Introduction.

For many analysts the concept of projective identification has become a pivotal reference point in their work, for like the older term ‘countertransference' which it partially supersedes, it addresses the analyst's often confusing clinical experience. To those who have embraced it, it probably represents the most important advance in theory in the last sixty years. To those more sceptical about its true explanatory value, it can seem like a Trojan horse that has taken over the analytic establishment. Its importance can be gauged by the number of publications devoted to it; it has been the subject of several conferences (e. g. Sandler et al, 1988) and numerous reviews (esp. Ogden , 1979, 1994).

Since its introduction by Melanie Klein (1946), the concept has been significantly modified and extended (Bion, 1959, 1962; Rosenfeld, 1971, 1987; Grotstein, 1981; Joseph, 1987; Ogden, 1979; etc). In this regard, Bion was an important contributor because he transformed the concept from one which referred to a primitive psychic mechanism (something the infant ‘did' to an ‘other') into one that delineated an interpersonal process in which the recipient's role (the receiver of the projections) was critically important. Also in part through Bion's work, the process came to be seen as a primitive form of communication between baby and mother, perhaps, indeed, the only one there was. This and other extensions of the term have tended to make it a portmanteau concept which embraces on the one hand the infant's primitive defences against intolerable distress, on the other, the foundations of normal infant-mother communication.

Where perhaps there is common ground between supporters and critics of the term is the area of non-verbal emotional communication. This includes both the need to have an account of infant-mother communication during the preverbal period and the need to understand a non-verbal type of communication between patient and analyst in which some understanding of mental content is conveyed without the mediation of words. I shall touch on both these issues in this paper but my reference to the literature will be focussed in the extreme. Rather than summarising the literature on the subject (which has been attempted on many occasions), I will examine in some detail one key paper (Bion, 1959), viewing it in counterpoint with a paper of Enid Balint's (1963) from the Independent tradition. Bion's paper included clinical material that led him towards new understanding of the original concept and in discussing this, I hope to demonstrate a certain confusion of ideas that continues to dog the subject to the present day.

Although the concept of projective identification was introduced by Melanie Klein in 1946, it did not for some time acquire the currency it now possesses. Being embedded in the matrix of Kleinian theory, the term was probably uncongenial to analysts of other persuasions. Winnicott, for example, for whom the mother's contribution was crucial, seems never to have used the term, while the Balints, who also emphasised the actual mother's importance, preferred a language of fusion to one of projection and identification - e.g. Michael Balint's (1968) “harmonious interpenetrating mix-up”. Moreover, Independent Group theorists tended towards a conception of mother-infant relatedness in which the infant had a more active capacity for communication than the Kleinians envisaged. For them, the notion that everything was determined by primitive instinctual processes would have been difficult to accept. Winnicott's entire body of theory was built on the premise of adaptive environmental provision, and although the Balints developed no such comprehensive accounts of early development, their writing is permeated with similar thinking.

Feedback, mirroring and maternal rêverie.

In her paper On being empty of oneself (Balint, E., 1963), Enid Balint focused on a clinical picture that would now be termed borderline or narcissistic disorder. As the title suggests, she singled out a particular kind of disturbed experience, the origin of which she placed very early in development. She considered it to be linked, “both in… nature and chronology, to the importance for the child of communication with his mother (op. cit. p. 40).” She noted “that much help in understanding the condition is provided by Kleinian ideas” to do with introjection and envy, but concludes that ultimately such ideas “seem insufficient, since the patient described … was troubled more by the lack of a self than of objects good or bad inside herself (op. cit. p. 41).” Balint elaborated the view that for satisfactory development, something needs to take place between mother and infant that precedes the projective and introjective events discussed by Klein. If this something does not occur, a failure in the sense of self will follow. In other words, Balint believed that if there is no sense of a containing self, there can equally be no sense of anything contained. There can be no primitive sense of ‘me' and ‘mine', no envy, nothing to project, and nothing to preserve inside. She considered that her patient Sarah existed in this pre-self state.

Reconstructing the patient's development, she noted that while on the surface things developed satisfactorily, “there was apparently a vitally important area where there was no reliable understanding between mother and daughter (op. cit. p. 50).” It seemed to her that the mother “responded more to her own preconceived ideas” than to what her baby actually felt, and she linked this with Sarah's sense of never feeling recognised. [Perhaps], she wrote, “Sarah's mother could not bear unhappiness or violence or fear in her child, did not respond to it, and tried to manipulate her so that everything wrong was either put right at once, or denied (op.cit. p. 50).” She developed this idea as follows:

Sarah's mother was impervious to any communication which was different from the picture she had of her daughter [the concept of imperviousness is one that I shall discuss below.] …In consequence, Sarah could not understand her mother's communications and felt that her mother never saw her as she was. Neither found an echo in the other… [She refers to Winnicott's (1945) idea of mother and child ‘living an experience together', and thinks of this] as the child finding an echo of herself coming from the mother; or as the mother accepting her child's as yet unorganised feelings and emotions and, by her reactions to them, enabling the child to organise them into a self. I propose to call this process a ‘feed-back', which starts in the child and acts as a stimulus on the mother, who must accept it and recognise that something has happened. Her recognition results in a kind of integration, and this is then reflected, fed back, to the child… This feed-back process presupposes an interaction between two active partners, which, I think, differentiates it from projection and introjection in which one of them is only a passive object (italics mine) (p. 51).”

These ideas prefigure – and echo - the writings of two other theoreticians. On the one hand, they resonate with Winnicott's seminal paper of four years later: Mirror role of mother and family in child development (Winnicott, 1967). On the other, they have marked resonance with Bion's earlier reformulation of projective identification (Bion, 1959), which saw the process in a more interactive way than Klein had envisaged.

Of these two convergences, I shall first consider Winnicott's mirroring . Although he did not use the word, Winnicott (1967) proposed that the mother's face, with its rich variety of emotional responses, was a principal means through which the preverbal infant obtained emotional ‘feedback' about himself. The mother's face is the child's first mirror, and what the infant sees in the mother's expression is related to what she perceives as the infant's ‘experience'. Like Enid Balint, Winnicott saw this mirroring as essential to growth of the infant self, and without it, he thought, the infant would have little sense of existing at all.

Bion (1959), on the other hand, took Klein's theory that the baby ‘projected' something in phantasy into the mother – a one way traffic - and transformed it into a two-way, or circular transaction, in which the mother's contribution, her “reverie”, was vital. As Balint and Winnicott were later to do, Bion had begun to regard the process he was describing as an interactive form of communication between infant and mother, the maternal contribution being vitally important to the infant's emotional growth.

In this sense, the ideas of these three writers show remarkable convergence. One could say that certain ideas were “in the air” and each picked them up in a different way. It would be wrong, however, to suppose that each writer was saying the same thing in a different way. For while the two Independents, Winnicott and Balint, thought in a similar way, Bion, who was still a Kleinian in spite of his innovations, theorised within a different semantic space. I will consider this difference in greater detail.

Projective identification and maternal failure .

“Mrs Klein,” wrote Mrs Spillius, “thought of projective identification as a phantasy in which bad parts of the self were split off from the rest of the self and, together with bad excrements, were projected into the mother or her breast to control or take possession of her in such a fashion that she was felt to become the bad self. Good parts of the self were projected too, she thought, leading to the enhancement of the ego and good object relations, providing the process was not carried to excess (Spillius, 1988, p. 81).”

From this, it is clear that in its original conception projective identification was driven by a sense of threat : that either the self would be overwhelmed by bad objects, or the integrity of good objects would be jeopardised if left in the subject's own keeping. This ‘threat/defence' aspect of the concept is carried over into Bion's new formulation and it serves to differentiate it from the process that Winnicott and Enid Balint had in mind. For them, the transactions that occurred between infant and mother (mirroring, feedback and recognition) were not part of a system of defence. They were not a function of aggression or envy and the infant was being neither evacuative, nor hostile and destructive. On the contrary, they were relatively free from such negative impulses, and if responded to by the mother in a satisfactory way, led to positive developments. The mother's interventions not only relieved the infant's anxiety but bestowed a sense of enhanced aliveness. In short, they met a primary need for communication and ( maternal ) recognition that were essential for the growth of the self.

There are thus two contrasting accounts of early events between mother and infant and it could be possible to see Bion's enlargement of Klein's conception as offering a kind of synthesis. However, it seems to me that this is only superficially the case, for while in Bion's revised concept the maternal element neutralises innate destructive potential , in the Independent view of Balint and Winnicott, it serves a primary need to feel connected with the mother . This latter view looks back to a more relational tradition. It recalls Suttie's ‘primary need for the mother' (Suttie 1935), and has affinity with Bowlby's concept of attachment (Bowlby, 1969) which was being formulated at this time. It also prefigures the more experimental work of Trevarthen (1979) who wrote of the infant's need for “companionship” with the primary carer. By contrast, Bion's account still has one foot in Klein's model in which the primary driving force is the infant's primitive aggression and the need to defend against this.

Thus it is not a case of the same process being described in different ways but a fundamental divergence in the way infancy is conceived. On the one hand, the baby is a bundle of primitive impulses which seek discharge and threaten the potential cohesion of the self; on the other, the baby is object-seeking and attachment-seeking, with a primary need for relatedness to the mother and recognition by her. Within this perspective, the development of the infant depends on adequate fulfilment of this relational need by attuned and mirroring feedback.

As we shall see, Bion's model retains a basic conception in which the infant communicates with the mother by projecting, even forcing, emotional experience into her through projective identification i.e. infant projection is followed by maternal identification with what is projected. It is as though in this model the mother discovers what the baby feels by finding herself feeling in a particular way (cf. the way the analyst finds himself feeling some surprising affect and then ‘realises' it ‘belongs' to the patient). Winnicott's model, by contrast, sees the infant as communicating through emotional signs . The infant's affect is displayed in such a way that the attuned mother naturally ‘reads' it. To be sure the infant may not ‘intend' the mother to ‘read' its emotion, yet probably the display is programmed in a way that anticipates such maternal response. In the first case it is as though the mother is not attending to the signs and the baby has to redouble its efforts to make her notice. In the second case, the mother is right there, searching for the signs, in a state that Winnicott called ‘primary maternal preoccupation'. In the first case it as though the mother does not realise that her baby can communicate; in the second, she is constantly scanning the situation for communicative signs.

This raises the interesting possibility that when the infant (or patient) seems to force emotional experience into the mother (analyst), this is already a response to the mother's (analyst's) failure to ‘read' the signs. If this were so, projection of something into the mother would not, as Bion argued, be the primary means through which the infant gained access to the mother's attention and “reverie” - the fundamental form of infant-mother communication the Kleinian school asserts – but a consequence of breakdown in more basic modes of mother-infant relatedness . Such a breakdown would result from failure in attuned response on the part of the mother, though other factors, including impairments in the infant, could also be involved. I use the term attuned response in a general way to refer to any means through which the mother conveys to her infant that its specific message has been received and attended to. This will often take the form of appropriate action, though may also be conveyed in the way the mother handles and responds to her infant. I am assuming that a mother who is able to get close to what her baby is experiencing through imaginative identification will respond to that baby in significantly different ways from a mother who cannot make such an emotional leap.

Maternal imperviousness.

If we take this point of view seriously, attempts to force emotionally aroused parts of the self into the mother (projective identification) constitute an understandable response to maternal failure; they are not an innate defence mechanism. On the contrary, they would be acts of desperation in the face of maternal imperviousness rather than a fundamental means of preserving endangered parts of the self. It would be as though the baby were saying to the mother: “I've been ‘telling' you how I feel but you haven't listened! I will make you hear me and force you to understand what I'm going through!” It follows from this that there would be an inverse relation between the need for projective identification and the mother's capacity for attunement. The presence of projective identification would attest to a failure of maternal attunement, and the violence of the projective process would correlate with the degree of maternal imperviousness with which the infant had to contend.

Maternal imperviousness is a concept that was current in the scientific literature of the 1960's. It derived from the study of schizophrenic families, and was part of a cluster of terms that included the idea of the schizophrenogenic mother. As far as I can ascertain, the term was first defined by Lee (1963) (1), and operationalised by Laing, Phillipson and Lee (1966). The term was important in Enid Balint's paper from which I have quoted, and Bion refers to something similar (see below) when he describes how he came to reframe the process of projective identification to include the idea of maternal responsiveness and reverie (Bion, 1959). It may be significant that Bion too was working with psychotic and severely borderline patients at this time. Indeed, it was through such work, and the evidence of maternal deficit in the experience of these patients, that he developed the now familiar idea of the mother as “container” of the infant's projections. Bion proposed that a ‘normal' mother can take in the infant's ‘projections' and work on them in her ‘reverie' so that the infant can take them back, or ‘re-introject' them, in a modified and more manageable form.

Bion's clinical material.

Bion's primary purpose in his 1959 paper was to illustrate the operation of projective identification in borderline psychotic patients and to illustrate how this was used by such patients to attack and destroy any process that made links between objects. This included not only the parental relationship but also linking processes in the mind through which meaning was established. He wrote: “I shall discuss phantasied attacks on the breast as the prototype of all attacks on objects that serve as a link and projective identification as the mechanism employed by the psyche to dispose of the ego fragments produced by its destructiveness.” In other words, his paper was to be a demonstration of the usefulness of Klein's concept in clinical work.

However, in the second half of the paper he offers material from one patient that seemed to suggest a different kind of understanding. Rather than being in the service of primary aggressive phantasies, projective identification seemed to be reactive to maternal failure. I shall quote Bion in detail, highlighting key phrases, though of necessity, I have condensed his account. He writes (p.312):

Throughout the analysis, the patient resorted to projective identification with a persistence suggesting it was a mechanism of which he had never been able sufficiently to avail himself ; the analysis afforded him an opportunity for the exercise of a mechanism of which he had been cheated … There were sessions which led me to suppose that the patient felt there was some object that denied him the use of projective identification … the patient felt that parts of his personality that he wished to repose in me were refused entry by me … [Note that Bion does not question here the patient's (infant's) basic need to use projective identification; he merely describes a situation in which the process has been frustrated].

[Earlier associations] showed an increasing intensity of emotions in the patient. This originated in what he felt was my refusal to accept parts of his personality. Consequently he strove to force them into me with increased desperation and violence. His behaviour, isolated from the context of the analysis, might have appeared to be an expression of primary aggression. …[as in the earlier cases], “but I quote this series because it shows the patient in a different light, his violence a reaction to what he felt was my hostile defensiveness.

This analytic situation built up in Bion's mind an early scenario:

I felt that the patient had experienced in infancy a mother who dutifully responded to the infant's emotional displays. The dutiful response had in it an element of impatient ‘I don't know what's the matter with the child.' My deduction was that in order to understand what the child had wanted the mother should have treated the infant's cry as more than a demand for her presence. From the infant's point of view, she should have taken into her, and thus experienced , the fear that the child was dying (2). It was this fear that the child could not contain [i.e. the fear that the mother could not recognise. How Bion knows that this was precisely the fear is not clear but possibly it could be inferred from the clinical material he had at his disposal]. He [therefore] strove to split it off, together with the part of the personality in which it lay, and project it into the mother. An understanding mother is able to experience the feeling of dread that this baby was trying to deal with by projective identification, and yet retain a balanced outlook. This patient had had to deal with a mother who could not tolerate experiencing such feelings and reacted… by denying their ingress… (Bion 1959, pp. 312-313, italics and underlinings mine).

It is clear from this that Bion is describing a particular kind of maternal failure in which the mother (analyst) seems impervious to the infant's (patient's) mental state and the infant (patient) reacts to this with increasingly violent forms of projective identification. Bion's position here is extremely close to the one I have already outlined though a detailed consideration of his argument betrays important differences. These lie in the way he views both the normal role and function of the mother, and the nature of the normal processes in the baby. These are issues that have always separated Kleinian and Independent Groups. The Kleinian group gives primacy to the infant's inner world of phantasy; Independents give greater weight to the actual environment. Such differences lead to critical questions: Does the mother only become alerted to her infant's state because the infant ‘projects' (in phantasy) into her, in which case the infant's inner world and projective defences are the prime movers? Or, does the mother play a more active role, scanning the infant for specific ‘signs' of distress in a pre-emptive way? In the first case, projection comes first, and the mother may or may not respond (understand what is the matter, or in Bion's terms, ‘process' the projection), depending on her receptivity. In the second case, the mother's scanning is already in place when the baby cries, her state being similar to that described by Winnicott as primary maternal preoccupation.

I think that in this paper, clinical Bion is closer to Winnicott than theoretical Bion allows. He understands about maternal (or analytic) failure of receptiveness and sees the infant's (analysand's) desperate rage as reactive to this. He also understands how maternal imperviousness provokes phantasies of forcing oneself into the other's attention and emotional holding. Theoretical Bion, however, clings to the older Klein schema: The infant's primary form of communication is ‘projective identification' and when this is thwarted, a more hostile and intrusive form of the process supervenes. “[The patient felt that projective identification ] was a mechanism of which he had never been able sufficiently to avail himself ; the analysis afforded him an opportunity for the exercise of a mechanism of which he had been cheated … There were sessions which led me to suppose that the patient felt there was some object that denied him the use of projective identification … the patient felt that parts of his personality that he wished to repose in me were refused entry by me … (op.cit. p. 312)”.

In conclusion it can be said that Bion's thought makes an important leap in realising that the baby's (patient's) behaviour is often reactive to the mother's (analyst's) failure to understand and respond. However, he never abandons the earlier Kleinian view that projective identification is the fundamental form of infant-mother relatedness. In Bion's formulation the primary onus remains on the infant to project into the mother, not on the mother to seek out the infant's signs of distress. For Bion, as for Klein, projective identification is primary ; the baby engages in it from the beginning and the mother's contribution merely facilitates or thwarts this fundamental process.

Primary maternal preoccupation and maternal imperviousness.

There are thus two elements in my argument: a mother who registers, or fails to register, the specifics of the infant's state through empathy and imaginative identification; and a baby who emits specific distress signals, while anticipating in some basic way that these will be responded to. The problem for theory is to state the relationship between these two elements.

The Kleinian view, even in Bion's modified form, asserts that the infant puts the experience of an emotional state into the mother through projective identification; the Independent view, represented by Enid Balint and Winnicott, turns the process on its head: it suggests that the mother anticipates and enters into the baby's state proactively through a process of imaginative identification . In the first case, the baby initiates and does all the work, ‘forcing' entry into the mother through projective identification; in the second, the mother leads and takes much of the load, feeling her way into the baby through a symbolically mediated process (see below).

Underlying these two accounts are different views of infant development, in particular, an implied disagreement about the infant's degree of experienced separateness from the mother and how this comes about. It has often been noted that infantile projective identification makes no sense unless the baby has a rudimentary sense of separateness: How else could a baby project something into an object unless that object was felt to be separate (Rosenfeld, 1971 [1988]; Sandler (1988)? The Kleinian view thus implies a sense of separateness from the beginning. By contrast, the Independent view (Winnicott and the Balints) considers that the infant has no sense of separateness for a significant period after birth (cf. also Mahler et al., 1975) (3).

In Winnicott's account, the key to the baby's unawareness of separateness is primary maternal preoccupation (1958) which cushions the impact of ‘reality' - in other words, the experience of frustration and difference. The mother holds these at bay by her adaptive response and only gradually introduces them in small and manageable doses. In this way, the separateness of ‘reality' is assimilated imperceptibly into experience through graduated maternal failure. It is central to this view that any sudden impingement of reality is potentially traumatic and part of the mother's role is to see that this does not happen.

Within this model of infant care, maternal attunement and maternal imperviousness work in opposite directions. Maternal attunement shields the baby from the harshness of the object world while maternal imperviousness throws the baby against it. In a direct and obtrusive way, it forces the object mother into the frame, and throws the baby into premature frustration and premature object awareness. Maternal attunement delays the development of separateness while maternal imperviousness catastrophically provokes it.

Winnicott was thus keenly aware of the potential trauma of sudden change. He had seen, for example, how damaging it could be to an infant if the mother became depressed, and he saw this in terms of the sudden loss of maternal adaptation (e.g. Winnicott, 1967). He was less obviously concerned with consistently impervious mothers (like those described in psychotic and schizophrenic families), who had never (or seldom) been able to respond accurately to their baby's needs. Sarah's mother, in Mrs Balint's case, and many of Bion's cases, seem to have been of this kind and it may be that Bion's ideas concerning the ubiquity of projective identification were shaped as much by this skewed sample of clinical experience as by the Kleinian theory he inherited.

Within this maternal adaptation framework, projective identification can be seen as the infant's attempt to overcome the imperviousness of the maternal object. The mother's habitual failure in adaptation throws the infant into premature awareness of separateness (Winnicott) and the infant must then find increasingly aggressive ways of forcing itself into the object's awareness. With each failure of the process, the infant becomes more desperate, more violent and full of rage, and it is not difficult to see how this could culminate in phantasies of entering and/or destroying the frustrating object in ways similar to those described by Klein. Finally, in so far as there may have been earlier experiences of at-oneness with the mother, however fleeting, it can be seen how efforts to breach the maternal imperviousness might constitute attempts to regain such a lost state in which parts of the self can again be experienced as ‘at one' with the maternal object.

The issue of ‘non-verbal communication'.

Any consideration of projective identification has to address the question of non-verbal communication for two reasons: first because clinically, the concept of projective identification is most often invoked when the analyst ‘picks up' on an affect in the session that was not being expressed in a verbal or obvious way; and secondly, because the concept is used to explain how the infant communicates affective states to the mother during the pre-verbal period. In the first case, the analyst discerns the affective state through an often incongruous aspect of his own experience in the session, and then tells himself that the patient has ‘projected it into him'. In similar fashion, the mother is deemed to become aware of what her infant is ‘experiencing' when the infant has ‘projected' this affect into her.

However, the problem with such ‘explanations' is that they do not explain anything at all. To explain a process of this kind we would have to be able to say how it was mediated and on this aspect analysts are usually silent. However, in so far as such experiences are accurately perceived, it is clear that the mediating mechanism must be some kind of non-verbal communication. The fact that the analyst cannot describe this is merely evidence of being taken unawares. It is in the nature of such experiences to be unexpected.

There is, however, a more intrinsic difficulty in discussing non-verbal communication. Communication as normally understood involves the mediation of symbols . These are structures which ‘carry' or ‘refer to' meanings, the words of language being the clearest example. When it comes to non-verbal processes the situation is more complicated. Affects, for example, are normally mediated by signs which are part of the expression of the emotion itself ( Darwin , 1872). When you look at me and see the expression of anger on my face, for example, you are able to ‘read' these signs, even though communicating my affect to you by means of my expression is not part of my intention. My angry expression is merely an accompaniment of my inner state and for this reason it cannot be said that I am conveying my experience in a symbolic way.

At other times, however, such intentional communication may be operating. For example, you might tell me something that I disapprove of and I might curl my lip in a particular way that conveys my distaste. In this case, my expression could be regarded as a conscious symbol because I intended to convey my distaste to you. The curling of the lip is a culturally shared non-verbal symbol of distaste.

While, however, the expression of affective signs (as opposed to symbols) is largely involuntary, and at most we may partially suppress them, their reading (symbolic processing) is not involuntary in the same way. While the potential for recognition of emotional expression (signs) is usually thought to be innate, the degree to which we become aware of, and respond to, another's feeling is a variable matter. To do so, we have to be ‘open' to the other's affective signs and perhaps on the look out for them. Winnicott's “good enough mother” could be thought of in this way, with “primary maternal preoccupation” being an accentuation of such normal receptivity. By contrast, an impervious mother is relatively ‘closed' to the recognition of affective signs.

To understand this fully, however, it is necessary to consider how a person normally deals with affective messages. This can be thought of in terms of Bion's processing ( a -function) through ‘reverie' but can more plainly be designated as a stage of symbolic elaboration. Thus, for the mother to ‘know' what is troubling her baby, she not only has to read the affective signs or signals, but interpret them by imagining herself into the baby's situation. She has to be aware of when the baby was last changed and fed, and whether it is teething or suffering with colic. She needs to bear in mind recent upheavals of routine or changes in carer, and she has to synthesise these possibilities into a ‘picture' or ‘story' about the baby. This is what I refer to as “ imaginative identification” because it is a process of symbolically imagining oneself into the other person's situation, of constructing a context in which a certain signal makes sense (4). It is quite different from projective identification, being a form of knowing, not a means of affecting the other person (a form of action).

To understand projective identification in these terms we would have to regard the person said to be employing this mechanism as emitting non-verbal affective signals (signs) through tone of voice, posture, and so on, while not knowing that he is doing this . In other words, he is impervious to his own affective state . He does not know anything about it because he has not symbolically processed it. This could be seen in terms of the attitude he originally internalised in childhood from his own impervious mother. His impervious attitude towards his own emotional state is a present day version of her attitude towards him in childhood. His imperviousness affects his awareness of his affect (he does not ‘know' about it, just as his mother did not), but not its underlying development when events set it in motion.

In these terms, the analyst can be seen as ‘picking up' the emotional signals which are very much there, though absent from the patient's awareness, and often too from his verbal communication. Moreover, it is not necessary to see the patient as ‘projecting' such signals ‘into' the analyst in the way that Kleinian theory, with its penumbra of aggressive intent, dictates. It could just as well be that the analyst is getting in touch with such signals because he is turned towards the patient (open to him) in a state of heightened receptivity.

That the analyst is emotionally turned towards the patient in such a special way could scarcely be contested. He holds the patient in mind and is consciously and unconsciously in touch with the vast body of ‘knowledge' he has built up about him. He is on the look out for emotional signals, scanning the field for affective ‘information'. In this respect, he is not so different from the mother Winnicott describes as in a state of ‘primary maternal preoccupation' – ‘primary analytic preoccupation' would convey something of the stance that every analyst would recognise. Like the mother in this state, the analyst relates imaginatively to every aspect of the ‘material', sensing what ‘might be the case' from all the available cues. In this way the unacknowledged affect signal of the patient comes to be ‘contained' (Bion) within a symbolic fabric provided at least in part by the analyst.

Conclusion.

By examining Bion's modification of Klein's theory of projective identification in the light of Winnicott's work I have developed an account of the clinical phenomenon of projective identification which frees it to some extent from its theoretical Kleinian matrix. I have suggested that we do not have to regard the infant's basic mode of communication with the mother (the emission of emotional signals) as a projective phenomenon but merely as an affective display which is linked to an inbuilt expectation of maternal response. In so far as the more projective state of affairs develops, I have constructed a scenario, closely following Bion's insights, in which increasingly aggressive and projective phantasies in the infant can be seen as meaningful reactions to maternal failure. This failure takes the form of maternal imperviousness, by which I mean a lack of receptiveness, and response to, the infant's emotional messages. I have argued that the clinical phenomena in question do not necessitate adherence to Kleinian assumptions of defence against primitive aggression but can more readily be understood as the infant's desperate responses to an object that has failed to understand and resonate with its disturbed state. My account suggests that projective identification (in the Kleinian ‘aggressive' sense) is not basic to infant functioning in the way that is often supposed, but the result of thwarted attempts to obtain response from the mother. I am assuming that the infant has an inborn propensity to communicate distress through non-verbal channels, and equally, an inborn expectation that these will be responded to in ways that lessen the disturbance.

That infant's engage in communicative and responsive interaction from birth is now attested by a huge volume of infant research (e. g. Stern, 1985). Such communication frequently has a mimetic quality and is often structured into complex sequences. Disturbance in expected rhythms and patterns creates confusion and disturbance in the infant, and probably the maternal input to such sequences contributes to the infant's sense of being held and contained in a maternal environment. I underline the existence of such communicative patterns between mother and infant, because it is often stated that projective identification itself is the main pathway of communication in infancy. In my view, this is not supported by the available evidence.

Footnotes .

•  Lee (1963) wrote: For smooth, adequate interaction to occur, each party must register the other's point of view. …Typically the parent fails to register his child's view, while the child does not register that his view has not been (and perhaps cannot be) registered… Most often the parent appears to remain impervious to the child's view because he feels it is uncomplimentary to him, or because it does not fit his value system… The parent insists that the child does believe what the parent feels the child “should” believe… The child feels as if he continuously runs into an invisible, solid glass wall (quoted in Watzlawick et al. 1968, italics mine).

•  The fact that Bion speaks of the mother taking into herself the baby's state illustrates the point I make below about theoretical bias. Bion could have talked of the mother (or analyst) imaginatively entering into the baby's (analysand's) experience - what I call ‘ imaginative identification' (see footnote 4). But Kleinian theory has already decided that the baby communicates through projection into the mother (i.e. projective identification), not through the mother imaginatively seeking out and entering into the baby's experience.

•  Winnicott, for example, spoke of the “merged in state”, Michael Balint (1968) of a “harmonious interpenetrating mix-up”.

•  Note that the analyst is constantly performing a similar function in relation to the patient – imaginatively contextualising the patient's utterances, and in those cases where he invokes projective identification, the patient's unacknowledged affective signs.

References

Balint, E. (1993). Before I was I. Psychoanalysis and the Imagination.

Ed. By J. Mitchell and M. Parsons. London : Free Association Books.

Balint, E (1963). On being empty of oneself. In Balint, E. Before I was I. pp. 39-55.

London : Free Association Books, 1993,

Balint, M. (1968). The Basic Fault: Therapeutic Aspects of Regression.

London : Tavistock.

Bion, W. (1959). Attacks on linking. In: Spillius (1988), pp. 87-101.

Bion, W. (1962). Learning from Experience. London : Heinemann.

Bowlby, J. (1969). Attachment and Loss. Vol. I., Attachment. London :

The Hogarth Press.

Darwin, C. (1872). The Expression of the Emotions in Man and Animals.

London : John Murray.

Grotstein, J.S. (1981). Splitting and Projective Identification. New York : Aronson. Klein, M. (1946). Notes on some schizoid mechanisms. In: The Writings of Melanie

Klein, vol. 3, 1-24.

Lee, A.R. (1963). Levels of imperviousness in schizophrenic families. Paper read at

Western Division Meeting of the American Psychiatric Association, San

Francisco, September 1963. Quoted in Watzlawick et al. (1968).

Laing, R. D., Phillipson, H., and Lee, A.R. (1966) Interpersonal Perception: A Theory

and Method of Research. New York : Springer.

Mahler, M., Pine, F., and Bergmann, A. (1975). The Psychological Birth of the

Human Infant. London : Hutchinson .

Ogden, T. (1979). On projective identification. Int. J. Psycho-Anal.

60: 357-353.

Ogden, T. (1994). Subjects of Analysis. London : Karnac.

Rosenfeld, H. (1971). Contribution to the psychology of psychotic states. In Melanie

Klein Today. Developments in Theory and Practice vol 1: Mainly Theory.

Ed. E. Spillius, pp. 117-137, London : Routledge, 1988.

Rosenfeld, H. (1987). Impasse and Interpretation. London : Tavistock.

Sandler, J. (1988). The concept of projective identification. In: Projection,

Identification, Projective Identification , Ed. J. Sandler. London : Karnac, 1988.

Spillius, E. B (1988). Melanie Klein Today. Developments in Theory and Practice.

Vol 1. Mainly Theory . Ed. E. Spillius. London : Routledge.

Stern, D. (1985). The Interpersonal World of the Infant. New York : Basic.

Suttie, I. (1935). The Origins of Love and Hate. London : Kegan Paul.

Trevarthen, C. (1979) Communication and cooperation in early infancy: A description

of primary intersubjectivity, in: M. Bullowa (Ed.), Before Speech ,

pp. 321-349, Cambridge : Cambridge University Press.

Watzlawick, P., Beavin, J.H., Jackson , D. D. The Pragmatics of Human

Communication. A Study of Interactional Patterns, Pathologies and

Paradoxes. London : Faber and Faber .

Winnicott, D.W. (1945). Primitive emotional development. In Winnicott, D.W.

(1958), pp. 145-156.

Winnicott, D. W. (1956). Primary maternal preoccupation. In: Winnicott, D. W.

(1958), pp. 300-305.

Winnicott, D. W. (1958). Collected Papers –Through Paediatrics to Psychoanalysis.

London : Tavistock.

Winnicott, D.W. ((1967). Mirror role of mother and family in child development. In:

Winnicott, D.W. (1971), pp. 111-118.

Winnicott, D. W. (1971). Playing and Reality. London : Tavistock.

Summary

In this chapter, I discuss the Kleinian concept of projective identification and examine the clinical material that led Bion to revise Klein's theory. Approaching the concept from a Winnicottian perspective, I suggest that the process is less fundamental to infant life than is often claimed, and can be understood, at least in part, as a reaction to maternal imperviousness. I question the view that it forms the basic mode of communication with the mother and suggest that primitive communication is better understood in terms of the emotional signs which are integral to emotional arousal. I argue that such signs are not projected into the mother by the baby but have an innate communicative intent and are ‘read' by an attuned mother who then imagines herself into the baby's situation (a process mediated by symbols). According to this view, the use of projective identification in adult life is evidence of failure in early relating rather than persistence of (or regression to) primitive defences.

Abstract

The Kleinian concept of projective identification is often invoked to explain non-verbal processes in the analytic situation and other close relationships and in this paper I examine some of the clinical material that led Bion to revise Klein's original formulation. Approaching the concept from a Winnicottian perspective, I suggest that the process is less fundamental to infant life than is often claimed and can be understood, at least in part, as a reaction to maternal imperviousness. I question the Kleinian assumption that it constitutes the basic mode of infant communication and suggest that primitive communication can better be understood by reference to the emotional signs or signals which form an integral part of emotional arousal. I suggest that such signs are not projected into a reluctant mother by the baby but noted and read from a position akin to ‘primary maternal preoccupation' (Winnicott) by a mother who imagines and feels herself into the baby's situation. From this perspective, projective identification can be seen as a pathological consequence of maternal failure rather than a fundamental defensive feature of infant life.

Biographical details .

Dr Kenneth Wright is a psychoanalyst working in the Independent tradition. He is also a general psychiatrist and qualified in adult and marital psychotherapy at the Tavistock Clinic in London . He is a patron of the Squiggle Foundation and works in private practice in Suffolk , England . His book Vision and Separation: Between Mother and Baby (London: Free Association Books,1991) was awarded the Margaret Mahler Literature Prize in 1992 (published in Italy as Visione e Separazione: Tra Madre e Bambino. Roma: Borla, 2000.) He lectures nationally and internationally and has published a number of papers on psychoanalysis, the creative arts and religion. Recent papers include: ‘To make experience sing.' In: Art Creativity Living, Ed. Lesley Caldwell, London : Karnac, 2000; ‘Preverbal experience and the intuition of the sacred.' In: Psychoanalysis and Religion in the Twenty-first Century: Competitors or Collaborators, Ed. David Black, London : Routledge, 2006.

mail

Kjt.wright@doctors.org.uk

.