And finally
psychoanalysis
itself has begun to catch up with bowlby.
Bowlby - Separation
-i-
Attachment and Loss
VOLUME II SEPARATION ANXIETY AND ANGER
John Bowlby
With Additional Notes by the Author
BASIC BOOKS
A Member of the Perseus Books Group
1
To THREE FRIENDS
Evan Durbin Eric Trist Robert Hinde
Copyright (C) 1973 by The Tavistock Institute of Human Relations Library of Congress Catalog Card Number: 70-78464
ISBN: 0-465-07691-2 Cloth
ISBN: 0-465-09716-2 Paper
Printed in the United States of America 99 RRD-H 30 29 28 27 26 25 24 -ii-
2
Contents
Foreword page vii Preface xi Acknowledgements xvii
PART I: SECURITY, ANXIETY, AND DISTRESS 1 Prototypes of Human Sorrow 3
Responses of young children to separation from mother 3 Conditions leading to intense responses 6 Conditions mitigating the intensity of responses 16 Presence or absence of mother figurer: a key variable 22
2 The Place of Separation and Loss in Psychopathology 25 Problem and perspective 25 Separation anxiety and other forms of anxiety 30 A challenge for theory 30
3 Behaviour with and without Mother: Humans 33 Naturalistic observations 33 Experimental Studies 39 Ontogeny of responses to separation 52
4 Behaviour with and without Mother: Non-human Primates 57 Naturalistic observations 57 Early experimental studies 60 Further studies by Hinde and Spencer-Booth 69
PART II: AN ETHOLOGICAL APPROACH TO
HUMAN FEAR
5 Basic Postulates in Theories of Anxiety and Fear 77
Anxiety allied to fear 77 Models of motivation and their effects on theory 79 Puzzling phobia or natural fear 83
6 Forms of Behaviour Indicative of Fear 87 An empirical approach 87 Withdrawal behaviour and attachment behaviour 89 Feeling afraid and its variants: feeling alarmed and feeling anxious 92
-iii-
7 Situations that Arouse Fear in Humans 96 A difficult field of study 96 Fear-arousing situations: the first year 99 Fear-arousing situations: the second and later years 105 Compound situations 118 Fear behaviour and the development of attachment 119 Natural clues to potential danger 124 Fear behaviour of non-human primates 127 Compound situations 134 Fear, attack, and exploration 136
3
9 Natural Clues to Danger and Safety 138
10 Natural Clues, Cultural Clues, and the Assessment of Danger 151
Clues of three kinds 151 Real danger: difficulties of assessment 153 'Imaginary' dangers 156 Cultural clues learnt from others 158 Continuing role of the natural clues 161 Behaviour in disaster 166
11 Rationalization, Misattribution, and Projection 169 Difficulties in identifying situations that arouse fear 169 Misattribution and the role of projection 172 The case of Schreber: a re-examination 174
12 Fear of Separation 178
Better safe than sorry
Potential danger of being alone
Potential safety of familiar companions and environment Maintaining a stable relationship with the familiar environment: a form of homeostasis 148
138 142 146
Hypotheses regarding its development Need for two terminologies
PART III: INDIVIDUAL DIFFERENCES IN
SUSCEPTIBILITY TO FEAR: ANXIOUS
178 182
-iv-
ATTACHMENT
13 Some Variables responsible for Individual Differences 187
Constitutional variables 187
Experiences and processes that reduce susceptibility to fear 191
Experiences and processes that increase susceptibility to fear196 14 Susceptibility to Fear and the Availability of Attachment Figures 201
Forecasting the availability of an attachment figure 201 Working models of attachment figures and of self 203 The role of experience in determining working models 207 A note on use of the terms 'mature' and 'immature' 209
15 Anxious Attachment and Some Conditions that Promote it 211 'Overdependency' or anxious attachment 211 Anxious attachment of children reared without a permanent mother figure 215 Anxious attachment after a period of separation or of daily substitute care 220 Anxious attachment following threats of abandonment or suicide
226 16 'Overdependency' and the Theory of Spoiling 237
Some contrasting theories 237
Studies of 'overdependency' and its antecedents 240 17 Anger, Anxiety, and Attachment 245
4
Anger: a response to separation 245 Anger: functional and dysfunctional 246 Anger, ambivalence, and anxiety 253
78 Anxious Attachment and the 'Phobias' of Childhood 258 Phobia, pseudophobia, and anxiety state 258 'School phobia' or school refusal 261 Two classical cases of childhood phobia: a reappraisal 283 Animal phobias in childhood 289
19 Anxious Attachment and 'Agoraphobia' 292 Symptomatology and theories of 'agoraphobia' 292 Pathogenic patterns of family interaction 299 'Agoraphobia', bereavement, and depression 309 A note on response to treatment 310
20 Omission, Suppression, and Falsification of Family Context 313 -v-
21 Secure Attachment and the Growth of Self-reliance 322 Personality development and family experience 322 Studies of adolescents and young adults 328 Studies of young children 330 Self-reliance and reliance on others 359
22 Pathways for the Growth of Personality 363 The nature of individual variation: alternative models 363 Developmental pathways and homeorhesis 366 One person's pathway: some determinants 369
APPENDICES
I Separation Anxiety: Review of Literature 375 II Psychoanalysis and Evolution Theory 399 III Problems of Terminology 404 Additional Notes 409 References 415 Additional References 436 Index 439
-vi-
5
Foreword
Truly revolutionary works that open up new conceptual vistas tend to straddle intellectual worlds. Their originality makes them difficult to locate in the historical context in which they are written, and their pervasive impact makes it difficult to appreciate just how formative they were in generating subsequent developments that have come to take on the quality of great obviousness.
John Bowlby was a psychoanalyst; he notes in his original preface that psychoanalysis provided him his first inspiration as well as the only serviceable vocabulary for writing about intense human experiences, trauma and separation. Retrospectively, we might imagine that his work on the powerful emotional bonds between children and their mothers would have been welcomed within the psychoanalytic world. It was not. It was certainly not ignored, however; the persuasiveness of Bowlby's presentation made that impossible.
Bowlby's initial work on attachment was given an airing in the most influential psychoanalytic publications of his day and was then attacked by the most prestigious analytic authorities. That he was virtually exiled by the psychoanalytic community was, ultimately, fortuitous. It led to the establishment around Bowlby's work of a research tradition into the nature of emotional attachment, separation and loss that has generated some of the most remarkable, reliable and provocative empirical data of the past fifty years.
What was it that got Bowlby into so much trouble? He introduced the project as a working out of the implications of observations of the responses to loss young children. It was Bowlby's commitment to the importance of his observations that was the problem. The psychoanalysis of Bowlby's day was an interpretive system that was almost hermetically sealed. Freud had provided the Rosetta Stone for deciphering the underlying structures of mind, and Melanie Klein had extended his psychic excavations to the deepest recesses. Armed with their decoding devices, psychoanalysts felt they could see beneath the superficial, behavioral levels of human interaction to the underlying instinctual conflicts and fantasies that generated their deeper meanings.
Bowlby had been asked by the World Health Organization in 1950 to study and offer advice on the plight of homeless children, real children in the real world. The ravaging impact of maternal deprivations and separations made an enormous impression on
-vii-
him -- to his surprise. It is a tribute to how much Bowlby's work has changed our consciousness that we must strain to imagine why he would have experienced the magnitude of such an impact as a surprise.
In the understanding of mid-twentieth century experts, small children had basic physical needs that required tending; the complex emotional relationships involving a unique sense of interpersonal connection evolved only later. In the conditioning paradigms that dominated American psychology in those days, the caregiver was a "secondary-reinforcer," who became important to the child only by virtue of being associatively linked with physical ministrations. And in the language of the prevailing psychoanalytic theory the mother was, similarly, a "need-gratifying object," whose significance developed gradually through her role in
6
satisfying drive pressures. Separations from mothers shouldn't matter terribly much, as long as the child's needs were taken care of.
But separations did matter, Bowlby discovered. Observations of actual children suggested that maternal deprivation was extremely traumatic. Bowlby s position was a little like that of the wife in the old story who discovers her philandering husband in bed with another woman. "Darling" pleads the husband, "who are you going to believe, me or your lying eyes? " Bowlby decided to believe his eyes.
Any good psychological theory applied cleverly can account for virtually any data. Surely the instinct theory of Bowlby's day could be stretched to accommodate his observations. But for Bowlby, the center of gravity of the data on trauma in real children in the real world did not match up with the center of gravity of Freud's drive theory, privileging impersonal, body- based needs and fantasies. So, as Bowlby announces in this work, a new instinct theory was required, in which the powerful emotional attachment between child and mother is understood not as derivative of more basic processes but as fundamental in itself.
This was no minor addition; it challenged psychoanalytic metapsychology at its core. Bowlby liked to recount experiences formative for him as a student listening to case presentations that emphasized unconscious, instinct-based fantasies at the British Psychoanalytic Society; he remembered one conference at which he felt moved to rise and state emphatically, "But there is such a thing as a bad mother! "
Bowlby was not working completely alone; he had conceptual fellow travelers, especially in the Interpersonal Psychiatry of the American Harry Stack Sullivan and the psychoanalytic contributions of other innovators like W. R. D. Fairbairn, Donald Winni-
-viii-
cott and Hans Loewald. Part of what drew so much fire in Bowlby's direction, however, was his clarity. Sullivan was a tortured, blocked writer. Fairbairn was often tedious and difficult. Winnicott was poetic and elusive. Loewald was extremely subtle and often obscure. Bowlby wrote with lucidity and power.
It was amply apparent from the very beginning and throughout that Bowlby regarded his contributions as a direct challenge to certain basic tenets of Freudian theory. And he had data on children in the real world to back them up. And he identified himself very much as a scientist, offering testable hypotheses. And, his links with other scientists, especially the ethologists of his day, made his position extremely persuasive. For the psychoanalytic establishment of the time, this was simply too much to bear. Bowlby became much more interested than the average psychoanalyst in what actually goes on between people in the real world, and the neighboring discipline of ethology provided powerful explanatory concepts for understanding what he had been observing in children's reactions to separation and loss.
Both Freud and Bowlby were extremely involved with Darwin's contribution (one of Bowlby's last works was a biography of Darwin), but their Darwins were very different. Freud's Darwin was part of the first wave of reaction to the extraordinary implications of the theory of evolution; one of Freud's projects was to work out the implications for human psychology of Darwin's demonstration of the continuity between, so-called lower and so- called higher forms of animal life. Freud's fascination with primitivism and his reliance on
7
bestial metaphors are thematic throughout his writings. And Freud's structural model of the psyche is a re-creation, on a microcosmic level, of Darwin's sweeping account of the evolution of species: lower level, primitive energy of the id is transformed by the reality- oriented ego into higher level, aim-inhibited resources for activities consistent with the cultural values of the superego. Ontogeny recapitulates phylogeny.
Bowlby draws on a different Darwin. Like the Ego psychologist Heinz Hartmann, Bowlby was most interested in what Darwin taught about animal adaptation to environmental conditions and niches. In this second volume of his Attachment and Loss trilogy, Bowlby actually refers to Freud as pre-Darwinian because Freud did not grasp the importance of the principle of "natural selection" in Darwin's theory of the evolution of species. Bowlby, like Darwin, was interested in what animals do to maximize their chances for survival. Whereas Freud's Darwin lent himself to the study of internality and unconscious, primitive states, Bowlby's Darwin lent himself
-ix-
to a behavioral analysis of what small children and mothers actually do with each other. Thus, among the most vivid of Bowlby's contributions is his account of the five component instincts that insure the baby's proximity to the mother--the underpinnings for both healthy attachment and traumatic separations and loss.
One advantage of this behavioral emphasis has been that Bowlby's ideas have been applied, with extraordinary effectiveness, to the empirical research tradition that Mary Ainsworth and Mary Main have done so much to develop. Another advantage is the ease with which Bowlby's observations have been adopted by popular pediatric practice and social policy planners to prioritize the personal, emotional dimensions of childcare. Despite our difficulties as a society in investing the resources necessary for psychologically healthy child- evelopment, we have come to take as axiomatic the principle of "bonding" between parents and children and the importance of emotional warmth and security in caregiving. Bowlby's work was central to this enriched understanding.
Until recently, the disadvantage of Bowlby's behavioral emphasis has been the relative underdevelopment of the psychodynamic dimension within attachment theory, which has made the bridge to other psychoanalytic theorizing more difficult. Bowlby's concept of "working models" had an overly schematic, mechanistic feel to it, which lacked the richness of psychoanalytic investigations of the inner world. But the more recent attachment literature (Fonagy) has taken a more inward turn in exploring the concomitants of secure and insecure attachments in the textures of conscious and unconscious subjectivity. And finally psychoanalysis itself has begun to catch up with bowlby. The recent relational turn in psychoanalysis ( Mitchell and Aron) has made bridges between Bowlby's work and contemporary psychoanalytic thought much more compelling.
Part of what makes Bowlby's early work so inspiring thirty years later is his blend of openness and persistence. His intellectual curiosity seems to have known no bounds, and he continued to draw upon many diverse sources for ideas and conceptual tools, including child- observation, ethology, systems theory, and information-processing. Bowlby also knew from early on that he was onto something very important, and his integrity and perseverance in pursuing what mattered, despite criticism, has benefitted us all.
8
Attachment and Loss has been one of the most influential works of this century. Stephen A. Mitchell New York City, October 1999
-x-
Preface
IN the preface to the first volume of this work I describe the circumstances in which it was begun. Clinical experience of disturbed children, research into their family backgrounds, and an opportunity, in 1950, to read the literature and to discuss problems of mental health with colleagues in several countries led me, in a report commissioned by the World Health Organization, to formulate a principle: 'What is believed to be essential for mental health is that the infant and young child should experience a warm, intimate and continuous relationship with his mother (or permanent mother-substitute) in which both find satisfaction and enjoyment' (Bowlby 1951). To support this conclusion evidence was presented for believing that many forms of psychoneurosis and character disorder are to be attributed either to deprivation of maternal care or to discontinuities in a child's relationship with his mother figure.
Though the contents of the report proved controversial at the time, most of the conclusions are now accepted. What has plainly been missing, however, is an account of the processes through which the many and varied ill effects attributed to maternal deprivation or to discontinuities in the mother-child bond are brought into being. It is this gap that my colleagues and I have since striven to fill. In doing so we have adopted a research strategy that we believe is still too little exploited in the field of psychopathology.
In their day-to-day work, whether with disturbed children, disturbed adults, or disturbed families, clinicians have of necessity to view causal processes backwards, from the disturbance of today back to the events and conditions of yesterday. Though this method has yielded many valuable insights into possible pathogenic events and into the kinds of pathological process to which they appear to give rise, as a research method it has grave limitations. To complement it, a method regularly adopted in other branches of medical research is, having identified a possible pathogen, to study its effects prospectively. If the pathogen has been correctly identified and the studies of its effects in the short and long term are skilfully executed, it then becomes possible to describe the processes set in train by the pathogenic agent and also the ways by which they lead to
-xi-
the various consequent conditions. In such studies attention must be paid not only to the processes set in train by the pathogen but also to the very many conditions, internal and external to the organism, that affect their course. Only then can some grasp be had of the particular processes, conditions, and sequences that lead from a potentially pathogenic occurrence to the particular types of disturbance with which the clinician was in the first place concerned.
In adopting a prospective research strategy my colleagues and I early became deeply impressed by the observations of our colleague, James Robertson, who had recorded, both on
9
paper and on film, how young children in their second and third years of life respond while away from home and cared for instead in a strange place by a succession of unfamiliar people, and also how they respond during and after return home to mother ( Robertson 1952; 1953; Robertson & Bowlby 1952). During the period away, perhaps in residential nursery or hospital ward, a young child is usually acutely distressed for a time and is not easily comforted. After his return home he is likely to be either emotionally detached from his mother or else intensely clinging; as a rule a period of detachment, either brief or long depending mainly on length of separation, precedes a period during which he becomes strongly demanding of his mother's presence. Should a child then come to believe, for any reason, that there is risk of a further separation he is likely to become acutely anxious.
Reflecting on these observations we concluded that 'loss of mother figure, either by itself or in combination with other variables yet to be clearly identified, is capable of generating responses and processes that are of the greatest interest to psychopathology'. Our reason for this belief was that the responses and processes observed seemed to be the same as those found to be active in older individuals who are still disturbed by separations they have suffered in early life. These comprise, on the one hand, a tendency to make intensely strong demands on others and to be anxious and angry when they are not met, a condition common in individuals labelled neurotic; and, on the other, a blockage in the capacity to make deep relationships, such as is present in affectionless and psychopathic personalities.
From the start an important and controversial issue has been the part played in the responses of children to separation from mother by variables other than that of separation per se;
-xii-
these include illness, the strange surroundings in which a child finds himself, the kind of substitute care he receives while away, the kind of relations he has both before and after the event. It is plain that these factors can greatly intensify, or in some cases mitigate, a child's responses. Yet evidence is convincing that presence or absence of mother figure is itself a condition of the greatest significance in determining a child's emotional state. The issue is already discussed in Chapter 2 of the first volume, where a description is given of some of the relevant findings, and is taken up again in the first chapter of this one, where attention is given to the results of a foster-care project undertaken in recent years by James and Joyce Robertson in which they 'sought to create a separation situation from which many of the factors that complicate institutional studies were eliminated; and in which the emotional needs of the children would be met as far as possible by a fully available substitute mother' ( Robertson & Robertson 1971). 1 Study of the Robertsons' findings has led to some modification of views expressed in earlier publications, in which insufficient weight was given to the influence of skilled care from a familiar substitute. In parallel with the empirical studies of my colleagues, I have myself been engaged in studying the theoretical and clinical implications of the data. In particular, I have been trying to sketch a schema able to comprehend data derived from a number of distinct sources:
observations of how young children behave during periods when they are away from mother and after they return home to her;
observations of how older subjects, children and adults, behave during and after a separation from a loved figure, or after a permanent loss;
observations of difficulties found during clinical work with children and adults who, during childhood or adolescence, have either experienced a long separation or a loss or had grounds to fear one; these include various forms of acute or chronic anxiety and
10
depression, and difficulties of every degree in making and maintaining close affectional bonds, whether with parent figures, with members of the opposite sex, or with own children.
____________________
1 In addition to their written report the Robertsons have published a series of films on the
children fostered, particulars of which are given in the list of references at the end of this volume.
-xiii-
First steps towards formulating a theoretical schema were taken in a series of papers published between 1958 and 1963. The present three-volume work1 is a further attempt at a formulation.
Volume 1, Attachment, is devoted to problems originally tackled in the first paper of the series, 'The Nature of the Child's Tie to his Mother' ( 1958b). In order effectively to discuss the empirical data regarding the development of that tie and to formulate a theory to account for it, it proved necessary to discuss first the whole problem of instinctive behaviour and how best to conceptualize it. In doing so I drew heavily on findings and ideas contributed by ethologists and also on ideas derived from control theory.
This, the second volume, deals mainly with problems of separation anxiety and covers ground originally tackled in two further papers of the original series, 'Separation Anxiety' ( 1960a) and 'Separation Anxiety: A Critical Review of the Literature' ( 1961a). Once again, in order to comprehend better the problems before us -- the distress occurring during a separation and the anxiety often evident after it -- it has proved desirable first to discuss a broad range of related phenomena and theory, notably the various forms of behaviour taken to be indicative of fear and the nature of the situations that commonly elicit fear. This discussion occupies Part II of the volume; it provides a background against which are considered, in Part III, the great differences in susceptibility to fear and anxiety that are found when one individual is compared with another. Since many of the data required for the completion of this task are missing, much extrapolation is necessary and the resulting picture is patchy. In some places it can be painted in detail, in others only impressionistically. The aim is to provide clinicians and others with principles on which they can base their actions, and research workers with problems to explore and hypotheses to test.
The third volume, Loss, will deal with problems of grief and mourning and with the defensive processes to which anxiety and loss can give rise. It will comprise a revision and amplification of material first published in the remaining papers of the earlier series -- 'Grief and Mourning in Infancy and Early Childhood' ( 1960b), 'Processes of Mourning' ( 1961b), and 'Pathological Mourning and Childhood Mourning' ( 1963).
____________________
1 In the preface to the first volume I refer only to a second volume. During further work,
however, it has become apparent that a third volume will be required. -xiv-
11
Meanwhile two colleagues, Colin Murray Parkes and Peter Marris, have written books in which they approach problems of loss in a way close to my own. The books are Bereavement by Parkes ( 1972) and Loss and Change by Marris (in press).
In the preface to the first volume it was explained that the frame of reference from which I start is that of psychoanalysis. The reasons are several. The first is that my early thinking on the subject was inspired by psychoanalytic work -- my own and others'. A second is that, despite all its limitations, psychoanalysis and its derivatives remain by far the most used of any present-day approach to psychopathology and psychotherapy. A third and most important is that, whereas many of the central concepts of my schema -- object relations (better termed affectional bonds), separation anxiety, mourning, defence, trauma, sensitive periods in early life -- are the stockin-trade of psychoanalytic thinking, until the last decade or two they have been given scant attention by other behavioural disciplines.
Nevertheless, although the initial frame of reference is that of psychoanalysis, there are many ways in which the theory advanced here differs from the classical theories advanced by Freud and elaborated by his followers. A number of these differences are described already in the first chapter of the earlier volume. Others are referred to throughout the present volume, notably in Chapters 2, 5, and 16.
Note to the Paperback Edition
A paperback edition of this volume has provided the opportunity to include a number of additional notes referring to some later published work. These notes appear on pp. 409 14; the points in the text to which they relate are indicated by a dagger (? ). Details of the additional references are given on pp. 436 -7, following the original references. The index has been considerably expanded.
-xv-
Acknowledgements
IN the first volume of this work I listed the many colleagues and friends without whose help over the years these volumes could not have been written; and it is a great pleasure to express to them all once again my very warmest thanks. My debt to them is deep and lasting.
In the preparation of this volume I am indebted especially for help given by Robert Hinde, Mary Salter Ainsworth, and David Hamburg, each of whom read drafts of all or most of the material and offered a great many valuable criticisms and suggestions. James Robertson scrutinized the first chapter and proposed a number of improvements. Others who have contributed in different ways are Christoph Heinicke, Colin Murray Parkes, and Philip Crockatt. To all of them I am deeply grateful for the time and trouble they have given.
To the preparation of the script my secretary, Dorothy Southern, has again brought her customary care and enthusiasm. Library services have again been provided with unfailing efficiency by Ann Sutherland, and editorial assistance, similarly, by Rosamund Robson. The index has been prepared with great care by Lilian Rubin. To each of them my warmest thanks are due.
12
The many bodies that have supported the research for which I have been responsible at the Tavistock Institute of Human Relations since 1948 are listed in the first volume. Throughout the time that this volume has been in preparation I have been a part-time member of the External Scientific Staff of the Medical Research Council.
For permission to quote from published material, thanks are due to the publishers, authors, and others listed below. Bibliographical details of all the works cited in the text are given in the list of references at the end of the volume.
George Allen & Unwin Ltd, London, and Aldine Publishing Co. , Chicago, in respect of Four Years Old in an Urban Community by J. and E. Newson; Dr I. C. Kaufman, Dr L. A. Rosenblum, and Science in respect of 'Depression in Infant Monkeys Separated from their Mothers' (copyright 1967 by the American Association for the Advancement of Science); Methuen & Co.
-xvii-
Ltd, London, in respect of "'Attachment and Exploratory Behaviour of One-year-olds in a Strange Situation'" by M. D. S. Ainsworth and B. A. Wittig, in Determinants of Infant Behaviour, Vol. 4, edited by B. M. Foss; Dr R. F. Peck and Dr R. J. Havighurst in respect of The Psychology of Character Development, published by John Wiley & Sons, Inc. , New York; University of Chicago Press, Chicago, in respect of The Structure of Scientific Revolutions by T. S. Kuhn; University of London Press Ltd, London, in respect of Truancy by M. J. Tyerman.
Acknowledgement is due also to Tavistock Publications Ltd, London, for permission to include, in Chapter 21 of this volume, material that appears in Support, Innovation, and Autonomy edited by R. Gosling; and to the Journal of Child Psychology and Psychiatry for permission to reproduce, as the basis of Appendix I, a paper that was first published in that journal in 1961.
-xviii-
13
Part I
SECURITY, ANXIETY, AND DISTRESS
-1-
Chapter 1
Prototypes of Human Sorrow
Unhappiness in a child accumulates because he sees no end to the dark tunnel. The thirteen weeks of a term might just as well be thirteen years.
GRAHAM GREENE, A Sort of Life
Responses of young children to separation from mother 1
A generation has now passed since Dorothy Burlingham and Anna Freud recorded their experiences of caring for infants and young children in the setting of a residential nursery. In two modest booklets published during the second world war ( Burlingham & Freud 1942; 1944) they describe the immense problem of providing for young children who are out of mother's care. In particular they emphasize how impossible it is in a nursery setting to provide a child with a substitute figure who can mother him as well as his own mother can. When the Hampstead Nurseries were reorganized so that each nurse could care for her own little group of children they tell how the children became strongly possessive of their nurse and acutely jealous whenever she gave attention to another child: 'Tony (3 1/2) . . . would not allow Sister Mary to use "his" hand for handling other children. Jim (2-3) would burst into tears whenever his "own" nurse left the room. Shirley (4) would become intensely depressed and disturbed when "her" Marion was absent for some reason. '
Why, it may be asked, should these children have become so strongly possessive of their nurse and so deeply distressed whenever she was missing? Was it, as some traditionalists might suppose, that they had been spoiled by having been given too much attention and allowed too much their own way? Or
____________________
1 Although throughout this book the text refers usually to 'mother' and not to 'mother figure',
it is to be understood that in every case reference is to the person who mothers a child and to whom he becomes attached. For most children, of course, this person is also his natural mother.
-3-
was it, by contrast, that since leaving home they had been subjected to too many changes of mother figure and/or had too limited access to whoever in the nursery was acting temporarily as their mother figure? On how we answer these questions turn all our practices of child- rearing.
14
Not only did children in these nurseries become intensely possessive and jealous of their 'own' nurse but they were also unusually prone to become hostile towards her or to reject her, or else to retreat into a state of emotional detachment, as the following records illustrate:
Jim was separated from a very nice and affectionate mother at 17 months and developed well in our nursery. During his stay he formed two strong attachments to two young nurses who successively took care of him. Though he was otherwise a well adjusted, active and companionable child, his behaviour became impossible where these attachments were concerned. He was clinging, over-possessive, unwilling to be left for a minute, and continually demanded something without being able to define in any way what it was he wanted. It was no unusual sight to see Jim lie on the floor sobbing and despairing. These reactions ceased when his favourite nurse was absent even for short periods. He was then quiet and impersonal.
And finally psychoanalysis itself has begun to catch up with bowlby. The recent relational turn in psychoanalysis ( Mitchell and Aron) has made bridges between Bowlby's work and contemporary psychoanalytic thought much more compelling.
Part of what makes Bowlby's early work so inspiring thirty years later is his blend of openness and persistence. His intellectual curiosity seems to have known no bounds, and he continued to draw upon many diverse sources for ideas and conceptual tools, including child- observation, ethology, systems theory, and information-processing. Bowlby also knew from early on that he was onto something very important, and his integrity and perseverance in pursuing what mattered, despite criticism, has benefitted us all.
8
Attachment and Loss has been one of the most influential works of this century. Stephen A. Mitchell New York City, October 1999
-x-
Preface
IN the preface to the first volume of this work I describe the circumstances in which it was begun. Clinical experience of disturbed children, research into their family backgrounds, and an opportunity, in 1950, to read the literature and to discuss problems of mental health with colleagues in several countries led me, in a report commissioned by the World Health Organization, to formulate a principle: 'What is believed to be essential for mental health is that the infant and young child should experience a warm, intimate and continuous relationship with his mother (or permanent mother-substitute) in which both find satisfaction and enjoyment' (Bowlby 1951). To support this conclusion evidence was presented for believing that many forms of psychoneurosis and character disorder are to be attributed either to deprivation of maternal care or to discontinuities in a child's relationship with his mother figure.
Though the contents of the report proved controversial at the time, most of the conclusions are now accepted. What has plainly been missing, however, is an account of the processes through which the many and varied ill effects attributed to maternal deprivation or to discontinuities in the mother-child bond are brought into being. It is this gap that my colleagues and I have since striven to fill. In doing so we have adopted a research strategy that we believe is still too little exploited in the field of psychopathology.
In their day-to-day work, whether with disturbed children, disturbed adults, or disturbed families, clinicians have of necessity to view causal processes backwards, from the disturbance of today back to the events and conditions of yesterday. Though this method has yielded many valuable insights into possible pathogenic events and into the kinds of pathological process to which they appear to give rise, as a research method it has grave limitations. To complement it, a method regularly adopted in other branches of medical research is, having identified a possible pathogen, to study its effects prospectively. If the pathogen has been correctly identified and the studies of its effects in the short and long term are skilfully executed, it then becomes possible to describe the processes set in train by the pathogenic agent and also the ways by which they lead to
-xi-
the various consequent conditions. In such studies attention must be paid not only to the processes set in train by the pathogen but also to the very many conditions, internal and external to the organism, that affect their course. Only then can some grasp be had of the particular processes, conditions, and sequences that lead from a potentially pathogenic occurrence to the particular types of disturbance with which the clinician was in the first place concerned.
In adopting a prospective research strategy my colleagues and I early became deeply impressed by the observations of our colleague, James Robertson, who had recorded, both on
9
paper and on film, how young children in their second and third years of life respond while away from home and cared for instead in a strange place by a succession of unfamiliar people, and also how they respond during and after return home to mother ( Robertson 1952; 1953; Robertson & Bowlby 1952). During the period away, perhaps in residential nursery or hospital ward, a young child is usually acutely distressed for a time and is not easily comforted. After his return home he is likely to be either emotionally detached from his mother or else intensely clinging; as a rule a period of detachment, either brief or long depending mainly on length of separation, precedes a period during which he becomes strongly demanding of his mother's presence. Should a child then come to believe, for any reason, that there is risk of a further separation he is likely to become acutely anxious.
Reflecting on these observations we concluded that 'loss of mother figure, either by itself or in combination with other variables yet to be clearly identified, is capable of generating responses and processes that are of the greatest interest to psychopathology'. Our reason for this belief was that the responses and processes observed seemed to be the same as those found to be active in older individuals who are still disturbed by separations they have suffered in early life. These comprise, on the one hand, a tendency to make intensely strong demands on others and to be anxious and angry when they are not met, a condition common in individuals labelled neurotic; and, on the other, a blockage in the capacity to make deep relationships, such as is present in affectionless and psychopathic personalities.
From the start an important and controversial issue has been the part played in the responses of children to separation from mother by variables other than that of separation per se;
-xii-
these include illness, the strange surroundings in which a child finds himself, the kind of substitute care he receives while away, the kind of relations he has both before and after the event. It is plain that these factors can greatly intensify, or in some cases mitigate, a child's responses. Yet evidence is convincing that presence or absence of mother figure is itself a condition of the greatest significance in determining a child's emotional state. The issue is already discussed in Chapter 2 of the first volume, where a description is given of some of the relevant findings, and is taken up again in the first chapter of this one, where attention is given to the results of a foster-care project undertaken in recent years by James and Joyce Robertson in which they 'sought to create a separation situation from which many of the factors that complicate institutional studies were eliminated; and in which the emotional needs of the children would be met as far as possible by a fully available substitute mother' ( Robertson & Robertson 1971). 1 Study of the Robertsons' findings has led to some modification of views expressed in earlier publications, in which insufficient weight was given to the influence of skilled care from a familiar substitute. In parallel with the empirical studies of my colleagues, I have myself been engaged in studying the theoretical and clinical implications of the data. In particular, I have been trying to sketch a schema able to comprehend data derived from a number of distinct sources:
observations of how young children behave during periods when they are away from mother and after they return home to her;
observations of how older subjects, children and adults, behave during and after a separation from a loved figure, or after a permanent loss;
observations of difficulties found during clinical work with children and adults who, during childhood or adolescence, have either experienced a long separation or a loss or had grounds to fear one; these include various forms of acute or chronic anxiety and
10
depression, and difficulties of every degree in making and maintaining close affectional bonds, whether with parent figures, with members of the opposite sex, or with own children.
____________________
1 In addition to their written report the Robertsons have published a series of films on the
children fostered, particulars of which are given in the list of references at the end of this volume.
-xiii-
First steps towards formulating a theoretical schema were taken in a series of papers published between 1958 and 1963. The present three-volume work1 is a further attempt at a formulation.
Volume 1, Attachment, is devoted to problems originally tackled in the first paper of the series, 'The Nature of the Child's Tie to his Mother' ( 1958b). In order effectively to discuss the empirical data regarding the development of that tie and to formulate a theory to account for it, it proved necessary to discuss first the whole problem of instinctive behaviour and how best to conceptualize it. In doing so I drew heavily on findings and ideas contributed by ethologists and also on ideas derived from control theory.
This, the second volume, deals mainly with problems of separation anxiety and covers ground originally tackled in two further papers of the original series, 'Separation Anxiety' ( 1960a) and 'Separation Anxiety: A Critical Review of the Literature' ( 1961a). Once again, in order to comprehend better the problems before us -- the distress occurring during a separation and the anxiety often evident after it -- it has proved desirable first to discuss a broad range of related phenomena and theory, notably the various forms of behaviour taken to be indicative of fear and the nature of the situations that commonly elicit fear. This discussion occupies Part II of the volume; it provides a background against which are considered, in Part III, the great differences in susceptibility to fear and anxiety that are found when one individual is compared with another. Since many of the data required for the completion of this task are missing, much extrapolation is necessary and the resulting picture is patchy. In some places it can be painted in detail, in others only impressionistically. The aim is to provide clinicians and others with principles on which they can base their actions, and research workers with problems to explore and hypotheses to test.
The third volume, Loss, will deal with problems of grief and mourning and with the defensive processes to which anxiety and loss can give rise. It will comprise a revision and amplification of material first published in the remaining papers of the earlier series -- 'Grief and Mourning in Infancy and Early Childhood' ( 1960b), 'Processes of Mourning' ( 1961b), and 'Pathological Mourning and Childhood Mourning' ( 1963).
____________________
1 In the preface to the first volume I refer only to a second volume. During further work,
however, it has become apparent that a third volume will be required. -xiv-
11
Meanwhile two colleagues, Colin Murray Parkes and Peter Marris, have written books in which they approach problems of loss in a way close to my own. The books are Bereavement by Parkes ( 1972) and Loss and Change by Marris (in press).
In the preface to the first volume it was explained that the frame of reference from which I start is that of psychoanalysis. The reasons are several. The first is that my early thinking on the subject was inspired by psychoanalytic work -- my own and others'. A second is that, despite all its limitations, psychoanalysis and its derivatives remain by far the most used of any present-day approach to psychopathology and psychotherapy. A third and most important is that, whereas many of the central concepts of my schema -- object relations (better termed affectional bonds), separation anxiety, mourning, defence, trauma, sensitive periods in early life -- are the stockin-trade of psychoanalytic thinking, until the last decade or two they have been given scant attention by other behavioural disciplines.
Nevertheless, although the initial frame of reference is that of psychoanalysis, there are many ways in which the theory advanced here differs from the classical theories advanced by Freud and elaborated by his followers. A number of these differences are described already in the first chapter of the earlier volume. Others are referred to throughout the present volume, notably in Chapters 2, 5, and 16.
Note to the Paperback Edition
A paperback edition of this volume has provided the opportunity to include a number of additional notes referring to some later published work. These notes appear on pp. 409 14; the points in the text to which they relate are indicated by a dagger (? ). Details of the additional references are given on pp. 436 -7, following the original references. The index has been considerably expanded.
-xv-
Acknowledgements
IN the first volume of this work I listed the many colleagues and friends without whose help over the years these volumes could not have been written; and it is a great pleasure to express to them all once again my very warmest thanks. My debt to them is deep and lasting.
In the preparation of this volume I am indebted especially for help given by Robert Hinde, Mary Salter Ainsworth, and David Hamburg, each of whom read drafts of all or most of the material and offered a great many valuable criticisms and suggestions. James Robertson scrutinized the first chapter and proposed a number of improvements. Others who have contributed in different ways are Christoph Heinicke, Colin Murray Parkes, and Philip Crockatt. To all of them I am deeply grateful for the time and trouble they have given.
To the preparation of the script my secretary, Dorothy Southern, has again brought her customary care and enthusiasm. Library services have again been provided with unfailing efficiency by Ann Sutherland, and editorial assistance, similarly, by Rosamund Robson. The index has been prepared with great care by Lilian Rubin. To each of them my warmest thanks are due.
12
The many bodies that have supported the research for which I have been responsible at the Tavistock Institute of Human Relations since 1948 are listed in the first volume. Throughout the time that this volume has been in preparation I have been a part-time member of the External Scientific Staff of the Medical Research Council.
For permission to quote from published material, thanks are due to the publishers, authors, and others listed below. Bibliographical details of all the works cited in the text are given in the list of references at the end of the volume.
George Allen & Unwin Ltd, London, and Aldine Publishing Co. , Chicago, in respect of Four Years Old in an Urban Community by J. and E. Newson; Dr I. C. Kaufman, Dr L. A. Rosenblum, and Science in respect of 'Depression in Infant Monkeys Separated from their Mothers' (copyright 1967 by the American Association for the Advancement of Science); Methuen & Co.
-xvii-
Ltd, London, in respect of "'Attachment and Exploratory Behaviour of One-year-olds in a Strange Situation'" by M. D. S. Ainsworth and B. A. Wittig, in Determinants of Infant Behaviour, Vol. 4, edited by B. M. Foss; Dr R. F. Peck and Dr R. J. Havighurst in respect of The Psychology of Character Development, published by John Wiley & Sons, Inc. , New York; University of Chicago Press, Chicago, in respect of The Structure of Scientific Revolutions by T. S. Kuhn; University of London Press Ltd, London, in respect of Truancy by M. J. Tyerman.
Acknowledgement is due also to Tavistock Publications Ltd, London, for permission to include, in Chapter 21 of this volume, material that appears in Support, Innovation, and Autonomy edited by R. Gosling; and to the Journal of Child Psychology and Psychiatry for permission to reproduce, as the basis of Appendix I, a paper that was first published in that journal in 1961.
-xviii-
13
Part I
SECURITY, ANXIETY, AND DISTRESS
-1-
Chapter 1
Prototypes of Human Sorrow
Unhappiness in a child accumulates because he sees no end to the dark tunnel. The thirteen weeks of a term might just as well be thirteen years.
GRAHAM GREENE, A Sort of Life
Responses of young children to separation from mother 1
A generation has now passed since Dorothy Burlingham and Anna Freud recorded their experiences of caring for infants and young children in the setting of a residential nursery. In two modest booklets published during the second world war ( Burlingham & Freud 1942; 1944) they describe the immense problem of providing for young children who are out of mother's care. In particular they emphasize how impossible it is in a nursery setting to provide a child with a substitute figure who can mother him as well as his own mother can. When the Hampstead Nurseries were reorganized so that each nurse could care for her own little group of children they tell how the children became strongly possessive of their nurse and acutely jealous whenever she gave attention to another child: 'Tony (3 1/2) . . . would not allow Sister Mary to use "his" hand for handling other children. Jim (2-3) would burst into tears whenever his "own" nurse left the room. Shirley (4) would become intensely depressed and disturbed when "her" Marion was absent for some reason. '
Why, it may be asked, should these children have become so strongly possessive of their nurse and so deeply distressed whenever she was missing? Was it, as some traditionalists might suppose, that they had been spoiled by having been given too much attention and allowed too much their own way? Or
____________________
1 Although throughout this book the text refers usually to 'mother' and not to 'mother figure',
it is to be understood that in every case reference is to the person who mothers a child and to whom he becomes attached. For most children, of course, this person is also his natural mother.
-3-
was it, by contrast, that since leaving home they had been subjected to too many changes of mother figure and/or had too limited access to whoever in the nursery was acting temporarily as their mother figure? On how we answer these questions turn all our practices of child- rearing.
14
Not only did children in these nurseries become intensely possessive and jealous of their 'own' nurse but they were also unusually prone to become hostile towards her or to reject her, or else to retreat into a state of emotional detachment, as the following records illustrate:
Jim was separated from a very nice and affectionate mother at 17 months and developed well in our nursery. During his stay he formed two strong attachments to two young nurses who successively took care of him. Though he was otherwise a well adjusted, active and companionable child, his behaviour became impossible where these attachments were concerned. He was clinging, over-possessive, unwilling to be left for a minute, and continually demanded something without being able to define in any way what it was he wanted. It was no unusual sight to see Jim lie on the floor sobbing and despairing. These reactions ceased when his favourite nurse was absent even for short periods. He was then quiet and impersonal.
Reggie, who had come to our house as a baby of 5 months, went home to his mother when he was 1 year 8 months, and has been with us ever since his return to the nursery 2 months later. While with us, he formed two passionate relationships to two young nurses who took care of him at different periods. The second attachment was suddenly broken at 2 years 8 months when his 'own' nurse married. He was completely lost and desperate after her departure, and refused to look at her when she visited him a fortnight later. He turned his head to the other side when she spoke to him, but stared at the door, which had closed behind her, after she had left the room. In the evening in bed he sat up and said: 'My very own Mary-Ann! But I don't like her. '
These observations, made in the pressure of wartime and recorded anecdotally with all too little detail, none the less cast a shaft of light on the nature of many forms of psychiatric disturbance. States of anxiety and depression that occur during
-4-
adult years, and also psychopathic conditions, can, it is held, be linked in a systematic way to the states of anxiety, despair, and detachment described by Burlingham & Freud, and subsequently by others, that are so readily engendered whenever a young child is separated for long from his mother figure, whenever he expects such a separation, and when, as sometimes happens, he loses her altogether. Whereas during later life it is often extremely difficult to trace how a person's disturbed emotional state is related to his experiences, whether they be those of his current life or those of his past, during the early years of childhood the relationship between emotional state and current or recent experience is often crystal clear. In these troubled states of early childhood, it is held, can be discerned the prototype of many a pathological condition of later years.
It is, of course, a commonplace that most children who have had experiences of these kinds recover and resume normal development, or at least they appear to do so. Not infrequently, therefore, doubts are expressed whether the psychological processes described are in reality related so intimately to personality disturbances of later life. Pending much further evidence, these are legitimate doubts. Nevertheless, reasons for holding to the thesis are strong. One is that data from many sources can be arranged and organized into a pattern that is internally consistent and consistent also with current biological theory. Another is that many clinicians and social workers find the resulting schema enables them to understand better the problems with which they are grappling and so to help their patients or clients more effectively.
15
Why some individuals should recover, largely or completely, from experiences of separation and loss while others seem not to is a central question, but one not easily answered. In living creatures variation of response is the rule and its explanation is often hard to fathom. Of all those who contract poliomyelitis less than 1 per cent develop paralysis, and only a fraction of 1 per cent remain crippled. Why one person should respond one way and another another remains obscure. To argue that, because 99 per cent recover, polio is a harmless infection would obviously be absurd. Similarly, in the field under consideration, to argue that because most individuals recover from the effects of a separation or loss these experiences are of no account would be equally absurd.
Nevertheless the problem of differential response remains
-5-
important. Conditions likely to be playing a part can be considered under two main heads: those intrinsic to or closely associated with the separation itself, notably the conditions in which a child is cared for while away from mother;
those present in the child's life over a longer period, notably his relations with parents during the months or years before and after the event.
Here we consider variables in category (a). Those in category (b) are discussed in the later chapters of Part III.
We start by reviewing observations of how children behave when cared for in one of two very different settings. The first is an ordinary residential nursery, in which a child finds himself in a strange place with strange people none of whom is sufficiently available to give him more than very limited mothering. The second is a foster home in which a child receives the fulltime and skilled care of a foster mother with whom he has become in some degree familiar beforehand.
Conditions leading to intense responses
In our early studies children were observed during stays in institutional settings and it was on the basis of these observations that the sequence of responses which we term protest, despair, and detachment came first to be delineated ( Robertson & Bowlby 1952). Since then two further studies have been conducted by colleagues in the Tavistock Child Development Research Unit, the first by Christoph Heinicke ( 1956) and the second by Heinicke & Ilse Westheimer ( 1966). Although in each of these investigations only a handful of children were observed (six in the first and ten in the second), the studies are unique for the care of their design and the amount of systematic observation. Moreover, for each sample of separated children a contrast group was selected and observed: in the first study it was a fairly well- matched group of children observed during their first weeks of attendance at a day nursery; in the second it was a similarly matched group of children observed while living in their own homes. Heinicke & Westheimer treat their data statistically and also describe in some detail the behaviour of individual children.
In the larger investigation ( 1966), work was conducted in three residential nurseries. Arrangements and facilities were fairly similar. In each, a child belonged to a defined group of
16
-6-
children and was cared for mainly by one or two nurses. Ample opportunities were available for free play either in large rooms or outdoors in a garden. Before a child entered the nursery, contact was made with the family by a psychiatric social worker ( Ilse Westheimer), who was also responsible either then or later for collecting full information about the family and the child. Arrival at the nursery was observed; and in the course of his stay a child was observed during free play on six occasions each week. Each of the two observers (one male, Christoph Heinicke, and one female, Elizabeth Wolpert) observed for a period of at least half an hour during each of the three sample periods into which the week was broken (Monday and Tuesday; Wednesday and Thursday; Friday, Saturday, and Sunday). The method used, of categorizing behavioural units in terms of agent, object, relation, mode, and intensity, had been used in the earlier study and had been shown to be reliable.
In addition to the categorized observations of free behaviour, similarly categorized observations were made of every child's behaviour in standardized doll-play sessions; and a number of other records of each child's stay in the nursery were kept.
It was originally intended to select the separated children in accordance with the five criteria used in the first study, namely: (i) that the child had had no previous separations of more than three days, (ii) that he fell within the age-limits of fifteen to thirty months, (iii) that he did not enter the nursery with a sibling, (iv) that he was living with both mother and father at the time the separation occurred, and (v) that there was no evidence that being placed in a nursery indicated a rejection by his parents. Because of the difficulty of obtaining cases, however, these criteria had to be modified to allow greater latitude.
Although most of the children had had either no separations or only very brief ones prior to the one being studied, in one case the length of previous separation was four weeks and in two it was three weeks. The age-range was slightly extended and ran from thirteen to thirty-two months, instead of from fifteen to thirty months. But the most marked departure from the previous criteria was that four of the children entered the residential nursery in the company of a sibling: in three cases this was a four-year-old sibling and in one case the sibling was younger. The remaining two criteria remained unmodified: each of the children was living with both mother and father at the time of separation, and there were no indications that he
-7-
was being rejected by the parents by being placed in the nursery.
The reason that the ten children studied were cared for in a residential nursery was that, in a family emergency, neither relatives nor friends were available to take temporary care of them. In the case of seven families, mother was to be away in hospital having a new baby. In two others mother was to be in hospital for some other form of medical attention. In the tenth case, the family became homeless.
Among much else in their book, Brief Separations ( 1966), Heinicke & Westheimer describe behaviour typical of the ten children during their time in the nursery, and, similarly, behaviour typical of the children after they had returned home. In the paragraphs that follow some of their principal findings are presented. Every one of the patterns reported had also been
17
observed and recorded by Robertson during his earlier, less systematic though more extensive, studies.
Behaviour during Separation
The children arrived at the nursery in the care of one or both parents. Four of them, brought by father, stayed close to him and seemed already subdued and anxious. Some of the others, who had come with mother or both parents, seemed more confident and were ready to explore the new environment. They ventured forth on short or long excursions and then returned.
When the moment came for the parent (s) to depart, crying or screaming was the rule. One child tried to follow her parents, demanding urgently where they were going, and finally had to be pushed back into the room by her mother. Another threw himself on the floor and refused to be comforted. Altogether eight of the children were crying loudly soon after their parents' departure. Bedtime was also an occasion for tears. The two who had not cried earlier screamed when put in a cot and could not be consoled. Some of the others whose initial crying had ceased broke into renewed sobs at bedtime.
Attachment and Loss
VOLUME II SEPARATION ANXIETY AND ANGER
John Bowlby
With Additional Notes by the Author
BASIC BOOKS
A Member of the Perseus Books Group
1
To THREE FRIENDS
Evan Durbin Eric Trist Robert Hinde
Copyright (C) 1973 by The Tavistock Institute of Human Relations Library of Congress Catalog Card Number: 70-78464
ISBN: 0-465-07691-2 Cloth
ISBN: 0-465-09716-2 Paper
Printed in the United States of America 99 RRD-H 30 29 28 27 26 25 24 -ii-
2
Contents
Foreword page vii Preface xi Acknowledgements xvii
PART I: SECURITY, ANXIETY, AND DISTRESS 1 Prototypes of Human Sorrow 3
Responses of young children to separation from mother 3 Conditions leading to intense responses 6 Conditions mitigating the intensity of responses 16 Presence or absence of mother figurer: a key variable 22
2 The Place of Separation and Loss in Psychopathology 25 Problem and perspective 25 Separation anxiety and other forms of anxiety 30 A challenge for theory 30
3 Behaviour with and without Mother: Humans 33 Naturalistic observations 33 Experimental Studies 39 Ontogeny of responses to separation 52
4 Behaviour with and without Mother: Non-human Primates 57 Naturalistic observations 57 Early experimental studies 60 Further studies by Hinde and Spencer-Booth 69
PART II: AN ETHOLOGICAL APPROACH TO
HUMAN FEAR
5 Basic Postulates in Theories of Anxiety and Fear 77
Anxiety allied to fear 77 Models of motivation and their effects on theory 79 Puzzling phobia or natural fear 83
6 Forms of Behaviour Indicative of Fear 87 An empirical approach 87 Withdrawal behaviour and attachment behaviour 89 Feeling afraid and its variants: feeling alarmed and feeling anxious 92
-iii-
7 Situations that Arouse Fear in Humans 96 A difficult field of study 96 Fear-arousing situations: the first year 99 Fear-arousing situations: the second and later years 105 Compound situations 118 Fear behaviour and the development of attachment 119 Natural clues to potential danger 124 Fear behaviour of non-human primates 127 Compound situations 134 Fear, attack, and exploration 136
3
9 Natural Clues to Danger and Safety 138
10 Natural Clues, Cultural Clues, and the Assessment of Danger 151
Clues of three kinds 151 Real danger: difficulties of assessment 153 'Imaginary' dangers 156 Cultural clues learnt from others 158 Continuing role of the natural clues 161 Behaviour in disaster 166
11 Rationalization, Misattribution, and Projection 169 Difficulties in identifying situations that arouse fear 169 Misattribution and the role of projection 172 The case of Schreber: a re-examination 174
12 Fear of Separation 178
Better safe than sorry
Potential danger of being alone
Potential safety of familiar companions and environment Maintaining a stable relationship with the familiar environment: a form of homeostasis 148
138 142 146
Hypotheses regarding its development Need for two terminologies
PART III: INDIVIDUAL DIFFERENCES IN
SUSCEPTIBILITY TO FEAR: ANXIOUS
178 182
-iv-
ATTACHMENT
13 Some Variables responsible for Individual Differences 187
Constitutional variables 187
Experiences and processes that reduce susceptibility to fear 191
Experiences and processes that increase susceptibility to fear196 14 Susceptibility to Fear and the Availability of Attachment Figures 201
Forecasting the availability of an attachment figure 201 Working models of attachment figures and of self 203 The role of experience in determining working models 207 A note on use of the terms 'mature' and 'immature' 209
15 Anxious Attachment and Some Conditions that Promote it 211 'Overdependency' or anxious attachment 211 Anxious attachment of children reared without a permanent mother figure 215 Anxious attachment after a period of separation or of daily substitute care 220 Anxious attachment following threats of abandonment or suicide
226 16 'Overdependency' and the Theory of Spoiling 237
Some contrasting theories 237
Studies of 'overdependency' and its antecedents 240 17 Anger, Anxiety, and Attachment 245
4
Anger: a response to separation 245 Anger: functional and dysfunctional 246 Anger, ambivalence, and anxiety 253
78 Anxious Attachment and the 'Phobias' of Childhood 258 Phobia, pseudophobia, and anxiety state 258 'School phobia' or school refusal 261 Two classical cases of childhood phobia: a reappraisal 283 Animal phobias in childhood 289
19 Anxious Attachment and 'Agoraphobia' 292 Symptomatology and theories of 'agoraphobia' 292 Pathogenic patterns of family interaction 299 'Agoraphobia', bereavement, and depression 309 A note on response to treatment 310
20 Omission, Suppression, and Falsification of Family Context 313 -v-
21 Secure Attachment and the Growth of Self-reliance 322 Personality development and family experience 322 Studies of adolescents and young adults 328 Studies of young children 330 Self-reliance and reliance on others 359
22 Pathways for the Growth of Personality 363 The nature of individual variation: alternative models 363 Developmental pathways and homeorhesis 366 One person's pathway: some determinants 369
APPENDICES
I Separation Anxiety: Review of Literature 375 II Psychoanalysis and Evolution Theory 399 III Problems of Terminology 404 Additional Notes 409 References 415 Additional References 436 Index 439
-vi-
5
Foreword
Truly revolutionary works that open up new conceptual vistas tend to straddle intellectual worlds. Their originality makes them difficult to locate in the historical context in which they are written, and their pervasive impact makes it difficult to appreciate just how formative they were in generating subsequent developments that have come to take on the quality of great obviousness.
John Bowlby was a psychoanalyst; he notes in his original preface that psychoanalysis provided him his first inspiration as well as the only serviceable vocabulary for writing about intense human experiences, trauma and separation. Retrospectively, we might imagine that his work on the powerful emotional bonds between children and their mothers would have been welcomed within the psychoanalytic world. It was not. It was certainly not ignored, however; the persuasiveness of Bowlby's presentation made that impossible.
Bowlby's initial work on attachment was given an airing in the most influential psychoanalytic publications of his day and was then attacked by the most prestigious analytic authorities. That he was virtually exiled by the psychoanalytic community was, ultimately, fortuitous. It led to the establishment around Bowlby's work of a research tradition into the nature of emotional attachment, separation and loss that has generated some of the most remarkable, reliable and provocative empirical data of the past fifty years.
What was it that got Bowlby into so much trouble? He introduced the project as a working out of the implications of observations of the responses to loss young children. It was Bowlby's commitment to the importance of his observations that was the problem. The psychoanalysis of Bowlby's day was an interpretive system that was almost hermetically sealed. Freud had provided the Rosetta Stone for deciphering the underlying structures of mind, and Melanie Klein had extended his psychic excavations to the deepest recesses. Armed with their decoding devices, psychoanalysts felt they could see beneath the superficial, behavioral levels of human interaction to the underlying instinctual conflicts and fantasies that generated their deeper meanings.
Bowlby had been asked by the World Health Organization in 1950 to study and offer advice on the plight of homeless children, real children in the real world. The ravaging impact of maternal deprivations and separations made an enormous impression on
-vii-
him -- to his surprise. It is a tribute to how much Bowlby's work has changed our consciousness that we must strain to imagine why he would have experienced the magnitude of such an impact as a surprise.
In the understanding of mid-twentieth century experts, small children had basic physical needs that required tending; the complex emotional relationships involving a unique sense of interpersonal connection evolved only later. In the conditioning paradigms that dominated American psychology in those days, the caregiver was a "secondary-reinforcer," who became important to the child only by virtue of being associatively linked with physical ministrations. And in the language of the prevailing psychoanalytic theory the mother was, similarly, a "need-gratifying object," whose significance developed gradually through her role in
6
satisfying drive pressures. Separations from mothers shouldn't matter terribly much, as long as the child's needs were taken care of.
But separations did matter, Bowlby discovered. Observations of actual children suggested that maternal deprivation was extremely traumatic. Bowlby s position was a little like that of the wife in the old story who discovers her philandering husband in bed with another woman. "Darling" pleads the husband, "who are you going to believe, me or your lying eyes? " Bowlby decided to believe his eyes.
Any good psychological theory applied cleverly can account for virtually any data. Surely the instinct theory of Bowlby's day could be stretched to accommodate his observations. But for Bowlby, the center of gravity of the data on trauma in real children in the real world did not match up with the center of gravity of Freud's drive theory, privileging impersonal, body- based needs and fantasies. So, as Bowlby announces in this work, a new instinct theory was required, in which the powerful emotional attachment between child and mother is understood not as derivative of more basic processes but as fundamental in itself.
This was no minor addition; it challenged psychoanalytic metapsychology at its core. Bowlby liked to recount experiences formative for him as a student listening to case presentations that emphasized unconscious, instinct-based fantasies at the British Psychoanalytic Society; he remembered one conference at which he felt moved to rise and state emphatically, "But there is such a thing as a bad mother! "
Bowlby was not working completely alone; he had conceptual fellow travelers, especially in the Interpersonal Psychiatry of the American Harry Stack Sullivan and the psychoanalytic contributions of other innovators like W. R. D. Fairbairn, Donald Winni-
-viii-
cott and Hans Loewald. Part of what drew so much fire in Bowlby's direction, however, was his clarity. Sullivan was a tortured, blocked writer. Fairbairn was often tedious and difficult. Winnicott was poetic and elusive. Loewald was extremely subtle and often obscure. Bowlby wrote with lucidity and power.
It was amply apparent from the very beginning and throughout that Bowlby regarded his contributions as a direct challenge to certain basic tenets of Freudian theory. And he had data on children in the real world to back them up. And he identified himself very much as a scientist, offering testable hypotheses. And, his links with other scientists, especially the ethologists of his day, made his position extremely persuasive. For the psychoanalytic establishment of the time, this was simply too much to bear. Bowlby became much more interested than the average psychoanalyst in what actually goes on between people in the real world, and the neighboring discipline of ethology provided powerful explanatory concepts for understanding what he had been observing in children's reactions to separation and loss.
Both Freud and Bowlby were extremely involved with Darwin's contribution (one of Bowlby's last works was a biography of Darwin), but their Darwins were very different. Freud's Darwin was part of the first wave of reaction to the extraordinary implications of the theory of evolution; one of Freud's projects was to work out the implications for human psychology of Darwin's demonstration of the continuity between, so-called lower and so- called higher forms of animal life. Freud's fascination with primitivism and his reliance on
7
bestial metaphors are thematic throughout his writings. And Freud's structural model of the psyche is a re-creation, on a microcosmic level, of Darwin's sweeping account of the evolution of species: lower level, primitive energy of the id is transformed by the reality- oriented ego into higher level, aim-inhibited resources for activities consistent with the cultural values of the superego. Ontogeny recapitulates phylogeny.
Bowlby draws on a different Darwin. Like the Ego psychologist Heinz Hartmann, Bowlby was most interested in what Darwin taught about animal adaptation to environmental conditions and niches. In this second volume of his Attachment and Loss trilogy, Bowlby actually refers to Freud as pre-Darwinian because Freud did not grasp the importance of the principle of "natural selection" in Darwin's theory of the evolution of species. Bowlby, like Darwin, was interested in what animals do to maximize their chances for survival. Whereas Freud's Darwin lent himself to the study of internality and unconscious, primitive states, Bowlby's Darwin lent himself
-ix-
to a behavioral analysis of what small children and mothers actually do with each other. Thus, among the most vivid of Bowlby's contributions is his account of the five component instincts that insure the baby's proximity to the mother--the underpinnings for both healthy attachment and traumatic separations and loss.
One advantage of this behavioral emphasis has been that Bowlby's ideas have been applied, with extraordinary effectiveness, to the empirical research tradition that Mary Ainsworth and Mary Main have done so much to develop. Another advantage is the ease with which Bowlby's observations have been adopted by popular pediatric practice and social policy planners to prioritize the personal, emotional dimensions of childcare. Despite our difficulties as a society in investing the resources necessary for psychologically healthy child- evelopment, we have come to take as axiomatic the principle of "bonding" between parents and children and the importance of emotional warmth and security in caregiving. Bowlby's work was central to this enriched understanding.
Until recently, the disadvantage of Bowlby's behavioral emphasis has been the relative underdevelopment of the psychodynamic dimension within attachment theory, which has made the bridge to other psychoanalytic theorizing more difficult. Bowlby's concept of "working models" had an overly schematic, mechanistic feel to it, which lacked the richness of psychoanalytic investigations of the inner world. But the more recent attachment literature (Fonagy) has taken a more inward turn in exploring the concomitants of secure and insecure attachments in the textures of conscious and unconscious subjectivity. And finally psychoanalysis itself has begun to catch up with bowlby. The recent relational turn in psychoanalysis ( Mitchell and Aron) has made bridges between Bowlby's work and contemporary psychoanalytic thought much more compelling.
Part of what makes Bowlby's early work so inspiring thirty years later is his blend of openness and persistence. His intellectual curiosity seems to have known no bounds, and he continued to draw upon many diverse sources for ideas and conceptual tools, including child- observation, ethology, systems theory, and information-processing. Bowlby also knew from early on that he was onto something very important, and his integrity and perseverance in pursuing what mattered, despite criticism, has benefitted us all.
8
Attachment and Loss has been one of the most influential works of this century. Stephen A. Mitchell New York City, October 1999
-x-
Preface
IN the preface to the first volume of this work I describe the circumstances in which it was begun. Clinical experience of disturbed children, research into their family backgrounds, and an opportunity, in 1950, to read the literature and to discuss problems of mental health with colleagues in several countries led me, in a report commissioned by the World Health Organization, to formulate a principle: 'What is believed to be essential for mental health is that the infant and young child should experience a warm, intimate and continuous relationship with his mother (or permanent mother-substitute) in which both find satisfaction and enjoyment' (Bowlby 1951). To support this conclusion evidence was presented for believing that many forms of psychoneurosis and character disorder are to be attributed either to deprivation of maternal care or to discontinuities in a child's relationship with his mother figure.
Though the contents of the report proved controversial at the time, most of the conclusions are now accepted. What has plainly been missing, however, is an account of the processes through which the many and varied ill effects attributed to maternal deprivation or to discontinuities in the mother-child bond are brought into being. It is this gap that my colleagues and I have since striven to fill. In doing so we have adopted a research strategy that we believe is still too little exploited in the field of psychopathology.
In their day-to-day work, whether with disturbed children, disturbed adults, or disturbed families, clinicians have of necessity to view causal processes backwards, from the disturbance of today back to the events and conditions of yesterday. Though this method has yielded many valuable insights into possible pathogenic events and into the kinds of pathological process to which they appear to give rise, as a research method it has grave limitations. To complement it, a method regularly adopted in other branches of medical research is, having identified a possible pathogen, to study its effects prospectively. If the pathogen has been correctly identified and the studies of its effects in the short and long term are skilfully executed, it then becomes possible to describe the processes set in train by the pathogenic agent and also the ways by which they lead to
-xi-
the various consequent conditions. In such studies attention must be paid not only to the processes set in train by the pathogen but also to the very many conditions, internal and external to the organism, that affect their course. Only then can some grasp be had of the particular processes, conditions, and sequences that lead from a potentially pathogenic occurrence to the particular types of disturbance with which the clinician was in the first place concerned.
In adopting a prospective research strategy my colleagues and I early became deeply impressed by the observations of our colleague, James Robertson, who had recorded, both on
9
paper and on film, how young children in their second and third years of life respond while away from home and cared for instead in a strange place by a succession of unfamiliar people, and also how they respond during and after return home to mother ( Robertson 1952; 1953; Robertson & Bowlby 1952). During the period away, perhaps in residential nursery or hospital ward, a young child is usually acutely distressed for a time and is not easily comforted. After his return home he is likely to be either emotionally detached from his mother or else intensely clinging; as a rule a period of detachment, either brief or long depending mainly on length of separation, precedes a period during which he becomes strongly demanding of his mother's presence. Should a child then come to believe, for any reason, that there is risk of a further separation he is likely to become acutely anxious.
Reflecting on these observations we concluded that 'loss of mother figure, either by itself or in combination with other variables yet to be clearly identified, is capable of generating responses and processes that are of the greatest interest to psychopathology'. Our reason for this belief was that the responses and processes observed seemed to be the same as those found to be active in older individuals who are still disturbed by separations they have suffered in early life. These comprise, on the one hand, a tendency to make intensely strong demands on others and to be anxious and angry when they are not met, a condition common in individuals labelled neurotic; and, on the other, a blockage in the capacity to make deep relationships, such as is present in affectionless and psychopathic personalities.
From the start an important and controversial issue has been the part played in the responses of children to separation from mother by variables other than that of separation per se;
-xii-
these include illness, the strange surroundings in which a child finds himself, the kind of substitute care he receives while away, the kind of relations he has both before and after the event. It is plain that these factors can greatly intensify, or in some cases mitigate, a child's responses. Yet evidence is convincing that presence or absence of mother figure is itself a condition of the greatest significance in determining a child's emotional state. The issue is already discussed in Chapter 2 of the first volume, where a description is given of some of the relevant findings, and is taken up again in the first chapter of this one, where attention is given to the results of a foster-care project undertaken in recent years by James and Joyce Robertson in which they 'sought to create a separation situation from which many of the factors that complicate institutional studies were eliminated; and in which the emotional needs of the children would be met as far as possible by a fully available substitute mother' ( Robertson & Robertson 1971). 1 Study of the Robertsons' findings has led to some modification of views expressed in earlier publications, in which insufficient weight was given to the influence of skilled care from a familiar substitute. In parallel with the empirical studies of my colleagues, I have myself been engaged in studying the theoretical and clinical implications of the data. In particular, I have been trying to sketch a schema able to comprehend data derived from a number of distinct sources:
observations of how young children behave during periods when they are away from mother and after they return home to her;
observations of how older subjects, children and adults, behave during and after a separation from a loved figure, or after a permanent loss;
observations of difficulties found during clinical work with children and adults who, during childhood or adolescence, have either experienced a long separation or a loss or had grounds to fear one; these include various forms of acute or chronic anxiety and
10
depression, and difficulties of every degree in making and maintaining close affectional bonds, whether with parent figures, with members of the opposite sex, or with own children.
____________________
1 In addition to their written report the Robertsons have published a series of films on the
children fostered, particulars of which are given in the list of references at the end of this volume.
-xiii-
First steps towards formulating a theoretical schema were taken in a series of papers published between 1958 and 1963. The present three-volume work1 is a further attempt at a formulation.
Volume 1, Attachment, is devoted to problems originally tackled in the first paper of the series, 'The Nature of the Child's Tie to his Mother' ( 1958b). In order effectively to discuss the empirical data regarding the development of that tie and to formulate a theory to account for it, it proved necessary to discuss first the whole problem of instinctive behaviour and how best to conceptualize it. In doing so I drew heavily on findings and ideas contributed by ethologists and also on ideas derived from control theory.
This, the second volume, deals mainly with problems of separation anxiety and covers ground originally tackled in two further papers of the original series, 'Separation Anxiety' ( 1960a) and 'Separation Anxiety: A Critical Review of the Literature' ( 1961a). Once again, in order to comprehend better the problems before us -- the distress occurring during a separation and the anxiety often evident after it -- it has proved desirable first to discuss a broad range of related phenomena and theory, notably the various forms of behaviour taken to be indicative of fear and the nature of the situations that commonly elicit fear. This discussion occupies Part II of the volume; it provides a background against which are considered, in Part III, the great differences in susceptibility to fear and anxiety that are found when one individual is compared with another. Since many of the data required for the completion of this task are missing, much extrapolation is necessary and the resulting picture is patchy. In some places it can be painted in detail, in others only impressionistically. The aim is to provide clinicians and others with principles on which they can base their actions, and research workers with problems to explore and hypotheses to test.
The third volume, Loss, will deal with problems of grief and mourning and with the defensive processes to which anxiety and loss can give rise. It will comprise a revision and amplification of material first published in the remaining papers of the earlier series -- 'Grief and Mourning in Infancy and Early Childhood' ( 1960b), 'Processes of Mourning' ( 1961b), and 'Pathological Mourning and Childhood Mourning' ( 1963).
____________________
1 In the preface to the first volume I refer only to a second volume. During further work,
however, it has become apparent that a third volume will be required. -xiv-
11
Meanwhile two colleagues, Colin Murray Parkes and Peter Marris, have written books in which they approach problems of loss in a way close to my own. The books are Bereavement by Parkes ( 1972) and Loss and Change by Marris (in press).
In the preface to the first volume it was explained that the frame of reference from which I start is that of psychoanalysis. The reasons are several. The first is that my early thinking on the subject was inspired by psychoanalytic work -- my own and others'. A second is that, despite all its limitations, psychoanalysis and its derivatives remain by far the most used of any present-day approach to psychopathology and psychotherapy. A third and most important is that, whereas many of the central concepts of my schema -- object relations (better termed affectional bonds), separation anxiety, mourning, defence, trauma, sensitive periods in early life -- are the stockin-trade of psychoanalytic thinking, until the last decade or two they have been given scant attention by other behavioural disciplines.
Nevertheless, although the initial frame of reference is that of psychoanalysis, there are many ways in which the theory advanced here differs from the classical theories advanced by Freud and elaborated by his followers. A number of these differences are described already in the first chapter of the earlier volume. Others are referred to throughout the present volume, notably in Chapters 2, 5, and 16.
Note to the Paperback Edition
A paperback edition of this volume has provided the opportunity to include a number of additional notes referring to some later published work. These notes appear on pp. 409 14; the points in the text to which they relate are indicated by a dagger (? ). Details of the additional references are given on pp. 436 -7, following the original references. The index has been considerably expanded.
-xv-
Acknowledgements
IN the first volume of this work I listed the many colleagues and friends without whose help over the years these volumes could not have been written; and it is a great pleasure to express to them all once again my very warmest thanks. My debt to them is deep and lasting.
In the preparation of this volume I am indebted especially for help given by Robert Hinde, Mary Salter Ainsworth, and David Hamburg, each of whom read drafts of all or most of the material and offered a great many valuable criticisms and suggestions. James Robertson scrutinized the first chapter and proposed a number of improvements. Others who have contributed in different ways are Christoph Heinicke, Colin Murray Parkes, and Philip Crockatt. To all of them I am deeply grateful for the time and trouble they have given.
To the preparation of the script my secretary, Dorothy Southern, has again brought her customary care and enthusiasm. Library services have again been provided with unfailing efficiency by Ann Sutherland, and editorial assistance, similarly, by Rosamund Robson. The index has been prepared with great care by Lilian Rubin. To each of them my warmest thanks are due.
12
The many bodies that have supported the research for which I have been responsible at the Tavistock Institute of Human Relations since 1948 are listed in the first volume. Throughout the time that this volume has been in preparation I have been a part-time member of the External Scientific Staff of the Medical Research Council.
For permission to quote from published material, thanks are due to the publishers, authors, and others listed below. Bibliographical details of all the works cited in the text are given in the list of references at the end of the volume.
George Allen & Unwin Ltd, London, and Aldine Publishing Co. , Chicago, in respect of Four Years Old in an Urban Community by J. and E. Newson; Dr I. C. Kaufman, Dr L. A. Rosenblum, and Science in respect of 'Depression in Infant Monkeys Separated from their Mothers' (copyright 1967 by the American Association for the Advancement of Science); Methuen & Co.
-xvii-
Ltd, London, in respect of "'Attachment and Exploratory Behaviour of One-year-olds in a Strange Situation'" by M. D. S. Ainsworth and B. A. Wittig, in Determinants of Infant Behaviour, Vol. 4, edited by B. M. Foss; Dr R. F. Peck and Dr R. J. Havighurst in respect of The Psychology of Character Development, published by John Wiley & Sons, Inc. , New York; University of Chicago Press, Chicago, in respect of The Structure of Scientific Revolutions by T. S. Kuhn; University of London Press Ltd, London, in respect of Truancy by M. J. Tyerman.
Acknowledgement is due also to Tavistock Publications Ltd, London, for permission to include, in Chapter 21 of this volume, material that appears in Support, Innovation, and Autonomy edited by R. Gosling; and to the Journal of Child Psychology and Psychiatry for permission to reproduce, as the basis of Appendix I, a paper that was first published in that journal in 1961.
-xviii-
13
Part I
SECURITY, ANXIETY, AND DISTRESS
-1-
Chapter 1
Prototypes of Human Sorrow
Unhappiness in a child accumulates because he sees no end to the dark tunnel. The thirteen weeks of a term might just as well be thirteen years.
GRAHAM GREENE, A Sort of Life
Responses of young children to separation from mother 1
A generation has now passed since Dorothy Burlingham and Anna Freud recorded their experiences of caring for infants and young children in the setting of a residential nursery. In two modest booklets published during the second world war ( Burlingham & Freud 1942; 1944) they describe the immense problem of providing for young children who are out of mother's care. In particular they emphasize how impossible it is in a nursery setting to provide a child with a substitute figure who can mother him as well as his own mother can. When the Hampstead Nurseries were reorganized so that each nurse could care for her own little group of children they tell how the children became strongly possessive of their nurse and acutely jealous whenever she gave attention to another child: 'Tony (3 1/2) . . . would not allow Sister Mary to use "his" hand for handling other children. Jim (2-3) would burst into tears whenever his "own" nurse left the room. Shirley (4) would become intensely depressed and disturbed when "her" Marion was absent for some reason. '
Why, it may be asked, should these children have become so strongly possessive of their nurse and so deeply distressed whenever she was missing? Was it, as some traditionalists might suppose, that they had been spoiled by having been given too much attention and allowed too much their own way? Or
____________________
1 Although throughout this book the text refers usually to 'mother' and not to 'mother figure',
it is to be understood that in every case reference is to the person who mothers a child and to whom he becomes attached. For most children, of course, this person is also his natural mother.
-3-
was it, by contrast, that since leaving home they had been subjected to too many changes of mother figure and/or had too limited access to whoever in the nursery was acting temporarily as their mother figure? On how we answer these questions turn all our practices of child- rearing.
14
Not only did children in these nurseries become intensely possessive and jealous of their 'own' nurse but they were also unusually prone to become hostile towards her or to reject her, or else to retreat into a state of emotional detachment, as the following records illustrate:
Jim was separated from a very nice and affectionate mother at 17 months and developed well in our nursery. During his stay he formed two strong attachments to two young nurses who successively took care of him. Though he was otherwise a well adjusted, active and companionable child, his behaviour became impossible where these attachments were concerned. He was clinging, over-possessive, unwilling to be left for a minute, and continually demanded something without being able to define in any way what it was he wanted. It was no unusual sight to see Jim lie on the floor sobbing and despairing. These reactions ceased when his favourite nurse was absent even for short periods. He was then quiet and impersonal.
And finally psychoanalysis itself has begun to catch up with bowlby. The recent relational turn in psychoanalysis ( Mitchell and Aron) has made bridges between Bowlby's work and contemporary psychoanalytic thought much more compelling.
Part of what makes Bowlby's early work so inspiring thirty years later is his blend of openness and persistence. His intellectual curiosity seems to have known no bounds, and he continued to draw upon many diverse sources for ideas and conceptual tools, including child- observation, ethology, systems theory, and information-processing. Bowlby also knew from early on that he was onto something very important, and his integrity and perseverance in pursuing what mattered, despite criticism, has benefitted us all.
8
Attachment and Loss has been one of the most influential works of this century. Stephen A. Mitchell New York City, October 1999
-x-
Preface
IN the preface to the first volume of this work I describe the circumstances in which it was begun. Clinical experience of disturbed children, research into their family backgrounds, and an opportunity, in 1950, to read the literature and to discuss problems of mental health with colleagues in several countries led me, in a report commissioned by the World Health Organization, to formulate a principle: 'What is believed to be essential for mental health is that the infant and young child should experience a warm, intimate and continuous relationship with his mother (or permanent mother-substitute) in which both find satisfaction and enjoyment' (Bowlby 1951). To support this conclusion evidence was presented for believing that many forms of psychoneurosis and character disorder are to be attributed either to deprivation of maternal care or to discontinuities in a child's relationship with his mother figure.
Though the contents of the report proved controversial at the time, most of the conclusions are now accepted. What has plainly been missing, however, is an account of the processes through which the many and varied ill effects attributed to maternal deprivation or to discontinuities in the mother-child bond are brought into being. It is this gap that my colleagues and I have since striven to fill. In doing so we have adopted a research strategy that we believe is still too little exploited in the field of psychopathology.
In their day-to-day work, whether with disturbed children, disturbed adults, or disturbed families, clinicians have of necessity to view causal processes backwards, from the disturbance of today back to the events and conditions of yesterday. Though this method has yielded many valuable insights into possible pathogenic events and into the kinds of pathological process to which they appear to give rise, as a research method it has grave limitations. To complement it, a method regularly adopted in other branches of medical research is, having identified a possible pathogen, to study its effects prospectively. If the pathogen has been correctly identified and the studies of its effects in the short and long term are skilfully executed, it then becomes possible to describe the processes set in train by the pathogenic agent and also the ways by which they lead to
-xi-
the various consequent conditions. In such studies attention must be paid not only to the processes set in train by the pathogen but also to the very many conditions, internal and external to the organism, that affect their course. Only then can some grasp be had of the particular processes, conditions, and sequences that lead from a potentially pathogenic occurrence to the particular types of disturbance with which the clinician was in the first place concerned.
In adopting a prospective research strategy my colleagues and I early became deeply impressed by the observations of our colleague, James Robertson, who had recorded, both on
9
paper and on film, how young children in their second and third years of life respond while away from home and cared for instead in a strange place by a succession of unfamiliar people, and also how they respond during and after return home to mother ( Robertson 1952; 1953; Robertson & Bowlby 1952). During the period away, perhaps in residential nursery or hospital ward, a young child is usually acutely distressed for a time and is not easily comforted. After his return home he is likely to be either emotionally detached from his mother or else intensely clinging; as a rule a period of detachment, either brief or long depending mainly on length of separation, precedes a period during which he becomes strongly demanding of his mother's presence. Should a child then come to believe, for any reason, that there is risk of a further separation he is likely to become acutely anxious.
Reflecting on these observations we concluded that 'loss of mother figure, either by itself or in combination with other variables yet to be clearly identified, is capable of generating responses and processes that are of the greatest interest to psychopathology'. Our reason for this belief was that the responses and processes observed seemed to be the same as those found to be active in older individuals who are still disturbed by separations they have suffered in early life. These comprise, on the one hand, a tendency to make intensely strong demands on others and to be anxious and angry when they are not met, a condition common in individuals labelled neurotic; and, on the other, a blockage in the capacity to make deep relationships, such as is present in affectionless and psychopathic personalities.
From the start an important and controversial issue has been the part played in the responses of children to separation from mother by variables other than that of separation per se;
-xii-
these include illness, the strange surroundings in which a child finds himself, the kind of substitute care he receives while away, the kind of relations he has both before and after the event. It is plain that these factors can greatly intensify, or in some cases mitigate, a child's responses. Yet evidence is convincing that presence or absence of mother figure is itself a condition of the greatest significance in determining a child's emotional state. The issue is already discussed in Chapter 2 of the first volume, where a description is given of some of the relevant findings, and is taken up again in the first chapter of this one, where attention is given to the results of a foster-care project undertaken in recent years by James and Joyce Robertson in which they 'sought to create a separation situation from which many of the factors that complicate institutional studies were eliminated; and in which the emotional needs of the children would be met as far as possible by a fully available substitute mother' ( Robertson & Robertson 1971). 1 Study of the Robertsons' findings has led to some modification of views expressed in earlier publications, in which insufficient weight was given to the influence of skilled care from a familiar substitute. In parallel with the empirical studies of my colleagues, I have myself been engaged in studying the theoretical and clinical implications of the data. In particular, I have been trying to sketch a schema able to comprehend data derived from a number of distinct sources:
observations of how young children behave during periods when they are away from mother and after they return home to her;
observations of how older subjects, children and adults, behave during and after a separation from a loved figure, or after a permanent loss;
observations of difficulties found during clinical work with children and adults who, during childhood or adolescence, have either experienced a long separation or a loss or had grounds to fear one; these include various forms of acute or chronic anxiety and
10
depression, and difficulties of every degree in making and maintaining close affectional bonds, whether with parent figures, with members of the opposite sex, or with own children.
____________________
1 In addition to their written report the Robertsons have published a series of films on the
children fostered, particulars of which are given in the list of references at the end of this volume.
-xiii-
First steps towards formulating a theoretical schema were taken in a series of papers published between 1958 and 1963. The present three-volume work1 is a further attempt at a formulation.
Volume 1, Attachment, is devoted to problems originally tackled in the first paper of the series, 'The Nature of the Child's Tie to his Mother' ( 1958b). In order effectively to discuss the empirical data regarding the development of that tie and to formulate a theory to account for it, it proved necessary to discuss first the whole problem of instinctive behaviour and how best to conceptualize it. In doing so I drew heavily on findings and ideas contributed by ethologists and also on ideas derived from control theory.
This, the second volume, deals mainly with problems of separation anxiety and covers ground originally tackled in two further papers of the original series, 'Separation Anxiety' ( 1960a) and 'Separation Anxiety: A Critical Review of the Literature' ( 1961a). Once again, in order to comprehend better the problems before us -- the distress occurring during a separation and the anxiety often evident after it -- it has proved desirable first to discuss a broad range of related phenomena and theory, notably the various forms of behaviour taken to be indicative of fear and the nature of the situations that commonly elicit fear. This discussion occupies Part II of the volume; it provides a background against which are considered, in Part III, the great differences in susceptibility to fear and anxiety that are found when one individual is compared with another. Since many of the data required for the completion of this task are missing, much extrapolation is necessary and the resulting picture is patchy. In some places it can be painted in detail, in others only impressionistically. The aim is to provide clinicians and others with principles on which they can base their actions, and research workers with problems to explore and hypotheses to test.
The third volume, Loss, will deal with problems of grief and mourning and with the defensive processes to which anxiety and loss can give rise. It will comprise a revision and amplification of material first published in the remaining papers of the earlier series -- 'Grief and Mourning in Infancy and Early Childhood' ( 1960b), 'Processes of Mourning' ( 1961b), and 'Pathological Mourning and Childhood Mourning' ( 1963).
____________________
1 In the preface to the first volume I refer only to a second volume. During further work,
however, it has become apparent that a third volume will be required. -xiv-
11
Meanwhile two colleagues, Colin Murray Parkes and Peter Marris, have written books in which they approach problems of loss in a way close to my own. The books are Bereavement by Parkes ( 1972) and Loss and Change by Marris (in press).
In the preface to the first volume it was explained that the frame of reference from which I start is that of psychoanalysis. The reasons are several. The first is that my early thinking on the subject was inspired by psychoanalytic work -- my own and others'. A second is that, despite all its limitations, psychoanalysis and its derivatives remain by far the most used of any present-day approach to psychopathology and psychotherapy. A third and most important is that, whereas many of the central concepts of my schema -- object relations (better termed affectional bonds), separation anxiety, mourning, defence, trauma, sensitive periods in early life -- are the stockin-trade of psychoanalytic thinking, until the last decade or two they have been given scant attention by other behavioural disciplines.
Nevertheless, although the initial frame of reference is that of psychoanalysis, there are many ways in which the theory advanced here differs from the classical theories advanced by Freud and elaborated by his followers. A number of these differences are described already in the first chapter of the earlier volume. Others are referred to throughout the present volume, notably in Chapters 2, 5, and 16.
Note to the Paperback Edition
A paperback edition of this volume has provided the opportunity to include a number of additional notes referring to some later published work. These notes appear on pp. 409 14; the points in the text to which they relate are indicated by a dagger (? ). Details of the additional references are given on pp. 436 -7, following the original references. The index has been considerably expanded.
-xv-
Acknowledgements
IN the first volume of this work I listed the many colleagues and friends without whose help over the years these volumes could not have been written; and it is a great pleasure to express to them all once again my very warmest thanks. My debt to them is deep and lasting.
In the preparation of this volume I am indebted especially for help given by Robert Hinde, Mary Salter Ainsworth, and David Hamburg, each of whom read drafts of all or most of the material and offered a great many valuable criticisms and suggestions. James Robertson scrutinized the first chapter and proposed a number of improvements. Others who have contributed in different ways are Christoph Heinicke, Colin Murray Parkes, and Philip Crockatt. To all of them I am deeply grateful for the time and trouble they have given.
To the preparation of the script my secretary, Dorothy Southern, has again brought her customary care and enthusiasm. Library services have again been provided with unfailing efficiency by Ann Sutherland, and editorial assistance, similarly, by Rosamund Robson. The index has been prepared with great care by Lilian Rubin. To each of them my warmest thanks are due.
12
The many bodies that have supported the research for which I have been responsible at the Tavistock Institute of Human Relations since 1948 are listed in the first volume. Throughout the time that this volume has been in preparation I have been a part-time member of the External Scientific Staff of the Medical Research Council.
For permission to quote from published material, thanks are due to the publishers, authors, and others listed below. Bibliographical details of all the works cited in the text are given in the list of references at the end of the volume.
George Allen & Unwin Ltd, London, and Aldine Publishing Co. , Chicago, in respect of Four Years Old in an Urban Community by J. and E. Newson; Dr I. C. Kaufman, Dr L. A. Rosenblum, and Science in respect of 'Depression in Infant Monkeys Separated from their Mothers' (copyright 1967 by the American Association for the Advancement of Science); Methuen & Co.
-xvii-
Ltd, London, in respect of "'Attachment and Exploratory Behaviour of One-year-olds in a Strange Situation'" by M. D. S. Ainsworth and B. A. Wittig, in Determinants of Infant Behaviour, Vol. 4, edited by B. M. Foss; Dr R. F. Peck and Dr R. J. Havighurst in respect of The Psychology of Character Development, published by John Wiley & Sons, Inc. , New York; University of Chicago Press, Chicago, in respect of The Structure of Scientific Revolutions by T. S. Kuhn; University of London Press Ltd, London, in respect of Truancy by M. J. Tyerman.
Acknowledgement is due also to Tavistock Publications Ltd, London, for permission to include, in Chapter 21 of this volume, material that appears in Support, Innovation, and Autonomy edited by R. Gosling; and to the Journal of Child Psychology and Psychiatry for permission to reproduce, as the basis of Appendix I, a paper that was first published in that journal in 1961.
-xviii-
13
Part I
SECURITY, ANXIETY, AND DISTRESS
-1-
Chapter 1
Prototypes of Human Sorrow
Unhappiness in a child accumulates because he sees no end to the dark tunnel. The thirteen weeks of a term might just as well be thirteen years.
GRAHAM GREENE, A Sort of Life
Responses of young children to separation from mother 1
A generation has now passed since Dorothy Burlingham and Anna Freud recorded their experiences of caring for infants and young children in the setting of a residential nursery. In two modest booklets published during the second world war ( Burlingham & Freud 1942; 1944) they describe the immense problem of providing for young children who are out of mother's care. In particular they emphasize how impossible it is in a nursery setting to provide a child with a substitute figure who can mother him as well as his own mother can. When the Hampstead Nurseries were reorganized so that each nurse could care for her own little group of children they tell how the children became strongly possessive of their nurse and acutely jealous whenever she gave attention to another child: 'Tony (3 1/2) . . . would not allow Sister Mary to use "his" hand for handling other children. Jim (2-3) would burst into tears whenever his "own" nurse left the room. Shirley (4) would become intensely depressed and disturbed when "her" Marion was absent for some reason. '
Why, it may be asked, should these children have become so strongly possessive of their nurse and so deeply distressed whenever she was missing? Was it, as some traditionalists might suppose, that they had been spoiled by having been given too much attention and allowed too much their own way? Or
____________________
1 Although throughout this book the text refers usually to 'mother' and not to 'mother figure',
it is to be understood that in every case reference is to the person who mothers a child and to whom he becomes attached. For most children, of course, this person is also his natural mother.
-3-
was it, by contrast, that since leaving home they had been subjected to too many changes of mother figure and/or had too limited access to whoever in the nursery was acting temporarily as their mother figure? On how we answer these questions turn all our practices of child- rearing.
14
Not only did children in these nurseries become intensely possessive and jealous of their 'own' nurse but they were also unusually prone to become hostile towards her or to reject her, or else to retreat into a state of emotional detachment, as the following records illustrate:
Jim was separated from a very nice and affectionate mother at 17 months and developed well in our nursery. During his stay he formed two strong attachments to two young nurses who successively took care of him. Though he was otherwise a well adjusted, active and companionable child, his behaviour became impossible where these attachments were concerned. He was clinging, over-possessive, unwilling to be left for a minute, and continually demanded something without being able to define in any way what it was he wanted. It was no unusual sight to see Jim lie on the floor sobbing and despairing. These reactions ceased when his favourite nurse was absent even for short periods. He was then quiet and impersonal.
Reggie, who had come to our house as a baby of 5 months, went home to his mother when he was 1 year 8 months, and has been with us ever since his return to the nursery 2 months later. While with us, he formed two passionate relationships to two young nurses who took care of him at different periods. The second attachment was suddenly broken at 2 years 8 months when his 'own' nurse married. He was completely lost and desperate after her departure, and refused to look at her when she visited him a fortnight later. He turned his head to the other side when she spoke to him, but stared at the door, which had closed behind her, after she had left the room. In the evening in bed he sat up and said: 'My very own Mary-Ann! But I don't like her. '
These observations, made in the pressure of wartime and recorded anecdotally with all too little detail, none the less cast a shaft of light on the nature of many forms of psychiatric disturbance. States of anxiety and depression that occur during
-4-
adult years, and also psychopathic conditions, can, it is held, be linked in a systematic way to the states of anxiety, despair, and detachment described by Burlingham & Freud, and subsequently by others, that are so readily engendered whenever a young child is separated for long from his mother figure, whenever he expects such a separation, and when, as sometimes happens, he loses her altogether. Whereas during later life it is often extremely difficult to trace how a person's disturbed emotional state is related to his experiences, whether they be those of his current life or those of his past, during the early years of childhood the relationship between emotional state and current or recent experience is often crystal clear. In these troubled states of early childhood, it is held, can be discerned the prototype of many a pathological condition of later years.
It is, of course, a commonplace that most children who have had experiences of these kinds recover and resume normal development, or at least they appear to do so. Not infrequently, therefore, doubts are expressed whether the psychological processes described are in reality related so intimately to personality disturbances of later life. Pending much further evidence, these are legitimate doubts. Nevertheless, reasons for holding to the thesis are strong. One is that data from many sources can be arranged and organized into a pattern that is internally consistent and consistent also with current biological theory. Another is that many clinicians and social workers find the resulting schema enables them to understand better the problems with which they are grappling and so to help their patients or clients more effectively.
15
Why some individuals should recover, largely or completely, from experiences of separation and loss while others seem not to is a central question, but one not easily answered. In living creatures variation of response is the rule and its explanation is often hard to fathom. Of all those who contract poliomyelitis less than 1 per cent develop paralysis, and only a fraction of 1 per cent remain crippled. Why one person should respond one way and another another remains obscure. To argue that, because 99 per cent recover, polio is a harmless infection would obviously be absurd. Similarly, in the field under consideration, to argue that because most individuals recover from the effects of a separation or loss these experiences are of no account would be equally absurd.
Nevertheless the problem of differential response remains
-5-
important. Conditions likely to be playing a part can be considered under two main heads: those intrinsic to or closely associated with the separation itself, notably the conditions in which a child is cared for while away from mother;
those present in the child's life over a longer period, notably his relations with parents during the months or years before and after the event.
Here we consider variables in category (a). Those in category (b) are discussed in the later chapters of Part III.
We start by reviewing observations of how children behave when cared for in one of two very different settings. The first is an ordinary residential nursery, in which a child finds himself in a strange place with strange people none of whom is sufficiently available to give him more than very limited mothering. The second is a foster home in which a child receives the fulltime and skilled care of a foster mother with whom he has become in some degree familiar beforehand.
Conditions leading to intense responses
In our early studies children were observed during stays in institutional settings and it was on the basis of these observations that the sequence of responses which we term protest, despair, and detachment came first to be delineated ( Robertson & Bowlby 1952). Since then two further studies have been conducted by colleagues in the Tavistock Child Development Research Unit, the first by Christoph Heinicke ( 1956) and the second by Heinicke & Ilse Westheimer ( 1966). Although in each of these investigations only a handful of children were observed (six in the first and ten in the second), the studies are unique for the care of their design and the amount of systematic observation. Moreover, for each sample of separated children a contrast group was selected and observed: in the first study it was a fairly well- matched group of children observed during their first weeks of attendance at a day nursery; in the second it was a similarly matched group of children observed while living in their own homes. Heinicke & Westheimer treat their data statistically and also describe in some detail the behaviour of individual children.
In the larger investigation ( 1966), work was conducted in three residential nurseries. Arrangements and facilities were fairly similar. In each, a child belonged to a defined group of
16
-6-
children and was cared for mainly by one or two nurses. Ample opportunities were available for free play either in large rooms or outdoors in a garden. Before a child entered the nursery, contact was made with the family by a psychiatric social worker ( Ilse Westheimer), who was also responsible either then or later for collecting full information about the family and the child. Arrival at the nursery was observed; and in the course of his stay a child was observed during free play on six occasions each week. Each of the two observers (one male, Christoph Heinicke, and one female, Elizabeth Wolpert) observed for a period of at least half an hour during each of the three sample periods into which the week was broken (Monday and Tuesday; Wednesday and Thursday; Friday, Saturday, and Sunday). The method used, of categorizing behavioural units in terms of agent, object, relation, mode, and intensity, had been used in the earlier study and had been shown to be reliable.
In addition to the categorized observations of free behaviour, similarly categorized observations were made of every child's behaviour in standardized doll-play sessions; and a number of other records of each child's stay in the nursery were kept.
It was originally intended to select the separated children in accordance with the five criteria used in the first study, namely: (i) that the child had had no previous separations of more than three days, (ii) that he fell within the age-limits of fifteen to thirty months, (iii) that he did not enter the nursery with a sibling, (iv) that he was living with both mother and father at the time the separation occurred, and (v) that there was no evidence that being placed in a nursery indicated a rejection by his parents. Because of the difficulty of obtaining cases, however, these criteria had to be modified to allow greater latitude.
Although most of the children had had either no separations or only very brief ones prior to the one being studied, in one case the length of previous separation was four weeks and in two it was three weeks. The age-range was slightly extended and ran from thirteen to thirty-two months, instead of from fifteen to thirty months. But the most marked departure from the previous criteria was that four of the children entered the residential nursery in the company of a sibling: in three cases this was a four-year-old sibling and in one case the sibling was younger. The remaining two criteria remained unmodified: each of the children was living with both mother and father at the time of separation, and there were no indications that he
-7-
was being rejected by the parents by being placed in the nursery.
The reason that the ten children studied were cared for in a residential nursery was that, in a family emergency, neither relatives nor friends were available to take temporary care of them. In the case of seven families, mother was to be away in hospital having a new baby. In two others mother was to be in hospital for some other form of medical attention. In the tenth case, the family became homeless.
Among much else in their book, Brief Separations ( 1966), Heinicke & Westheimer describe behaviour typical of the ten children during their time in the nursery, and, similarly, behaviour typical of the children after they had returned home. In the paragraphs that follow some of their principal findings are presented. Every one of the patterns reported had also been
17
observed and recorded by Robertson during his earlier, less systematic though more extensive, studies.
Behaviour during Separation
The children arrived at the nursery in the care of one or both parents. Four of them, brought by father, stayed close to him and seemed already subdued and anxious. Some of the others, who had come with mother or both parents, seemed more confident and were ready to explore the new environment. They ventured forth on short or long excursions and then returned.
When the moment came for the parent (s) to depart, crying or screaming was the rule. One child tried to follow her parents, demanding urgently where they were going, and finally had to be pushed back into the room by her mother. Another threw himself on the floor and refused to be comforted. Altogether eight of the children were crying loudly soon after their parents' departure. Bedtime was also an occasion for tears. The two who had not cried earlier screamed when put in a cot and could not be consoled. Some of the others whose initial crying had ceased broke into renewed sobs at bedtime.
