Projective identification

Klein's paper first appeared in the International Journal of Psycho-Analysis in 1946 and was reprinted, revised, in Developments in Psycho-Analysis^[2] (1952). It describes the months that precede the depressive position she had set out in the 1930s. In this phase the infant's anxiety is mainly persecutory: a fear of annihilation arising from the death instinct within, which is deflected outward and met again as persecution. Its chief defence is splitting. The object, at first the breast, is divided into a good and a bad one, and since the ego cannot split the object without being split itself, the same division runs through the self.

Projective identification is the phantasy that builds on this splitting. Split-off parts of the self are forced into the mother in order to harm, possess and control her, and she comes to be felt as those parts. When the hated parts are projected into her, she is felt to contain them and to be the bad self, and Klein takes the identification that follows as the prototype of an aggressive object relation. Good parts of the self are projected too. Within limits this is necessary for good object relations, but in excess it depletes the ego. Klein treats excessive splitting and projective identification together as what weakens the ego in the earliest position, and she argues that when persecutory anxiety and splitting are too strong the depressive position cannot be worked through, so that the fixation points of schizophrenia lie in this first position.

The name came in two steps. The term appears only once in the 1946 paper; the sentence that explicitly proposes it as a name for these processes was added in the revised version of 1952. The same paper takes the word schizoid from Ronald Fairbairn, whose account of schizoid phenomena Klein discusses at length, accepting much of it while objecting to his neglect of anxiety, aggression and the death instinct.

The New Dictionary of Kleinian Thought (2011), whose entry the Melanie Klein Trust reproduces, defines projective identification as an unconscious phantasy in which aspects of the self or of an internal object are split off and attributed to an external object. What is projected may be felt by the projector to be either good or bad. The phantasy need not be accompanied by any action, but it may come with "evocative behaviour unconsciously intended to induce the recipient" to feel and act in accordance with it. The entry also gives such phantasies acquisitive as well as attributive properties: the projector may not only get rid of parts of the mind but also, in phantasy, enter the other in order to take qualities they want, so that projective and introjective phantasies operate together.

On its relation to projection, the same entry says that British Kleinians tacitly treat projection and projective identification as the same thing, regarding the latter as an enrichment or extension of Freud's concept of projection. In Klein's 1946 account the projection is followed by an identification: the mother is felt to contain the projected parts and to be them.

Rosenfeld's Analysis of a Schizophrenic State with Depersonalization^[3] (1947) was the first clinical paper to use Klein's newly named concept. In his account the psychotic transference is carried by projective identification: the patient puts split-off parts of the self into the analyst, who is then experienced as those parts rather than as a separate person. The 1947 paper was collected with others written between 1946 and 1964 in Psychotic States^[4] (1965), which also contains his paper On the Psychopathology of Narcissism^[5] (1964). There the omnipotent narcissistic object relation rests on projective and introjective identification: parts of the self are put into the object, and the object is taken in as if it were the self, so that no separate object is acknowledged.

Rosenfeld's posthumous book Impasse and Interpretation^[6] (1987) deals with the analyst's contribution to successful and unsuccessful treatment and with the sources of therapeutic impasse, and gives one of its four parts to the influence of projective identification on the analyst's task, another to the influence of narcissism on it. In his later work Rosenfeld sorted the forms of projective identification by motive: communicative projections that the analyst can contain, as against evacuative and controlling ones aimed at denying separateness and dependence.

In Notes on Symbol Formation^[7] (1957) Segal used projective identification to account for disturbances of symbol formation in psychotic patients. When projective identification is excessive, part of the ego is identified with the object, and the symbol, itself a creation of the ego, is then identified with the object it stands for; this is the symbolic equation. Later, after Bion, Segal separated the pathological form of projective identification from its normal, communicative use.

In Attacks on Linking^[8] (1959) Bion describes the attacks that the psychotic part of the personality makes, in phantasy, on anything it feels to be a link between two objects. He places these attacks within Klein's paranoid-schizoid position and her account of splitting and projective identification, but adds a distinction of his own. Projective identification is not only a means of evacuating unwanted parts of the self. In its normal form it is the infant's earliest way of communicating: the infant puts into the mother feelings it cannot bear, such as the fear that it is dying, so that they can be modified there and taken back in a form it can tolerate. When the mother, and later the analyst, will not receive these projections, the infant resorts to projective identification with ever greater force and frequency, and its development deteriorates. Bion therefore traces attacks on linking to two sources together: an inborn disposition to destructiveness, hatred and envy, and an object that denies the patient the normal use of projective identification. The outcome is a primitive superego that forbids projective identification, and an internal object that intensifies the emotions it was needed to modify.

In Learning from Experience^[9] (1962) the same model enters a theory of thinking. Sense impressions and emotional experiences that alpha-function does not transform remain as beta-elements, undigested facts fit only to be evacuated by projective identification or discharged in action. Alpha-function develops in the infant through the mother's capacity for reverie: she takes in the states the infant projects into her, such as a fear of dying, and returns them in a form it can bear, and the infant introjects the function along with them. When she cannot, what returns to the infant is nameless dread. In the book's later chapters Bion abstracts this mother–infant model into the relation of container and contained, a concept that derives from projective identification and that his later books develop.

Meltzer followed the concept from his first book, The Psycho-Analytical Process^[10] (1967), to The Claustrum^[11] (1992). In the first, projective identification organises the gathering of the transference, and massive projective identification produces geographical confusions between inside and outside, self and object, which the analysis then has to sort out. In the second he separates its intrusive, omnipotent form from the communicative projective identification described by Bion. He keeps the name projective identification for communicative projection into a receptive container, and calls the omnipotent phantasy of forcing one's way into the internal mother to possess her qualities intrusive identification. The inner world it produces is the claustrum, divided into the compartments of the head-breast, the genital and the rectum, in which the intruder lives in claustrophobic dread of expulsion.

In Explorations in Autism^[12] (1975), the report of a research group of child psychotherapists he led, Meltzer described a way of relating that diverges from projective identification. Projective identification needs an object with an inside to project into. Adhesive identification belongs to a two-dimensional world without inner space, in which the self sticks to the object's surface and imitates it instead of entering it.

In Borderline Conditions and Pathological Narcissism^[13] (1975) Kernberg counted projective identification among the primitive defences, centred on splitting, that mark borderline personality organization. Borderline patients put into others what they cannot bear in themselves while still feeling it and trying to control the other, and Kernberg used the concept to read their transferences. A chapter of the same book argues for treating the analyst's whole emotional response, the countertransference, as a source of knowledge about such patients.

Ogden, in On Projective Identification^[14] (1979) and Projective Identification and Psychotherapeutic Technique^[15] (1982), described projective identification as at once an unconscious fantasy, a mode of communication, a primitive object relation and a pathway for psychological change. He set out three phases: a fantasy of putting part of oneself into another person; interpersonal pressure that leads the recipient to feel and act in line with it; and the recipient's processing of what was induced, which the projector can then take back in modified form. In Subjects of Analysis^[16] (1994) he recast projective identification as a form of the analytic third, the subjugating third, in which the separate subjectivities of analyst and patient are partly subsumed and then reclaimed.

The authors above distinguished forms of projective identification by what is projected and why. Klein held that good parts of the self are projected as well as bad ones, and that within limits this is necessary for good object relations, while in excess it depletes the ego. Bion separated a normal, communicative projective identification from the evacuation of unwanted parts, and Segal, after him, separated its pathological form from its normal use. Rosenfeld distinguished communicative projections, which the analyst can contain, from evacuative and controlling ones. Meltzer kept the name for the communicative form and gave the omnipotent one a name of its own. The New Dictionary of Kleinian Thought distinguishes attributive from acquisitive phantasies, and Ogden's account adds the recipient's part, in which what was induced is processed and can be taken back.

Projective identification derives from projection and splitting and belongs to the paranoid-schizoid position. The symbolic equation, narcissistic object relations, the psychotic transference, attacks on linking, container and contained, and the analytic third all derive from it. Intrusive identification and the claustrum name Meltzer's omnipotent form of it and the inner world that form produces, while adhesive identification describes a mode of relating that diverges from it. Bion's alpha-function, beta-elements and reverie describe what happens to projected states when they are, or are not, received and transformed.