During the early 1950s
Robertson
and I had generalized the sequence of responses seen in young children during temporary separation from mother as those of protest, despair, and detachment (Robertson and Bowlby, 1952).
A-Secure-Base-Bowlby-Johnf
Requested to contribute to a United Nations study of the needs of homeless children, Ronald Hargreaves1 decided to appoint a short- term consultant to report on the mental health aspects of the problem and, knowing of my in- terest in the field, invited me to undertake the task.
For me this was a golden opportunity.
After five years as an army psychiatrist, I had returned to child psychiatry determined to explore further the problems I had begun working on before the war; and I had already appointed as my first re- search assistant James Robertson, a newly quali- fied psychiatric social worker who had worked with Anna Freud in the Hampstead Nurseries during the war.
The six months I spent with the World Health Organisation in 1950 gave me the chance not only to read the literature and to discuss it with the authors, but also to meet many others in Europe and the United States with experience of the field. Soon after the end of my contract I submitted my report, which was published early in 1951 as a
64/362
? ? ? WHO monograph entitled Maternal Care and Mental Health. In it I reviewed the far from neg- ligible evidence then available regarding the ad- verse influences on personality development of inadequate maternal care during early childhood, called attention to the acute distress of young children who find themselves separated from those they know and love, and made recommend- ations of how best to avoid, or at least mitigate, the short- and long-term ill effects. During the next few years this report was translated into a dozen other languages and appeared also in a cheap abridged edition in English.
Influential though the written word may often be, it has nothing like the emotional impact of a movie. Throughout the 1950s Rene? Spitz's early film Grief: A Peril in Infancy (1947), and James Robertson's A Two-Year-Old Goes to Hospital (1952) together had an enormous influence. Not only did they draw the attention of professional workers to the immediate distress and anxiety of young children in an institutional setting but they proved powerful instruments for promoting changes in practice. In this field Robertson was to play a leading part (e. g. Robertson, 1958, 1970).
65/362
? ? ? Although by the end of the 1950s a great many of those working in child psychiatry and psycho- logy and in social work, and some also of those in paediatrics and sick children's nursing, had ac- cepted the research findings and were imple- menting change, the sharp controversy aroused by the early publications and films continued. Psychiatrists trained in traditional psychiatry and psychologists who adopted a learning-theory ap- proach never ceased to point to the deficiencies of the evidence and to the lack of an adequate ex- planation of how the types of experience implic- ated could have the effects on personality devel- opment claimed. Many psychoanalysts, in addi- tion, especially those whose theory focused on the role of fantasy in psychopathology to the relative exclusion of the influence of real life events, re- mained unconvinced and sometimes very critical. Meanwhile, research continued. For example, at Yale Sally Provence and Rose Lipton, were mak- ing a systematic study of institutionalized infants in which they compared their development with that of infants living in a family (Provence and Lipton, 1962). At the Tavistock members of my small research group were active collecting fur- ther data on the short-term effects on a young
66/362
? ? ? child of being in the care of strange people in a strange place for weeks and sometimes months at a time (see especially the studies by Christoph Heinicke, 1956 and, with Ilse Westheimer, 1966), whilst I addressed myself to the theoretical prob- lems posed by our data.
Meanwhile the field was changing. One import- ant influence was the publication in 1963 by the World Health Organisation of a collection of art- icles in which the manifold effects of the various types of experience covered by the term 'depriva- tion of maternal care' were reassessed. Of the six articles, by far the most comprehensive was by my colleague Mary Ainsworth (1962). In it she not only reviewed the extensive and diverse evid- ence and considered the many issues that had given rise to controversy but also identified a large number of problems requiring further research.
A second important influence was the publica- tion, beginning during the late fifties, of Harry Harlow's studies of the effects of maternal deprivation on rhesus monkeys; and once again film played a big part. Harlow's work in the Un- ited States had been stimulated by Spitz's reports. In the United Kingdom complementary studies
67/362
? ? ? by Robert Hinde had been stimulated by our work at the Tavistock. For the next decade a stream of experimental results from those two scientists (see summaries in Harlow and Harlow, 1965 and Hinde and Spencer-Booth, 1971), com- ing on top of the Ainsworth review, undermined the opposition. Thereafter nothing more was heard of the inherent implausibility of our hypo- theses; and criticism became more constructive.
Much, of course, remained uncertain. Even if the reality of short-term distress and behavioural disturbance is granted, what evidence is there, it was asked, that the ill effects can persist? What features of the experience, or combination of fea- tures, are responsible for the distress? And, should it prove true that in some cases ill effects do persist, how is that to be accounted for? How does it happen that some children seem to come through very unfavourable experiences relatively unharmed? How important is it that a child should be cared for most of the time by one prin- cipal caregiver? In less developed societies it was claimed (wrongly as it turns out) that multiple mothering is not uncommon. In addition to all these legitimate questions, moreover, there were misunderstandings. Some supposed that
68/362
? ? ? advocates of the view that a child should be cared for most of the time by a principal mother-figure held that that had to be the child's natural moth- er--the so-called blood-tie theory. Others sup- posed that, in advocating that a child should 'ex- perience a warm intimate and continuous rela- tionship with his mother (or permanent mother- substitute)', proponents were prescribing a re- gime in which a mother had to care for her child 24 hours a day, day in and day out, with no res- pite. In a field in which strong feelings are aroused and almost everyone has some sort of vested interest, clear unbiased thinking is not al- ways easy.
A NEW LOOK AT THEORY
The monograph Maternal Care and Mental Health is in two parts. The first reviews the evid- ence regarding the adverse effects of maternal deprivation, the second discusses means for pre- venting it. What was missing, as several reviewers pointed out, was any explanation of how experi- ences subsumed under the broad heading of ma- ternal deprivation could have the effects on per- sonality development of the kinds claimed. The reason for this omission was simple: the data
69/362
? ? ? were not accommodated by any theory then cur- rent and in the brief time of my employment by the World Health Organisation there was no pos- sibility of developing a new one.
The child's tie to his mother
At that time it was widely held that the reason a child develops a close tie to his mother is that she feeds him. Two kinds of drive are postulated, primary and secondary. Food is thought of as primary; the personal relationship, referred to as 'dependency', as secondary. This theory did not seem to me to fit the facts. For example, were it true, an infant of a year or two should take read- ily to whomever feeds him and this clearly was not the case. An alternative theory, stemming from the Hungarian school of psychoanalysis, postulated a primitive object relation from the beginning. In its best-known version, however, the one advocated by Melanie Klein, mother's breast is postulated as the first object, and the greatest emphasis is placed on food and orality and on the infantile nature of 'dependency'. None of these features matched my experience of children.
70/362
? ? ? But if the current dependency theories were in- adequate, what was the alternative?
During the summer of 1951 a friend mentioned to me the work of Lorenz on the following re- sponses of ducklings and goslings. Reading about this and related work on instinctive behaviour re- vealed a new world, one in which scientists of high calibre were investigating in non-human species many of the problems with which we were grappling in the human, in particular the relat- ively enduring relationships that develop in many species, first between young and parents and later between mated pairs, and some of the ways in which these developments can go awry. Could this work, I asked myself, cast light on a problem central to psychoanalysis, that of 'instinct' in humans?
Next followed a long phase during which I set about trying to master basic principles and to ap- ply them to our problems, starting with the nature of the child's tie to his mother. Here Lorenz's work on the following response of duck- lings and goslings (Lorenz, 1935) was of special interest. It showed that in some animal species a strong bond to an individual motherfigure could develop without the intermediary of food: for
71/362
? ? ? these young birds are not fed by parents but feed themselves by catching insects. Here then was an alternative model to the traditional one, and one that had a number of features that seemed pos- sibly to fit the human case. Thereafter, as my grasp of ethological principles increased and I ap- plied them to one clinical problem after another, I became increasingly confident that this was a promising approach. Thus, having adopted this novel point of view, I decided to 'follow it up through the material as long as the application of it seems to yield results' (to borrow a phrase of Freud's).
From 1957, when The Nature of the Child's Tie to his Mother was first presented, through 1969 when Attachment appeared, until 1980 with the publication of Loss I concentrated on this task. The resulting conceptual framework2 is designed to accommodate all those phenomena to which Freud called attention--for example love rela- tions, separation anxiety, mourning, defence, an- ger, guilt, depression, trauma, emotional detach- ment, sensitive periods in early life--and so to of- fer an alternative to the traditional metapsycho- logy of psychoanalysis and to add yet another to the many variants of the clinical theory now
72/362
? ? ? extant. How successful these ideals will prove only time will tell.
As Kuhn has emphasized, any novel conceptual framework is difficult to grasp, especially so for those long familiar with a previous one. Of the many difficulties met with in understanding the framework advocated, I describe only a few. One is that, instead of starting with a clinical syn- drome of later years and trying to trace its origins retrospectively, I have started with a class of childhood traumata and tried to trace the se- quelae prospectively. A second is that, instead of starting with the private thoughts and feelings of a patient, as expressed in free associations or play, and trying to build a theory of personality development from those data, I have started with observations of the behaviour of children in cer- tain sorts of defined situation, including records of the feelings and thoughts they express, and have tried to build a theory of personality devel- opment from there. Other difficulties arise from my use of concepts such as control system (in- stead of psychic energy) and developmental path- way (instead of libidinal phase), which, although now firmly established as key concepts in all the biological sciences, are still foreign to the
73/362
? ? ? thinking of a great many psychologists and clinicians.
Having discarded the secondary-drive, depend- ency theory of the child's tie to his mother, and also the Kleinian alternative, a first task was to formulate a replacement. This led to the concept of attachment behaviour with its own dynamics distinct from the behaviour and dynamics of either feeding or sex, the two sources of human motivation for long widely regarded as the most fundamental. Strong support for this step soon came from Harlow's finding that, in another primate species--rhesus macaques--infants show a marked preference for a soft dummy 'mother', despite its providing no food, to a hard one that does provide it (Harlow and Zimmermann, 1959).
Attachment behaviour is any form of behaviour that results in a person attaining or maintaining proximity to some other clearly identified indi- vidual who is conceived as better able to cope with the world. It is most obvious whenever the person is frightened, fatigued, or sick, and is as- suaged by comforting and caregiving. At other times the behaviour is less in evidence. Neverthe- less for a person to know that an attachment fig- ure is available and responsive gives him a strong
74/362
? ? ? and pervasive feeling of security, and so encour- ages him to value and continue the relationship. Whilst attachment behaviour is at its most obvi- ous in early childhood, it can be observed throughout the life cycle, especially in emergen- cies. Since it is seen in virtually all human beings (though in varying patterns), it is regarded as an integral part of human nature and one we share (to a varying extent) with members of other spe- cies. The biological function attributed to it is that of protection. To remain within easy access of a familiar individual known to be ready and willing to come to our aid in an emergency is clearly a good insurance policy--whatever our age.
By conceptualizing attachment in this way, as a fundamental form of behaviour with its own in- ternal motivation distinct from feeding and sex, and of no less importance for survival, the beha- viour and motivation are accorded a theoretical status never before given them--though parents and clinicians alike have long been intuitively aware of their importance. As already emphas- ized, the terms 'dependency' and 'dependency need' that have hitherto been used to refer to them have serious disadvantages. In the first
75/362
? ? ? place 'dependency' has a perjorative flavour; in the second it does not imply an emotionally charged relationship to one or a very few clearly preferred individuals; and in the third no valu- able biological function has ever been attributed to it.
It is now 30 years since the notion of attach- ment was first advanced as a useful way of con- ceptualizing a form of behaviour of central im- portance not only to clinicians and to develop- mental psychologists but to every parent as well. During that time attachment theory has been greatly clarified and amplified. The most notable contributors have been Robert Hinde who, in ad- dition to his own publications (e. g. 1974), has constantly guided my own thinking, and Mary Ainsworth who, starting in the late 50s, has pion- eered empirical studies of attachment behaviour both in Africa (1963, 1967) and in the USA (Ainsworth and Wittig, 1969; Ainsworth et al. , 1978), and has also helped greatly to develop the- ory (e. g. 1969, 1982). Her work, together with that of her students and others influenced by her (which has expanded dramatically since this lec- ture was given and is described in some detail in Lecture 7), has led attachment theory to be
76/362
? ? ? widely regarded as probably the best supported theory of socio-emotional development yet avail- able (Rajecki, Lamb, and Obmascher, 1978; Rut- ter, 1980; Parkes and Stevenson-Hinde, 1982; Sroufe, 1986).
Because my starting point in developing theory was observations of behaviour, some clinicians have assumed that the theory amounts to no more than a version of behaviourism. This mis- take is due in large part to the unfamiliarity of the conceptual framework proposed and in part to my own failure in early formulations to make clear the distinction to be drawn between an at- tachment and attachment behaviour. To say of a child (or older person) that he is attached to, or has an attachment to, someone means that he is strongly disposed to seek proximity to and con- tact with that individual and to do so especially in certain specified conditions. The disposition to behave in this way is an attribute of the attached person, a persisting attribute which changes only slowly over time and which is unaffected by the situation of the moment. Attachment behaviour, by contrast, refers to any of the various forms of behaviour that the person engages in from time
77/362
? ? ? to time to obtain and/or maintain a desired proximity.
There is abundant evidence that almost every child habitually prefers one person, usually his mother-figure, to whom to go when distressed but that, in her absence, he will make do with someone else, preferably someone whom he knows well. On these occasions most children show a clear hierarchy of preference so that, in extremity and with no one else available, even a kindly stranger may be approached. Thus, whilst attachment behaviour may in differing circum- stances be shown to a variety of individuals, an enduring attachment, or attachment bond, is confined to very few. Should a child fail to show such clear discrimination, it is likely he is severely disturbed.
The theory of attachment is an attempt to ex- plain both attachment behaviour, with its episod- ic appearance and disappearance, and also the enduring attachments that children and other in- dividuals make to particular others. In this theory the key concept is that of behavioural system. This is conceived on the analogy of a physiologic- al system organized homeostatically to ensure that a certain physiological measure, such as
78/362
? ? ? body temperature or blood pressure, is held between appropriate limits. In proposing the concept of a behavioural system to account for the way a child or older person maintains his re- lation to his attachment figure between certain limits of distance or accessibility, no more is done than to use these well-understood principles to account for a different form of homeostasis, namely one in which the set limits concern the organism's relation to clearly identified persons in, or other features of, the environment and in which the limits are maintained by behavioural instead of physiological means.
In thus postulating the existence of an internal psychological organization with a number of highly specific features, which include represent- ational models of the self and of attachment fig- ure(s), the theory proposed can be seen as having all the same basic properties as those that charac- terize other forms of structural theory, of which the variants of psychoanalysis are some of the best known, and that differentiate them so sharply from behaviourism in its many forms. Historically attachment theory was developed as a variant of object-relations theory.
79/362
? ? ? The reason why in this account I have given so much space to the concept and theory of attach- ment is that, once those principles are grasped, there is little difficulty in understanding how the many other phenomena of central concern to clinicians are explained within the framework proposed.
Separation anxiety
For example, a new light is thrown on the prob- lem of separation anxiety, namely anxiety about losing, or becoming separated from, someone loved. Why 'mere separation' should cause anxi- ety has been a mystery. Freud wrestled with the problem and advanced a number of hypotheses (Freud, 1926; Strachey, 1959). Every other lead- ing analyst has done the same. With no means of evaluating them, many divergent schools of thought have proliferated.
The problem lies, I believe, in an unexamined assumption, made not only by psychoanalysts but by more traditional psychiatrists as well, that fear is aroused in a mentally healthy person only in situations that everyone would perceive as in- trinsically painful or dangerous, or that are per- ceived so by a person only because of his having
80/362
? ? ? become conditioned to them. Since fear of separ- ation and loss does not fit this formula, analysts have concluded that what is feared is really some other situation; and a great variety of hypotheses have been advanced.
The difficulties disappear, however, when an ethological approach is adopted. For it then be- comes evident that man, like other animals, re- sponds with fear to certain situations, not be- cause they carry a high risk of pain or danger, but because they signal an increase of risk. Thus, just as animals of many species, including man, are disposed to respond with fear to sudden move- ment or a marked change in level of sound or light because to do so has survival value, so are many species, including man, disposed to re- spond to separation from a potentially caregiving figure and for the same reasons.
When separation anxiety is seen in this light, as a basic human disposition, it is only a small step to understand why it is that threats to abandon a child, often used as a means of control, are so very terrifying. Such threats, and also threats of suicide by a parent, are, we now know, common causes of intensified separation anxiety. Their ex- traordinary neglect in traditional clinical theory
81/362
? ? ? is due, I suspect, not only to an inadequate theory of separation anxiety but to a failure to give prop- er weight to the powerful effects, at all ages, of real-life events.
Not only do threats of abandonment create in- tense anxiety but they also arouse anger, often also of intense degree, especially in older children and adolescents. This anger, the function of which is to dissuade the attachment figure from carrying out the threat, can easily become dys- functional. It is in this light, I believe, that we can understand such absurdly paradoxical behaviour as the adolescent, reported by Burnham (1965), who, having murdered his mother, exclaimed, 'I couldn't stand to have her leave me. '
Other pathogenic family situations are readily understood in terms of attachment theory. One fairly common example is when a child has such a close relationship with his mother that he has difficulty in developing a social life outside the family, a relationship sometimes described as symbiotic. In a majority of such cases the cause of the trouble can be traced to the mother who, hav- ing grown up anxiously attached as a result of a difficult childhood, is now seeking to make her own child her attachment figure. So far from the
82/362
? ? ? child being over-indulged, as is sometimes asser- ted, he is being burdened with having to care for his own mother. Thus, in these cases, the normal relationship of attached child to caregiving par- ent is found to be inverted.
Mourning
Whilst separation anxiety is the usual response to a threat or some other risk of loss, mourning is the usual response to a loss after it has occurred. During the early years of psychoanalysis a num- ber of analysts identified losses, occurring during childhood or in later life, as playing a causal role in emotional disturbance, especially in depressive disorders; and by 1950 a number of theories about the nature of mourning, and other re- sponses to loss, had been advanced. Moreover, much sharp controversy had already been en- gendered. This controversy, which began during the 30s, arose from the divergent theories about infant development that had been elaborated in Vienna and London. Representative examples of the different points of view about mourning are those expressed in Helene Deutsch's Absence of Grief (1937) and Melanie Klein's Mourning and its Relation to Manic-Depressive States (1940).
83/362
? ? ? Whereas Deutsch held that, due to inadequate psychic development, children are unable to mourn, Klein held that they not only can mourn but do. In keeping with her strong emphasis on feeding, however, she held that the object mourned was the lost breast; and, in addition, she attributed a complex fantasy-life to the in- fant. Opposite though these theoretical positions are, both were constructed using the same meth- odology, namely by inferences about earlier phases of psychological development based on observations made during the analysis of older, and emotionally disturbed, subjects. Neither the- ory had been checked by direct observation of how ordinary children of different ages respond to a loss.
Approaching the problem prospectively, as I did, led me to different conclusions.
During the early 1950s Robertson and I had generalized the sequence of responses seen in young children during temporary separation from mother as those of protest, despair, and detachment (Robertson and Bowlby, 1952). A few years later, when reading a study by Marris (1958) of how widows respond to loss of husband, I was struck by the similarity of the responses he describes to
84/362
? ? ? those of young children. This led me to a system- atic study of the literature on mourning, espe- cially the mourning of healthy adults. The se- quence of responses that commonly occur, it be- came clear, was very different from what clinical theorists had been assuming. Not only does mourning in mentally healthy adults last far longer than the six months often suggested in those days, but several component responses widely regarded as pathological were found to be common in healthy mourning. These include an- ger, directed at third parties, the self, and some- times at the person lost, disbelief that the loss has occurred (misleadingly termed denial), and a tendency, often though not always unconscious, to search for the lost person in the hope of re- union. The clearer the picture of mourning re- sponses in adults became, the clearer became their similarities to the responses observed in childhood. This conclusion, when first advanced (Bowlby, 1960, 1961), was much criticized; but it has now been amply supported by a number of subsequent studies (e. g. Parkes, 1972; Kliman, 1965; Furman, 1974; Raphael, 1982).
Once an accurate picture of healthy mourning has been obtained, it becomes possible to identify
85/362
? ? ? features that are truly indicative of pathology. It becomes possible also to discern many of the con- ditions that promote healthy mourning and those that lead in a pathological direction. The belief that children are unable to mourn can then be seen to derive from generalizations that had been made from the analyses of children whose mourning had followed an atypical course. In many cases this had been due either to the child never having been given adequate information about what had happened, or else to there having been no one to sympathize with him and help him gradually come to terms with his loss, his yearning for his lost parent, his anger, and his sorrow.
Defensive processes
The next step in this reformulation of theory was to consider how defensive processes could best be conceptualized, a crucial step since defensive pro- cesses have always been at the heart of psycho- analytic theory. Although as a clinician I have in- evitably been concerned with the whole range of defences, as a research worker I have directed my attention especially to the way a young child be- haves towards his mother after a spell in a
86/362
? ? ? hospital or residential nursery unvisited. In such circumstances it is common for a child to begin by treating his mother almost as though she were a stranger, but then, after an interval, usually of hours or days, to become intensely clinging, anxious lest he lose her again, and angry with her should he think he may. In some way all his feel- ing for his mother and all the behaviour towards her we take for granted, keeping within range of her and most notably turning to her when frightened or hurt, have suddenly vanished--only to reappear again after an interval. That was the condition James Robertson and I termed detach- ment and which we believed was a result of some defensive process operating within the child.
Whereas Freud in his scientific theorizing felt confined to a conceptual model that explained all phenomena, whether physical or biological, in terms of the disposition of energy, today we have available conceptual models of much greater vari- ety. Many draw on such interrelated concepts as organization, pattern, and information; while the purposeful activities of biological organisms can be conceived in terms of control systems struc- tured in certain ways. With supplies of physical energy available to them, these systems become
87/362
? ? ? active on receipt of certain sorts of signal and in- active on receipt of signals of other sorts. Thus the world of science in which we live is radically different from the world Freud lived in at the turn of the century, and the concepts available to us immeasurably better suited to our problems than were the very restricted ones available in his day.
If we return now to the strange detached beha- viour a young child shows after being away for a time with strange people in a strange place, what is so peculiar about it is, of course, the absence of attachment behaviour in circumstances in which we would confidently expect to see it. Even when he has hurt himself severely, such a child shows no sign of seeking comfort. Thus signals that would ordinarily activate attachment behaviour are failing to do so. This suggests that in some way and for some reason these signals are failing to reach the behavioural system responsible for attachment behaviour, that they are being blocked off, and the behavioural system itself is thereby immobilized. What this means is that a system controlling such crucial behaviour as at- tachment can in certain circumstances be rendered either temporarily or permanently
88/362
? ? ? incapable of being activated, and with it the whole range of feeling and desire that normally accompanies it is rendered incapable of being aroused.
In considering how this deactivation might be effected I turn to the work of the cognitive psy- chologists (e. g. Norman, 1976; Dixon, 1971, 1981) who, during the past 20 years, have revolution- ized our knowledge of how we perceive the world and how we construe the situations we are in. Amongst much else that is clinically congenial, this revolution in cognitive theory not only gives unconscious mental processes the central place in mental life that analysts have always claimed for them, but presents a picture of the mental appar- atus as being well able to shut off information of certain specified types and of doing so selectively without the person being aware of what is happening.
In the emotionally detached children described earlier and also, I believe, in adults who have de- veloped the kind of personality that Winnicott (1960) describes as 'false self' and Kohut (1977) as 'narcissistic', the information being blocked off is of a very special type. So far from its being the routine exclusion of irrelevant and potentially
89/362
? ? ? distracting information that we engage in all the time and that is readily reversible, what are being excluded in these pathological conditions are the signals, arising from both inside and outside the person, that would activate their attachment be- haviour and that would enable them both to love and to experience being loved. In other words, the mental structures responsible for routine se- lective exclusion are being employed--one might say exploited--for a special and potentially patho- logical purpose. This form of exclusion I refer to--for obvious reasons--as defensive exclusion, which is, of course, only another way of describ- ing repression. And, just as Freud regarded re- pression as the key process in every form of de- fence, so I see the role of defensive exclusion. 3 A fuller account of this, an information-processing approach to the problem of defence, in which de- fences are classified into defensive processes, de- fensive beliefs, and defensive activities, is given in an early chapter of Loss (Bowlby, 1980).
An alternative framework
During the time it has taken to develop the con- ceptual framework described here Margaret Mahler has been concerned with many of the
90/362
? ? ? same clinical problems and some of the same fea- tures of children's behaviour; and she also has been developing a revised conceptual framework to account for them, set out fully in her book The Psychological Birth of the Human Infant (Mahler, Pine, and Bergman, 1975). To compare alternative frameworks is never easy, as Kuhn (1962) emphasizes, and no attempt is made to do so here. Elsewhere (e. g. Bowlby, 1981) I describe what I believe to be some of the strengths of the framework I favour, including its close related- ness to empirical data, both clinical and develop- mental, and its compatibility with current ideas in evolutionary biology and neurophysiology; whilst what I see as the shortcomings of Mahler's framework are trenchantly criticized by Peterfre- und (1978) and Klein (1981).
In brief, Mahler's theories of normal develop- ment, including her postulated normal phases of autism and symbiosis, are shown to rest not on observation but on preconceptions based on tra- ditional psychoanalytic theory and, in doing so, to ignore almost entirely the remarkable body of new information about early infancy that has been built up from careful empirical studies over the past two decades. Although some of the
91/362
? ? ? clinical implications of Mahler's theory are not very different from those of attachment theory, and her concept of return to base to 'refuel' is similar to that of use of an attachment figure as a secure base from which to explore, the key con- cepts with which the two frameworks are built are very different.
RESEARCH
Nothing has been so rewarding as the immense amount of careful research to which the early work on maternal deprivation has given rise. The literature is now enormous and far beyond the compass of an account of this sort to summarize. Fortunately, moreover, it is unnecessary since a comprehensive and critical review of the field has been published by Rutter (1979) who concludes by referring to the 'continuing accumulation of evidence showing the importance of deprivation and disadvantage on children's psychological de- velopment' and expressing the view that the ori- ginal arguments 'have been amply confirmed'. A principal finding of recent work is the extent to which two or more adverse experiences interact so that the risk of a psychological disturbance fol- lowing is multiplied, often many times over. An
92/362
? ? ? example of this interactive effect of adverse ex- periences is seen in the findings of Brown and Harris (1978) derived from their studies of de- pressive disorders in women. (During the last decade this group has published many further findings of the greatest interest, see Harris (1988). )
Not only is there this strongly interactive effect of adverse experiences but there is an increased likelihood for someone who has had one adverse experience to have another. For example, 'people brought up in unhappy or disrupted homes are more likely to have illegitimate children, to be- come teenage mothers, to make unhappy mar- riages, and to divorce' (Rutter, 1979). Thus ad- verse childhood experiences have effects of at least two kinds. First they make the individual more vulnerable to later adverse experiences. Se- condly they make it more likely that he or she will meet with further such experiences. Whereas the earlier adverse experiences are likely to be wholly independent of the agency of the individual con- cerned, the later ones are likely to be the con- sequences of his or her own actions, actions that spring from those disturbances of personality to which the earlier experiences have given rise.
93/362
? ? ? Of the many types of psychological disturbance that are traceable, at least in part, to one or an- other pattern of maternal deprivation, the effects on parental behaviour and thereby on the next generation are potentially the most serious. Thus a mother who, due to adverse experiences during childhood, grows up to be anxiously attached is prone to seek care from her own child and thereby lead the child to become anxious, guilty, and perhaps phobic (see review in Bowlby, 1973). A mother who as a child suffered neglect and fre- quent severe threats of being abandoned or beaten is more prone than others to abuse her child physically (DeLozier, 1982), resulting in the adverse effects on the child's developing person- ality recorded, amongst others, by George and Main (1979). Systematic research into the effects of childhood experiences on the way mothers and fathers treat their children has only just begun and seems likely to be one of the most fruitful of all fields for further research. Other research leads are described in a recent symposium edited by Parkes and Stevenson-Hinde (1982).
My reason for giving so much space in this ac- count to the development of theory is not only because it has occupied so much of my time but
94/362
? ? ? because, as Kurt Lewin remarked long ago, 'There is nothing so practical as a good theory', and, of course, nothing so handicapping as a poor one. Without good theory as a guide, research is likely to be difficult to plan and to be unproduct- ive, and findings are difficult to interpret. Without a reasonably valid theory of psychopath- ology, therapeutic techniques tend to be blunt and of uncertain benefit. Without a reasonably valid theory of aetiology, systematic and agreed measures of prevention will never be supported. My hope is that in the long term the greatest value of the theory proposed may prove to be the light it throws on the conditions most likely to promote healthy personality development. Only when those conditions are clear beyond doubt will parents know what is best for their children and will communities be willing to help them provide it.
1 Ronald Hargreaves's premature death in 1962, when professor of psychiatry at Leeds, was a grievous loss to preventive psychiatry.
2 This is the term Thomas Kuhn (1974) now uses to re- place 'paradigm', the term he used in his earlier work (Kuhn, 1962).
95/362
? ? ? 3 As Spiegel (1981) points out, my term 'defensive ex- clusion' carries a meaning very similar to Sullivan's term 'selective inattention'.
3
PSYCHOANALYSIS AS ART AND SCIENCE
During the summer of 1978 I was invited to give a number of lectures in Canada. Among the in- vitations was one from the Canadian Psycho- analytic Society to give their academic lecture to the annual meeting of the Society to be held in Quebec City. The topic I selected is one which had concerned me for some years, and about which I believe there is still a great deal of con- fused thinking.
In taking as my theme psychoanalysis as art and science I want to draw attention to what I believe to be two very different aspects of our discip- line--the art of psychoanalytic therapy and the science of psychoanalytic psychology--and in do- ing so to emphasize, on the one hand, the dis- tinctive value of each and, on the other, the gulf that divides them--in regard both to the contrast- ing criteria by which each should be judged and
? 97/362
? ? ? the very different mental outlook that each de- mands. In emphasizing these distinctions, I can- not help regretting that the word psychoanalysis came early to be used ambiguously as Freud him- self described it. 'While it was originally the name of a particular therapeutic method', he writes in his autobiography (1925), 'it has now also become the name of a science--the science of unconscious mental processes'.
The distinction I am drawing, of course, is not confined to psychoanalysis. It applies in every field in which the practice of a profession or a craft gives birth to a body of scientific know- ledge--the blacksmith to metallurgy, the civil en- gineer to soil mechanics, the farmer to plant physiology, and the physician to the medical sci- ences. In each of these fields the roles differenti- ate. On the one hand are the practitioners, on the other the scientists, with a limited number of in- dividuals attempting to combine both roles. As history shows, this process of differentiation of- ten proves painful and misunderstandings are frequent. Since I believe differentiation is bound to come also in our own field, and is perhaps already overdue, let us consider some of the diffi- culties and misunderstandings to which it may all
98/362
? ? ? too easily give rise in the hope of avoiding them or mitigating their consequences.
I start by contrasting the roles of practitioner and research scientist and do so under three headings, using the case of medicine as an example.
FOCUS OF STUDY
The aim of the practitioner is to take into account as many aspects as he can of each and every clin- ical problem with which he is called upon to deal. This requires him not only to apply any scientific principle that appears relevant but also to draw on such personal experience of the condition as he may have acquired and, especially, to attend to that unique combination of features met with in each patient. Knowing how greatly patients dif- fer, the experienced clinician recognizes that a form of treatment well suited to one would be totally inappropriate to another.
Taking all factors into account and giving each its due weight is the art of clinical judgement.
The outlook of the research scientist is quite different. In his efforts to discern general pat- terns underlying individual variety he ignores the particular and strives to simplify, risking thereby
99/362
? ? ? over- simplification. If he is wise he will probably concentrate attention on a limited aspect of a limited problem. If in making his selection he proves sagacious, or simply lucky, he may not only elucidate the problem selected but also de- velop ideas applicable to a broader range. If his selection proves unwise or unlucky he may merely end up knowing more and more about less and less. That is the risk every researcher runs. The art of research lies in selecting a limited manageable problem and the methods that will best help solve it. This brings me to my second point.
MODES OF ACQUIRING INFORMATION
In the methods available to him for acquiring in- formation the practitioner has certain great ad- vantages over the research scientist but also cer- tain great disadvantages. Let us start with the advantages.
In his role of giving help the practitioner is per- mitted access to information of certain kinds that remain closed to the scientist: as a friend of mine is fond of saving, it's only surgeons who are al- lowed to cut you open to see what's inside. In an analogous way it is only by treating a patient
100/362
? ? ? therapeutically that a psychoanalyst is given ac- cess to much of importance going on in a person's mind. In both professions, moreover, practition- ers are permitted to intervene in specified ways and privileged to observe what the consequences of such interventions are. These are immense ad- vantages and psychoanalysts have not been slow to exploit them.
Yet no science can prosper for long without en- listing new methods to cross-check on observa- tions made and on hypotheses born of older methods. Here the research scientist is likely to have the advantage. In the medical sciences, physiologists and pathologists have made im- mense advances by means of animal experi- ments, tissue culture, biochemical analyses, and a thousand other ingenious techniques. Indeed, it is a hallmark of a creative scientist that he devises new means by which phenomena, perhaps already well studied by other methods, can be ob- served in some new way.
It is in this area, I believe, that the ambiguous use of the word psychoanalysis has done greatest harm. For it has led some analysts to suppose that the only method of enquiry appropriate for the advancement of psychoanalytic science is that
101/362
? ? ? of treating a patient psychoanalytically. Since I believe this to be a profound misunderstanding I shall be saying a good deal more about it. Before doing so, however, I want to say a word about the place of scepticism and faith in the respective worlds of scientist and practitioner.
SCEPTICISM AND FAITH
In his day-to-day work it is necessary for a scient- ist to exercise a high degree of criticism and self- criticism: and in the world he inhabits neither the data nor the theories of a leader, however ad- mired personally he may be, are exempt from challenge and criticism. There is no place for authority.
The same is not true in the practice of a profes- sion. If he is to be effective a practitioner must be prepared to act as though certain principles and certain theories were valid; and in deciding which to adopt he is likely to be guided by those with ex- perience from whom he learns. Since, moreover, there is a tendency in all of us to be impressed whenever the application of a theory appears to have been successful, practitioners are at special risk of placing greater confidence in a theory than the evidence available may justify.
102/362
? ? ? From the standpoint of clinical practice this is no bad thing. On the contrary, there is abundant evidence that the great majority of patients are helped by the faith and hope that a practitioner brings to his work; whilst it is often the very lack of these qualities that makes so many excellent research workers sadly ill-suited to be therapists.
Yet, though faith in the validity of particular data and in particular theories is out of place in a scientist, I do not wish to imply that he is nothing but a sceptic. On the contrary, his whole way of living is founded on faith, faith that in the long run the best route to reliable knowledge is the ap- plication of scientific method.
I am, of course, aware that there are many psy- choanalysts who do not share this faith and who believe that the types of problem with which we deal lie far outside the scope of science. This is a view I respect, though I do not share: nor, of course, did Freud. Yet even those of us who are most enthusiastic about applying scientific meth- od in our field must recognize that there may well be problems that it can never solve. We simply do not know. Our task, as I see it, is to apply our method as skilfully as we can, on the one hand believing that the area of reliable knowledge will
103/362
? ? ? thereby be expanded and, on the other, accepting that there are likely always to remain still greater areas lying beyond the scope of any existing mode of scientific enquiry.
To many of you, I am afraid, engaged in thera- peutic practice but also hoping to contribute to the advance of psychoanalytic science, the con- trasts I am drawing between the roles of practi- tioner and scientist will hardly be welcome. Yet I believe that it is only by recognizing these differ- ences and acting accordingly that the strengths of each role can be used to fullest advantage--or that any one person can occupy both of them with any hope of success. As practitioners we deal in complexity; as scientists we strive to simplify. As practitioners we use theory as a guide; as sci- entists we challenge that same theory. As practi- tioners we accept restricted modes of enquiry; as scientists we enlist every method we can.
Earlier I remarked on the need for every devel- oping science to devise new methods for obtain- ing data. The reason for this is that, however pro- ductive any one method may be, it is bound to have its limitations, whilst there is always a pro- spect that some other method may compensate for them. Thus the new method may be in no way
104/362
? ? ? superior to the old; indeed it may have great lim- itations. Its usefulness lies simply in the fact that its strengths and its limitations are different. Per- haps I can illustrate the point by reference to my own work.
When I qualified in psychoanalysis in 1937, members of the British Society were occupied in exploring the fantasy worlds of adults and chil- dren, and it was regarded as almost outside the proper interest of an analyst to give systematic attention to a person's real experiences. That was a time when Freud's famous about-turn of 1897 regarding the aetiology of hysteria1 had led to the view that anyone who places emphasis on what a child's real experiences may have been, and per- haps still are being, was regarded as pitifully nai? ve. Almost by definition it was assumed that anyone interested in the external world could not be interested in the internal world, indeed was al- most certainly running away from it.
To me as a biologist this contrast of internal with external, of organism with environment, never appealed. Furthermore, as a psychiatrist engaged in work with children and families and deeply influenced by the insights of two analytic- ally oriented social workers, I was daily
105/362
?
The six months I spent with the World Health Organisation in 1950 gave me the chance not only to read the literature and to discuss it with the authors, but also to meet many others in Europe and the United States with experience of the field. Soon after the end of my contract I submitted my report, which was published early in 1951 as a
64/362
? ? ? WHO monograph entitled Maternal Care and Mental Health. In it I reviewed the far from neg- ligible evidence then available regarding the ad- verse influences on personality development of inadequate maternal care during early childhood, called attention to the acute distress of young children who find themselves separated from those they know and love, and made recommend- ations of how best to avoid, or at least mitigate, the short- and long-term ill effects. During the next few years this report was translated into a dozen other languages and appeared also in a cheap abridged edition in English.
Influential though the written word may often be, it has nothing like the emotional impact of a movie. Throughout the 1950s Rene? Spitz's early film Grief: A Peril in Infancy (1947), and James Robertson's A Two-Year-Old Goes to Hospital (1952) together had an enormous influence. Not only did they draw the attention of professional workers to the immediate distress and anxiety of young children in an institutional setting but they proved powerful instruments for promoting changes in practice. In this field Robertson was to play a leading part (e. g. Robertson, 1958, 1970).
65/362
? ? ? Although by the end of the 1950s a great many of those working in child psychiatry and psycho- logy and in social work, and some also of those in paediatrics and sick children's nursing, had ac- cepted the research findings and were imple- menting change, the sharp controversy aroused by the early publications and films continued. Psychiatrists trained in traditional psychiatry and psychologists who adopted a learning-theory ap- proach never ceased to point to the deficiencies of the evidence and to the lack of an adequate ex- planation of how the types of experience implic- ated could have the effects on personality devel- opment claimed. Many psychoanalysts, in addi- tion, especially those whose theory focused on the role of fantasy in psychopathology to the relative exclusion of the influence of real life events, re- mained unconvinced and sometimes very critical. Meanwhile, research continued. For example, at Yale Sally Provence and Rose Lipton, were mak- ing a systematic study of institutionalized infants in which they compared their development with that of infants living in a family (Provence and Lipton, 1962). At the Tavistock members of my small research group were active collecting fur- ther data on the short-term effects on a young
66/362
? ? ? child of being in the care of strange people in a strange place for weeks and sometimes months at a time (see especially the studies by Christoph Heinicke, 1956 and, with Ilse Westheimer, 1966), whilst I addressed myself to the theoretical prob- lems posed by our data.
Meanwhile the field was changing. One import- ant influence was the publication in 1963 by the World Health Organisation of a collection of art- icles in which the manifold effects of the various types of experience covered by the term 'depriva- tion of maternal care' were reassessed. Of the six articles, by far the most comprehensive was by my colleague Mary Ainsworth (1962). In it she not only reviewed the extensive and diverse evid- ence and considered the many issues that had given rise to controversy but also identified a large number of problems requiring further research.
A second important influence was the publica- tion, beginning during the late fifties, of Harry Harlow's studies of the effects of maternal deprivation on rhesus monkeys; and once again film played a big part. Harlow's work in the Un- ited States had been stimulated by Spitz's reports. In the United Kingdom complementary studies
67/362
? ? ? by Robert Hinde had been stimulated by our work at the Tavistock. For the next decade a stream of experimental results from those two scientists (see summaries in Harlow and Harlow, 1965 and Hinde and Spencer-Booth, 1971), com- ing on top of the Ainsworth review, undermined the opposition. Thereafter nothing more was heard of the inherent implausibility of our hypo- theses; and criticism became more constructive.
Much, of course, remained uncertain. Even if the reality of short-term distress and behavioural disturbance is granted, what evidence is there, it was asked, that the ill effects can persist? What features of the experience, or combination of fea- tures, are responsible for the distress? And, should it prove true that in some cases ill effects do persist, how is that to be accounted for? How does it happen that some children seem to come through very unfavourable experiences relatively unharmed? How important is it that a child should be cared for most of the time by one prin- cipal caregiver? In less developed societies it was claimed (wrongly as it turns out) that multiple mothering is not uncommon. In addition to all these legitimate questions, moreover, there were misunderstandings. Some supposed that
68/362
? ? ? advocates of the view that a child should be cared for most of the time by a principal mother-figure held that that had to be the child's natural moth- er--the so-called blood-tie theory. Others sup- posed that, in advocating that a child should 'ex- perience a warm intimate and continuous rela- tionship with his mother (or permanent mother- substitute)', proponents were prescribing a re- gime in which a mother had to care for her child 24 hours a day, day in and day out, with no res- pite. In a field in which strong feelings are aroused and almost everyone has some sort of vested interest, clear unbiased thinking is not al- ways easy.
A NEW LOOK AT THEORY
The monograph Maternal Care and Mental Health is in two parts. The first reviews the evid- ence regarding the adverse effects of maternal deprivation, the second discusses means for pre- venting it. What was missing, as several reviewers pointed out, was any explanation of how experi- ences subsumed under the broad heading of ma- ternal deprivation could have the effects on per- sonality development of the kinds claimed. The reason for this omission was simple: the data
69/362
? ? ? were not accommodated by any theory then cur- rent and in the brief time of my employment by the World Health Organisation there was no pos- sibility of developing a new one.
The child's tie to his mother
At that time it was widely held that the reason a child develops a close tie to his mother is that she feeds him. Two kinds of drive are postulated, primary and secondary. Food is thought of as primary; the personal relationship, referred to as 'dependency', as secondary. This theory did not seem to me to fit the facts. For example, were it true, an infant of a year or two should take read- ily to whomever feeds him and this clearly was not the case. An alternative theory, stemming from the Hungarian school of psychoanalysis, postulated a primitive object relation from the beginning. In its best-known version, however, the one advocated by Melanie Klein, mother's breast is postulated as the first object, and the greatest emphasis is placed on food and orality and on the infantile nature of 'dependency'. None of these features matched my experience of children.
70/362
? ? ? But if the current dependency theories were in- adequate, what was the alternative?
During the summer of 1951 a friend mentioned to me the work of Lorenz on the following re- sponses of ducklings and goslings. Reading about this and related work on instinctive behaviour re- vealed a new world, one in which scientists of high calibre were investigating in non-human species many of the problems with which we were grappling in the human, in particular the relat- ively enduring relationships that develop in many species, first between young and parents and later between mated pairs, and some of the ways in which these developments can go awry. Could this work, I asked myself, cast light on a problem central to psychoanalysis, that of 'instinct' in humans?
Next followed a long phase during which I set about trying to master basic principles and to ap- ply them to our problems, starting with the nature of the child's tie to his mother. Here Lorenz's work on the following response of duck- lings and goslings (Lorenz, 1935) was of special interest. It showed that in some animal species a strong bond to an individual motherfigure could develop without the intermediary of food: for
71/362
? ? ? these young birds are not fed by parents but feed themselves by catching insects. Here then was an alternative model to the traditional one, and one that had a number of features that seemed pos- sibly to fit the human case. Thereafter, as my grasp of ethological principles increased and I ap- plied them to one clinical problem after another, I became increasingly confident that this was a promising approach. Thus, having adopted this novel point of view, I decided to 'follow it up through the material as long as the application of it seems to yield results' (to borrow a phrase of Freud's).
From 1957, when The Nature of the Child's Tie to his Mother was first presented, through 1969 when Attachment appeared, until 1980 with the publication of Loss I concentrated on this task. The resulting conceptual framework2 is designed to accommodate all those phenomena to which Freud called attention--for example love rela- tions, separation anxiety, mourning, defence, an- ger, guilt, depression, trauma, emotional detach- ment, sensitive periods in early life--and so to of- fer an alternative to the traditional metapsycho- logy of psychoanalysis and to add yet another to the many variants of the clinical theory now
72/362
? ? ? extant. How successful these ideals will prove only time will tell.
As Kuhn has emphasized, any novel conceptual framework is difficult to grasp, especially so for those long familiar with a previous one. Of the many difficulties met with in understanding the framework advocated, I describe only a few. One is that, instead of starting with a clinical syn- drome of later years and trying to trace its origins retrospectively, I have started with a class of childhood traumata and tried to trace the se- quelae prospectively. A second is that, instead of starting with the private thoughts and feelings of a patient, as expressed in free associations or play, and trying to build a theory of personality development from those data, I have started with observations of the behaviour of children in cer- tain sorts of defined situation, including records of the feelings and thoughts they express, and have tried to build a theory of personality devel- opment from there. Other difficulties arise from my use of concepts such as control system (in- stead of psychic energy) and developmental path- way (instead of libidinal phase), which, although now firmly established as key concepts in all the biological sciences, are still foreign to the
73/362
? ? ? thinking of a great many psychologists and clinicians.
Having discarded the secondary-drive, depend- ency theory of the child's tie to his mother, and also the Kleinian alternative, a first task was to formulate a replacement. This led to the concept of attachment behaviour with its own dynamics distinct from the behaviour and dynamics of either feeding or sex, the two sources of human motivation for long widely regarded as the most fundamental. Strong support for this step soon came from Harlow's finding that, in another primate species--rhesus macaques--infants show a marked preference for a soft dummy 'mother', despite its providing no food, to a hard one that does provide it (Harlow and Zimmermann, 1959).
Attachment behaviour is any form of behaviour that results in a person attaining or maintaining proximity to some other clearly identified indi- vidual who is conceived as better able to cope with the world. It is most obvious whenever the person is frightened, fatigued, or sick, and is as- suaged by comforting and caregiving. At other times the behaviour is less in evidence. Neverthe- less for a person to know that an attachment fig- ure is available and responsive gives him a strong
74/362
? ? ? and pervasive feeling of security, and so encour- ages him to value and continue the relationship. Whilst attachment behaviour is at its most obvi- ous in early childhood, it can be observed throughout the life cycle, especially in emergen- cies. Since it is seen in virtually all human beings (though in varying patterns), it is regarded as an integral part of human nature and one we share (to a varying extent) with members of other spe- cies. The biological function attributed to it is that of protection. To remain within easy access of a familiar individual known to be ready and willing to come to our aid in an emergency is clearly a good insurance policy--whatever our age.
By conceptualizing attachment in this way, as a fundamental form of behaviour with its own in- ternal motivation distinct from feeding and sex, and of no less importance for survival, the beha- viour and motivation are accorded a theoretical status never before given them--though parents and clinicians alike have long been intuitively aware of their importance. As already emphas- ized, the terms 'dependency' and 'dependency need' that have hitherto been used to refer to them have serious disadvantages. In the first
75/362
? ? ? place 'dependency' has a perjorative flavour; in the second it does not imply an emotionally charged relationship to one or a very few clearly preferred individuals; and in the third no valu- able biological function has ever been attributed to it.
It is now 30 years since the notion of attach- ment was first advanced as a useful way of con- ceptualizing a form of behaviour of central im- portance not only to clinicians and to develop- mental psychologists but to every parent as well. During that time attachment theory has been greatly clarified and amplified. The most notable contributors have been Robert Hinde who, in ad- dition to his own publications (e. g. 1974), has constantly guided my own thinking, and Mary Ainsworth who, starting in the late 50s, has pion- eered empirical studies of attachment behaviour both in Africa (1963, 1967) and in the USA (Ainsworth and Wittig, 1969; Ainsworth et al. , 1978), and has also helped greatly to develop the- ory (e. g. 1969, 1982). Her work, together with that of her students and others influenced by her (which has expanded dramatically since this lec- ture was given and is described in some detail in Lecture 7), has led attachment theory to be
76/362
? ? ? widely regarded as probably the best supported theory of socio-emotional development yet avail- able (Rajecki, Lamb, and Obmascher, 1978; Rut- ter, 1980; Parkes and Stevenson-Hinde, 1982; Sroufe, 1986).
Because my starting point in developing theory was observations of behaviour, some clinicians have assumed that the theory amounts to no more than a version of behaviourism. This mis- take is due in large part to the unfamiliarity of the conceptual framework proposed and in part to my own failure in early formulations to make clear the distinction to be drawn between an at- tachment and attachment behaviour. To say of a child (or older person) that he is attached to, or has an attachment to, someone means that he is strongly disposed to seek proximity to and con- tact with that individual and to do so especially in certain specified conditions. The disposition to behave in this way is an attribute of the attached person, a persisting attribute which changes only slowly over time and which is unaffected by the situation of the moment. Attachment behaviour, by contrast, refers to any of the various forms of behaviour that the person engages in from time
77/362
? ? ? to time to obtain and/or maintain a desired proximity.
There is abundant evidence that almost every child habitually prefers one person, usually his mother-figure, to whom to go when distressed but that, in her absence, he will make do with someone else, preferably someone whom he knows well. On these occasions most children show a clear hierarchy of preference so that, in extremity and with no one else available, even a kindly stranger may be approached. Thus, whilst attachment behaviour may in differing circum- stances be shown to a variety of individuals, an enduring attachment, or attachment bond, is confined to very few. Should a child fail to show such clear discrimination, it is likely he is severely disturbed.
The theory of attachment is an attempt to ex- plain both attachment behaviour, with its episod- ic appearance and disappearance, and also the enduring attachments that children and other in- dividuals make to particular others. In this theory the key concept is that of behavioural system. This is conceived on the analogy of a physiologic- al system organized homeostatically to ensure that a certain physiological measure, such as
78/362
? ? ? body temperature or blood pressure, is held between appropriate limits. In proposing the concept of a behavioural system to account for the way a child or older person maintains his re- lation to his attachment figure between certain limits of distance or accessibility, no more is done than to use these well-understood principles to account for a different form of homeostasis, namely one in which the set limits concern the organism's relation to clearly identified persons in, or other features of, the environment and in which the limits are maintained by behavioural instead of physiological means.
In thus postulating the existence of an internal psychological organization with a number of highly specific features, which include represent- ational models of the self and of attachment fig- ure(s), the theory proposed can be seen as having all the same basic properties as those that charac- terize other forms of structural theory, of which the variants of psychoanalysis are some of the best known, and that differentiate them so sharply from behaviourism in its many forms. Historically attachment theory was developed as a variant of object-relations theory.
79/362
? ? ? The reason why in this account I have given so much space to the concept and theory of attach- ment is that, once those principles are grasped, there is little difficulty in understanding how the many other phenomena of central concern to clinicians are explained within the framework proposed.
Separation anxiety
For example, a new light is thrown on the prob- lem of separation anxiety, namely anxiety about losing, or becoming separated from, someone loved. Why 'mere separation' should cause anxi- ety has been a mystery. Freud wrestled with the problem and advanced a number of hypotheses (Freud, 1926; Strachey, 1959). Every other lead- ing analyst has done the same. With no means of evaluating them, many divergent schools of thought have proliferated.
The problem lies, I believe, in an unexamined assumption, made not only by psychoanalysts but by more traditional psychiatrists as well, that fear is aroused in a mentally healthy person only in situations that everyone would perceive as in- trinsically painful or dangerous, or that are per- ceived so by a person only because of his having
80/362
? ? ? become conditioned to them. Since fear of separ- ation and loss does not fit this formula, analysts have concluded that what is feared is really some other situation; and a great variety of hypotheses have been advanced.
The difficulties disappear, however, when an ethological approach is adopted. For it then be- comes evident that man, like other animals, re- sponds with fear to certain situations, not be- cause they carry a high risk of pain or danger, but because they signal an increase of risk. Thus, just as animals of many species, including man, are disposed to respond with fear to sudden move- ment or a marked change in level of sound or light because to do so has survival value, so are many species, including man, disposed to re- spond to separation from a potentially caregiving figure and for the same reasons.
When separation anxiety is seen in this light, as a basic human disposition, it is only a small step to understand why it is that threats to abandon a child, often used as a means of control, are so very terrifying. Such threats, and also threats of suicide by a parent, are, we now know, common causes of intensified separation anxiety. Their ex- traordinary neglect in traditional clinical theory
81/362
? ? ? is due, I suspect, not only to an inadequate theory of separation anxiety but to a failure to give prop- er weight to the powerful effects, at all ages, of real-life events.
Not only do threats of abandonment create in- tense anxiety but they also arouse anger, often also of intense degree, especially in older children and adolescents. This anger, the function of which is to dissuade the attachment figure from carrying out the threat, can easily become dys- functional. It is in this light, I believe, that we can understand such absurdly paradoxical behaviour as the adolescent, reported by Burnham (1965), who, having murdered his mother, exclaimed, 'I couldn't stand to have her leave me. '
Other pathogenic family situations are readily understood in terms of attachment theory. One fairly common example is when a child has such a close relationship with his mother that he has difficulty in developing a social life outside the family, a relationship sometimes described as symbiotic. In a majority of such cases the cause of the trouble can be traced to the mother who, hav- ing grown up anxiously attached as a result of a difficult childhood, is now seeking to make her own child her attachment figure. So far from the
82/362
? ? ? child being over-indulged, as is sometimes asser- ted, he is being burdened with having to care for his own mother. Thus, in these cases, the normal relationship of attached child to caregiving par- ent is found to be inverted.
Mourning
Whilst separation anxiety is the usual response to a threat or some other risk of loss, mourning is the usual response to a loss after it has occurred. During the early years of psychoanalysis a num- ber of analysts identified losses, occurring during childhood or in later life, as playing a causal role in emotional disturbance, especially in depressive disorders; and by 1950 a number of theories about the nature of mourning, and other re- sponses to loss, had been advanced. Moreover, much sharp controversy had already been en- gendered. This controversy, which began during the 30s, arose from the divergent theories about infant development that had been elaborated in Vienna and London. Representative examples of the different points of view about mourning are those expressed in Helene Deutsch's Absence of Grief (1937) and Melanie Klein's Mourning and its Relation to Manic-Depressive States (1940).
83/362
? ? ? Whereas Deutsch held that, due to inadequate psychic development, children are unable to mourn, Klein held that they not only can mourn but do. In keeping with her strong emphasis on feeding, however, she held that the object mourned was the lost breast; and, in addition, she attributed a complex fantasy-life to the in- fant. Opposite though these theoretical positions are, both were constructed using the same meth- odology, namely by inferences about earlier phases of psychological development based on observations made during the analysis of older, and emotionally disturbed, subjects. Neither the- ory had been checked by direct observation of how ordinary children of different ages respond to a loss.
Approaching the problem prospectively, as I did, led me to different conclusions.
During the early 1950s Robertson and I had generalized the sequence of responses seen in young children during temporary separation from mother as those of protest, despair, and detachment (Robertson and Bowlby, 1952). A few years later, when reading a study by Marris (1958) of how widows respond to loss of husband, I was struck by the similarity of the responses he describes to
84/362
? ? ? those of young children. This led me to a system- atic study of the literature on mourning, espe- cially the mourning of healthy adults. The se- quence of responses that commonly occur, it be- came clear, was very different from what clinical theorists had been assuming. Not only does mourning in mentally healthy adults last far longer than the six months often suggested in those days, but several component responses widely regarded as pathological were found to be common in healthy mourning. These include an- ger, directed at third parties, the self, and some- times at the person lost, disbelief that the loss has occurred (misleadingly termed denial), and a tendency, often though not always unconscious, to search for the lost person in the hope of re- union. The clearer the picture of mourning re- sponses in adults became, the clearer became their similarities to the responses observed in childhood. This conclusion, when first advanced (Bowlby, 1960, 1961), was much criticized; but it has now been amply supported by a number of subsequent studies (e. g. Parkes, 1972; Kliman, 1965; Furman, 1974; Raphael, 1982).
Once an accurate picture of healthy mourning has been obtained, it becomes possible to identify
85/362
? ? ? features that are truly indicative of pathology. It becomes possible also to discern many of the con- ditions that promote healthy mourning and those that lead in a pathological direction. The belief that children are unable to mourn can then be seen to derive from generalizations that had been made from the analyses of children whose mourning had followed an atypical course. In many cases this had been due either to the child never having been given adequate information about what had happened, or else to there having been no one to sympathize with him and help him gradually come to terms with his loss, his yearning for his lost parent, his anger, and his sorrow.
Defensive processes
The next step in this reformulation of theory was to consider how defensive processes could best be conceptualized, a crucial step since defensive pro- cesses have always been at the heart of psycho- analytic theory. Although as a clinician I have in- evitably been concerned with the whole range of defences, as a research worker I have directed my attention especially to the way a young child be- haves towards his mother after a spell in a
86/362
? ? ? hospital or residential nursery unvisited. In such circumstances it is common for a child to begin by treating his mother almost as though she were a stranger, but then, after an interval, usually of hours or days, to become intensely clinging, anxious lest he lose her again, and angry with her should he think he may. In some way all his feel- ing for his mother and all the behaviour towards her we take for granted, keeping within range of her and most notably turning to her when frightened or hurt, have suddenly vanished--only to reappear again after an interval. That was the condition James Robertson and I termed detach- ment and which we believed was a result of some defensive process operating within the child.
Whereas Freud in his scientific theorizing felt confined to a conceptual model that explained all phenomena, whether physical or biological, in terms of the disposition of energy, today we have available conceptual models of much greater vari- ety. Many draw on such interrelated concepts as organization, pattern, and information; while the purposeful activities of biological organisms can be conceived in terms of control systems struc- tured in certain ways. With supplies of physical energy available to them, these systems become
87/362
? ? ? active on receipt of certain sorts of signal and in- active on receipt of signals of other sorts. Thus the world of science in which we live is radically different from the world Freud lived in at the turn of the century, and the concepts available to us immeasurably better suited to our problems than were the very restricted ones available in his day.
If we return now to the strange detached beha- viour a young child shows after being away for a time with strange people in a strange place, what is so peculiar about it is, of course, the absence of attachment behaviour in circumstances in which we would confidently expect to see it. Even when he has hurt himself severely, such a child shows no sign of seeking comfort. Thus signals that would ordinarily activate attachment behaviour are failing to do so. This suggests that in some way and for some reason these signals are failing to reach the behavioural system responsible for attachment behaviour, that they are being blocked off, and the behavioural system itself is thereby immobilized. What this means is that a system controlling such crucial behaviour as at- tachment can in certain circumstances be rendered either temporarily or permanently
88/362
? ? ? incapable of being activated, and with it the whole range of feeling and desire that normally accompanies it is rendered incapable of being aroused.
In considering how this deactivation might be effected I turn to the work of the cognitive psy- chologists (e. g. Norman, 1976; Dixon, 1971, 1981) who, during the past 20 years, have revolution- ized our knowledge of how we perceive the world and how we construe the situations we are in. Amongst much else that is clinically congenial, this revolution in cognitive theory not only gives unconscious mental processes the central place in mental life that analysts have always claimed for them, but presents a picture of the mental appar- atus as being well able to shut off information of certain specified types and of doing so selectively without the person being aware of what is happening.
In the emotionally detached children described earlier and also, I believe, in adults who have de- veloped the kind of personality that Winnicott (1960) describes as 'false self' and Kohut (1977) as 'narcissistic', the information being blocked off is of a very special type. So far from its being the routine exclusion of irrelevant and potentially
89/362
? ? ? distracting information that we engage in all the time and that is readily reversible, what are being excluded in these pathological conditions are the signals, arising from both inside and outside the person, that would activate their attachment be- haviour and that would enable them both to love and to experience being loved. In other words, the mental structures responsible for routine se- lective exclusion are being employed--one might say exploited--for a special and potentially patho- logical purpose. This form of exclusion I refer to--for obvious reasons--as defensive exclusion, which is, of course, only another way of describ- ing repression. And, just as Freud regarded re- pression as the key process in every form of de- fence, so I see the role of defensive exclusion. 3 A fuller account of this, an information-processing approach to the problem of defence, in which de- fences are classified into defensive processes, de- fensive beliefs, and defensive activities, is given in an early chapter of Loss (Bowlby, 1980).
An alternative framework
During the time it has taken to develop the con- ceptual framework described here Margaret Mahler has been concerned with many of the
90/362
? ? ? same clinical problems and some of the same fea- tures of children's behaviour; and she also has been developing a revised conceptual framework to account for them, set out fully in her book The Psychological Birth of the Human Infant (Mahler, Pine, and Bergman, 1975). To compare alternative frameworks is never easy, as Kuhn (1962) emphasizes, and no attempt is made to do so here. Elsewhere (e. g. Bowlby, 1981) I describe what I believe to be some of the strengths of the framework I favour, including its close related- ness to empirical data, both clinical and develop- mental, and its compatibility with current ideas in evolutionary biology and neurophysiology; whilst what I see as the shortcomings of Mahler's framework are trenchantly criticized by Peterfre- und (1978) and Klein (1981).
In brief, Mahler's theories of normal develop- ment, including her postulated normal phases of autism and symbiosis, are shown to rest not on observation but on preconceptions based on tra- ditional psychoanalytic theory and, in doing so, to ignore almost entirely the remarkable body of new information about early infancy that has been built up from careful empirical studies over the past two decades. Although some of the
91/362
? ? ? clinical implications of Mahler's theory are not very different from those of attachment theory, and her concept of return to base to 'refuel' is similar to that of use of an attachment figure as a secure base from which to explore, the key con- cepts with which the two frameworks are built are very different.
RESEARCH
Nothing has been so rewarding as the immense amount of careful research to which the early work on maternal deprivation has given rise. The literature is now enormous and far beyond the compass of an account of this sort to summarize. Fortunately, moreover, it is unnecessary since a comprehensive and critical review of the field has been published by Rutter (1979) who concludes by referring to the 'continuing accumulation of evidence showing the importance of deprivation and disadvantage on children's psychological de- velopment' and expressing the view that the ori- ginal arguments 'have been amply confirmed'. A principal finding of recent work is the extent to which two or more adverse experiences interact so that the risk of a psychological disturbance fol- lowing is multiplied, often many times over. An
92/362
? ? ? example of this interactive effect of adverse ex- periences is seen in the findings of Brown and Harris (1978) derived from their studies of de- pressive disorders in women. (During the last decade this group has published many further findings of the greatest interest, see Harris (1988). )
Not only is there this strongly interactive effect of adverse experiences but there is an increased likelihood for someone who has had one adverse experience to have another. For example, 'people brought up in unhappy or disrupted homes are more likely to have illegitimate children, to be- come teenage mothers, to make unhappy mar- riages, and to divorce' (Rutter, 1979). Thus ad- verse childhood experiences have effects of at least two kinds. First they make the individual more vulnerable to later adverse experiences. Se- condly they make it more likely that he or she will meet with further such experiences. Whereas the earlier adverse experiences are likely to be wholly independent of the agency of the individual con- cerned, the later ones are likely to be the con- sequences of his or her own actions, actions that spring from those disturbances of personality to which the earlier experiences have given rise.
93/362
? ? ? Of the many types of psychological disturbance that are traceable, at least in part, to one or an- other pattern of maternal deprivation, the effects on parental behaviour and thereby on the next generation are potentially the most serious. Thus a mother who, due to adverse experiences during childhood, grows up to be anxiously attached is prone to seek care from her own child and thereby lead the child to become anxious, guilty, and perhaps phobic (see review in Bowlby, 1973). A mother who as a child suffered neglect and fre- quent severe threats of being abandoned or beaten is more prone than others to abuse her child physically (DeLozier, 1982), resulting in the adverse effects on the child's developing person- ality recorded, amongst others, by George and Main (1979). Systematic research into the effects of childhood experiences on the way mothers and fathers treat their children has only just begun and seems likely to be one of the most fruitful of all fields for further research. Other research leads are described in a recent symposium edited by Parkes and Stevenson-Hinde (1982).
My reason for giving so much space in this ac- count to the development of theory is not only because it has occupied so much of my time but
94/362
? ? ? because, as Kurt Lewin remarked long ago, 'There is nothing so practical as a good theory', and, of course, nothing so handicapping as a poor one. Without good theory as a guide, research is likely to be difficult to plan and to be unproduct- ive, and findings are difficult to interpret. Without a reasonably valid theory of psychopath- ology, therapeutic techniques tend to be blunt and of uncertain benefit. Without a reasonably valid theory of aetiology, systematic and agreed measures of prevention will never be supported. My hope is that in the long term the greatest value of the theory proposed may prove to be the light it throws on the conditions most likely to promote healthy personality development. Only when those conditions are clear beyond doubt will parents know what is best for their children and will communities be willing to help them provide it.
1 Ronald Hargreaves's premature death in 1962, when professor of psychiatry at Leeds, was a grievous loss to preventive psychiatry.
2 This is the term Thomas Kuhn (1974) now uses to re- place 'paradigm', the term he used in his earlier work (Kuhn, 1962).
95/362
? ? ? 3 As Spiegel (1981) points out, my term 'defensive ex- clusion' carries a meaning very similar to Sullivan's term 'selective inattention'.
3
PSYCHOANALYSIS AS ART AND SCIENCE
During the summer of 1978 I was invited to give a number of lectures in Canada. Among the in- vitations was one from the Canadian Psycho- analytic Society to give their academic lecture to the annual meeting of the Society to be held in Quebec City. The topic I selected is one which had concerned me for some years, and about which I believe there is still a great deal of con- fused thinking.
In taking as my theme psychoanalysis as art and science I want to draw attention to what I believe to be two very different aspects of our discip- line--the art of psychoanalytic therapy and the science of psychoanalytic psychology--and in do- ing so to emphasize, on the one hand, the dis- tinctive value of each and, on the other, the gulf that divides them--in regard both to the contrast- ing criteria by which each should be judged and
? 97/362
? ? ? the very different mental outlook that each de- mands. In emphasizing these distinctions, I can- not help regretting that the word psychoanalysis came early to be used ambiguously as Freud him- self described it. 'While it was originally the name of a particular therapeutic method', he writes in his autobiography (1925), 'it has now also become the name of a science--the science of unconscious mental processes'.
The distinction I am drawing, of course, is not confined to psychoanalysis. It applies in every field in which the practice of a profession or a craft gives birth to a body of scientific know- ledge--the blacksmith to metallurgy, the civil en- gineer to soil mechanics, the farmer to plant physiology, and the physician to the medical sci- ences. In each of these fields the roles differenti- ate. On the one hand are the practitioners, on the other the scientists, with a limited number of in- dividuals attempting to combine both roles. As history shows, this process of differentiation of- ten proves painful and misunderstandings are frequent. Since I believe differentiation is bound to come also in our own field, and is perhaps already overdue, let us consider some of the diffi- culties and misunderstandings to which it may all
98/362
? ? ? too easily give rise in the hope of avoiding them or mitigating their consequences.
I start by contrasting the roles of practitioner and research scientist and do so under three headings, using the case of medicine as an example.
FOCUS OF STUDY
The aim of the practitioner is to take into account as many aspects as he can of each and every clin- ical problem with which he is called upon to deal. This requires him not only to apply any scientific principle that appears relevant but also to draw on such personal experience of the condition as he may have acquired and, especially, to attend to that unique combination of features met with in each patient. Knowing how greatly patients dif- fer, the experienced clinician recognizes that a form of treatment well suited to one would be totally inappropriate to another.
Taking all factors into account and giving each its due weight is the art of clinical judgement.
The outlook of the research scientist is quite different. In his efforts to discern general pat- terns underlying individual variety he ignores the particular and strives to simplify, risking thereby
99/362
? ? ? over- simplification. If he is wise he will probably concentrate attention on a limited aspect of a limited problem. If in making his selection he proves sagacious, or simply lucky, he may not only elucidate the problem selected but also de- velop ideas applicable to a broader range. If his selection proves unwise or unlucky he may merely end up knowing more and more about less and less. That is the risk every researcher runs. The art of research lies in selecting a limited manageable problem and the methods that will best help solve it. This brings me to my second point.
MODES OF ACQUIRING INFORMATION
In the methods available to him for acquiring in- formation the practitioner has certain great ad- vantages over the research scientist but also cer- tain great disadvantages. Let us start with the advantages.
In his role of giving help the practitioner is per- mitted access to information of certain kinds that remain closed to the scientist: as a friend of mine is fond of saving, it's only surgeons who are al- lowed to cut you open to see what's inside. In an analogous way it is only by treating a patient
100/362
? ? ? therapeutically that a psychoanalyst is given ac- cess to much of importance going on in a person's mind. In both professions, moreover, practition- ers are permitted to intervene in specified ways and privileged to observe what the consequences of such interventions are. These are immense ad- vantages and psychoanalysts have not been slow to exploit them.
Yet no science can prosper for long without en- listing new methods to cross-check on observa- tions made and on hypotheses born of older methods. Here the research scientist is likely to have the advantage. In the medical sciences, physiologists and pathologists have made im- mense advances by means of animal experi- ments, tissue culture, biochemical analyses, and a thousand other ingenious techniques. Indeed, it is a hallmark of a creative scientist that he devises new means by which phenomena, perhaps already well studied by other methods, can be ob- served in some new way.
It is in this area, I believe, that the ambiguous use of the word psychoanalysis has done greatest harm. For it has led some analysts to suppose that the only method of enquiry appropriate for the advancement of psychoanalytic science is that
101/362
? ? ? of treating a patient psychoanalytically. Since I believe this to be a profound misunderstanding I shall be saying a good deal more about it. Before doing so, however, I want to say a word about the place of scepticism and faith in the respective worlds of scientist and practitioner.
SCEPTICISM AND FAITH
In his day-to-day work it is necessary for a scient- ist to exercise a high degree of criticism and self- criticism: and in the world he inhabits neither the data nor the theories of a leader, however ad- mired personally he may be, are exempt from challenge and criticism. There is no place for authority.
The same is not true in the practice of a profes- sion. If he is to be effective a practitioner must be prepared to act as though certain principles and certain theories were valid; and in deciding which to adopt he is likely to be guided by those with ex- perience from whom he learns. Since, moreover, there is a tendency in all of us to be impressed whenever the application of a theory appears to have been successful, practitioners are at special risk of placing greater confidence in a theory than the evidence available may justify.
102/362
? ? ? From the standpoint of clinical practice this is no bad thing. On the contrary, there is abundant evidence that the great majority of patients are helped by the faith and hope that a practitioner brings to his work; whilst it is often the very lack of these qualities that makes so many excellent research workers sadly ill-suited to be therapists.
Yet, though faith in the validity of particular data and in particular theories is out of place in a scientist, I do not wish to imply that he is nothing but a sceptic. On the contrary, his whole way of living is founded on faith, faith that in the long run the best route to reliable knowledge is the ap- plication of scientific method.
I am, of course, aware that there are many psy- choanalysts who do not share this faith and who believe that the types of problem with which we deal lie far outside the scope of science. This is a view I respect, though I do not share: nor, of course, did Freud. Yet even those of us who are most enthusiastic about applying scientific meth- od in our field must recognize that there may well be problems that it can never solve. We simply do not know. Our task, as I see it, is to apply our method as skilfully as we can, on the one hand believing that the area of reliable knowledge will
103/362
? ? ? thereby be expanded and, on the other, accepting that there are likely always to remain still greater areas lying beyond the scope of any existing mode of scientific enquiry.
To many of you, I am afraid, engaged in thera- peutic practice but also hoping to contribute to the advance of psychoanalytic science, the con- trasts I am drawing between the roles of practi- tioner and scientist will hardly be welcome. Yet I believe that it is only by recognizing these differ- ences and acting accordingly that the strengths of each role can be used to fullest advantage--or that any one person can occupy both of them with any hope of success. As practitioners we deal in complexity; as scientists we strive to simplify. As practitioners we use theory as a guide; as sci- entists we challenge that same theory. As practi- tioners we accept restricted modes of enquiry; as scientists we enlist every method we can.
Earlier I remarked on the need for every devel- oping science to devise new methods for obtain- ing data. The reason for this is that, however pro- ductive any one method may be, it is bound to have its limitations, whilst there is always a pro- spect that some other method may compensate for them. Thus the new method may be in no way
104/362
? ? ? superior to the old; indeed it may have great lim- itations. Its usefulness lies simply in the fact that its strengths and its limitations are different. Per- haps I can illustrate the point by reference to my own work.
When I qualified in psychoanalysis in 1937, members of the British Society were occupied in exploring the fantasy worlds of adults and chil- dren, and it was regarded as almost outside the proper interest of an analyst to give systematic attention to a person's real experiences. That was a time when Freud's famous about-turn of 1897 regarding the aetiology of hysteria1 had led to the view that anyone who places emphasis on what a child's real experiences may have been, and per- haps still are being, was regarded as pitifully nai? ve. Almost by definition it was assumed that anyone interested in the external world could not be interested in the internal world, indeed was al- most certainly running away from it.
To me as a biologist this contrast of internal with external, of organism with environment, never appealed. Furthermore, as a psychiatrist engaged in work with children and families and deeply influenced by the insights of two analytic- ally oriented social workers, I was daily
105/362
?
