DURING THE EARLY decades of psychoanalysis very little was known about the development either of a child’s cognitive abilities or of his relations with his parents during his first two or three years of life. As a result a variety of rather arbitrary guesses were made, many of them much influenced by the assumption that a child’s interest in persons is necessarily secondary to, and derived from, his desire for food. Thus at one extreme it was confidently believed that a child as old as two years is still so dominated by his physiological needs that he switches his affections promptly to whoever at the moment is meeting them. At an opposite extreme complex cognitive abilities, and relationships with the breast as a part-object, were attributed to infants no more than a few months old. On the basis of these ideas two or more quite different theories of social development were elaborated by psychoanalysts, leading inevitably to widely divergent theories of mourning.
Today, thanks to the systematic studies of a steadily increasing band of developmental psychologists, it is no longer necessary to rely on guesses. Whilst a great deal still needs to be learned, a reliable outline of cognitive and socio-emotional development during the earliest years is becoming available. In this chapter I indicate briefly how children’s responses to separation and loss can be considered in the light of these findings.
In giving attention to cognitive development I am drawing first and foremost on the work of Piaget (see especially The Construction of Reality in the Child, 1937 and Play, Dreams and Imitation in Childhood, 1951) and also on the work of Bower (1974), whose imaginative experiments have led him not only strongly to support Piaget’s concepts but also to elaborate and clarify them. Bower’s work suggests that during the first year of life an infant is appreciably more advanced in his cognitive development than Piaget initially supposed and much other work points also in that direction (see the critical review by Gratch, 1977). By cognitive development is meant the steps through which an infant progresses which result in his behaviour no longer being dependent exclusively on immediate stimulus input but becoming, instead, guided by rules that enable him to combine perceptual information with information from memory. By means of these rules he becomes able to predict more or less accurately what is likely to happen in his world and to plan and respond accordingly.
Piaget was the first to point out that it is not until half way through the first year of life that an infant makes any attempt to search for some interesting object he has seen disappear. Before that age not only does he make no attempt to search for it but when, later, the same object reappears he treats it as though it were a different one (as judged by latency of response and allied measures). Out of sight seems, therefore, to be truly out of mind. The reason for this, Bower suggests, is that prior to five months an infant seems able to identify an object by means of only two of its many possible identifying characteristics, namely either that it remains in the same place or else that it follows a consistent trajectory. Such obvious features as the object’s size, shape and colour on which adults usually rely to identify particular objects are not used during these early months.
From about five months onwards, however, a marked change occurs. Henceforward size, shape and colour come to have significance. One result of this is that an object that reappears after an interval is recognized and treated as being the same object. Another is that it is credited with having a continuing existence even when out of sight, as is shown by its being searched for. Admittedly, for many months an infant’s searching is sadly ill-guided and he makes many strange mistakes. For example, at first, even after being allowed to watch where an object is being hidden, perhaps under a cloth, he will none the less search for it not where he saw it put, but either in the place where he had last seen it or else in a place where he had found it previously. Later, having come to realize that missing objects can be hidden in any of a great variety of places, he still has difficulty in realizing that hidden objects may be moved from one place to another inside a container. For example, if a coin is hidden under one of two similar cups and the cups are then transposed in his presence he will nevertheless look for the coin under the wrong cup, namely the one which is now in the place where he had last seen the coin. Only after he has reached fifteen months or more is it likely that he will be able to solve this problem.
Thus, even after an infant has begun searching for an object he has seen disappear, it is only step by step that his knowledge of where to look for it improves and results finally in an efficient performance. From these observations Piaget and other students of cognitive development infer that, during the later months of the first year and the early months of the second, infants are becoming increasingly able to conceive of an object as an entity that exists independently of themselves. During these stages of development an infant is said to be achieving the concept of object permanence.
Piaget’s own experiments were done mainly with small objects like keys and matchboxes, and the question arises whether his conclusions regarding the development of the concept of an object apply equally to the development of the concept of a person. Evidence at present available suggests that in principle they do though there are certain differences; for example, an infant’s knowledge of persons develops rather earlier than his knowledge of things, as Piaget himself confidently predicted. What is especially interesting, however, is that, also as Piaget predicted, his skill in searching for and finding persons in successively more difficult situations seems to progress step by step through all the same stages as does his skill in searching for things (Décarie 1965; Bell 1970; Brossard in Décarie 1974).
An infant’s attachment to a discriminated mother-figure is developing fast between the fourth and seventh month. Perceptual recognition comes early; but because before five months his concept of object is still primitive, some strange behaviour is observed. For example, as we have seen, evidence suggests that an infant of about twelve weeks thinks that an object is defined either by the place it occupies or else by the trajectory it follows, not both. As yet, therefore, he seems not to realize that one object may be first in one place and then move to another; nor that it cannot be in two places simultaneously. Evidence in keeping with these ideas is that when an infant of less than five months is shown multiple images of his mother (presented by means of some optical device) he is not disturbed but interacts happily with each of the ‘mothers’ in turn. Should, however, one or two of the mother-images be replaced by those of strangers, he knows well whom he prefers. Once past the age of five months, by contrast, his grasp of the properties of objects is such that the sight of multiple images of his mother becomes highly disturbing: by now he knows he has but one mother and that she cannot be visually present in two, or three, places at once (although as we see later he is not always certain about this).
From these and other observations it can be inferred that during these middle months of the first year an infant is developing some elemental representation of his mother-figure. Yet for him to be able to recognize her when present is not the same as his being able to recall her when absent; and there is reason to think that the latter capacity is not developing until the final months of the first year. An illustration of this developmental step is a finding, reported by Schaffer (1971) and already described in Chapter 3 of Volume II. When an infant of six months is placed with his mother sitting out of sight behind him and is then confronted with a strange object he behaves as though his mother were not there. When, by contrast, an infant of twelve months is placed in the same situation he habitually turns round to refer to her before deciding how to respond. Thus, within these admittedly narrow limits of space and time, one-year-olds brought up at home have no difficulty in knowing where their visually absent mother is or in utilizing that knowledge.
In keeping with this, by the last quarter of the first year a secure infant whose mother is responsive to his signals, is happy to play on his own for a time evidently aware that his mother, though visually absent, is available nearby should he want her (Stayton and Ainsworth 1973).
Not unexpectedly infants differ greatly in the age at which they develop all these cognitive abilities, whether applied to inanimate objects or to so important a person as mother. As regards the latter, some infants are already showing a limited degree of skill in finding her by seven months and can solve all or most of the problems by nine. Others are slower by several months (Bell 1970; Brossard in Décarie 1974). Also not unexpectedly, the age at which an infant develops these abilities is much influenced by his experiences. An infant whose mother is responsive to his signals and engages in plenty of social interaction with him is likely to be more advanced than is one whose mother is less attentive. Because infants differ so much in the age during which they are achieving the concept of person permanence or, more precisely, mother permanence, any statements linking stages of cognitive or emotional development to chronological age have to be treated with caution.
As a child develops his concept of person permanence he becomes increasingly capable of representing to himself the whereabouts and doings of absent persons. Thus, during the early months of the second year, a healthily developing child is becoming able to draw on his general knowledge in order to deduce where a vanished person may have got to and how he got there. In illustration of this achievement Piaget cites an incident in which one of his children, Laurent, a few days short of eighteen months, was asked, successively, where absent members of the family were. In reply he pointed each time to where he supposed them to be, evidently influenced in his opinion either by where they had been an hour previously or else by knowledge of their habitual occupations.
From observations such as these cognitive psychologists conclude that most children aged eighteen months and older, who have been reared in attentive homes, are able not only to represent the external world symbolically but also to manipulate their representations. By so doing a child can recapitulate actions of the past and anticipate actions of the future, including reaching a solution to a problem by purely cognitive means and without resort to action. These cognitive achievements, Piaget believes, and with him many linguists, provide a child with a necessary (though perhaps not a sufficient) basis to start comprehending and producing language (Cromer 1974).1
Although the work of Piaget, and of others working in the same tradition, suggests that a child is not capable of recalling and using his representational model of the world in all these more complex ways before the middle of the second year, it also shows that he is capable none the less of various embryonic degrees of representation throughout the preceding twelve months. Thus, it is extremely misleading to speak as though a child’s representational model of his attachment figure is absent before a certain age or stage of development and present thereafter. Instead, the model is to be thought of as developing during the middle months of the first year from which time it is available for recognition and elementary search and, as the months pass, is becoming increasingly available also for recall and for cognitive operations. This way of conceptualizing early development I believe to have much more explanatory potential than those traditionally advanced by psychoanalysts. The serious shortcomings of the much invoked concept of ‘libidinal object constancy’ are discussed shortly.
Fields about which we still know too little are the length of an infant’s or young child’s memory span and the conditions that enable him to recognize or recall significant people and places after varying lapses of time. What is known, however, suggests that during the early months an infant has a memory span for visual information a good deal longer than is sometimes supposed (see reviews by Cohen and Gelber 1975, and Olson 19762). For example, Fagan (1973) presents evidence that an infant of five months shown the photograph of a face for only two minutes can still recognize it when presented to him two weeks later. In the light of that it is hardly surprising that Bower (1974) should have observed that infants aged five and six months brought into the laboratory for a second visit often reveal that they remember what happened during their visit of a day or so earlier and begin rehearsing their responses before testing begins; nor that Ainsworth and her colleagues (Ainsworth et al. 1978) should have observed that infants aged twelve months put through the strange situation sequence for a second time two weeks after the first clearly foresaw what was going to happen and responded accordingly. As regards the development of spontaneous recall our information is still negligible and it would therefore be unwise to draw conclusions. In particular, until far more knowledge is available it is rash to conclude, as some clinicians do, that a child younger than eighteen months is wholly unable without reminders from adults to recall persons and places after an interval longer than a day or two.
In conclusion, it should be noted that experimental studies of young children’s skill in recognizing items after a lapse of time would lead us confidently to expect that a child’s capacity to recognize and recall his mother would be developing weeks and probably months in advance of his capacity to recognize and recall anything or anyone else. The reasons for this lie partly in the fact that she has far greater emotional salience for him than does anything else and partly in the fact that he has far greater and far more varied experience of interacting with her—through sight, sound, smell and touch—than he has with anything or anyone else.
In discussing problems of childhood mourning many psychoanalysts invoke the concept of ‘object constancy’, sometimes expanded to ‘libidinal object constancy’. Since I believe this to be a most unsatisfactory concept, I do not use it. As Fraiberg (1969) points out, the term has come to be employed in a number of quite different ways, the variations reflecting in part the mixed parentage of the concept and in part a shift in the meaning given to the word ‘constancy’.
The concept was introduced by Hartmann (1952) in connection with the contrast between what was, and by some psychoanalysts still is, believed to be a phase of development when an infant has no interest in any ‘object’ (person) except at the moments when it (she) is satisfying his physiological needs, and a much later phase when he is thought to become emotionally attached to a discriminated person: in Hartmann’s words, ‘there is a long way from the object that exists only as long as it is need-satisfying to that form of satisfactory object relations that includes object constancy . . .’. The following year Hartmann (1953) linked his new concept to Piaget’s already established concept of object permanence. Partly as a result of this linkage and partly for other reasons psychoanalysts have come to use the term object constancy in no less than three distinct ways.
(a) One usage is simply to equate object constancy with Piaget’s object permanence. This is how Spitz (1957) used it and also how it is used by Furman (1974).
(b) A second usage rejects any linkage to cognitive psychology and reserves the term to denote ‘the child’s capacity to keep up object cathexis irrespective of frustration or satisfaction’, a phase postulated to be contrasted sharply with a previous phase during which a child is held to consider the object as ‘non-existent, unnecessary’ whenever ‘no need or libidinal wish is present . . .’ (A. Freud 1968). This usage is in keeping with Hartmann’s original proposal and is adopted by Anna Freud and those influenced by her.
(c) A third usage, which grows out of the second but is not the same, applies object constancy to the stage of development when a child can ‘remain away from the mother for some length of time and still function with emotional poise, provided he is in a fairly familiar environment’ (Mahler 1966). In Mahler’s scheme of development this capacity is regarded as developing during the fourth sub-phase of separation-individuation, which extends from about 25 to 36 months. In addition to Mahler herself, this usage is adopted by her associates Pine (1974) and McDevitt (1975).
As a result of these different usages the age at which a child is held to attain libidinal object constancy varies from six months to later in his third year.
Let us consider these three usages in the light of the scheme of development proposed in this work.
Since the first usage is synonymous with the terms ‘object permanence’ and ‘person permanence’ already established in the field of cognitive psychology, the new term is redundant. Furthermore, use of object constancy in this sense risks confusion with the term ‘perceptual constancy’ which refers to the ability to perceive an object as staying the same size, shape and colour despite changes in its orientation and lighting that alter how it appears to the eye. (According to Bower (1974) this ability is already in evidence by 22 weeks.)
The second usage assumes two distinct phases of development, an early phase lasting well into the second year during which the ‘object’ is believed to exist for a child only so long as it is need-satisfying and a later phase, that of object constancy, when that is no longer so. Since, however, the assumption that there are two such phases is out of keeping with the evidence, there is no need to introduce a special term.
The concept to which the third usage applies looks at first sight to be an equivalent of the concept of secure attachment as it is manifested in the behaviour of children towards the end of their third year and which I have discussed in earlier volumes, notably in Chapter 21 of Volume II. It is, however, a little different.
In Mahler’s thinking, the assumption tends to be made that it is only when a child is becoming able to sustain short separations with equanimity, e.g. a morning in a playgroup, that we can properly credit him with having developed the capacity to evoke mental representations of his missing mother (e.g. Mahler 1966). This seems to imply that as soon as he is able to evoke a mental representation of her he will be able to sustain brief separations with equanimity. I see no evidence that these two developmental steps occur simultaneously. On the contrary, the evidence shows that the capacity to evoke a representational model develops independently of the ability to sustain separations of the kind proposed and that it usually precedes the latter by a year or two; indeed in the case of pathological development it may do so by an indefinitely long period. This means that, although the capacity to evoke a representational model is necessary if a child who is approaching his third birthday is to sustain such separations with equanimity, it is far from being a sufficient condition (a point also made by McDevitt 1975). For conditions to be sufficient not only has the current external situation to be a familiar one and the child himself healthy and unfatigued but the model of the missing mother that he evokes must represent her as being reliably accessible and also well disposed towards him. This is a development that evidence shows is dependent not only on the maturation of certain cognitive skills but also on the form the child’s model of his mother takes, which is in turn dependent in high degree on how she treats him. Thus the development of secure attachment is conceived not simply as a maturational stage but as a step along certain only of the array of developmental pathways that are initially available to a child.3
Although for purposes of theory building the differences between these formulations are of considerable consequence, for purposes of treating patients they are not necessarily so. It is of interest, for example, that the therapeutic principles advocated by Fleming (1975) and which she derives from Mahler’s developmental scheme are extremely close to those I have myself derived from attachment theory (Bowlby 1977).
When data on how infants and very young children respond to the temporary absence of mother-figure are examined in the light of the findings described earlier no incompatibilities are apparent. Indeed, each set of data illumines the other.
At about six months of age an infant, if mobile, will attempt to follow his mother out of the room and will greet her on her return. Yet at that age there is no reason to suppose that during her absence he has access to whatever germinal representation he may be developing of her. As an illustration of this (though no proof) we note that in the experiment by Schaffer (1971) referred to earlier, infants of six months behaved quite differently to those of twelve months. When confronted by a strange object not one of these younger infants turned round to refer to his mother standing immediately behind him. Instead, each seemed totally absorbed by the object before him and totally oblivious of his mother’s proximity.
As another illustration we note that when infants of twenty-six weeks and less are placed in a strange place without mother they appear to accept strangers as mother-substitutes without noticeable change in level of responsiveness and show little or none of the protest and fretting typical of the slightly older child. From the age of seven months onwards however a child in this situation not only notices the change but, by protest and crying and also by persistent fretting and rejection of the strange nurses, indicates his intense dislike of it. Furthermore, on returning home after periods in hospital lasting as long as three weeks about half those aged 7 to 9 months and almost all those over ten months become extremely clinging and cry excessively whenever mother is absent.4 These observations are clearly consistent with the hypothesis that during the third quarter of the first year an infant’s representational model of his mother is becoming readily accessible to him for purposes of comparison during her absence and for recognition after her return. They are consistent also with the view that during these months an infant is developing a capacity to conceive of his mother as a person existing independently of himself.
Before he can talk the only evidence we have that a child is thinking about his absent parent derives from observations of his behaviour. Since, however, only very few observers have been alert to the relevance of such observations our knowledge remains scanty. The few records available are therefore of much interest. One, observed during a longitudinal study undertaken by Margaret Mahler, is reported by McDevitt (1975). It records the fairly typical behaviour of an infant who had achieved some degree of person permanence and who was left by her mother in what (for the infant) was a fairly strange place.
One morning, when Donna was nine months and two weeks old, her expression suddenly became very solemn as she watched her mother leave the room. When the door closed, she started fussing; immediately she sat herself down and mouthed a toy. During the half hour her mother was gone, Donna was constantly on the verge of tears, could not be distracted easily, and looked at the door over and over again, often with a worried expression.5
During the second year of life there is no lack of records of children left in hospital or nursery watching the door through which a parent has departed, and doing so persistently for several days in the evident hope of seeing him or her return through it. An example is the sixteen-month-old Dawn who, left in a residential nursery, stood for days near the door fretting inconsolably (Chapter 24). Similarly, the hopes raised both in the seventeen-month-old Jane when on her fifth day away she noticed the gate of her parents’ garden and in the sixteen-month-old Dawn when on her thirteenth day in the nursery she apparently mistook the female observer for her mother (described in the previous chapter) are wholly in keeping with what present knowledge of early cognitive development might lead us to expect.
Earlier in the chapter it was noted that observation shows that when searching for a missing object an infant under twelve months is extremely likely to search for it wherever he last saw it or else where he last found it, and that he behaves similarly with his missing mother. Although this tendency to locate persons in special places, and to find it difficult to think of them as being anywhere else, diminishes during development, it seems not to disappear altogether. Indeed common experience suggests that it stays with many people throughout life.6 To its persistence, I suggest, can be traced certain common features of mourning.
One is the strong tendency for a mourner, of any age, to have the vivid experience of seeing or hearing a dead relative in a place he used often to be in. Another is the possibly universal tendency for a mourner to think of the departed as being located somewhere definite—perhaps in the grave, or in heaven, or in one of his favourite haunts, or, as in Japan, in a special shrine—and why being able to do so commonly brings a sense of stability and comfort. Yet a further feature may be the perplexing tendency of a mourner sometimes to think of the person lost as being in both of two places simultaneously. An example is a bereaved child who both knows that his parent has gone for ever and also expects him or her to return shortly.
In his discussion of the development of person permanence Piaget (1937) has reported observations that seem to throw much light on how such incompatible belief systems may develop. The observations that follow7 concern his younger daughter, Lucienne.
At 15 months Lucienne is in the garden with her mother. Her father arrives. She sees him come, recognizes him and smiles. Her mother then asks her: ‘Where is Papa?’ Curiously enough, instead of pointing straight to him Lucienne turns towards the window of his office where she is accustomed to seeing him and points there. A moment later the experiment is repeated. Though she has just seen her father no more than a yard or so away yet, when her mother says ‘Papa’, Lucienne at once turns again towards his office.
Three months later, when aged 18 months, Lucienne behaves in a similar way, this time in reference to her elder sister. For a week Jacqueline had been unwell and confined to bed, and Lucienne had visited her there. This day Jacqueline was able to get up and so Lucienne has been playing with her downstairs. Despite this, however, Lucienne later climbs the stairs, clearly expecting to find Jacqueline still in her bedroom.
Similar episodes occur when Lucienne is 2½ years old and again at 3½.
On the latter occasion, after seeing her godfather depart at the end of a visit, Lucienne returns to the house and goes straight to the room in which he had slept. ‘I want to see if godfather has left,’ she announces. Then, entering the room alone, she assures herself, ‘Yes, he has gone.’
Not only do these observations demonstrate how closely tied to a particular location a person is in the world of a young child but they show also how easily this leads to a person being conceived as being in two places virtually at once. Furthermore, they call sharply in question the notions that such ‘splits in the ego’ are necessarily as pathological as it is customary for clinical theorists to suppose and that their origins must go back to the earliest months of life. On the contrary, Piaget’s observations and conclusions strongly support the view, expressed in earlier chapters, that incompatible beliefs of this sort are of normal occurrence at all ages and that whether they lead to pathology turns on how much opportunity a bereaved child or older person has of discussing his uncertainties with an understanding and trusted companion.
In the light of these ideas and of the observations reported earlier, there seem good grounds for attributing a germinal capacity for mourning to young children at least from sixteen months onwards. This implies that, as in the case of Jane, they have the ability to construct and to retain an image of their absent mother, to distinguish mother from foster-mother and to know well whom they prefer. It implies too that they relinquish the missing figure only reluctantly and that, when given the opportunity, become attached to the new figure only gradually.
At first it seems likely that the length of time during which the distinction between old and new and preference for the old can be maintained is a matter of weeks rather than months; and until about the second birthday doing so may require the active co-operation of the foster-mother. The reason for conceptualizing the responses of children of this age in the same terms as those of older children and adults is that the ontogenetic continuity of the responses is thereby emphasized so that differences of response at different ages can then be studied as variations on a common theme.
How best to characterize the responses of children aged between about six months and sixteen months, however, remains a puzzle. If the cognitive psychologists are right in believing that a child of less than about seventeen months has only the most limited capacity for symbolic functioning the term mourning may be inappropriate. Yet throughout this age-range a child shows manifest distress when his familiar attachment figure is missing and as the months pass searches for her with an increasing degree of competence. Even when he is receiving skilled foster-care, distress at separation and elements of search are present; and after reunion he shows a degree of recovery, the speed and extent of which turns both on the length of the separation and the conditions of care during it. Because the distress is so clearly a response to the absence of a highly discriminated individual there is, at the least, good reason to continue the practice of referring to it as grieving, as has for long been customary see (Chapter 1).
In considering these matters I am of course aware that much of the debate about whether very young children mourn arose as a result of the opinions I expressed in papers published during the late nineteen-fifties and early sixties. In those papers my aim was to emphasize that a child’s attachment to a mother-figure develops during the first year of life, to a large extent independently of whoever feeds him, that after six months he is overtly distressed when he loses his mother and that, whatever the differences may be when the responses of the earliest years are compared to those of later ones, the similarities are both evident and important. Nothing in the subsequent debate—neither empirical observation nor theoretical argument—leads me to change that opinion.
Where my position has changed is that I now give much more weight than formerly to the influence on a child’s responses of the conditions in which he is cared for whilst he is away from his mother, whether the separation be temporary or permanent. With regard to this the work of the Robertsons with children in their second and third years has been especially valuable by calling attention to the mitigating effects of good foster-care. Yet here again I am struck as much by the similarities in the ways that very young children, older children and adults respond as by any differences. For it is by no means only the responses to loss of very young children that are influenced by the family conditions that obtain after the loss. The same is true, as we saw in Chapters 15, 19 and 21, in regard to older children and adolescents; and the same again, as we saw in Chapter 10, in regard to adults. Indeed, in preparing this volume nothing has impressed me more deeply than the evidence showing the pervasive influence at all ages of the pattern of a human being’s family life on the way he responds to loss.
1 Because a child’s use of language lags far behind his use of non-verbal modes of representation there is a persistent tendency for adults to underestimate a young child’s cognitive capacities.
2 After reviewing the evidence Olson concludes that ‘the simplest generalization is that infants in the range of 3 to 6 months do not forget visual stimuli very rapidly if they have had sufficiently long initial exposure and if there is relatively little specific interference’.
3 A child developing compulsive self-reliance may also show a capacity to sustain brief separations with what appears to be equanimity; but the representational model of his mother that he is presumed to have developed is of course a very different one.
4 These observations, which are reported by H. R. Schaffer (Schaffer 1958; Schaffer and Callender 1959), are consistent with those reported by Spitz (1946a) in his pioneer work. See also Yarrow (1963) on the responses of infants after transfer to adopting parents.
5 Although McDevitt notes that Donna was more sensitive than other children to her mother’s comings and goings, he regards the form of her response as typical of the phase of development in which she was.
6 Experiments by von Wright et al. (1975) suggest that it may be a basic characteristic of human information processing that the location of an item is coded and stored in memory, routinely and automatically, with other information about the item. The experiments, using pictures, were done with children and young adults at several age-levels from 5 years to 18–23 years, with similar results at all age-levels. It is possible therefore that a strong association of person with place is a special example of a general tendency.
7 The accounts given are abstracts of material presented in Observation 51 on pp. 58–9 of the English translation of 1955.