioned stimulus as 
a decisive factor in determining the quality and strength of the conditioned 
reflex effect. 

However, one physiological phenomenon which has considerablv 
shaken this widespread idea was described in detail in Pavlov's laboratory. 
I refer to the phenomenon of the dyitaiiiic paneni (Pavlov, 1932). 

As is well known, a definite and precise sequence of the selfsame con- 
ditioned stimuli, trained without any alterations over a long period of 
time, becomes the principal factor determining the quality and strength of 
the conditioned reactions. Conversely, the role of the itidiviihial condi- 
tioned stimulus is eliminated and the latter serves only as a noii-spccific 
iiiipetiis for the appearance of the conditioned reflex, whereas the quality 
and nature of the conditioned reaction does not depend on any condi- 
tioned stimulus in particular. This stimulus does not even have to be a 
conditioned stimulus, but may be an entirely new, i.e. indifferent (in 
Pavlovian terminology) stimulus. Nevertheless, it will evoke the condi- 
tioned reaction characteristic of the absent conditioned stimulus which 
was always used in the former experiments at tiic given point of the (dynamic 
pattern (Pavlov, 1932). 

Thus these experiments already revealed that the conditioned reflex, as 
regards its quality, strength and time'of appearance, is a synthetic result of 
the action of the conditioned stimulus and the preceding action of a 
greater sum of af]:erent stimuli representing the conditions of the experiment 
as a whole. 

A direct EEG analysis of the processes of the cerebral cortex conducted 
in our laboratory has shown that a sound stimulus, applied at the point of 
the dynamic pattern where light was always used, causes a desynchroniza- 
tion of cortical electrical activity, although applied at its usual place 
15 seconds previously it did not produce such desynchronization (Fig. 2) 
(Anokhin, 1956). Thus a real sound cannot alter the electrical activity of the 
cerebral cortex, this activity alternating according to the preceding 
training of the dynamic pattern. 

This dependence of each conditioned reflex on the synthetic nature of 
the external afferent influences was particularly clearly revealed in a 
special experiment conducted in our laboratory. The selfsame comhtioneil 
stinnihis (bell) was reinforceti by food in the morning and by an electro- 
cutaneous pain stimulus in the evening (Laptev, 1937). In the end the dog 
elaborates a perfectly clear differentiation: experimented upon in the 
morning it responds to the bell with pronounced food reflexes, whereas 


p. K. ANOKHIN 


193 


in the evening in response to the same bell in the saiiw chamber the animal 
shows a pronounced defensive reaction. 

In this experiment the difference in the reactions clearly does not depend 
on the conditioned stimulus which has retained only the uoii-spccijic 
trigger action because both reactions emerge only in response to the action 


^^ 


J9e<S^ 


J¥^ 


Fig. 2 

A. EEG of the dynamic pattern of 'Bell-Light-Siren'. The regular light ( + ) is suddenly 
replaced by a bell, but a desynchronization corresponding to the application of light at this 
place is evident just the same. 

B. Application of all the stimuli of the dynamic pattern ( j ) is discontinued. Further develop- 
ment of cerebral activity in total conformity with the structure of the pattern can be seen. 
Complete desynchronization can be seen in place of the absent light (+). 


of the bell, whereas in the intervals between the conditioned stimuli the 
animal sits quietly. 

Nor are the conditions of the experiment as a whole a differentiating 
factor since they are the same in both experiments and only the tunc oj 
day, as one of the components of the conditions of the experiment, is the 
determining factor. 

If we take into consideration the fact that the time of day becomes an 


194 BRAIN MECHANISMS AND LEARNING 

active factor the luomcnt the dog is placed in the stand and the reaction 
emerges only at the moment the bell comes into play it will become clear 
that the time of day, being a special stimulus, determines the quahty of 
the reaction because of the creation of a specific subthreshold dommant 
state ot the central nervous system (according to Uchtomsky) which 
spreads to the working apparatuses owing to the trigger action of the 
conditioned stimulus. 

The synthetic nature oi the relations between the conditioned reflex 
and the external conditions ot the experiment, both as a whole and of its 
individual components, may also be revealed under the usual conditions. 
Suffice it to transfer the animal, that until then had responded with good 
conditioned alimentary reflexes, to other surroundings (for example, a 
lecture-hall for demonstration to students) and the conditioned stimulus 
loses its conditioned reflex effect. In this case the synthesis elaborated from 
the tonic afferent action of the usual experimental situation and the trigger 
action ot the conditioned stimulus is disturbed and the conditioned reflex 
disappears. 

The following question arose in our laboratory: does this process of the 
synthesis of the situational and trigger afferent influences have any 
preferable localization in the cerebral cortex? 

The experiments of our associate Shumilina, conducted over many 
years and consisting in extirpation of the frontal lobes in dogs who had 
conditioned reflexes elaborated in the two-sided stand, have shown that 
the frontal lobes have a very definite relation to this function of synthesis 
of the afferent stimulations of different quality. During the experiment all 
the normal intact animals in the two-sided stand behave rather mono- 
tonously, with certain insignificant variations. Since the different condi- 
tioned stimuli are reinforced by a single portion of food on the different 
sides of the stand, the animals elaborate an expedient habit of sitting 
quietly during the intervals between the conditioned stimuli /// the niiddh' 
of the stand. 

After removal of the frontal parts (six to eight according to Brodman) 
all the intact animals begin to behave in a characteristic way immediately 
following the operation — they keep running from one feeding trough to 
the other as long as the experiment lasts. We have named these movements 
'pendular' (Fig. 3). 

It might be assumed that these movements were one of the forms of 
'motor restlessness' described by many authors after they had removed the 
frontal areas of the animals' brains. However, a direct analysis of the 
mechanism of these pendulous movements has shown them to be directly 


p. K. ANOKHIN 


195 


\*(#f In 


Seen 


\ \ \ \ « «< 


■* f I" 


-MT 


■■■ —1 wmmmmmWf'^im 


Fr;. ,^ 

A. Normal discrimination of the sides of the stand in response to corres- 
ponding conditioned stimuli. Stable movements to the left and right sides 
and quiet attitude in the middle of the stand during the interval between 
the stimulations. 

B. Continuous 'pcndular' runs to the opposite sides of the stand alter 
removal of the frontal divisions of the cerebral cortex. Stops only during 
feeding. Conditioned secretion is the same as before the operation. 

dcpcudciit o\\ the nature ot the aflerent intiuences suflered by the animal 
under the conditions of the given experiment. 

For example, if we elaborate in the animal conditioned reflexes only to 
one side of the stand, i.e. set up the conditions of Pavlov's classical method, 

o 


jg6 BRAIN MECHANISMS AND LEARNING 

after the removal of the frontal parts of the cerebral cortex this animal 
behaves perfectly quietly during the intervals between the applications of 
the conditioned stimuli: it sits on oiiv side of the stand and performs no pendn- 
lar nioi'cnieiits. 

However, if we feed this quiet animal trom the opposite feeding 
trough but once, it immediately begms to perform the well-known 
continuous pendulous movements to both sides of the stand. 

These experiments suggest that the 'pendular' movements are a direct 
result of the situation stimulation which includes conditions of alternate 
choice and behaviour. 

Why then does the intact animal sit quietly in the middle ot the stand 
during the intervals between the applications ot the conditioned stimuli? 

The interest of this phenomenon consists precisely in the fact that the 
numerous stimulations of the experimental conditions (situational afteren- 
tation) affect the nervous system all the time, whereas the conditioned 
reaction manifests itself only at the moment the conditioned stimulus 
begins to act (trigger ali'erentation). At the same time the extreme change 
in the experimental situation convinces us that the situational stimuli form 
an organic component part of the afferent integral. Thus it becomes clear 
that all forms of afferent excitations affecting the animal's nervous system 
at the given moment undergo a synthetic processing and that only after 
this stage does the efferent complex of working excitations begin to form. 

In recent years, neurophysiological studies have convinced us that all 
the afferent influences on the organism come to the cerebral cortex along 
two channels: (i) the specific or lemniscus pathway and (2) the non- 
specific, i.e. the brain stem reticular formation. The latter excitation is an 
indispensable condition for any interaction and interconnection of 
specific excitations on the level of the cerebral cortex. 

Specially for the physiology of the conditioned refiex this means that 
any form of association of stimulations acting on the organism simul- 
taneously or successively becomes possible only if the activating effect of 
the reticular formation has spread to the substrate of the cerebral cortex. 
The constant mediator between the new external conditions and the 
associative activity of the cerebral cortex and subcortical structures is the 
orienting-investigatory reaction of the animal taking place under an 
uncommonly high activation of the apparatus of the brain stem reticular 
formation (Heniandez-Peon, Shumilina, Havlicck). 

The physiological and biological purport of the orienting-investigatory 
reaction was very well revealed by Pavlov himself (Pavlov, 1925). It has 
now been shown that it occurs under a constant excitation not only of 


p. K. ANOKHIN 197 

the cerebral cortex but also under an efferent excitation of the peripheral 
apparatus of the sense organs (Granit, Dell, Livingston). Owing 
to the rapid alternation of this efferent stream from one analyser to 
another the afferent synthesis stage precedmg the very conditioned 
reaction is considerably enriched by afferent impulses, something that 
determines a finer and more adequate (for the given conditions) formation 
of effector processes of the conditioned reflex. 

Naturally, after the stage oi elaboration of the conditioned reaction this 
primary period of the multifarious efferent influences on the sense organs, 
increasing the excitability of the latter, is in large measure eliminated, 
owing to which the apparatus of the very conditioned reflex are simplified 
during the stage of its complete automation (see Fig. 4). 

Electrophysiological studies of the orienting and investigatory reaction 
have shown that it is able continuously to maintain a high level of excita- 
tion in the cerebral cortex and at the same time maintain in an active 
state the newly formed temporary bonds (Anokhin, 1957; Karazina, 
1957). Recently Professor Lindsley published a remarkable paper in wliich 
direct stimulation of the brain stem reticular formation demonstrated the 
tremendous role played by the reticular formation in the discriminative 
function of the cerebral cortex for rhythmic flashes of light (Lindsley, 
1958). 

Thus we believe it completely proved that the power basis for the all- 
rouiui synthesis of the iiiiiiieroiis external and internal, situational and trio(^er 
afferent influences on the cerebral cortex is the ascendino oenerali::ed actiration 
of the brain stem reticular formation. 

The role of the reticular formation in refining afferent analysis also 
manifests itself in the fact that at this moment an alternate increase in the 
excitability of various receptor structures at the periphery takes place 
(Granit, 1957; Dell, 1957; Snyakin, 1958). 

All told this makes it absolutely necessary, in analysing conditioned 
reflex adaptations, to set apart an independent stage of afferent synthesis. 

The significance of this stage consists first of all in the fact that before its 
completion the flow of effector excitations to the functioning organs 
cannot be formed. The composition of these effector excitations and, 
consequently, the nature of the adaptive act itself are directly dependent 
on the way in which this stage of the afferent synthesis is completed. 

At this point I must emphasize that the decisive role of the afferent 
function, as a 'creative' function of the cerebral cortex, was repeatecily 
indicated by Pavlov (Pavlov, 1928). The proof of this decisive role of the 
afferent function o{ the cerebral cortex is. in our opinion, the afferent 


198 


BRAIN MECHANISMS AND LEARNING 


synthesis stage which, as wc shall sec below, determines all the subsequent 
stages in the formation of the behaviour of man and animals. 


Fig. 4 
General scheme illustrating the role of the orienting-investigatory 
reaction in the formation of the conditioned reaction, i.e. in the 
establishment of the chain of afferent traces from the stimulus to 
the acceptor of action. 

A. At each stage of the animal's movement towards the 
reinforcing factor the orientmg-investigatory reaction contributes 
to the production of return afferentation. 

B. Final form of bonds. The conditioned signal aids in the almost 
instant spread of excitations along the afferent traces fixed in the 
past by means of the orienting-investigatory reaction. 


2. SPECIFIC ACTIVATING EFFECT ON THE CEREBRAL CORTEX AND PROCESSES 
OF AFFERENT SYNTHESIS 

The specific activating function of the brain stem reticular formation 
plays a special part in the processes of afferent synthesis (Anokhin, 1958). 


p. K. ANOKHIN 


199 


There is a widespread opinion that the activating effect of the reticular 
formation on the cerebral cortex is of a non-specific nature. According to 
this view, the non-specitic influence on the cerebral cortex is always of the 


.,4hiHH#WV^^ 


-1 '\i-.y v- 


B 


iiiililiiiiitiiililTi i iiniiiiiiii i iiiiiiiiiiiiiii i imiint,,,, , 


'r^^^-rf,j"^ff(rW 


Fig. 5 

A. Normal EEG of the rabbit in an experiment without a pain reiiiforceiiwnl. 

B. EEG of the same rabbit after application of sereral pain reinforcements durintj elaboration of a 
conditioned defensive reflex. Very strong desynchronization of electrical activity can be seen 
not only at the moment of application of the conditioned stimuli but also during the intervals 
between them. 

C. After chlorpromazin injection. 

same physiological nature regardless of the reactions of the integral 
organism taking place in the given situation (Moruzzi and Magoun, 1949; 
Magoun, 1952; Magoun, 1958; ct ah). 


200 BRAIN MECHANISMS AND LEARNING 

This point of view undoubtedly fnids justification in the facihtating 
effect which develops in the cortical cells for the excitations coming to the 
cortex along the specific lemniscus system. This conclusion is also favoured 
by the phenomenon of 'awakening' or desynchronization of the 
cortical electrical activity which is always of a generalized nature and 
seems to be independent oi the biological quality of the unfolding 
reactions. 

However, along with these remarkable generalizations, our laboratory 
has obtained tacts attesting that the non-specific activating effect of the 
brain stem reticular formation on the cerebral cortex is or<^iviical]y connected 
with the hioh\i^ical specificity of the (^ii'en conditioned reaction. It proved 
possible to block by means of chlorpromazinc (amuiasinc) the activating 
effect of the reticular formation for the conditioned defensive reaction, at 
the same time retaining the activating effect on the cortex for the alinien- 
tary reaction, i.e. for a reaction of another biological quality. At this 
moment the animal can greedily eat the food offered to it (Gavlichck, 
1958) (Fig. 5). 

Similarly blocking the defensive activation of cortical electrical activity 
by chlorpromazinc fails to eliminate the animal's waking state which, as is 
well known, is also maintained by the activating effect of the brain stem 
reticular formation on the cerebral cortex (Fig. 6). 

Thus it was demonstrated beyond all doubt that there arc several 
activating influences on the cerebral cortex and that each of theni may he 
hhuked indii'idnally because of the different chemical specificity of the nerrous 
substrate on iidiich each of these bioh-x^ically different reactions occur. 

What is the significance of this varying biological specificity of the 
activating influences on the cerebral cortex occurring during the afferent 
synthesis which precedes the formation of the conditioned reaction itself ■ 

In the first place it contributes to the sek-ctii'e rise in the excitability 
('facihtation') of the neural elements in the cerebral cortex, which in the 
given animal were historically associated, by the principle of conditioned 
coupling, with the inborn subcortical reaction of the given biological 
quality (positive or negative). 

Owing to this selective involvement of the cortical elements the 
mobilization of the numerous cortical systemic connections, which help 
to complete the stage of afferent synthesis, proceeds much faster and more 
efficiently. 

Our view of the specific, i.e. selective, activating influence of the 
reticular formation on the cerebral cortex with the perceptibk oeiieralized 
desynchronization of its eh'cfrical activity fits in with the remarkable experi- 


p. K. ANOKHIN 201 


II TON iCO I X ,/ 


-II 79 TONE 500 


^Aj^f^^llyVV^ 


B 


~A^^M-^J'fAA/\iW\/^Ar^W\^^ — 


'jy^V/^VVV/*VW\v-Vw^-^^Vs 


■'a 5J TOKE POOD 


f; r ^ r i 1 ] a f i f i f i fi 1 f , fi f ' ^JL fL - J ti — fi — fl — L . 

Fig. 6 

A. Shows the presence in the reticular formation of a specific rhythm of electrical activity 
(4-6/scc.) corresponding to the animal's defensive state. 

B. Elimination of the specific reaction by an injection of chlorpromazine. Application of the 
conditioned defensive stimulus no longer produces the generalized activation of the EEC 

C. Cortical and subcortical BEG of the food conditioned stimulus (tone). 


202 


BRAIN MECHANISMS AND LEARNING 


incuts performed by Jasper, Kogan and our own collaborator Polyantsev 
(Jasper, 1957; Kogan, 1958; Polyantsev, 1959) (Fig- 7)- 

These authors have shown that, when cortical electrical activity is in a 
state of generalized dcsynchronization, the various individual cortical 
elements are in absolutely different states of activity. Some of them may be 
excited, others inhibited, while still others may change the form ot their 
activity several times during the entire desynchronized state. These data 
can be understood only in the sense that each activated state of the 


500^^ 


ZOj^y 


vvy^^v^ 


B 


300/«v 


20;iy 


Fig. 7 

A. Simultaneous lead-oft" with the same electrode, i.e. from the same point of the cerebral 
cortex, of slow EEG activity and impulse activity. The apparent lack of coincidence between 
the two forms of activity shows a certain independence of the cellular impulse activity on the 
EEG. 

B. An even more demonstrative lack of coincidence is revealed by leading off^thc two indices 
from the same point of the reticular formation. Here the EEG does not have the usual desyn- 
chronization at all but the cellular impulse activity emerges just the same. 


cerebral cortex (according to the desynchronization index) has a corres- 
ponding system of selective excitation of cortical bonds. It is just this that 
constitutes the physiological basis for the extensively ramified process ot 
afferent synthesis. 

This first stage in the formation of the physiological architecture of the 
conditioned refiex may be represented schematically as in Fig. cS. 

This Figure shows that the afferent synthesis, as a physiological process, 
draws its energy from the ascending activating influences of the brain 
stem reticular formation. These influences reacli the cerebral cortex in the 


p. K. ANOKHIN 


203 


form of facilitating influences both at the moment ot the action oi the 
external stimulations and, especially, at the moment of the appearance of 
the orienting-invcstigatory reaction. 


Fig. 8 
Schematic representation of the constituent 
processes of the afferent synthesis stage. The 
scheme shows the constant activating influ- 
ence of the brain stem reticular formation 
on the afferent synthesis. It manifests itself 
during the action of the experimental situ- 
ation as a whole and durmg the episodic 
action of the conditioned stimulus. 
A^, A-. Analysers through which the 
situation stimuh produce their effect, 
a^. Collateral action of the same stunuli 
through the reticular formation. 
A. Analyser of the conditioned stinuilus. 
a. Collateral action of the conditioned 
stimulus throutrh the reticular formation. 


3. PHYSIOLOGICAL MECHANISM.S FORMING THE APPARATUS OF THE ACCEPTOR 

OF ACTION 

In this part of our report we shall describe a new mechanism ot the 
conditioned reaction noticeci and claborateti in our laboratory. In its 
physiological essence this mechanism is the end result of the afferent 


204 BRAIN MECHANISMS AND LEARNING 

synthesis stage, like the subsequent formation of the effector apparatus of 
the conditioned reaction itself However, it forms, in large measure as an 
independent, physiological formation and has, as we shall see later, a 
special afferent significance. 

hi its physiological content this apparatus consists essentially of the 
afferent excitations which in their totality reflect precisely the sum of 
afferent excitations that must enter the cerebral cortex only at the end of 
the reflex action. It follows that the acceptor of action, as an afferent 
reflection of the results of future action, is a physiological apparatus of 
so-called 'anticipation'. 

The tremendous importance of this apparatus in the behaviour of 
animals consists primarily in the fact that forming immediately after the 
end of the afferent synthesis stage it considerably forestalls the process of 
formation of the reflex action as well as the completion of this action. 

The acceptor of action makes its appearance so soon after the end of the 
afferent synthesis that it may be argued that this apparatus is a direct and 
immediate result of precisely the afferent synthesis stage. 

From a psychological point of view this moment, i.e. the end of the 
afferent synthesis corresponds to the emergence of the 'idea', 'intention' or 
'aim' to perform the given action. 

I believe it is necessary to observe at this point that the various reports 
that I have pubhshed on the physiological mechanisms of this stage in the 
development of conditioned reflex actions, have shown that many 
investigators do not as yet adequately understand that m a psychological 
sense the emergence of an 'intention' to perform some action /'.■; an abso- 
lutely indispensable stage which antedates the action itself and that, con- 
sequently, we, physiologists of the nervous system and higher nervous 
activity, must strive to analyse the physiological correlates of this 'intention'. 

The physiological content of this apparatus consists of the entire past 
afferent experience of the animal or man concretely related to the given 
behaviour pattern. Basically it is a certain integration of the afferent 
impulses w hich arise from the results of some reflex action or act of 
behaviour. 

For example, grasping a pitcher of water is connected with the reception 
of a series of aftcrcnt influences of a varying modality. 

Our brain receives specific tactile impulses reflecting the form and 
mechanical properties of the pitcher, temperature impulses, kmaesthetic 
impulses suggesting weight and, consequently the fact that the pitcher is 
filleci with water, etc. The impulses also include a visual afferentation 
suggesting the movement of the hand towards the pitcher. 


p. K. ANOKHIN 205 

The aggregate ot all these afferent impulses arises only when the pitcher 
is grasped and, consequently, may be referred to as affcrciitcitioii oj the 
results of the action, in the true sense of these words. 

From the morpho-physiological point of view such an aggregate of 
afferent influences torms in the cerebral cortex and subcortical apparatus 
a system of strong bonds which, by virtue of a number of repetitions of 
the given act acquires properties ot an organized, integral afferent forma- 
tion. It is precisely this system of bonds that becomes active owing to the 
rapid selective spread of excitations at the moment the afferent synthesis 
stage is being completed and that constitutes the physiological basis of the 
acceptor of action. 

At this point we are consciously leavnig out ot consitieration the 
aggregate oi the internal and external nitluences which, after the afferent 
synthesis stage, have, on the whole, conditioned the emergence of the very 
intention 'to grasp the pitcher'. 

It stands to reason that this intention is only a separate stage in a series of 
other 'intentions' which are tinally completed through the stage of a 
subjective 'quenching ot thirst' by a drop in the osmotic blood 
pressure. 

At this point it is important to inne diat, as soon as the afferent synthesis 
stage was completed, the cerebral cortex immeciiately exhibited an active 
excitation ot the entire system of the aforesaid afferent bonds which had 
become consolidated on the basis ot the past experience ot the afterent 
results of the act of grasping the pitcher. 

In other words, the afferent apparatus, which we ha\'c named the 
acceptor of action, torms under any action oi a condititMied stimulus 
before this reflex action has occurred and consequently constitutes an 
absolutely integral part of the cyclic architecture of any conditioned 
reflex action. 

It may be assumed that the anticipatory formation ot the acceptor of 
action has become the decisive factor in the organization ot the adaptive 
behaviour ot animals because it eliminates chaos in the choice and elabora- 
tion ot the individual acts ot this behaviour (Fig. 9). 

Containing all the afferent restiks of the past reintorcements, i.e. con- 
taining at the moment ot action ot the conditioned stimulation the 
afferent results only ot the tuture (!) action, the acceptor ot action becomes 
a peculiar control apparatus which establishes a physiological correspond- 
ence between the completed action and the initial "intention' to pertorm 
this action. 

There are several torms oi experimental proot ot the existence ot such 


206 


BRAIN MECHANISMS AND LEARNING 


an afferent apparatus which forestalls the appearance of the conditioned 
rellex action, as well as its afferent results. 

The first and most demonstrative form of experiment consists in the 
fact that the experiiiwiitcr suddtiil)' replaces the qtiality of the uiicoiiditioiied 
alimentary reiiiforceiiient. 

hi conducting these experiments we reasoned as follows: it the prepared 
conditioned excitation of the afferent cells in the cortical representation of 
the unconditioned centre precisely reflects the properties ot the Jiitiire 


Fig. 9 


CS. 


Stage of formation of the 'acceptor of action' consisting of the afferent traces of past reinforce- 
ments. At this stage the reflex action itself has not formed as yet. 


return unconditioned excitation and the normally elaborated behaviour of 
animals is based on this adequacy, the latter must iniallibly change if we 
replace the unconditioned stimulus. Owing to this replacement the 
anticipatory conditioned excitation in the additional afterent apparatus 
would be of one quality (on the basis of the former reinforcements) while 
the real unconditioned stimulus would suddenly (!) be of another quality 
and, consequently, because of such a combination of the conditions the 
composition of the nervous impulses of the return afterentation coming to 
the cerebral cortex from the real unconditioned stimulus would not 
correspond to the conditioned excitation prepared there. What will the 
actual behaviour of the animal in this case be? 


p. K. ANOKHIN 207 

This idea was applied in our laboratory (Anokhin and Strczh, Soi'iet 
Journal of Physiolo(^y, No. 5, 1933) on the basis of a bilateral alimentary 
reinforcement allowing these peculiarities of higher nervous activity to 
be demonstrated. 

The experiment was conducted in the followuig manner. Only two 
conditioned secretomotor reflexes were elaborated in the animal: to the 
tone of 'A' with a reinforcement on the right side of the stand and to the 
tone of 'F' with a reinforcement on the left side. Both reflexes were rein- 
forced by 20 g. of dry bread and were sufficiently well consolidated. After 
a brief latent period the animal would rush to the corresponding side of 
the stand and wait there for the unconditioned stimulus. At this stage of 
the experiments the annual no longer exhibited any erroneous motor 
reactions. 

At the beginning ot an expernncntal day dried meat was placed in one 
of the plates on the left side and thus, against the background of the usual 
dried-bread reinforcements, the animal was to receive a nieat reinforce- 
ment to one of the regular stimulations with the tone of 'F'. On the basis 
of the aforementioned peculiarities of the afferent apparatus ot conditioned 
excitation we niust assume that tor some brict space ot time the new un- 
conditioned stimulations ii'liicli do not coincide in their visnal, olfactory and 
qnstntory qualities with the already enien^ed conditioned excitation must lead to 
a lack ot coincidence in the two excitations and then to the development 
of an orienting-investigatory reaction. The latter shcnild be the more 
strongly pronounced the more the prepared conditioned afferent excita- 
tions and the available afferent excitations trom the true unconditioned 
stimulus tail to coincide. 

These expectations were justified by experiment. When the uncondi- 
tioned stimulus is thus replaced it usually gives rise to an orienting- 
investigatory reaction which, depending on the strength ot the stimula- 
tory action ot the suddenly used inadequate unconditioned stimulus, 
either changes to an active alimentary reaction (when bread is replaced by 
meat) or retards the alimentary reaction and the animal even retuses the 
food (if the meat is replaced by bread). 

The toregoing experiment enabled us to observe both tonus ot the 
reaction. 

The second form of the experiment aimed at proving the existence of 
the acceptor of action is based on the peculiarities ot the experimental 
method in the two-sided stand proposed by us (sec Fig. 1). 

Several conditioned secretomotor reflexes firmly tixed by training are 
elaborated in the animal. To each of the stimuli used at any given moment 


208 BRAIN MECHANISMS AND LEARNING 

the animal responds with a positive motor ahmentary reaction to a 
feeding-trough either on the right or left side according to the side on 
which the animal was fed during the application of the given conditioned 
stimulus. 

Let us assume that with a flash of light the ciog was always fed from the 
right-hand trough, whereas \\'ith the sound of the tone of 'C it was given 
food in the left-hand trough. According to the law of the conditioned 
reflex, the animal responds to this form of experiment by elaborating 
precise conditioned motor reactions to the right and left sides of the stand. 

As soon as these conditioned reactions are firmly fixed the food is 
suddenly (!) given to the flash of hght on the left rather than on the right 
side of the stand. 

hi other words, an operation is performed or, in the terminology of the 
Pavlovian laboratory, the conditioned reflex is 'reshaped'. 

hi this form of experiment it is possible to see that the animal frequently 
fails to take food from the trough on the side that does not correspond to 
the conditioned signal. For a long time it looks alternately to the right and 
left feeding-troughs (orienting-investigatory reaction) and begins to eat 
the food only after a long latent period. Additional investigations of the 
animars respiratory component, when the side on which the food rein- 
forcement is made is suddenly changed, show the animal to exhibit an 
uncommonly pronounced oricntihg-investigatory reaction with a 
strongly marked inspiratory tonus in breathing. 

The only cause that could be tliought of for this pronounced reaction 
was the lack of corrcspoiidciice hctwccu the sudden feediui^ {return affereiitation) 
and the already prepared complex of afferent excitations in the cerebral cortex in 
accordance with the place of the old and usual feedino (acceptor of action). Tlic 
same conclusion is suggested by the fact that both food reinforcements, 
on the right as well as on the left sides, are essentially of the same signifi- 
cance to the animal, the distances from both feeding-troughs also being 
the same. 

The only remaining cause is the lack of correspondence between the 
already prepared acceptor of action and the afferent influences emerging 
when the animal is fed on the inadequate side. 

The method of sudden replacement of the rehiforccment may be used 
in most diverse variations, especially in experiments c:>n human adults and 
children. For example, by showing some daintv to a child and placing this 
dainty, so that the child sees it, in a 'problem box' from which the child 
extracts it, it is possible finally to establish adequate correlations between 
the sight of the daiiity and the subsequent reinforcement wliich occurs 


p. K. ANOKHIN 209 

when the child opens the box. However, if a (laiiity of another quality is 
imperceptibly placed in the box, upon opening the box the child imme- 
diately develops a reaction of 'surprise' ('What is it?', according to Pavlov) 
which occurs with different variations depending on a number of condi- 
tions. The child may carefully examine the box from all sides, shake it 
in the hope of receiving the expected reini(n-ceinent, etc. It is perfectly 
clear that all these actions are a direct result ot the lack of correspondence 
between tlie fixed afferent excitations in the form of the acceptor of actio)! and the 
afferent iujiitences coniino to the central nerrons system from the inadequate 
appearance of the nen' dainty when the box is opened. 

The third way of proving the emergence of afferent excitations in the 
acceptor of action forestalling the formation of the action itself is the 
electroencephalographic method. As stated previously the systems of 
afferent excitations forming part of the acceptor of action are mobilized 
much more rapidly than the effector part of action. At times, a very com- 
plex reflex action is formed. It follows that the ascertainment of the prepara- 
tory excitations in the area of the cerebral cortex, ndiicli must, in the future, 
receive the return afferentation from the results of the action, should become one 
of the methods of studying the physiological mechanisms of tlie acceptor 
of action. 

A typical example of the anticipatory spread of excitations over the 
cerebral cortex is the generally known, so-called, electroencephalographic 
'conditioned reflex' which manifests itself when sound and light are 
combined. As is well known, after a number of such combinations 
the sound alone causes a desynchronization of the a-rhythm in the 
occipital area of the cortex before the appearance of the lioht. Here the 
sound sets off a chain of excitations in the cerebral cortex, which spreading 
rapidly reaches the elements in the visual cortex that will not receive 
adequate visual excitation from the periphery until a few seconds later 
(Jasper, 1937; Livanov, 1937; Karazina, 1957; and many others). 

Any acceptor of action for any, even very complex acts of behaviour, is 
formed absolutely according to the same type. Any act of behaviour is 
represented in the cortex by the development of an uninterrupted chain of 
afferent excitations received from the individual stages of this act to the 
final reinforcing factor inclusive. It is precisely this chain of traces of the 
past afferent excitations that is a peculiar 'conductor' for the rapid spread 
of the excitations from the conditioned stimulus to the system of excita- 
tions udiich is the afferent reflection of the results of the as yet forthcomino 
action. 

Exaniples of such a spread of the afferent excitations may now be 


210 


BRAIN MECHANISMS AND LEARNING 


obtained from most diverse papers in which the authors themselves did 
not intend to study precisely this problem. 

We may cite data from the work oi Jasper who points out that in a 
number of cases the secretomotor area of the cerebral cortex exhibits a 
depression of its rhythm much ahead of the real movement of the arm, 
this desvnchronization sometmics appearing in the person tested at the 
thought of the forthcoming movement alone ( lasper, 1958). 


A 


o — o 


o — o 


'il' 


8. 


• — o 

II 

• — o 

II 

• --o 

--0 


1 


d' 

c' 

*-^' 


Fig. 10 
Schematic representation of the 'anticipatory' spread of excitation 
and its significance in the formation of the acceptor of action. 

A, a, h, c, d, -^ successively developing action of the external stimu- 
lations of which d is the reinforcement with food. Orienting reactions 
a^, b^, c^ and d^ emerge correspondingly. 

B. Correlation of the processes after training. The action of the one 
initial stimulus a is enough for the process of excitation immediately 
to spread along the afferent traces to d which is the 'acceptor of action' 
in this case. 

We fmd indications of similar facts 111 the studies oi Gastaut concerned 
with depressing the rolandic rhythm (Gastaut, 1957). 

hi his recent detailed review of the stuciies of the reticular formation 
O'Leary also cited data to the effect that the 'idea' of the forthcoming 
movement alone may lead to a desynchronization of the electrical activity 
of the cerebral cortex (Fig. 10). 

All the data culled from the literature, as well as our own observations, 
indicate that after the application of a stimulus or the reading of instruc- 
tions the selective impulse process of excitation, which ensures the 


p. K. ANOKHIN 211 


formation of the conditioned reflex, spreads with uncommon speed 
through all the chains ot the past afferent stimulations which reflected the 
continuity in the development ot some act ot behaviour. This process of 
the anticipatory propagation of afferent excitations until the moment the 
acceptor of action is formed may be shown diagramatically as in Fig. 9. 

4. FORMATION OF THE CONDITIONED REACTION EFFECTOR APPARATUS 

The stage of formation oi the conditioned reaction effector apparatus 
is directly dependent on the course or end of the afferent synthesis. The 
latter is always an integral whole and contains in its composition both 
somatic and vegetative components (motor component, respiratory 
component, cardiac component, vascular component, hormonal com- 
ponent, etc.). 

The characteristic feature ot this effector integration is the fact that each 
of its functioning peripheral components is harmoniously related to the 
other components and together they constitute the given act oi behaviour 
or the conditioned reflex. 

For example, the respiratory component ot the conditioned reflex was 
subjected to systematic analysis fc^r the tirst time in our laboratory 
(Balakhm, 1930-35; Polezhayev, 1952; Makarov, 195(S; ct ah). It was 
shown to be ot an entirely specitic character corresponding to the bio- 
Ic^gical quality of the animal's given reaction as a whole. In the defensive 
conditioned reflex the respiratory component exhibits a high inspiratory 
tonus and frequent respiratory movements of the thorax. 

In a well-fixed alimentary conditioned reflex the respiratory rhythm is, 
on the contrary, quiet and only at the moment when the conditioned 
stimulus is applied docs the orienting-investigatory reaction temporarily 
raise the activity of the respiratory component (Fig. 11). 

A comparative evaluation of the secretory, motor, respiratory and 
cardiovascular components of the conditioned reflex shows that they are 
all harmoniously adapted to the biological signiticance of the whole 
reaction. If the reaction requires general activity and tension (as, for 
example, the conditioned defensive reaction) all the vegetative com- 
ponents, as many as there are, unite into an integrated whole, each ot them 
corresponding to the tasks ot the given adaptive act. This means that the 
intensity of the vegetative components of the conditioned reflex, which 
provide the entire reaction with power resources, is directly dependent on 
the degree of the fortlicoiiiiin^ expenditure of these resources. 

This state, revealed particularly clearly as early as 1935 in the studies of 

p 


212 


BRAIN MECHANISMS AND LEARNING 


our associate Balakin, was recently confirmed by special forms of experi- 
ments. 

In our laboratory Dr Kasyanov has shown that the respiratory com- 
ponent of the conditioned reflex is the first conditioned reflex component 
to form in the effector pathways (Kasyanov, 1950). The cardiovascular 
complex reaches the effector pathways almost simultaneously with the 
respiratory component. Gantt's studies have show-n, however, that the 
cardiac component of the conditioned reflex appears somewhat ahead of 
the respiratory component (Gantt, I95-)- 


wvv>**iyv*^«-*«^ 


B 


A 


Fig. II 
Dissociation between the generalized and local excitation of the motor apparatus of the dog 
after bilateral extirpation of the sensorimotor areas of the cerebral cortex. The generalized 
reaction (A) is on hand, while the local raising of the hind limb is totally absent (B). 

The recent studies conducted by Shidlovsky in our laboratory with the 
aid of up-to-date cardiographic apparatus ('Cardiovar', 'Barovar', etc.) 
have confirmed our former observations and have shown the respiratory 
component to reach the effector pathways somewhat ahead of the cardiac 
component (Shidlovsky, 1959). 

It should be noted, however, that these insignificant differences in the 
appearance of the vegetative excitations in the peripheral organs, where 
they are detected by suitable apparatus, are of no fundamental importance. 
They may be due to different lengths of the pathways the excitations must 
traverse from the centres to corresponding organs, to the number o'l 
synapses in these pathways, and, lastly, to the peculiarities of the recording 
apparatus. 

But some aspects of this phenomenon, constant for all types of condi- 
tioned reactions, are undoubtedly important. For example, the vegetative 


V. K. ANOKIIIN 213 

components in general reach the tei'uimal effector components hi the 
form of components of specific quahties (secretion for the aUnientary 
conditioned reaction, movement for the defensive conditioned reaction) 
before the conditioned reflex reaction manifests itself. The other important 
aspect of what we call Vegetative outstripping' consists in the fact that all 
these components in their totality are of a conditioned refiex nature and 
reflect precisely the energy requirements of the forthcoming conciitioned 
reflex action. 

Shidlovsky devoted a special experiment to this problem. He recorded 
the cardiovascular components of the conditioned alimentary reaction in 
two different situations: in one case, the animal had to overcome a small 
obstacle to get at the food received as reinforcement, thus exerting a 
muscular effort, while in the other case the food was brought directly to 
the animal's muzzle. Thus, in the first case, the conditioned signal warned 
the animal not only of the forthcoming feeding but also of certain uiiisciihv 
efforts it had to exert before receiving the food. 

hi total conformity with these conditions the respiratory and cardio- 
vascular components of the conditioned reaction are vigorously activated 
in the first case. Conversely, in the second case, the same vegetative 
components manifest no such changes and deviate but slightly from the 
level of the resting state (Fig. 12). 

All the experiments performed in this direction clearly show that the 
conditioned reaction, as a manifestation of the integral organism, atlects 
the periphery through very numerous terminal neurones involving most 
diverse somatic and vegetative components. 

An evaluation of the physiological composition of this integral etlerent 
formation should take into account its three physiological peculiarities: 

[a) it is a direct result of the afferent synthesis stage; 

[b) it forms during the entire course of elaboration of the given concii- 
tioned reflex (see below) ; 

(f) it has a vertical physiological architecture, since it infallibly includes 
cortical and subcortical components. 

The last of the foregoing propositions is clearly demonstrated by the 
very participation of vegetative components in the conditioned reflex. 
Since these components reflect the integral character of the total reaction 
and are formed through the terminal pathways of the hypothalamus and 
the brain stem reticular formation, it is possible to chart the general 
distribution of the nervous impulses within the entire efferent stream of 
excitations, including the functioning organs. 


214 


BRAIN MECHANISMS AND LEARNING 


^. 


\rr^ 


~0. TH. ...^Y^^^^ ' 


F O R E L I M B E M G 


h *' ' P' ' — 1- 


Xfm 


Lji^ ( «^ni4^iiiiii H« 4"" 


RESPIRATION 


E C G LEAD 1 


ART. BLOOD PRESSURE 


J}J),^t4)ll,^^^H^4^^ w y **'**'^^ 


SALIVARY DROP^ 


Mm 


AUDIT S T I M 


Ip 


AF O O D 

Fig. 12A 
Illustration of the dependence of the vegetative components of the conditioned reflex on the 
forthcoming energy expenditures signalled by the conditioned stimulus. 
A. Reaction to the conditioned stimulus with the animal in close proxnuity to the food. 


p. K. ANOKHIN 


215 


E C G LEAD 1 


^4||,l>#H)t>>>i*WHtll**lli 


ART BLOOD PRESSURE 


SALIVARY DROPS 

|ll|MlllHilllHIIIIIIHllll 


B 


AUDIT S T I M 


»I 


ilL 


Fig. 12B 
B. Reaction to the conditioned stimulus when overcoming a difficulty before taking the food. 
The latter case shows much greater activity of all the vegetative components of the conditioned 
reaction. 


2l6 


BRAIN MECHANISMS AND LEARNING 

L.s.m. 


/^7'VVVv^/Vv^^wv^n(v>ft^^'VV^^vvw^«vv^^ 


o.i. 
n.v.p.L.L. 

n.v.p.L r 
EKG 

2f]0/xl/ 1 bee 

Fig. 12C 
C. Confrontation of dirtcrcnt components of a response to 
direct stimulation of the brain stem reticular formation. 
Abbreviations: l.s.m. — left sensorimotor cortex, r.s.m. — right 
sensorimotor cortex, t.l. — left temporal cortex, t.r. — right 
temporal cortex, o.s. — occipitalis sinistra, o.d. — occipitalis dex- 
tra, n.v.p.1.1. — n. ventralis posterior lateralis, left, n.v.p.l.r. — n. 
ventralis posterior lateralis, right, EKG, respir. — respiration. 


The latest data on the representation ot vegetative tunctions in the 
cerebral cortex once more emphasize the importance of precisely the verti- 
cal plan in the structure of the effector complex ot excitations in the condi- 
tioned reflex. I am referrnig primarily to Papez's view of the 'visceral 
cortex' and to the studies of a number of other authors directed towards 
the same aspect of the subject (Papez, ips.S; Maclean, 1954; Bard, 194^"^; 
Green, 1958; Adcy, 1958). 

On the basis of authentic facts indicating representation of vegetative 


p. K. ANOKHIN 217 

functions in the limbic, orbital, girus cinguli and other parts of the cere- 
bral cortex it might be assumed