Clearly that is not so, neither in man nor in any other
mammalian
species.
A-Secure-Base-Bowlby-Johnf
?
espouse.
This is a point at which analysts of dif- ferent schools diverge.
For some the events thought to be important might refer to feeding and weaning and fantasies about them during the earliest months of life; for others to toilet training or witnessing the primal scene during the second year; for others again to Oedipal situations and wishes during the third or fourth.
In the case of Mrs G the analyst drew on his knowledge of re- sponses of young children to events surrounding a prolonged separation from mother during sev- eral of the early years.
It is well recognized that not every child who has been in an institution for 18 months during his fifth and sixth years develops psychologically along the kind of pathway followed by Mrs G. In her case other factors almost certainly entered in. In considering what they might have been, I am influenced by remarks Mrs G made during the fi- nal phases of the analysis, for example, her fear lest her analyst should 'kick her out' or 'put her away' and her memory of how determined she had been that her mother should not 'get rid' of her. This suggests that as a method of disciplin- ing her daughter the mother may have repeatedly used threats to send her back to the institution,
157/362
? ? ? threats we know from other evidence are far from rare and that not only have a terrifying effect on a young child but are likely also to generate intense hatred in him. The better an analyst's knowledge of childhood conditions likely to lead to disturbed development, the better can he understand and help his patients.
Inevitably a patient's spontaneous or guided recollections of his childhood are of only suggest- ive value as evidence bearing on theories of per- sonality development. What a patient tells us about his childhood and especially what an ana- lyst subsequently reports his patient to have said are probably influenced as much or more by the analyst's preconceptions as by anything the pa- tient may in fact have said or done; this is why I regard the systematic study by direct observation of children developing within different patterns of family care as indispensable for progress. Yet I also believe that observations made during ther- apy still have considerable research potential, al- though that potential will not be realized unless studies are conducted along far more systematic lines than have hitherto been usual and data ob- tained during therapy are constantly compared with data derived from other sources.
158/362
? ? ? The research strength of the therapeutic situ- ation lies not in what it tells us about the patient's past but in what it tells us about disturbances of personality functioning in the present, especially, I would claim, disturbances in a person's capacit- ies to make secure attachments and the condi- tions in which these disturbances become ameli- orated. The case of Mrs G can be used as an intro- duction to a research proposal, since there is much both in the personality disturbance she presents and in the course of her analysis that we have reason to believe are fairly typical.
Drawing on case reports already in the literat- ure it would be possible to make a number of generalizations which could then be treated as predictions to be tested in further therapeutic work with patients showing similar clinical fea- tures. All such predictions, which would be con- ditional on the particular pattern of therapy to be followed,3 could be couched in terms of what can be observed at first hand. They would include statements about how a patient would be expec- ted to behave towards the analyst, the topics he would be expected to talk about or, in particular, avoid talking about, the affect he would be expec- ted to show or not to show, and in what
159/362
? ? ? situations. Of special interest would be changes in behaviour, topic, and affect that would be ex- pected to occur in relation to certain types of cur- rent event, both those occurring in the patient's everyday life and those occurring within the ana- lysis. Events of the latter sort would include how the analyst behaves, what he says, and how he says it, with particular reference to interruptions in the analysis due to holidays, sickness, or other circumstances. Tape recording of sessions to avoid biased reporting would, of course, be necessary.
By following the procedures proposed it would be possible over a period of time to gather com- parable bodies of data from two sources. One body of data would be gathered by means of dir- ect observation of the development and pattern- ing of affectional bonds during infancy and child- hood in children experiencing different types of care. The other would be gathered, also by direct observation of changes in the patterning of affec- tional bonds, during the course of a certain type of therapy. Provided that the conceptual frame- works used in making both sets of observations and the questions each set addresses are the
160/362
? ? ? same, findings could be compared and develop- mental hypotheses tested.
This is but one way in which psychoanalysis as a body of knowledge about personality develop- ment and psychopathology might move towards becoming the natural science Freud always inten- ded it to be.
The conceptual framework sketched here serves, I believe, to accommodate a substantial proportion of the data that psychoanalysis has se- lected as within its domain and already guides productive research programmes (e. g. Parkes and Stevenson-Hinde, 1982). The framework has the advantage of being compatible with evolutionary biology and neurophysiology and promises great- er economy and internal consistency than do tra- ditional ones. Nevertheless what its strengths and weaknesses will prove to be cannot be known without extensive testing of its powers to solve problems not yet studied, which include, for ex- ample, those of sexual development and devi- ation, and a far more intensive examination than has yet been possible of its usefulness in solving the problems already given attention.
Finally let us consider the issues raised by those who maintain that psychoanalysis is not
161/362
? ? ? and never can be a natural science. Their argu- ment, we find, stems from the belief that scientif- ic method is inseparable from logical positivism and reductionism. Although confidently and of- ten dogmatically held earlier this century, this model of science is now discarded and its place taken by a philosophy of evolutionary epistemo- logy (Popper, 1972; Latakos, 1974). This holds that all knowledge is conjectural and that science progresses through new theories coming to re- place older ones when it becomes clear that a new theory is able to make sense of a greater circle of phenomena than are comprehended and ex- plained by an older one and is able also to predict new phenomenena more accurately. 4 This same method is held to be applicable in all theoretical or generalizing sciences, whether they deal with physical, biological, or social phenomena. Fur- thermore, since to understand phenomena at any one level of complexity demands concepts appro- priate to that level, the notion that complex phe- nomena must be explained in terms of concepts appropriate to a lower level of complexity, namely reductionism, is now recognized as falla- cious. How these new ideas apply to our field is
162/362
? ? ? ably discussed by Holt (1981), Blight (1981), and Radford (1983).
Invaluable though scientific method is as a way of obtaining relatively reliable knowledge, resolv- ing differences of opinion and making useful pre- dictions, its limitations are nonetheless consider- able. One is that science deals in generalities but has little to say about singular specific events. This is a vital distinction, as the physicist, Weis- skopf (1981), points out. In the physical sciences it does not matter since physicists and engineers have no interest in the future of any particular atom or molecule. Once we move to the biological sciences, however, it does matter because biolo- gists are frequently interested in individual or- ganisms, all of which differ. In certain of the hu- man sciences, moreover, the individual example is the very essence of the case. History, for in- stance, whether it deals with societies or persons or ideas, is always concerned with an appallingly complex sequence of highly specific interacting events which no amount of science can enable us to explain adequately, let alone predict. Thus the distinction between the natural sciences and the historical sciences is not that they use a different method of obtaining knowledge but that the
163/362
? ? ? problems that they strive to understand and the criteria they adopt are quite different. One is con- cerned to formulate general laws in terms of probabilities, the other to understand singular specific events in as much detail as possible. The distinction is central to the whole argument.
Under the label 'psychoanalysis', it is clear, two complementary disciplines are striving to live and develop. In so far as we are trying to under- stand the general principles accounting for per- sonality development and psychopathology, ne- cessary, for example, if we are to know what forms of child care tend to produce what sorts of personality formation, we adopt the criteria of the natural sciences. And we do the same when we are trying to understand the essential features of effective therapy. In these fields we are dealing with statistical probabilities. In so far as we are concerned to understand the personal problems of a given individual and what events may have contributed to their development, necessary if we are to help him (though far from sufficient), we adopt the criteria of the historical sciences. Each approach contributes to our understanding; but, as I have pointed out in the previous lecture, only
164/362
? ? ? if we are clear-headed about what belongs to each, shall we be able to make progress.
1 For Freud's adherence to Lamarckian ideas see volume 3, chapter 10, of Ernest Jones's biography of Freud (Jones, 1957). For the influence of Haeckel's bio- genetic law see James Strachey's long editorial foot- note to his translation of Freud's Moses and Monothe- ism (SE 23,102) and especially Frank Sulloway's en- quiry into the origins of Freud's metapsychology (Sul- loway, 1979).
2 The account given here is derived from the contribu- tion by Thomas Mintz to a symposium organized by the American Psychoanalytic Association on the effects on adults of object loss during the first five years of life (Mintz, 1976).
3 The technique of analysis adopted by Mintz appears to have much in com-mon with that adopted in the Un- ited Kingdom by Donald Winnicott; see the account by Guntrip (1975).
4 The criterion of falsifiability on which Popper formerly laid great emphasis is no longer given so much prominence; though the constant comparison of theoretically derived predictions with an ever increas- ing array of observed data remains central.
5
VIOLENCE IN THE FAMILY
In the spring of 1983 I was invited to give the thirty-first annual Karen Horney Lecture by the Association for the Advancement of Psychoana- lysis at the conference it was holding in New York City. A principal reason why I selected viol- ence in the family as my theme was that re- search using the perspective of attachment the- ory was beginning to throw a shaft of light on a tragic but puzzling problem which, until recent years, had gone almost completely unrecognized by all those in the mental health field, not exclud- ing myself.
INTRODUCTION
It seems to me that as psychoanalysts and psy- chotherapists we have been appallingly slow to wake up to the prevalence and far-reaching con- sequences of violent behaviour between members of a family, and especially the violence of parents. As a theme in the analytic literature and in
? 166/362
? ? ? training programmes it has been conspicuous by its absence. Yet there is now abundant evidence not only that it is much commoner than we had hitherto supposed but that it is a major contribut- ory cause of a number of distressing and puzzling psychiatric syndromes. Since, moreover, violence breeds violence, violence in familes tends to per- petuate itself from one generation to the next.
Why family violence as a causal factor in psy- chiatry should have been so neglected by clini- cians--though, of course, not by social work- ers--would be a study in itself and cannot be entered on here. But the concentration in analytic circles on fantasy and the reluctance to examine the impact of real-life events has much to answer for. Ever since Freud made his famous, and in my view disastrous, volte-face in 1897, when he de- cided that the childhood seductions he had be- lieved to be aetiologically important were nothing more than the products of his patients' imagina- tions, it has been extremely unfashionable to at- tribute psychopathology to real-life experiences. It is not an analyst's job, so the conventional wis- dom has gone, to consider how a patient's par- ents may really have treated him, let alone to en- tertain the possibility, even probability, that a
167/362
? ? ? particular patient may have been the target for the violent words and violent deeds of one or both his parents. To focus attention on such pos- sibilities, I have often been told, is to be seduced by our patients' prejudiced tales, to take sides, to make scapegoats of perfectly decent parents. And in any case, it is asserted, to do so could be of no help to the patient, would in fact be antithera- peutic. It was indeed largely because the adverse behaviour of parents towards their children was such a taboo subject in analytic circles when I was starting my professional work that I decided to focus my research on the effects on children of real-life events of another sort, namely separa- tion and loss.
Of course, Karen Horney, in whose honour we are meeting today, did not share those prejudices. On the contrary, she is quite explicit in attribut- ing many of her patients' problems to the adverse influences they had met with as children. As she writes in the opening pages of her book, Neurosis and Human Growth (1951), these adverse influ- ences 'boil down to the fact that the people in the environment are too wrapped up in their own neuroses to be able to love the child, or even to conceive of him as the particular individual he is
168/362
? ? ? . . . '; and she goes on to list some of the many ways in which, unfortunately, parents can exert a harmful influence on their children. But I know well that these views were not always well re- ceived by her colleagues.
Today the scene is changing--though still far too slowly. For example, no one with eyes to see can any longer doubt that all too many children are battered by their parents, either verbally or physically or both, nor that all too many women are battered by husband or boyfriend. Moreover, our horror that parents can behave so is nowadays mitigated by our increasing knowledge of the kind of childhoods these parents have themselves had. Whilst horror at their acts is in- evitable, greater understanding of how they have come to behave in these violent ways evokes com- passion rather than blame. So far from wishing to scapegoat parents we wish to help them. So far from refusing to see that parents sometimes en- gage in horrific behaviour, we seek ways to suc- cour the casualties, old as well as young, psycho- logical as well as physical. Above all we seek ways of preventing violent patterns developing in new families. Let us hope that the policy of head-in- sand has had its day.
169/362
? ? ? CONCEPTUAL FRAMEWORK
In seeking to understand the more extreme ex- amples of violence in the family it is useful to consider first what we know about the more mod- erate and everyday example of members of a fam- ily getting angry with each other. Young chil- dren--and often older ones too--are commonly jealous of the attention mother gives the new baby. Lovers quarrel when one thinks the other is looking elsewhere--and the same is true after marriage. Moreover a woman may get very angry with her child if he does something dangerous, like running into the roadway, and also with her husband if he risks life or limb by taking unne- cessary risks. Thus we take it for granted that, when a relationship to a special loved person is endangered, we are not only anxious but are usu- ally angry as well. As responses to the risk of loss, anxiety and anger go hand in hand. It is not for nothing that they have the same etymological root.
In the situations described anger is often func- tional. When child or spouse behaves danger- ously, an angry protest is likely to deter. When a lover's partner strays, a sharp reminder of how much he or she cares may work wonders. When a
170/362
? ? ? child finds himself relatively neglected in favour of the new baby, assertion of his claims may re- dress the balance. Thus in the right place, at the right time, and in right degree, anger is not only appropriate but may be indispensable. It serves to deter from dangerous behaviour, to drive off a rival, or to coerce a partner. In each case the aim of the angry behaviour is the same--to protect a relationship which is of very special value to the angry person.
This being so, it is necessary to be clear why certain specific relationships, often called libidin- al relationships, should become so very import- ant in the lives of each one of us.
In his attempts to solve this problem Freud looked to the physics and biology of his day. Li- bidinal relationships, he proposed, were con- sequent on the individual's needs for food and sex. Subsequently, to account for some of the more puzzling manifestations of anger, he stepped outside biology to propose a death in- stinct. These hypotheses, framed in terms of the accumulation and discharge of psychic energies, led to a metapsychology so remote from clinical observation and experience that a great many analytically oriented clinicians have, implicitly or
171/362
? ? ? explicitly, abandoned it. To the resulting vacuum one reaction has been the development of a school of thought that has decided not only to di- vorce psychoanalysis from biology but to dis- pense altogether with scientific method and in- stead to espouse hermeneutics. Another and op- posite reaction has been to explore the principles found useful in modern biology, principles totally different from those of Freud's day, to see wheth- er they are more in tune with our clinical obser- vations and might therefore be used to construct a new metapsychology, or conceptual framework as it would now be called. That is the course that I and a number of others are following.
The specific relationships, threats to which may arouse anger, are of three main types: rela- tionships with a sexual partner (boyfriend, girl- friend, or spouse), relationships with parents, and relationships with offspring. Each type of re- lationship is shot through with strong emotion. In high degree indeed, a person's whole emotion- al life--the underlying tone of how he feels--is determined by the state of these long-term, com- mitted relationships. As long as they are running smoothly he is content; when they are threatened he is anxious and perhaps angry; when he has
172/362
? ? ? endangered them by his own actions he feels guilty; when they are broken he feels sad; and when they are resumed he is joyful.
In helping us understand why the state of these relationships should have such a profound effect on a person's feeling life, two branches of modern biology--ethology and evolution theory--are ex- tremely illuminating. Not only do all three of the relationships in question have counterparts in a wide range of other species but all three are in- timately concerned with the vital biological func- tions of reproduction and, especially, the survival of young. It is therefore more than likely that a human being's powerful propensity to make these deep and long-term relationships is the result of a strong gene-determined bias to do so, a bias that has been selected during the course of evolution. Within this frame of reference a child's strong propensity to attach himself to his mother and his father, or to whomever else may be caring for him, can be understood as having the function of reducing the risk of his coming to harm. For to stay in close proximity to, or in easy communica- tion with, someone likely to protect you is the best of all possible insurance policies. Similarly a parent's concern to care for his or her offspring
173/362
? ? ? plainly has the function of contributing to the child's survival. That success in the maintenance of these long-term relationships should usually bring satisfaction and contentment, and that fail- ure should bring frustration, anxiety, and some- times despair are, on this reading, the prizes and penalties selected during evolution to guide us in our activities.
It is within this evolutionary perspective that I believe we can understand how angry behaviour between members of a family can often be func- tional. For, as I said earlier, in the right place, at the right time, and in right degree, anger can serve to maintain these vitally important long- term relationships. But, as is very obvious, anger can be overdone. My thesis is simply that a great deal of the maladaptive violence met with in fam- ilies can be understood as the distorted and exag- gerated versions of behaviour that is potentially functional, especially attachment behaviour on the one hand and caregiving behaviour on the other.
There is now a considerable literature on the nature of the child's tie to his mother, tradition- ally referred to as dependency and now conceived in terms of attachment and careseeking. But in
174/362
? ? ? general terms, attachment behaviour results in one individual, usually the one who is less able to cope, maintaining proximity to, and/or commu- nication with, another individual, seen as better able to cope. This behaviour is elicited especially by pain, fatigue, or anything frightening, and also by the caregiver being or appearing to be, inac- cessible. Although conceived as being in part pre- programmed, there is now abundant evidence that the particular pattern in which attachment behaviour becomes organized during develop- ment is much influenced by how it is responded to by a child's principal caregivers, in the huge majority of cases his mother and father. Briefly, it seems clear that sensitive loving care results in a child developing confidence that others will be helpful when appealed to, becoming increasingly self-reliant and bold in his explorations of the world, co-operative with others, and also--a very important point--sympathetic and helpful to oth- ers in distress. Conversely, when a child's attach- ment behaviour is responded to tardily and un- willingly and is regarded as a nuisance, he is likely to become anxiously attached, that is, ap- prehensive lest his caregiver be missing or un- helpful when he needs her and therefore
175/362
? ? ? reluctant to leave her side, unwillingly and anxiously obedient, and unconcerned about the troubles of others. Should his caregivers, in addi- tion, actively reject him, he is likely to develop a pattern of behaviour in which avoidance of them competes with his desire for proximity and care, and in which angry behaviour is apt to become prominent. I shall say more about this later.
One further point about attachment behaviour I wish to emphasize is that it is a characteristic of human nature throughout our lives--from the cradle to the grave. Admittedly it is usually less intense and less demanding in adolescents and adults than it is in earlier years. Yet an urgent de- sire for love and care is natural enough when a person is anxious or distressed. It is therefore most unfortunate that, due to misleading theory, the pejorative adjectives 'infantile' and 'regressive' are now so widely current in clinical circles. They are words I never use.
Whilst the systematic study of attachment be- haviour, and especially the conditions influencing how it develops, has been in progress for twenty years, the systematic study of caregiving, or par- enting, and how it develops is only in its begin- nings. The approach I regard as the most
176/362
? ? ? promising is again an ethological one. This as- sumes that in humans as well as in other species parenting behaviour, like attachment behaviour, is in some degree preprogrammed and therefore ready to develop along certain lines when condi- tions elicit it. This means that in the ordinary course of events the parent of a baby experiences a strong urge to behave in certain typical sorts of way, for example, to cradle the infant, to soothe him when he cries, to keep him warm, protected, and fed. Such a viewpoint, of course, does not im- ply that the appropriate behaviour patterns manifest themselves complete in every detail from the first.
Clearly that is not so, neither in man nor in any other mammalian species. All the detail is learned, some of it during interaction with babies and children, much of it through ob- servation of how other parents behave, starting during the parent-to-be's own childhood and the way his parents treated him and his siblings.
RESEARCH FINDINGS
In considering what is now known of the indi- viduals involved in family violence, and the cir- cumstances in which it takes place, I start with women who physically assault their children and,
177/362
? ? ? next, with the effects on the children of such as- saults. My reason for doing so is that in these two areas the research findings seem most adequate. And in the light of those findings we can proceed to consider what is known about men who batter wives or children, a problem area just as import- ant but at present less well researched.
The findings of the many studies of women known to have battered their children show con- siderable agreement (see review by Spinetta and Rigler, 1972). Though probably commoner among families of lower socio-economic status, child abuse occurs also in middle-class families where it is likely to be hidden behind a fac? ade of ultra-respectability.
On the surface abusing individuals vary from being cold, rigid, obsessional, and censorious to being passive, unhappy, and disorganized. Yet emotionally they have much in common. Among features reported as especially frequent among abusive mothers we find the following: prone to periods of intense anxiety punctuated by out- bursts of violent anger, they are said to be im- pulsive and 'immature'. Although their 'depend- ency needs' are described as exceptionally strong, they are extremely distrustful and consequently
178/362
? ? ? unable or unwilling to make close relationships. Socially they are isolated. Having no one else to turn to, many of them seek care and comfort from one of their own children whom they treat as though they were much older than they are (Morris and Gould, 1963).
As regards childhood experience many re- searchers have noted that most of such women have had a miserable childhood and, as one pair of researchers put it, 'have been deprived of basic mothering' (Steele and Pollock, 1968). An appre- ciable minority were themselves battered as chil- dren. 1
To anyone thinking in terms of attachment the- ory the notion at once suggests itself that these woman are suffering from an extreme degree of anxious attachment and consequently that exper- iences of long or repeated separations and/or of being repeatedly threatened with abandonment would be a common feature of their childhoods. In a relatively small study these hypotheses have now been tested by Pauline DeLozier (1982), working in Los Angeles. Her samples consisted of 18 working-class women known to have assaulted their children physically and 18 others from the same socio-economic class, and matched for age
179/362
? ? ? (most in their thirties) and number of children, who had not done so. All subjects were given a semi-structured interview and a questionnaire to complete and were tested on the Hansburg Se- paration Anxiety Test (Hansburg, 1972). In the latter a series of pictures depicting scenes either of a child leaving his parents or of parents leaving the child is shown and the subject invited to de- scribe what she would feel and do in the situation concerned.
The results of the Separation Anxiety Test showed, as expected, that most of the abusing mothers were extremely sensitive to any type of separation situation, even the most everyday and commonplace, with responses indicative of high levels of anxiety and/or anger. The responses showed, in addition, that, while these women yearned for care, all they expected was rejection. Supporting another of the initial hypotheses was a high incidence of responses indicating anxious concern for the welfare of parents. For all these features the incidences in the control group, al- though not negligible, were significantly lower. For example, whereas 12 of the 18 abusing moth- ers were rated as showing the highest degree of
180/362
? ? ? anxious attachment, only two of the control group were so rated.
As regards the childhood experiences of these women, the results, as reported in interview and qiestionnaire, supported some of the initial hypo- theses but did not support others. For example, in view of previous reports DeLozier had expec- ted to find a significantly higher incidence of sep- arations from parents in the childhoods of the ab- usive mothers but this was not found. By con- trast, her expectation that a high proportion of these women would have experienced their par- ents as having repeatedly threatened to abandon them was fulfilled, a finding in keeping with the view that repeated threats to abandon are as pathogenic as actual separations and probably more so (Bowlby, 1973). In a similiar way, al- though actual violence from a parent had appar- ently not been common, many of the abusing mothers had suffered repeated threats of being beaten, maimed, or even killed.
Another striking feature of the childhoods of the abusing mothers, also in keeping with expect- ations, was that only a minority of them (seven) had felt they could turn to mother for help when in distress. Amongst those who could not were
181/362
? ? ? some who described someone else, a relative or neighbour, who would have been helpful; but four of the 18 had known of absolutely no one. Amongst the controls, by contrast, all but three had felt they could turn to mother, and each of the others had at least known of someone else.
Thus, unlike a girl who grows up in an ordinary reasonably happy home who can rely on her mother in an emergency, a majority of these wo- men could never do so. Indeed, and again as pre- dicted, for many of them the relationship of daughter to mother had been reversed and it was they who had been expected to care for the par- ent. 2
Given the childhood experiences these women had had, it is not difficult to understand why they had grown up as they had. Threats to abandon a child make her (or him) intensely anxious about any separation, however routine it might appear to others, and also intensely angry with her par- ent for threatening her so. Moreover failures to respond helpfully when a child is in distress, combined with repeated and impatient rejec- tions, lead her to be deeply suspicious of every- one else. Thus, whilst constantly yearning for the love and care she has never had, she has no
182/362
? ? ? confidence she will ever receive it; and she will mistrust any offer she may receive. Small wonder therefore, if when a woman with this background becomes a mother, that there are times when, in- stead of being ready to mother her child, she looks to her child to mother her. Small wonder too if when her child fails to oblige and starts cry- ing, demanding care and attention, that she gets impatient and angry with it.
It is against this background, I believe, that a mother's violent assaults on a child can be under- stood. Although I have never treated a woman who actually assaulted her child physically, I have treated one who came perilously close to doing so.
The reason I began seeing this woman, whom I will call Mrs Q, was that the doctor at the well- baby clinic she attended was concerned about her son, aged 18 months; he was refusing to eat and was losing weight. When I saw them both, it was at once apparent that Mrs Q was intensely anxious and depressed and had been so since the boy's birth. On enquiry I found that she was terri- fied lest her son die and was therefore pestering him to eat. She also told me that she had some- times had impulses to throw the baby out of the
183/362
? ? ? window. Only much later did she tell me that on occasion she became hysterical, smashed the dishes, and battered the baby's pram. She was in- tensely anxious talking to me and fully expected me to turn on her angrily. On my suggestion she came to see me for psychotherapy once a week.
The picture she gave me of her childhood, told reluctantly in fragments but always consistently, was one I now know to be typical. She recalled bitter quarrels between her parents in which they assaulted one another and threatened murder, and how her mother would repeatedly bring pres- sure on the family by threatening to desert. On two occasions Mrs Q had returned from school to find her mother with her head in the gas oven, and at other times her mother would pretend to have deserted by disappearing for half a day. Nat- urally Mrs Q grew up terrified that if she did any- thing wrong her mother would go. Moreover things were made even worse by her mother in- sisting that she breathe not a word about these terrifying events to anyone outside the home.
Mrs Q, who had been a skilled technician be- fore marriage, was known as a very helpful neigh- bour and did all she could to be a good wife and mother--in which for the most part she
184/362
? ? ? succeeded. Yet she was subject to these violent and destructive outbursts which frightened and puzzled her and about which she felt intense shame.
After some time I had little doubt that the angry outbursts were to be understood as the ex- pression of intense anger which, generated ini- tially and over many years by her mother's re- peated threats to abandon the family, had early in her life become directed away from her mother and towards less dangerous targets. Terrified then and later of ever expressing her anger dir- ectly, she redirected3 it towards something which, or someone who, could not retaliate. As a child, Mrs Q recalled, she had sometimes re- treated to her room and attacked her dolls. Now it was her crockery, the pram, and almost but not quite the baby. Each current outburst, I suspec- ted, was triggered by her mother who, dominat- ing and interfering as ever, still visited her daughter every day.
This explanation fits what facts we have and has the merit, not always appreciated in clinical circles, of being simple. Not surprisingly, other workers in the field have also proposed it (e. g. Feinstein, Paul, and Pettison, 1964). In other
185/362
? ? ? cases, it is clear, a husband has violently abused his wife and she, violently angry in return, has re- directed it against their child.
In turning next to the effects on the personality development of children who are assaulted, we have to bear in mind that the physical assaults are not the only episodes of hostility from parents that these children have experienced. In very many cases indeed the physical assaults are but the tip of an iceberg--the manifest signs of what have been repeated episodes of angry rejection, verbal as well as physical. In most cases therefore the psychological effects can be regarded as the outcome of prolonged hostile rejection and neg- lect. Nevertheless the experiences of individual children can vary greatly. A few, for example, may receive reasonably good care and only very occasionally suffer an outburst of parental viol- ence. For these reasons it is no surprise that the socio-emotional development of the children var- ies also. Here I describe findings that appear to be fairly typical.
Those who have observed such children in their homes or elsewhere describe them variously as depressed, passive, and inhibited, as 'depend- ent' and anxious, and also as angry and
186/362
? ? ? aggressive (Martin and Rodeheffer, 1980). Gaensbauer and Sands (1979), in endorsing this picture, emphasize how disturbing to a caregiver such behaviour can be. The children fail to parti- cipate in play and show little or no enjoyment. Expression of feeling is often so low key that it is easy to overlook, or else is ambiguous and con- trary. Crying may be prolonged and unresponsive to comforting; anger is easily aroused, intense, and not readily resolved. Once established, these patterns tend to persist.
An issue much discussed in the literature is the degree to which an infant's prematurity, ill- health, or difficult temperament may have con- tributed to a mother's problems and so ultimately to its having been ill-treated. In some cases these factors play a part, but they do so only when a mother reacts unfavourably to the baby and thereby sets up a vicious circle. 4 Such a sequence, of course, is all too likely to occur when a mother has herself had a difficult childhood, has grown up emotionally disturbed, and has little or no emotional support or help after her baby is born.
Towards his or her parent an abused toddler often shows a striking picture of frozen watchfulness, hyper-alert for what might happen.
187/362
? ? ? But some show also an unusual sensitivity to the needs of their parents (Malone, 1966). There are, in fact, good reasons for thinking that some chil- dren learn early that it is possible to placate a dis- turbed and potentially violent mother by constant attention to her wishes. 5
In a nursery setting battered infants and tod- dlers have a reputation for finding it difficult to make relationships, either with caregivers or with other children, and also for being very aggressive. In recent years these observations have been con- firmed and extended by more systematic research which has focused attention both on the particu- lar patterns of behaviour shown and on details of the situations in which each occurs. The findings that follow stem from a study being conducted at Berkeley by Main and George (George and Main, 1979; Main and George, 1985).
Their aim was to compare the behaviour in a day-nursery setting of two groups of children in the age-range 1 to 3 years. One group of ten were known to have been physically assaulted by a par- ent. The other group of ten were matched for all relevant variables but had not been assaulted; they were, however, in nurseries set up to care for children from families known to be under stress.
188/362
? ? ? To obtain data each child was observed for four periods each of 30 minutes on four different days spread over three weeks. Observers were instruc- ted to record all socially relevant behaviour, in- cluding such small movements as head-turning or stepping backwards.
In analysing the data, the children's social be- haviour was divided into four categories: ap- proach, avoidance, approach-avoid, and aggres- sion. It was also divided according to whom the behaviour was directed--another child or a care- giver. Another distinction was between behaviour initiated by the observed child and behaviour which occurred in response to a friendly ap- proach by another child or adult. Results are ex- pressed in terms of the mean number of incidents of a particular type of behaviour of the children in each group or else of the number of children in each group who showed that type of behaviour.
As regards occasions when a child initiated so- cial contact, either with another child or a care- giver, no appreciable differences were observed between the children in each group. By contrast, very striking differences were observed in the ways the children responded to a friendly ap- proach from the other. Characteristic responses
189/362
? ? ? of the abused children were either to take straight avoiding action or else to show both approach and avoidance behaviour, either in quick succes- sion or in some combination of the two. Examples are: 'she creeps towards him but sud- denly veers away', and 'she crawls towards the caregiver but with head averted'. Thus, when the overtures come from a caregiver, the abused tod- dlers were three times more likely than the con- trols to take avoiding action; whilst seven of them, compared to only one, showed the curious combination of alternation of approach and avoidance. When the overtures came from other children, the differences were even more marked. For example, whereas none of the controls showed approach-avoidance, all ten of the abused toddlers did so.
Aggressive behaviour was fairly common in both groups of toddlers, though, as predicted, it was significantly more so in the abused group. Not only did the abused toddlers assault other children twice as often as the controls, but five of them assaulted or threatened to assault an adult, behaviour seen in none of the controls. In addi- tion the abused toddlers were notable for a par- ticularly disagreeable type of aggression, termed
190/362
? ? ? 'harassment' (Manning, Heron, and Marshall, 1978). This consists of malicious behaviour which appears to have the sole intent of making the vic- tim show distress. Almost always it occurs sud- denly without any evident cause and so contrasts with hostility which occurs in reaction to a pro- vocation. Such attacks, coming unpredictably out of the blue, are frightening and invite retaliation. Clinical studies, referred to later, report them to be directed especially towards an adult to whom the child is becoming attached.
In view of the behaviour so far described it is not surprising to find that abused toddlers are singularly unsympathetic to age-mates in dis- tress. The studies of Zahn-Waxler and Radke- Yarrow have shown that infants and pre-school children who have affectionate and caring par- ents commonly express concern when another child is distressed and often make moves to com- fort him or her (Zahn-Waxler, Radke-Yarrow, and King, 1979). This type of behaviour was also seen at least once in five of the control children in the Main and George study; but on no occasion did any of the abused toddlers show the slightest hint of it. Instead, and unlike the controls, they reacted with some combination of fear, distress,
191/362
? ? ? or anger; and three behaved hostilely to the cry- ing child. For example, one small boy of 2 years 8 months slapped a little girl who was crying, ex- claiming repeatedly 'Cut it out, cut it out'. He pro- ceeded then to pat her back and next to hiss at her with bared teeth; before anyone could inter- vene, his patting had turned to beating.
My reasons for giving so much attention to these observations of young children will, I am sure, be apparent. They show with unmistakable clarity how early in life certain characteristic pat- terns of social behaviour--some hopeful for the future, others ominous--become established. They leave no doubts either about what types of family experience influence development in one direction or another. Again and again we see de- tails in the behaviour of a toddler, or in what he says, that are plainly straight replicas of how that toddler has himself been treated. Indeed the tendency to treat others in the same way that we ourselves have been treated is deep in human nature; and at no time is it more evident than in the earliest years. All parents please note!
Firm evidence of how these children develop must await an appropriately designed longitudin- al study. There is evidence that, if conditions of
192/362
? ? ? care improve, some recover sufficiently to pass for normal (Lynch and Roberts, 1982); others do not. Some have suffered serious brain damage and become diagnosed as mentally handicapped (e. g. Martin and Rodeheffer, 1980). For many others adverse conditions of care continue. Fur- thermore, once a child has developed the types of disagreeable behaviour described, it is not easy for an adult, whether parent, foster-parent, or professional, to give him the continuous affec- tionate care he needs, whilst treating such chil- dren by psychotherapy is extremely taxing. The sudden unprovoked attacks, which in older chil- dren can easily be damaging, are especially hard to take.
Some of these emotionally disturbed children, we know, reach psychiatric clinics where the ori- gin of their condition, I suspect, more often than not goes unrecognized. Amongst those who have treated these children, some of them psychotic, and who have traced the source of their troubles are Stroh (1974), Bloch (1978), and Hopkins (1984). Each notes the extreme degree of ambi- valence to be expected: one moment the child is hugging the therapist, the next he is kicking her. During adolescence and early adult life, some,
193/362
? ? ? mostly male, become diagnosed as aggressive psychopaths and/or violent delinquents (e. g. Far- rington, 1978). Others, especially females per- haps, are found to be suffering from multiple per- sonality (Bliss, 1980). Once psychiatrists become aware of the profound and far-reaching effects of childhood abuse and rejection, and the extent to which relevant information is suppressed and falsified by parents and overlooked by clinicians, many more cases are sure to be identified.
A significant proportion of rejected and abused children grow up to perpetuate the cycle of family violence by continuing to respond in social situ- ations with the very same patterns of behaviour that they had developed during early childhood.
A type of response found to be characteristic of many abusive parents, and of a kind which we have already seen to be characteristic of abused toddlers, is reported, for example, by Frodi and Lamb (1980). In a laboratory study, in which videotapes of crying infants were shown, abusive mothers were found to respond to a crying infant with less sympathy than did a group of non-abus- ive mothers and also with more annoyance and anger. Furthermore these same adverse re- sponses were shown by the abusive mothers even
194/362
? ? ? when they were shown video tapes of smiling in- fants, which suggested they disliked any form of interaction with an infant.
Let us turn now, rather belatedly, to the beha- viour of men who ill-treat girlfriend or wife.
Two of my social work colleagues at the Tavis- tock, Janet Mattinson and Ian Sinclair (1979), de- scribe a man, Mr S, who was apt, inexplicably and unpredictably, to attack his wife. At the time when he asked for an interview his wife had re- cently left him; she had just had their first baby.
It is well recognized that not every child who has been in an institution for 18 months during his fifth and sixth years develops psychologically along the kind of pathway followed by Mrs G. In her case other factors almost certainly entered in. In considering what they might have been, I am influenced by remarks Mrs G made during the fi- nal phases of the analysis, for example, her fear lest her analyst should 'kick her out' or 'put her away' and her memory of how determined she had been that her mother should not 'get rid' of her. This suggests that as a method of disciplin- ing her daughter the mother may have repeatedly used threats to send her back to the institution,
157/362
? ? ? threats we know from other evidence are far from rare and that not only have a terrifying effect on a young child but are likely also to generate intense hatred in him. The better an analyst's knowledge of childhood conditions likely to lead to disturbed development, the better can he understand and help his patients.
Inevitably a patient's spontaneous or guided recollections of his childhood are of only suggest- ive value as evidence bearing on theories of per- sonality development. What a patient tells us about his childhood and especially what an ana- lyst subsequently reports his patient to have said are probably influenced as much or more by the analyst's preconceptions as by anything the pa- tient may in fact have said or done; this is why I regard the systematic study by direct observation of children developing within different patterns of family care as indispensable for progress. Yet I also believe that observations made during ther- apy still have considerable research potential, al- though that potential will not be realized unless studies are conducted along far more systematic lines than have hitherto been usual and data ob- tained during therapy are constantly compared with data derived from other sources.
158/362
? ? ? The research strength of the therapeutic situ- ation lies not in what it tells us about the patient's past but in what it tells us about disturbances of personality functioning in the present, especially, I would claim, disturbances in a person's capacit- ies to make secure attachments and the condi- tions in which these disturbances become ameli- orated. The case of Mrs G can be used as an intro- duction to a research proposal, since there is much both in the personality disturbance she presents and in the course of her analysis that we have reason to believe are fairly typical.
Drawing on case reports already in the literat- ure it would be possible to make a number of generalizations which could then be treated as predictions to be tested in further therapeutic work with patients showing similar clinical fea- tures. All such predictions, which would be con- ditional on the particular pattern of therapy to be followed,3 could be couched in terms of what can be observed at first hand. They would include statements about how a patient would be expec- ted to behave towards the analyst, the topics he would be expected to talk about or, in particular, avoid talking about, the affect he would be expec- ted to show or not to show, and in what
159/362
? ? ? situations. Of special interest would be changes in behaviour, topic, and affect that would be ex- pected to occur in relation to certain types of cur- rent event, both those occurring in the patient's everyday life and those occurring within the ana- lysis. Events of the latter sort would include how the analyst behaves, what he says, and how he says it, with particular reference to interruptions in the analysis due to holidays, sickness, or other circumstances. Tape recording of sessions to avoid biased reporting would, of course, be necessary.
By following the procedures proposed it would be possible over a period of time to gather com- parable bodies of data from two sources. One body of data would be gathered by means of dir- ect observation of the development and pattern- ing of affectional bonds during infancy and child- hood in children experiencing different types of care. The other would be gathered, also by direct observation of changes in the patterning of affec- tional bonds, during the course of a certain type of therapy. Provided that the conceptual frame- works used in making both sets of observations and the questions each set addresses are the
160/362
? ? ? same, findings could be compared and develop- mental hypotheses tested.
This is but one way in which psychoanalysis as a body of knowledge about personality develop- ment and psychopathology might move towards becoming the natural science Freud always inten- ded it to be.
The conceptual framework sketched here serves, I believe, to accommodate a substantial proportion of the data that psychoanalysis has se- lected as within its domain and already guides productive research programmes (e. g. Parkes and Stevenson-Hinde, 1982). The framework has the advantage of being compatible with evolutionary biology and neurophysiology and promises great- er economy and internal consistency than do tra- ditional ones. Nevertheless what its strengths and weaknesses will prove to be cannot be known without extensive testing of its powers to solve problems not yet studied, which include, for ex- ample, those of sexual development and devi- ation, and a far more intensive examination than has yet been possible of its usefulness in solving the problems already given attention.
Finally let us consider the issues raised by those who maintain that psychoanalysis is not
161/362
? ? ? and never can be a natural science. Their argu- ment, we find, stems from the belief that scientif- ic method is inseparable from logical positivism and reductionism. Although confidently and of- ten dogmatically held earlier this century, this model of science is now discarded and its place taken by a philosophy of evolutionary epistemo- logy (Popper, 1972; Latakos, 1974). This holds that all knowledge is conjectural and that science progresses through new theories coming to re- place older ones when it becomes clear that a new theory is able to make sense of a greater circle of phenomena than are comprehended and ex- plained by an older one and is able also to predict new phenomenena more accurately. 4 This same method is held to be applicable in all theoretical or generalizing sciences, whether they deal with physical, biological, or social phenomena. Fur- thermore, since to understand phenomena at any one level of complexity demands concepts appro- priate to that level, the notion that complex phe- nomena must be explained in terms of concepts appropriate to a lower level of complexity, namely reductionism, is now recognized as falla- cious. How these new ideas apply to our field is
162/362
? ? ? ably discussed by Holt (1981), Blight (1981), and Radford (1983).
Invaluable though scientific method is as a way of obtaining relatively reliable knowledge, resolv- ing differences of opinion and making useful pre- dictions, its limitations are nonetheless consider- able. One is that science deals in generalities but has little to say about singular specific events. This is a vital distinction, as the physicist, Weis- skopf (1981), points out. In the physical sciences it does not matter since physicists and engineers have no interest in the future of any particular atom or molecule. Once we move to the biological sciences, however, it does matter because biolo- gists are frequently interested in individual or- ganisms, all of which differ. In certain of the hu- man sciences, moreover, the individual example is the very essence of the case. History, for in- stance, whether it deals with societies or persons or ideas, is always concerned with an appallingly complex sequence of highly specific interacting events which no amount of science can enable us to explain adequately, let alone predict. Thus the distinction between the natural sciences and the historical sciences is not that they use a different method of obtaining knowledge but that the
163/362
? ? ? problems that they strive to understand and the criteria they adopt are quite different. One is con- cerned to formulate general laws in terms of probabilities, the other to understand singular specific events in as much detail as possible. The distinction is central to the whole argument.
Under the label 'psychoanalysis', it is clear, two complementary disciplines are striving to live and develop. In so far as we are trying to under- stand the general principles accounting for per- sonality development and psychopathology, ne- cessary, for example, if we are to know what forms of child care tend to produce what sorts of personality formation, we adopt the criteria of the natural sciences. And we do the same when we are trying to understand the essential features of effective therapy. In these fields we are dealing with statistical probabilities. In so far as we are concerned to understand the personal problems of a given individual and what events may have contributed to their development, necessary if we are to help him (though far from sufficient), we adopt the criteria of the historical sciences. Each approach contributes to our understanding; but, as I have pointed out in the previous lecture, only
164/362
? ? ? if we are clear-headed about what belongs to each, shall we be able to make progress.
1 For Freud's adherence to Lamarckian ideas see volume 3, chapter 10, of Ernest Jones's biography of Freud (Jones, 1957). For the influence of Haeckel's bio- genetic law see James Strachey's long editorial foot- note to his translation of Freud's Moses and Monothe- ism (SE 23,102) and especially Frank Sulloway's en- quiry into the origins of Freud's metapsychology (Sul- loway, 1979).
2 The account given here is derived from the contribu- tion by Thomas Mintz to a symposium organized by the American Psychoanalytic Association on the effects on adults of object loss during the first five years of life (Mintz, 1976).
3 The technique of analysis adopted by Mintz appears to have much in com-mon with that adopted in the Un- ited Kingdom by Donald Winnicott; see the account by Guntrip (1975).
4 The criterion of falsifiability on which Popper formerly laid great emphasis is no longer given so much prominence; though the constant comparison of theoretically derived predictions with an ever increas- ing array of observed data remains central.
5
VIOLENCE IN THE FAMILY
In the spring of 1983 I was invited to give the thirty-first annual Karen Horney Lecture by the Association for the Advancement of Psychoana- lysis at the conference it was holding in New York City. A principal reason why I selected viol- ence in the family as my theme was that re- search using the perspective of attachment the- ory was beginning to throw a shaft of light on a tragic but puzzling problem which, until recent years, had gone almost completely unrecognized by all those in the mental health field, not exclud- ing myself.
INTRODUCTION
It seems to me that as psychoanalysts and psy- chotherapists we have been appallingly slow to wake up to the prevalence and far-reaching con- sequences of violent behaviour between members of a family, and especially the violence of parents. As a theme in the analytic literature and in
? 166/362
? ? ? training programmes it has been conspicuous by its absence. Yet there is now abundant evidence not only that it is much commoner than we had hitherto supposed but that it is a major contribut- ory cause of a number of distressing and puzzling psychiatric syndromes. Since, moreover, violence breeds violence, violence in familes tends to per- petuate itself from one generation to the next.
Why family violence as a causal factor in psy- chiatry should have been so neglected by clini- cians--though, of course, not by social work- ers--would be a study in itself and cannot be entered on here. But the concentration in analytic circles on fantasy and the reluctance to examine the impact of real-life events has much to answer for. Ever since Freud made his famous, and in my view disastrous, volte-face in 1897, when he de- cided that the childhood seductions he had be- lieved to be aetiologically important were nothing more than the products of his patients' imagina- tions, it has been extremely unfashionable to at- tribute psychopathology to real-life experiences. It is not an analyst's job, so the conventional wis- dom has gone, to consider how a patient's par- ents may really have treated him, let alone to en- tertain the possibility, even probability, that a
167/362
? ? ? particular patient may have been the target for the violent words and violent deeds of one or both his parents. To focus attention on such pos- sibilities, I have often been told, is to be seduced by our patients' prejudiced tales, to take sides, to make scapegoats of perfectly decent parents. And in any case, it is asserted, to do so could be of no help to the patient, would in fact be antithera- peutic. It was indeed largely because the adverse behaviour of parents towards their children was such a taboo subject in analytic circles when I was starting my professional work that I decided to focus my research on the effects on children of real-life events of another sort, namely separa- tion and loss.
Of course, Karen Horney, in whose honour we are meeting today, did not share those prejudices. On the contrary, she is quite explicit in attribut- ing many of her patients' problems to the adverse influences they had met with as children. As she writes in the opening pages of her book, Neurosis and Human Growth (1951), these adverse influ- ences 'boil down to the fact that the people in the environment are too wrapped up in their own neuroses to be able to love the child, or even to conceive of him as the particular individual he is
168/362
? ? ? . . . '; and she goes on to list some of the many ways in which, unfortunately, parents can exert a harmful influence on their children. But I know well that these views were not always well re- ceived by her colleagues.
Today the scene is changing--though still far too slowly. For example, no one with eyes to see can any longer doubt that all too many children are battered by their parents, either verbally or physically or both, nor that all too many women are battered by husband or boyfriend. Moreover, our horror that parents can behave so is nowadays mitigated by our increasing knowledge of the kind of childhoods these parents have themselves had. Whilst horror at their acts is in- evitable, greater understanding of how they have come to behave in these violent ways evokes com- passion rather than blame. So far from wishing to scapegoat parents we wish to help them. So far from refusing to see that parents sometimes en- gage in horrific behaviour, we seek ways to suc- cour the casualties, old as well as young, psycho- logical as well as physical. Above all we seek ways of preventing violent patterns developing in new families. Let us hope that the policy of head-in- sand has had its day.
169/362
? ? ? CONCEPTUAL FRAMEWORK
In seeking to understand the more extreme ex- amples of violence in the family it is useful to consider first what we know about the more mod- erate and everyday example of members of a fam- ily getting angry with each other. Young chil- dren--and often older ones too--are commonly jealous of the attention mother gives the new baby. Lovers quarrel when one thinks the other is looking elsewhere--and the same is true after marriage. Moreover a woman may get very angry with her child if he does something dangerous, like running into the roadway, and also with her husband if he risks life or limb by taking unne- cessary risks. Thus we take it for granted that, when a relationship to a special loved person is endangered, we are not only anxious but are usu- ally angry as well. As responses to the risk of loss, anxiety and anger go hand in hand. It is not for nothing that they have the same etymological root.
In the situations described anger is often func- tional. When child or spouse behaves danger- ously, an angry protest is likely to deter. When a lover's partner strays, a sharp reminder of how much he or she cares may work wonders. When a
170/362
? ? ? child finds himself relatively neglected in favour of the new baby, assertion of his claims may re- dress the balance. Thus in the right place, at the right time, and in right degree, anger is not only appropriate but may be indispensable. It serves to deter from dangerous behaviour, to drive off a rival, or to coerce a partner. In each case the aim of the angry behaviour is the same--to protect a relationship which is of very special value to the angry person.
This being so, it is necessary to be clear why certain specific relationships, often called libidin- al relationships, should become so very import- ant in the lives of each one of us.
In his attempts to solve this problem Freud looked to the physics and biology of his day. Li- bidinal relationships, he proposed, were con- sequent on the individual's needs for food and sex. Subsequently, to account for some of the more puzzling manifestations of anger, he stepped outside biology to propose a death in- stinct. These hypotheses, framed in terms of the accumulation and discharge of psychic energies, led to a metapsychology so remote from clinical observation and experience that a great many analytically oriented clinicians have, implicitly or
171/362
? ? ? explicitly, abandoned it. To the resulting vacuum one reaction has been the development of a school of thought that has decided not only to di- vorce psychoanalysis from biology but to dis- pense altogether with scientific method and in- stead to espouse hermeneutics. Another and op- posite reaction has been to explore the principles found useful in modern biology, principles totally different from those of Freud's day, to see wheth- er they are more in tune with our clinical obser- vations and might therefore be used to construct a new metapsychology, or conceptual framework as it would now be called. That is the course that I and a number of others are following.
The specific relationships, threats to which may arouse anger, are of three main types: rela- tionships with a sexual partner (boyfriend, girl- friend, or spouse), relationships with parents, and relationships with offspring. Each type of re- lationship is shot through with strong emotion. In high degree indeed, a person's whole emotion- al life--the underlying tone of how he feels--is determined by the state of these long-term, com- mitted relationships. As long as they are running smoothly he is content; when they are threatened he is anxious and perhaps angry; when he has
172/362
? ? ? endangered them by his own actions he feels guilty; when they are broken he feels sad; and when they are resumed he is joyful.
In helping us understand why the state of these relationships should have such a profound effect on a person's feeling life, two branches of modern biology--ethology and evolution theory--are ex- tremely illuminating. Not only do all three of the relationships in question have counterparts in a wide range of other species but all three are in- timately concerned with the vital biological func- tions of reproduction and, especially, the survival of young. It is therefore more than likely that a human being's powerful propensity to make these deep and long-term relationships is the result of a strong gene-determined bias to do so, a bias that has been selected during the course of evolution. Within this frame of reference a child's strong propensity to attach himself to his mother and his father, or to whomever else may be caring for him, can be understood as having the function of reducing the risk of his coming to harm. For to stay in close proximity to, or in easy communica- tion with, someone likely to protect you is the best of all possible insurance policies. Similarly a parent's concern to care for his or her offspring
173/362
? ? ? plainly has the function of contributing to the child's survival. That success in the maintenance of these long-term relationships should usually bring satisfaction and contentment, and that fail- ure should bring frustration, anxiety, and some- times despair are, on this reading, the prizes and penalties selected during evolution to guide us in our activities.
It is within this evolutionary perspective that I believe we can understand how angry behaviour between members of a family can often be func- tional. For, as I said earlier, in the right place, at the right time, and in right degree, anger can serve to maintain these vitally important long- term relationships. But, as is very obvious, anger can be overdone. My thesis is simply that a great deal of the maladaptive violence met with in fam- ilies can be understood as the distorted and exag- gerated versions of behaviour that is potentially functional, especially attachment behaviour on the one hand and caregiving behaviour on the other.
There is now a considerable literature on the nature of the child's tie to his mother, tradition- ally referred to as dependency and now conceived in terms of attachment and careseeking. But in
174/362
? ? ? general terms, attachment behaviour results in one individual, usually the one who is less able to cope, maintaining proximity to, and/or commu- nication with, another individual, seen as better able to cope. This behaviour is elicited especially by pain, fatigue, or anything frightening, and also by the caregiver being or appearing to be, inac- cessible. Although conceived as being in part pre- programmed, there is now abundant evidence that the particular pattern in which attachment behaviour becomes organized during develop- ment is much influenced by how it is responded to by a child's principal caregivers, in the huge majority of cases his mother and father. Briefly, it seems clear that sensitive loving care results in a child developing confidence that others will be helpful when appealed to, becoming increasingly self-reliant and bold in his explorations of the world, co-operative with others, and also--a very important point--sympathetic and helpful to oth- ers in distress. Conversely, when a child's attach- ment behaviour is responded to tardily and un- willingly and is regarded as a nuisance, he is likely to become anxiously attached, that is, ap- prehensive lest his caregiver be missing or un- helpful when he needs her and therefore
175/362
? ? ? reluctant to leave her side, unwillingly and anxiously obedient, and unconcerned about the troubles of others. Should his caregivers, in addi- tion, actively reject him, he is likely to develop a pattern of behaviour in which avoidance of them competes with his desire for proximity and care, and in which angry behaviour is apt to become prominent. I shall say more about this later.
One further point about attachment behaviour I wish to emphasize is that it is a characteristic of human nature throughout our lives--from the cradle to the grave. Admittedly it is usually less intense and less demanding in adolescents and adults than it is in earlier years. Yet an urgent de- sire for love and care is natural enough when a person is anxious or distressed. It is therefore most unfortunate that, due to misleading theory, the pejorative adjectives 'infantile' and 'regressive' are now so widely current in clinical circles. They are words I never use.
Whilst the systematic study of attachment be- haviour, and especially the conditions influencing how it develops, has been in progress for twenty years, the systematic study of caregiving, or par- enting, and how it develops is only in its begin- nings. The approach I regard as the most
176/362
? ? ? promising is again an ethological one. This as- sumes that in humans as well as in other species parenting behaviour, like attachment behaviour, is in some degree preprogrammed and therefore ready to develop along certain lines when condi- tions elicit it. This means that in the ordinary course of events the parent of a baby experiences a strong urge to behave in certain typical sorts of way, for example, to cradle the infant, to soothe him when he cries, to keep him warm, protected, and fed. Such a viewpoint, of course, does not im- ply that the appropriate behaviour patterns manifest themselves complete in every detail from the first.
Clearly that is not so, neither in man nor in any other mammalian species. All the detail is learned, some of it during interaction with babies and children, much of it through ob- servation of how other parents behave, starting during the parent-to-be's own childhood and the way his parents treated him and his siblings.
RESEARCH FINDINGS
In considering what is now known of the indi- viduals involved in family violence, and the cir- cumstances in which it takes place, I start with women who physically assault their children and,
177/362
? ? ? next, with the effects on the children of such as- saults. My reason for doing so is that in these two areas the research findings seem most adequate. And in the light of those findings we can proceed to consider what is known about men who batter wives or children, a problem area just as import- ant but at present less well researched.
The findings of the many studies of women known to have battered their children show con- siderable agreement (see review by Spinetta and Rigler, 1972). Though probably commoner among families of lower socio-economic status, child abuse occurs also in middle-class families where it is likely to be hidden behind a fac? ade of ultra-respectability.
On the surface abusing individuals vary from being cold, rigid, obsessional, and censorious to being passive, unhappy, and disorganized. Yet emotionally they have much in common. Among features reported as especially frequent among abusive mothers we find the following: prone to periods of intense anxiety punctuated by out- bursts of violent anger, they are said to be im- pulsive and 'immature'. Although their 'depend- ency needs' are described as exceptionally strong, they are extremely distrustful and consequently
178/362
? ? ? unable or unwilling to make close relationships. Socially they are isolated. Having no one else to turn to, many of them seek care and comfort from one of their own children whom they treat as though they were much older than they are (Morris and Gould, 1963).
As regards childhood experience many re- searchers have noted that most of such women have had a miserable childhood and, as one pair of researchers put it, 'have been deprived of basic mothering' (Steele and Pollock, 1968). An appre- ciable minority were themselves battered as chil- dren. 1
To anyone thinking in terms of attachment the- ory the notion at once suggests itself that these woman are suffering from an extreme degree of anxious attachment and consequently that exper- iences of long or repeated separations and/or of being repeatedly threatened with abandonment would be a common feature of their childhoods. In a relatively small study these hypotheses have now been tested by Pauline DeLozier (1982), working in Los Angeles. Her samples consisted of 18 working-class women known to have assaulted their children physically and 18 others from the same socio-economic class, and matched for age
179/362
? ? ? (most in their thirties) and number of children, who had not done so. All subjects were given a semi-structured interview and a questionnaire to complete and were tested on the Hansburg Se- paration Anxiety Test (Hansburg, 1972). In the latter a series of pictures depicting scenes either of a child leaving his parents or of parents leaving the child is shown and the subject invited to de- scribe what she would feel and do in the situation concerned.
The results of the Separation Anxiety Test showed, as expected, that most of the abusing mothers were extremely sensitive to any type of separation situation, even the most everyday and commonplace, with responses indicative of high levels of anxiety and/or anger. The responses showed, in addition, that, while these women yearned for care, all they expected was rejection. Supporting another of the initial hypotheses was a high incidence of responses indicating anxious concern for the welfare of parents. For all these features the incidences in the control group, al- though not negligible, were significantly lower. For example, whereas 12 of the 18 abusing moth- ers were rated as showing the highest degree of
180/362
? ? ? anxious attachment, only two of the control group were so rated.
As regards the childhood experiences of these women, the results, as reported in interview and qiestionnaire, supported some of the initial hypo- theses but did not support others. For example, in view of previous reports DeLozier had expec- ted to find a significantly higher incidence of sep- arations from parents in the childhoods of the ab- usive mothers but this was not found. By con- trast, her expectation that a high proportion of these women would have experienced their par- ents as having repeatedly threatened to abandon them was fulfilled, a finding in keeping with the view that repeated threats to abandon are as pathogenic as actual separations and probably more so (Bowlby, 1973). In a similiar way, al- though actual violence from a parent had appar- ently not been common, many of the abusing mothers had suffered repeated threats of being beaten, maimed, or even killed.
Another striking feature of the childhoods of the abusing mothers, also in keeping with expect- ations, was that only a minority of them (seven) had felt they could turn to mother for help when in distress. Amongst those who could not were
181/362
? ? ? some who described someone else, a relative or neighbour, who would have been helpful; but four of the 18 had known of absolutely no one. Amongst the controls, by contrast, all but three had felt they could turn to mother, and each of the others had at least known of someone else.
Thus, unlike a girl who grows up in an ordinary reasonably happy home who can rely on her mother in an emergency, a majority of these wo- men could never do so. Indeed, and again as pre- dicted, for many of them the relationship of daughter to mother had been reversed and it was they who had been expected to care for the par- ent. 2
Given the childhood experiences these women had had, it is not difficult to understand why they had grown up as they had. Threats to abandon a child make her (or him) intensely anxious about any separation, however routine it might appear to others, and also intensely angry with her par- ent for threatening her so. Moreover failures to respond helpfully when a child is in distress, combined with repeated and impatient rejec- tions, lead her to be deeply suspicious of every- one else. Thus, whilst constantly yearning for the love and care she has never had, she has no
182/362
? ? ? confidence she will ever receive it; and she will mistrust any offer she may receive. Small wonder therefore, if when a woman with this background becomes a mother, that there are times when, in- stead of being ready to mother her child, she looks to her child to mother her. Small wonder too if when her child fails to oblige and starts cry- ing, demanding care and attention, that she gets impatient and angry with it.
It is against this background, I believe, that a mother's violent assaults on a child can be under- stood. Although I have never treated a woman who actually assaulted her child physically, I have treated one who came perilously close to doing so.
The reason I began seeing this woman, whom I will call Mrs Q, was that the doctor at the well- baby clinic she attended was concerned about her son, aged 18 months; he was refusing to eat and was losing weight. When I saw them both, it was at once apparent that Mrs Q was intensely anxious and depressed and had been so since the boy's birth. On enquiry I found that she was terri- fied lest her son die and was therefore pestering him to eat. She also told me that she had some- times had impulses to throw the baby out of the
183/362
? ? ? window. Only much later did she tell me that on occasion she became hysterical, smashed the dishes, and battered the baby's pram. She was in- tensely anxious talking to me and fully expected me to turn on her angrily. On my suggestion she came to see me for psychotherapy once a week.
The picture she gave me of her childhood, told reluctantly in fragments but always consistently, was one I now know to be typical. She recalled bitter quarrels between her parents in which they assaulted one another and threatened murder, and how her mother would repeatedly bring pres- sure on the family by threatening to desert. On two occasions Mrs Q had returned from school to find her mother with her head in the gas oven, and at other times her mother would pretend to have deserted by disappearing for half a day. Nat- urally Mrs Q grew up terrified that if she did any- thing wrong her mother would go. Moreover things were made even worse by her mother in- sisting that she breathe not a word about these terrifying events to anyone outside the home.
Mrs Q, who had been a skilled technician be- fore marriage, was known as a very helpful neigh- bour and did all she could to be a good wife and mother--in which for the most part she
184/362
? ? ? succeeded. Yet she was subject to these violent and destructive outbursts which frightened and puzzled her and about which she felt intense shame.
After some time I had little doubt that the angry outbursts were to be understood as the ex- pression of intense anger which, generated ini- tially and over many years by her mother's re- peated threats to abandon the family, had early in her life become directed away from her mother and towards less dangerous targets. Terrified then and later of ever expressing her anger dir- ectly, she redirected3 it towards something which, or someone who, could not retaliate. As a child, Mrs Q recalled, she had sometimes re- treated to her room and attacked her dolls. Now it was her crockery, the pram, and almost but not quite the baby. Each current outburst, I suspec- ted, was triggered by her mother who, dominat- ing and interfering as ever, still visited her daughter every day.
This explanation fits what facts we have and has the merit, not always appreciated in clinical circles, of being simple. Not surprisingly, other workers in the field have also proposed it (e. g. Feinstein, Paul, and Pettison, 1964). In other
185/362
? ? ? cases, it is clear, a husband has violently abused his wife and she, violently angry in return, has re- directed it against their child.
In turning next to the effects on the personality development of children who are assaulted, we have to bear in mind that the physical assaults are not the only episodes of hostility from parents that these children have experienced. In very many cases indeed the physical assaults are but the tip of an iceberg--the manifest signs of what have been repeated episodes of angry rejection, verbal as well as physical. In most cases therefore the psychological effects can be regarded as the outcome of prolonged hostile rejection and neg- lect. Nevertheless the experiences of individual children can vary greatly. A few, for example, may receive reasonably good care and only very occasionally suffer an outburst of parental viol- ence. For these reasons it is no surprise that the socio-emotional development of the children var- ies also. Here I describe findings that appear to be fairly typical.
Those who have observed such children in their homes or elsewhere describe them variously as depressed, passive, and inhibited, as 'depend- ent' and anxious, and also as angry and
186/362
? ? ? aggressive (Martin and Rodeheffer, 1980). Gaensbauer and Sands (1979), in endorsing this picture, emphasize how disturbing to a caregiver such behaviour can be. The children fail to parti- cipate in play and show little or no enjoyment. Expression of feeling is often so low key that it is easy to overlook, or else is ambiguous and con- trary. Crying may be prolonged and unresponsive to comforting; anger is easily aroused, intense, and not readily resolved. Once established, these patterns tend to persist.
An issue much discussed in the literature is the degree to which an infant's prematurity, ill- health, or difficult temperament may have con- tributed to a mother's problems and so ultimately to its having been ill-treated. In some cases these factors play a part, but they do so only when a mother reacts unfavourably to the baby and thereby sets up a vicious circle. 4 Such a sequence, of course, is all too likely to occur when a mother has herself had a difficult childhood, has grown up emotionally disturbed, and has little or no emotional support or help after her baby is born.
Towards his or her parent an abused toddler often shows a striking picture of frozen watchfulness, hyper-alert for what might happen.
187/362
? ? ? But some show also an unusual sensitivity to the needs of their parents (Malone, 1966). There are, in fact, good reasons for thinking that some chil- dren learn early that it is possible to placate a dis- turbed and potentially violent mother by constant attention to her wishes. 5
In a nursery setting battered infants and tod- dlers have a reputation for finding it difficult to make relationships, either with caregivers or with other children, and also for being very aggressive. In recent years these observations have been con- firmed and extended by more systematic research which has focused attention both on the particu- lar patterns of behaviour shown and on details of the situations in which each occurs. The findings that follow stem from a study being conducted at Berkeley by Main and George (George and Main, 1979; Main and George, 1985).
Their aim was to compare the behaviour in a day-nursery setting of two groups of children in the age-range 1 to 3 years. One group of ten were known to have been physically assaulted by a par- ent. The other group of ten were matched for all relevant variables but had not been assaulted; they were, however, in nurseries set up to care for children from families known to be under stress.
188/362
? ? ? To obtain data each child was observed for four periods each of 30 minutes on four different days spread over three weeks. Observers were instruc- ted to record all socially relevant behaviour, in- cluding such small movements as head-turning or stepping backwards.
In analysing the data, the children's social be- haviour was divided into four categories: ap- proach, avoidance, approach-avoid, and aggres- sion. It was also divided according to whom the behaviour was directed--another child or a care- giver. Another distinction was between behaviour initiated by the observed child and behaviour which occurred in response to a friendly ap- proach by another child or adult. Results are ex- pressed in terms of the mean number of incidents of a particular type of behaviour of the children in each group or else of the number of children in each group who showed that type of behaviour.
As regards occasions when a child initiated so- cial contact, either with another child or a care- giver, no appreciable differences were observed between the children in each group. By contrast, very striking differences were observed in the ways the children responded to a friendly ap- proach from the other. Characteristic responses
189/362
? ? ? of the abused children were either to take straight avoiding action or else to show both approach and avoidance behaviour, either in quick succes- sion or in some combination of the two. Examples are: 'she creeps towards him but sud- denly veers away', and 'she crawls towards the caregiver but with head averted'. Thus, when the overtures come from a caregiver, the abused tod- dlers were three times more likely than the con- trols to take avoiding action; whilst seven of them, compared to only one, showed the curious combination of alternation of approach and avoidance. When the overtures came from other children, the differences were even more marked. For example, whereas none of the controls showed approach-avoidance, all ten of the abused toddlers did so.
Aggressive behaviour was fairly common in both groups of toddlers, though, as predicted, it was significantly more so in the abused group. Not only did the abused toddlers assault other children twice as often as the controls, but five of them assaulted or threatened to assault an adult, behaviour seen in none of the controls. In addi- tion the abused toddlers were notable for a par- ticularly disagreeable type of aggression, termed
190/362
? ? ? 'harassment' (Manning, Heron, and Marshall, 1978). This consists of malicious behaviour which appears to have the sole intent of making the vic- tim show distress. Almost always it occurs sud- denly without any evident cause and so contrasts with hostility which occurs in reaction to a pro- vocation. Such attacks, coming unpredictably out of the blue, are frightening and invite retaliation. Clinical studies, referred to later, report them to be directed especially towards an adult to whom the child is becoming attached.
In view of the behaviour so far described it is not surprising to find that abused toddlers are singularly unsympathetic to age-mates in dis- tress. The studies of Zahn-Waxler and Radke- Yarrow have shown that infants and pre-school children who have affectionate and caring par- ents commonly express concern when another child is distressed and often make moves to com- fort him or her (Zahn-Waxler, Radke-Yarrow, and King, 1979). This type of behaviour was also seen at least once in five of the control children in the Main and George study; but on no occasion did any of the abused toddlers show the slightest hint of it. Instead, and unlike the controls, they reacted with some combination of fear, distress,
191/362
? ? ? or anger; and three behaved hostilely to the cry- ing child. For example, one small boy of 2 years 8 months slapped a little girl who was crying, ex- claiming repeatedly 'Cut it out, cut it out'. He pro- ceeded then to pat her back and next to hiss at her with bared teeth; before anyone could inter- vene, his patting had turned to beating.
My reasons for giving so much attention to these observations of young children will, I am sure, be apparent. They show with unmistakable clarity how early in life certain characteristic pat- terns of social behaviour--some hopeful for the future, others ominous--become established. They leave no doubts either about what types of family experience influence development in one direction or another. Again and again we see de- tails in the behaviour of a toddler, or in what he says, that are plainly straight replicas of how that toddler has himself been treated. Indeed the tendency to treat others in the same way that we ourselves have been treated is deep in human nature; and at no time is it more evident than in the earliest years. All parents please note!
Firm evidence of how these children develop must await an appropriately designed longitudin- al study. There is evidence that, if conditions of
192/362
? ? ? care improve, some recover sufficiently to pass for normal (Lynch and Roberts, 1982); others do not. Some have suffered serious brain damage and become diagnosed as mentally handicapped (e. g. Martin and Rodeheffer, 1980). For many others adverse conditions of care continue. Fur- thermore, once a child has developed the types of disagreeable behaviour described, it is not easy for an adult, whether parent, foster-parent, or professional, to give him the continuous affec- tionate care he needs, whilst treating such chil- dren by psychotherapy is extremely taxing. The sudden unprovoked attacks, which in older chil- dren can easily be damaging, are especially hard to take.
Some of these emotionally disturbed children, we know, reach psychiatric clinics where the ori- gin of their condition, I suspect, more often than not goes unrecognized. Amongst those who have treated these children, some of them psychotic, and who have traced the source of their troubles are Stroh (1974), Bloch (1978), and Hopkins (1984). Each notes the extreme degree of ambi- valence to be expected: one moment the child is hugging the therapist, the next he is kicking her. During adolescence and early adult life, some,
193/362
? ? ? mostly male, become diagnosed as aggressive psychopaths and/or violent delinquents (e. g. Far- rington, 1978). Others, especially females per- haps, are found to be suffering from multiple per- sonality (Bliss, 1980). Once psychiatrists become aware of the profound and far-reaching effects of childhood abuse and rejection, and the extent to which relevant information is suppressed and falsified by parents and overlooked by clinicians, many more cases are sure to be identified.
A significant proportion of rejected and abused children grow up to perpetuate the cycle of family violence by continuing to respond in social situ- ations with the very same patterns of behaviour that they had developed during early childhood.
A type of response found to be characteristic of many abusive parents, and of a kind which we have already seen to be characteristic of abused toddlers, is reported, for example, by Frodi and Lamb (1980). In a laboratory study, in which videotapes of crying infants were shown, abusive mothers were found to respond to a crying infant with less sympathy than did a group of non-abus- ive mothers and also with more annoyance and anger. Furthermore these same adverse re- sponses were shown by the abusive mothers even
194/362
? ? ? when they were shown video tapes of smiling in- fants, which suggested they disliked any form of interaction with an infant.
Let us turn now, rather belatedly, to the beha- viour of men who ill-treat girlfriend or wife.
Two of my social work colleagues at the Tavis- tock, Janet Mattinson and Ian Sinclair (1979), de- scribe a man, Mr S, who was apt, inexplicably and unpredictably, to attack his wife. At the time when he asked for an interview his wife had re- cently left him; she had just had their first baby.
