Dallas Society for Psychoanalytic Psychology
4516 Lovers Lane #446 Dallas, TX 75225-6993






















This paper was originally published at Psyche Matters and is available on the DSPP web site with the permission of the web editor for Psyche Matters.   This paper has been made available online with the kind permission of Peter Fonagy, PhD, FBA.  Do not duplicate without permission of the author. The author may be reached at p.fonagy@ucl.ac.uk

The Process of Change and the Change of Processes:
What Can Change in a ‘Good’ Analysis'

Peter Fonagy, PhD, FBA

Keynote Address to the Spring Meeting of Division 39 of the American Psychological Association, New York 16th April 1999

Horizontal Line


It would be a brave or foolhardy man (or woman) who would stand in front of an audience of psychoanalysts and psychotherapists and announce that he (or she) had the "definitive model" of therapeutic action. Fred Pine (1998) suggested that it is no longer useful to look for a single model of therapeutic action in psychoanalysis, that the mechanisms will be individualized according to patient and therapist. I feel it would be shirking my responsibilities, however, were I just to give you a "state-of-the-art" review of recent and not so recent developments. Instead I shall attempt a compromise between scholarship and arrogance and present a simple personal heuristic that helps me understand how patients progress through psychoanalytic treatment.

I shall distinguish three types of psychic change linked to each other loosely across the course of treatment, each working through a focus on the relationship with the therapist. They are: (1) Intersubjective representational shifts; (2) Changes of mental processes and (3) Changes in mental representations. Before I become turgid and academic, let me introduce you to one of my first private analytic cases: Mr. A.

Introducing Mr. A.

Mr. A started his analysis in a rage that Kohut (1972) himself would have been proud of. The couch was invented to humiliate and belittle people who needed help. The analyst’s silence was a deliberate mockery. My talking about: "working together" on his problems was a calculated insult. I wanted to work "on him", or more likely "in him". Working with him suggested a partnership which was clearly a million miles from his perception of my intention in recommending the couch. His assessment with Professor X had been civilised and urbane, whilst he could tell that I was a second-rater who lived in the shadows of great men. The paper tissue on the pillow provided a magnet around which the iron filings of his shame and sense of being ridiculed were all aligned. Did I think that he would infect my other patients? Did I think he had head lice? Or, even worse, did I delude myself into some pathetic medical identity by the pretence of sterile conditions?

I knew nothing about Mr. A at this stage, except that he was 32 and well to do. The referral was from a very senior colleague who had seen and liked him, and pressured me quite hard to promise to take him, sight unseen as it were. He came late for his initial interview and said very little about his background except that his childhood had made him into an inadequate man suffering from anxiety and depression and frequently from an overwhelming sense of inadequacy in competitive relationships with other men. He could pay reasonable fees. I was relatively inexperienced - who was I to say that Professor X was wrong in suggesting that Mr. A would be a good case for me to learn to work independently as an analyst. In fact I recall feeling extremely grateful to my senior colleague after the referral. By the end of Mr. A’s first session I was ready to kill Mr. A and my senior colleague along with him. In fact I had a formulation. My senior colleague had exactly the same narcissistic problems as Mr. A, so naturally he saw him as a charming man. As far as I was concerned, I would have been happy for them to enjoy each others’ salubrious company.

For the best part of a whole year of five times weekly analysis Mr. A maintained me in such a quasi-homicidal state. Of course I managed some interpretive work with him but I found it hard to keep my wish to punish and humiliate him at bay. On a Friday session he was particularly boastful, listing the properties he owned and suggested that my consulting room could be moved, with advantage, to one of his large houses which was situated in a neighbourhood where many successful psychiatrists practised. I managed to squeeze out an interpretation about him wanting me close to him over the weekend and also under his control, so he could avoid the humiliation of having to miss me. In response he assured me that if that had indeed been his intention then he would simply have bought the house I was in. But in reality, he was quite fed up with my monotonous whining, was grateful for the respite and considered extending it by taking an unscheduled break early the coming week. I persisted in what I myself at the time realised was a feeble, easily deflected approach. "You are frightened of the helplessness which allowing yourself to become attached to someone faces you with. You don’t need to have complete control and buy this house as long as you can arrange to come and go as you please." He responded hitting hard and below the belt by some margin: "Look, if you could afford to buy this house, then you would not be renting one of its shabbiest rooms. Just because you can’t, there is no point in you getting irritated, just because you know I could."

This bit of repartee was typical of my work with him. Working on his experiences in the here and now came naturally to Mr. A, as there appeared to be few people besides himself who he was interested in. I began to think of him as "Teflon man" as none of my attempts to reach him appeared to stick. I repeatedly tried to address his sense of shame and vulnerability which I believed, as I am sure you will agree, partly motivated his obnoxious defensive behaviour. He was apparently quite unable to hear my comments and often, quite rudely, started talking in the middle of carefully sculptured interventions. He would make me feel alternately enfeebled and furious, and at some level I was deeply puzzled as to why he was coming.

Maybe at this stage you deserve an explanation as to why Mr. A is disrupting our discourse on therapeutic action, as quite obviously little useful analytic work was being done. The challenge which Mr. A set psychoanalytic scholarship, going way beyond my shabby consulting room, was the improvement in Mr. A’s personal happiness and relationships that accompanied our sorry attempt at starting a psychoanalytic process. His depression disappeared, his anxieties receded, his collaboration with his male colleagues, including his competitors, improved, he even started what seemed from a distance to be a reasonable relationship with a woman who was his equal - a radical change from his previous attempts at forming attachments with prostitutes or uneducated women whom he delicately referred to as "scrubbers".

It seems to me that cases of unaccountable improvement, such as happened with Mr. A, are perhaps more common than psychoanalytic reports of successful treatments might lead us to suspect. But even if there were just one Mr. A, the benefit he gained from his less than adequate treatment might justify scrutiny. So, let’s try and do this systematically. What might the books say?

The question of outcome aims

One approach, perhaps most characteristic of the French school of psychoanalysis, might be that symptomatic improvement is a largely irrelevant aspect of the therapeutic action of psychoanalysis. However, already in 1965 Wallerstein noted that whilst analysts adopted a therapeutic stance which could be characterised as "without memory or desire" (Bion, 1967), behind this lay the dramatic ambition to fundamentally alter the patient’s personality organisation (p.749). A lack of concern with clinical outcome seems curious at a time when evidence-based medicine (Kerridge, Lowe, & Henry, 1998; Sackett, Rosenberg, Gray, Haynes, & Richardson, 1996) is forcing all mental health practitioners to state their therapeutic goals and the expected outcome of their interventions.

Let me take a short digression on the issue of outcome, after all it could be somewhat presumptuous to discuss the therapeutic action of a treatment that we do not know to have any therapeutic effect. It is often said that there are no studies on the effectiveness of psychoanalysis and long-term psychodynamic psychotherapy. In fact this is not true. In a recent review of outcome studies in psychoanalysis and long-term psychodynamic psychotherapy undertaken by the International Psychoanalytic Association (Fonagy et al., 1999), four case record studies, thirteen naturalistic pre-post or quasi-experimental studies, nine follow-up studies and nine experimental studies were identified. In addition, six substantial process-outcome studies were reviewed. The report runs to 320 pages and is available at cost from the International Psychoanalytic Association, or if you have time from the web on the IPA’s website. While I am in plugging mode, I should mention that this review would not have been possible without the Psychoanalytic Electronic Publishing CD-Rom which contains the complete text of all major psychoanalytic journals from the beginning of their publication.

There have been quite comprehensive reviews of such studies carried out before (e.g.Bachrach, Galatzer-Levy, Skolnikoff, & Waldron, 1991; Crits-Christoph, 1992; Lazar, 1997; Roth & Fonagy, 1996) but they tend to come to quite varied conclusions. It is of course easy to be critical of psychoanalytic outcome research: there are no studies which definitively show psychoanalysis to be more effective than an active placebo or even than an alternative treatment. In fact, there are no methods available that would unequivocally indicate the existence of a psychoanalytic process. Most studies have significant limitations which lead critics of the discipline to discount their results. Amongst the most common problems are inadequate specification of the treatment procedures, lack of control on selection biases in sampling, unstandardised methods of assessment, lack of statistical power, etc. Notwithstanding these and other severe limitations, many of the studies are impressive: they report results which other psychotherapies have not been able to achieve, particularly with intractable disorders; some show very long-term benefits; results tend to be highly consistent across studies; and some include samples larger than those of most better controlled treatment trials. So whereas it is true that the outcome of psychoanalysis is not established, some suggestive conclusions may be drawn:

Psychoanalysis benefits the majority of those who are offered this treatment (Fonagy et al., 1999) and brings the functioning of a clinical group to the level of the normal population (e.g.Keller, Westhoff, Dilg, & Rohner, 1998; Sandell et al., 1997; Teufel & Volk, 1988). Longer treatments have better outcomes (e.g.Erle, 1979; Erle & Goldberg, 1984; Erle & Goldberg, 1979), and intensive psychoanalytic treatment is generally more effective than psychoanalytic psychotherapy (e.g.Heinicke, 1965; Heinicke & Ramsey-Klee, 1986; Rudolf, Manz, & Ori, 1994) - although its superiority is sometimes only apparent on long-term follow-up (Sandell et al., 1997). Psychoanalysis can lead to a reduction in health care related use and expenditure (Dührssen, 1962; Keller et al., 1998), which is maintained for some further years (Breyer, Heinzel, & Klein, 1997). Long-term psychoanalytic therapy can reduce symptomatology in severe personality disorder such as BPD (Bateman & Fonagy, in press; Monsen, Odland, Faugli, Daae, & Eilersten, 1995a; Monsen, Odland, Faugli, Daae, & Eilersten, 1995b) and these improvements are largely maintained (Stevenson & Meares, 1992; Stevenson & Meares, 1995). Psychoanalysis may also be an effective treatment for severe psychosomatic disorders (Fonagy & Moran, 1991; Moran, Fonagy, Kurtz, Bolton, & Brook, 1991; Moran & Fonagy, 1987; Rudolf, 1991).

Process aims

After this little excursion into science, let us return to the real world. Having said that it is inappropriate to dismiss therapeutic improvement as irrelevant, particularly as most patients come with precisely this concern, it should also be acknowledged that Mr. A’s analysis, despite considerable symptomatic improvement, categorically failed to achieve any of the ambitious goals which psychoanalytic authors have specified. These are normally termed "process aims" and are to be distinguished from "outcome aims" (Kennedy & Moran, 1991). A thought experiment which contrasts Mr. A’s analytic achievement with desirable process goals is sobering to say the least.

Hans Loewald would have been disappointed that Mr. A’s ego development had not been "set in motion" (Loewald, 1960, p.224), as would Peter Gray (1990) since Mr. A achieved, as far as one could tell, "no new conscious solutions to ego conflict". Charles Brenner (1976) might have found it challenging to identify in which, if any, of the compromise formations which characterised Mr. A’s psychic world, he had found more effective and adaptive solutions. Equally, Donald Winnicott might have been hard pressed to find in Mr. A’s analytic material any sign of growth towards the true self, any increased capacity for play or any increase in his sense of continuity of being (Winnicott, 1962). Authors with a Kleinian orientation would have been particularly disappointed by Mr. A. The truth about himself continued to elude him (Grinberg, 1980), and he did not reintegrate aspects of himself previously lost through projective identification (Steiner, 1989). But Steven Mitchell (1997) would have even less to celebrate as quite clearly all my attempts to generate meaning within a dialectic matrix of interpersonal tension had failed. Perhaps most disappointed with my work would have been Fred Busch and Christopher Bollas, as Mr. A’s capacity to engage in self-analysis remained frankly appalling (Busch, 1995) and his freedom of thought was as restricted as it had always been (Bollas, 1992). What Mr. A so clearly illustrated to me, at an early stage of my psychoanalytic career, was that process and outcome aims of psychoanalytic therapy were at best loosely coupled and at worst appeared to be unrelated to one another.

Alternative accounts

This is of course not news. Had Mr. A been a different kind of patient, I might have been persuaded that his improvement was the product of his unconscious need to comply with my unconscious need for him to improve. The notion of an as-if analysis hardly applies to a therapeutic situation when analysis is hard to identify in the first place. I have similar concerns about the so called "transference cure", which Glover (1931) attributed to suggestion. Therapies which "are all backed by strong transference authority, which means that by sharing the guilt with the suggestionist and by borrowing strength from the suggestionist's super-ego, a new substitution product is accepted by the patient's ego. The new 'therapeutic symptom construction' has become, for the time, ego-syntonic" (1931, p. 406).

Suggestion was of course Freud’s ‘bete noir’ (e.g. Freud, 1921). But the best example of transference cure that I know is provided indirectly by the discoverer of psychoanalysis himself. Sterba (1951) fascinated by Bruno Walter’s (1946) autobiographical account of his treatment by Freud went to interview the world famous conductor after the war. Walter consulted Freud about what appeared to be a hysterical paralysis of his arms that prevented him from conducting or playing the piano and did not respond to physical treatments. Walter anticipated an extensive period of "soul-searching", particularly with regard to his infantile sexual fantasies. Instead, Freud recommended that he should visit Sicily, which was "more Hellenic than Greece" (p.164) and should use his eyes rather than his hand. On his return, reporting to Freud that he felt no better, Freud now recommended what most psychologists would recognise as ‘exposure with response prevention’. Basically he told Walter to conduct a couple of operas and observe that his disabilty did not impair the music. Walter went ahead inspired by the strength of the transference, conducted with either one of his hands and sometimes just his head, and eventually went on to become one of the best-loved conductors of the century. Walter (1946) writes: "I endeavored to adapt my conducting technique to the weakness of my arm without impairing the musical effect. So, by dint of much effort and confidence, by learning and forgetting, I finally succeeded in finding my way back to my profession" (pp. 167-8).

Whatever we make of Freud’s approach to this case, it suggestion is clearly a qualitatively different form of psychic change from that which we see in somebody who is ostensibly in analysis, but positively resists the analytic process - and yet shows marked improvement. Nevertheless, one must admire the clinical flexibility of the inventor of psychoanalysis who, at least in some of his treatments, was far more ready to "throw away the book" (to use a phrase of Leon Hoffman, 1994) than most of his followers could allow themselves.

Some plausible answers for Mr. A’s improvement lie outside of psychoanalysis. Alterations in life circumstances must account for many of the changes we observe during analytic treatment. There may have been important biological changes (no one monitored Mr. A’s serotonin during this period). Also, some disorders simply disappear with time. Then there are the so-called "non-specifics" of therapy (‘hope’ Frank, 1988; ‘empathy, warmth and genuineness’ Rogers, 1951; ‘therapeutic alliance’ Safran, Crocker, McMain, & Murray, 1990).

But in all these cases we should insist on a parallel intra-psychic account. If our theory of pathology is to be considered truly comprehensive, improvements which occur independently of the psychoanalytic experience should be accommodated as readily as those changes which we believe ourselves to be instigators of. . I believe the concept of intersubjective representational shifts accounts for some such cases of ‘unexplained improvement’.

Intersubjective representational shifts

So what can we learn from Mr. A’s successful analytic failure?The lack of a stable sense of self is a central difficulty for narcissistic patients, particularly at the borderline end of this spectrum. We have written elsewhere about the role of reflective function in self organisation (Fonagy & Target, 1997). Reflective function, we believe is the psychological process that maintains our "intentional stance" – Daniel Dennet’s (1978) phrase for the interpretation of behaviour in terms of underlying mental states (beliefs, desires, wishes, feelings, thoughts etc). The symbolic representation of mental states may be seen as a prerequisite for a sense of identity. These representations form the core of a sense of psychological self. As Marcia Cavell (1994) has thoughtfully demonstrated, self-knowing self states do not arise intrinsically from within the mind (see Figure 1). They are internalised from the other who knows and mirrors the self. At the core of all our selves, then, is an image our object created of us as intentional beings.

In Mr. A’s case, parental self-absorption probably precluded the development of an authentic organic self image, built around internalised representations of self-states. His mother was a most peculiar woman. She was the only parent of any adult patient I ever treated who attempted to come to her son’s sessions. One day Mrs A just turned up at reception. The receptionist, a most charming and friendly woman who knew all my patients and who was well-used to most of the peculiarities of a psychoanalytic practice, phoned my office in a confused state, hardly able to control her panic mixed with irritation, saying: "Oh, Peter, I don’t know what to do! A Mrs A is here, determined to see you. She says you have her child in treatment. She can’t mean Mr. A, can she?" Mrs A had arrived, determined to see me because I had her child in treatment. Mr. A was 34. My curiosity drove me downstairs to meet this large, bejeweled woman who, when denied access to my consulting room, insisted on telling me in the hallway about her failed third marriage, which in her view was undoubtedly upsetting her son. It took every ounce of my over-long analytic training to maintain what passed for neutrality and yet assert the privacy of Mr. A’s analytic treatment.

If, as indeed I suspect was the case for Mr. A, the object is unable or unwilling to reflect the child’s internal state, or projects her own internal state onto the child, intentional states will not be symbolically bound and the developmental basis of the self structure will be absent (see Figure 2). The weakness of such a self-image leaves the child with affect which remains unlabelled and confusing -- presumably what Bion considered "unmetabolised" or uncontained (Bion, 1962). The building blocks of reflective function, mentalised self-states, are absent. The turmoil that results will make the child even more desperate to seek closeness, to find some organising structure for his affect, in whatever form this might be available. The child will then be willing to take in reflections from the object which do not map onto anything within his own experience. This will lead to the internalisation of representations of the parent’s state, rather than of a usable version of the child’s own experience (see Figure 3). This creates an alien experience within the self, based on representations of the other within the self. There may be similarities here to the "alien objects" that Ron Britton (1998) described, on the basis of clinical work with patients such as these.

Once internalised, the alien presence interferes with the relationship between thought and identity: ideas or feelings are experienced as one’s own which do not seem to belong to the self. The alien self destroys the coherence of self or identity, which the intentional stance demands. The experience of consistency can only be restored by constant and intense projection (see Figure 4). Understanding this process is vital clinically, because – in contrast to the neurotic case – the projection is not motivated by superego pressures, but by the need to reestablish the continuity of self experience.

Let me give a brief illustration from Mr. A’s analysis. About 18 months into his treatment, there was a brief period when at times I thought I was getting through to him. For example, there was a dream where he felt lost like a child running from empty room to empty room in an enormous confusing palace. I used the dream to help him understand his terror of helplessness at feeling lost in his rigidly compartmentalised mind. He felt pressured by the analysis to find feelings and thoughts – yet all he experienced was emptiness.

Needless to say, such an interpretation could not stick to ‘Teflon man’ for long. He started the next session by announcing that he thought I probably had a very small penis. By this time, I was quite used to what I privately thought of as his "Blitzkrieg" of my self esteem, so I did not even inquire how he thought he knew. He, however, enthusiastically pursued the topic, announcing confidently that his observation of the ratio of the size of my hands to the rest of my body clearly revealed that I had a good reason for feeling inferior to other men. Further, in his view, my becoming an analyst was an attempt to protect myself from the terror that genuine competition with other men might face me with.

Here I am only giving you a taste of his audacity. In reality, his assault on my masculinity and self-esteem lasted half the session, and was, as usual, astute as well as cruel. Eventually, something inside me snapped and I found myself saying: "You know, although you have never talked about this, I guess you must sometimes feel extremely uncertain of your maleness. Perhaps it is you who are worried about the size of your penis". The atmosphere in the room became electric and I shamefully remember thinking triumphantly: – "I got you Teflon man!" After a brief silence, Mr. A replied, more sad than anxious or angry: "So, that is your best shot?" and then added: "Do you think you would be good enough to give me the name of an analyst who can help me?" In fact, in my mind, I had had the list ready for some months, but I responded continuing the persecution: "You are so keen to win a competition between us, that you seem willing to totally sacrifice our joint work". The phrase "joint work", used quite unconsciously by me, got to him and the familiar track of my thinly disguised attempts at humiliating him by "fake therapy lingo" was played one more time.

Interestingly, while I felt quite depleted by this last encounter, Mr. A returned for the next session feeling better than ever. Clearly, I had simply become a vehicle for the alien part of Mr. A’s self - to use Kleinian jargon, the split-off and disowned parts of himself. While I fulfilled this function he felt calm and experienced a welcome degree of coherence, almost a sense of identity. When he was able to push our relationship close to the brink, provoking me to become what seemed to both of us to be self-centered and conceited, he felt reassured in seeing this noxious alien self outside and, with the intrapsychic deportation complete, could pursue his own fragile but benevolent agenda outside the analysis.

Of course, as always, there was a price to pay: he was stuck with me. Outside of his analysis, Mr. A could find no-one who would allow themselves to be used in this way. His obnoxious manipulativeness had repeatedly disrupted his relationships. It made him dependent on the analysis – not attached, he cared little about me at this stage – but reliant on someone being available for manipulation and distortion. His unconscious fantasy was that such projected parts of him should be left inside the object, and he was ready to leave me as soon as he was persuaded that the projection had been successful. This was the moment where analysis was possible but also the moment to move to the next analyst. If and when I tried to fight the projection, his depression and anxiety would worsen. He would close up and once again become the Teflon man.

People like Mr. A are only reachable, I believe, at the moment when their externalisation of the alien other is felt to be complete. The trouble is, as in the session above, that these are not the moments when we are likely to function effectively as analysts. Anthony Bateman (1996) described the same phenomenon sensitively in the context of Rosenfeld’s (1964) distinction between thick-skinned and thin-skinned narcissists. Extending Rosenfeld’s description, Bateman argued that narcissists alternate between the thick-skinned (derogatory and grandiose) and thin-skinned (vulnerable and self-loathing) states. Bateman shows that interpretive work can only be done with such patients at the dangerous moments when they are in movement from one state to the other. I think what Bateman describes is the successful externalisation of the alien self (derogatory or self hating). The externalisation of the persecutory object within the self leaves the patient able to listen and even experience concern. The analyst needs to be in a dual functioning mode, to infer and create a coherent representation of the patient’s true self, separate from but concurrently with any countertransference enactment (see Figure 5). However, this is where therapy often fails, because as soon as the patient hears anything other than what he projected onto his analytic object, he must be once again on the alert. He might be risking the return of his laboriously ejected introject.

To summarize: Intersubjective representational shifts may appear to be therapeutic and are associated with symptomatic relief. Their therapeutic significance is not in symptom relief but in making the patient’s true self momentarily accessible to therapeutic work. These moments are brief and their importance should not be exaggerated until progress has been made in the second of the domains for us to consider this afternoon.

Changes in mental processes and the recovery of reflective function

Strangely, whilst psychoanalysts have long recognised that all mind is representation, they have been curiously uninterested in the mechanisms which generate and organise these: mental processes. Mental representations are the products of mental processes. So, a mental process is the violin from which the melody of mental representation originates. The distinction is well established both in philosophy of mind and cognitive science (Bolton & Hill, 1996; Mandler, 1985).

The notion that mental processes are as vulnerable to the vicissitudes of conflict as mental representation is implicit in many psychoanalytic writings. Hanna Segal (1978), for example, illustrated how thinking puts a limit on the omnipotence of fantasy and can therefore be despised and fervently resisted. The Ecole Psicosomatique of Paris described patients who inhibit mental processes critical to the generation and recognition of affect who are believed to commonly present with somatic disorders and whose experience of life appears concrete, devoid of sentiment or affect and insist on seeing things "as they are" (De M'Uzan, 1974; Marty, 1990; McDougall, 1974; 1986). Their history tends to be one of having been overwhelmed by affect at a time when their capacity to deliberately exclude affect-laden representation is not yet available. Prior to the development of representational capacities which can selectively exclude ideas associated with unmanageable feelings, these individuals may have coped by disengaging or inhibiting mental processes critical to the generation and recognition of affect. Others have described this clinical picture as alexithymia (Sifneos, 1977).

Mary Target and I have argued that patients with severe personality disorder inhibit one particular aspect of the normal development of mental processes - their reflective function - and therefore have little reliable access to an accurate picture of their own mental experience, their representational world. They are unable to take a ‘step back’, and respond flexibly and adaptively to the symbolic, meaningful qualities of other people’s behaviour. Instead, they find themselves caught in fixed patterns of attribution, rigid stereotypes of response, non-symbolic, instrumental use of affect – mental patterns that are not amenable to either reflection or modulation. They inhibit their capacity to think in terms of thoughts and feelings in themselves and in others, prototypically as an adaptation to severe or chronic maltreatment.

Mr. A, as a vulnerable child, confronted with an undoubtedly intrusive and probably grossly self-preoccupied maternal object, could not develop a coherent image of his own mental state which could have served as the basis for an understanding of others. Although apparently somewhat reflective, his rigidity betrayed his incapacity truly to reason with mental states. One of the most striking aspects of Mr. A’s treatment was the quality of rigidity that imbued his representational world. His tendency to hold on to a specific point of view went far beyond that which might be associated with habitual patterns of defence. Like other patients, Mr. A organised the analytic relationship to conform to his unconscious expectations. But, for Mr. A these expectations were experienced with the full force of reality and alternative ways of viewing things were either dismissed out of hand, or entered into in superficial and meaningless ways. Equally striking was that lack of consistency between representations apparently caused little distress. The object, cherished as the source of salvation one day, could become the source of damnation the next. And thus his entire representational system seemed both in constant flux, and immutable to lasting change.

Ron Britton’s distinction between ‘knowledge’ and ‘belief’ is helpful here (Britton, 1995). Mr A did not believe in his superiority – he knew it. Belief entails uncertainty, and the knowledge that a mental state is just a way of constructing experience, not reality itself. He lived in a world of psychic equivalence (Fonagy, 1995) (Fonagy & Target, 1996; Target & Fonagy, 1996) where he believed that all he thought of did actually exist in the physical world. And all that existed in the physical world, Mr. A had certain knowledge of.

What brings about change in reflective function?

Mr. A did eventually show change in terms of process aims as well as outcome aims. Change was slow, and marked clearly by a worsening of his depression, as he became increasingly aware of undesirable aspects of his personality. There was no specific point I could mark with the prefix of "turning". (I always thought that Nina Coltart’s (1986) phrase "slouching towards Bethlehem" was more evocative of the undramatic drama involved). No clever interpretations were made that I could single out as delivering the change. Yet change there was, perhaps most noticeable in the change of atmosphere of the sessions . After three years of endurance I learned to talk with ‘Teflon man’. Just a brief illustration.

First thing in the morning, Mr. A brought his bombastic self into the session. "We have taken over this company that buys businesses in trouble, finds anything that’s worth having and liquidates the rest. All I need now is a recession, and the government is working on that for me." "I guess it is quite nice to feel that a whole lot of us are working for you, even if we are not aware of it." Mr. A laughed but stayed silent. I said: "I bet you are asset stripping the analysis as we speak." He said: "How did you know? I was just thinking that probably the only thing worth having round here is your couch. But even that has a slight tear in it." I said: "Maybe it can’t take the wear and tear." He became gloomy: "Maybe there is nothing worth having here." I said: "You know, perhaps you want to take over to protect rather than to destroy." He said: "Are you losing you grip? You never flatter me!" I said: "I am not flattering you now. I think you are terrified about what will happen to me. If you wear me thin, who will protect you from the predatory part of you that is only too ready to sell you short?" He said, as if that was what we had been talking about all along: "You looked really pale and tired to me this morning. Are you sure you are up to this? Maybe the guy before me gave you a tough time." Pointing out the obvious, I stated: "You were the last patient yesterday evening and the first one this morning. I wonder who we are talking about here?" He said: "Oops. I walked into that one, didn’t I?"

The ongoing discussions on therapeutic action have generally acknowledged that the psychic change in individuals who manifest developmental deviations, impairments, deficits or underlying structural deficiencies fit poorly with the neurotic model of therapeutic action. From the late 70s a number of writers have shifted the emphasis from ‘structural change’ as the focus of therapeutic action to the transaction between patient and analyst as a curative experience, and the early mother-child relationship as the most appropriate analogue for the therapeutic encounter. Developmental processes are invoked by those who link therapeutic action to the holding environment (Modell, 1976), separation-individuation (Stolorow & Lachmann, 1978) (Blatt & Behrends, 1987), a sense of union with the primary object (Loewald, 1979), social referencing (Viederman, 1991), empathy (Emde, 1990), or other aspects of developmental processes (Goodman, 1977; Schlessinger & Robbins, 1983).

These developmental models cannot give a satisfactory account of therapeutic action with the difficult patient. Sam Abrams (1990) makes the obvious point that the developmental sequences in psychoanalysis and the biological constraints imposed on normal development make for discrepancies which should invalidate the developmental metaphor. Christopher Bollas (1987) noted that analysts were in no sense parents of the analysand, rather they possessed the "generative paradigmatic skills that reach the child element in the adult analysand" (p. 115). Linda Mayes and Donald Spence (Mayes & Spence, 1994) make the important but counter-intuitive observation that the developmental metaphor applies more to relatively well-endowed adults who historically probably had the benefit of the kind of care-giving experiences that re-emerge in the transference, whilst the group of patients with whom these metaphors are most often used simply do not have the capacities which might make the developmental metaphor applicable.

So what was it that caused a shift in Mr. A’s psychic reality? I do not believe there is a single answer to this question. I believe that reflective processes are enhanced by work in the transference, which highlights differences in perspective between self and other. The transference also focuses the patient on the analyst’s mental state as he tries to conceive of the patient’s beliefs and desires. The repeated experience of finding himself in the mind of his therapist, not only enhances self representation but also removes the patient’s fear of looking. Believing, as opposed to knowing, allows uncertainty rather than reliance on rigidity, inaccuracy and ultimately vulnerability, from which grandiosity can only afford momentary protection.

In what I believe to be an appropriate use of the developmental metaphor, the analyst performs the function of the object who enters into the child’s pretend play, creating a transitional sphere of relatedness. Here thoughts and emotions may be played with while also experienced as real; the phobic avoidance of mentalisation, gradually and tentatively, gives way to reflective function. Humour is often a critical, and under-rated, component. Long before Mr. A could accept interpretations, he was able to laugh and to make me laugh. More generally, a degree of flexibility and mutual adaptation is required between patient and analyst. In the treatment of these patients, as Hoffman (1994) pointed out, there is a dialectic between being ready to "throw away the book" and "keeping it". The principle aim of the process must be to make the world of feelings and ideas safe for the patient once more.

Optimally, the patient’s mental work recapitulates that of the analyst. The analyst’s thinking, even if initially neither understood nor appreciated by the patient, continually challenges the patient’s mind stimulating a need to conceive of ideas in new ways. What is crucial, then, is the active engagement of one mind with another, inconceivable without empathy, holding and containment. Yet none of these is directly responsible for the therapeutic action.

Representational change

It is only when mental processes have been to some measure freed that change in representational structure is possible. Technically, this is a key stumbling block for beginning clinicians who understandably aim to modify specific representations as soon as their insights permit. I can do no better than offer a cautionary tale from my own practice. I vividly remember my first analytic experience with a borderline patient, Mr. B. Early in his analysis, following a lengthy discussion of his anxieties concerning competitiveness, I ventured to point out that these may be related to unresolved conflicts about his sexual competition with his father as a little boy. Mr. B seemed thoughtful about my interpretation and returned proudly the next day with an account of a dream where he and his father were fighting. He had a knife and after a struggle managed to cut his father’s penis off which he held up victoriously, reminding himself of the Statue of Liberty. By then I had the presence of mind to make the more appropriate interpretation that his anxiety the day before concerned his feeling of being in competition with me and now, having witnessed my inadequacy he could, indeed afford to feel triumphant. While clearly reducing his anxiety momentarily, these and other interpretations had little impact on his ways of seeing things.

Mr. A’s analysis offers clear examples of representational change. As the analysis progressed, his picture of himself in dreams gradually shifted from the empty palace of the beginning of the analysis, to a frightened mouse in its third year and two years later, at the end of the analysis, to a man with a physical handicap. Perhaps most striking was the way his images of past figures were revised. He had almost never discussed his parents in his analysis without denigration in the first year, but in the second year "discovered" his father as someone he admired for his tenacity and fair-mindedness. A frequently recalled episode where he had felt humiliated and ridiculed by a grandfather who had played a trick on him, gradually took on a far more benign colouring. He now experienced this episode as illustrating the man’s wish to surprise and tease him rather than mock and humiliate.

We could summarise such representational changes under three headings: (1). Self, object and relationship representations acquired enhanced integrity and coherence. (2). Person representations were increasingly seen in relationship to one another rather than as isolated figures. Ultimately Mr. A could talk about father and mother as having a marriage, where each parent had their own separate mind, with feelings and ideas. (3). Finally, totally new object representations emerged. In general, we may say that Mr. A’s representational system was restructured so that previously isolated, incompatible and unelaborated representations of the mental states self and object ceased to be pathogens.

By creating and elaborating a mental world for his internal objects these could change from part objects to whole structures. Understanding the other in mental state terms requires integrating assumed intentions in a coherent manner. The initial solution for the child, given the imperative to arrive at coherent object representations, is to split the representation of the other into several coherent subsets of intentions (Gergely, 1997), primarily an idealised and a persecutory identity. Splitting enables the individual to create mentalised images of others but these are inaccurate, over-simplified and allow for only an illusion of mentalised inter-personal interchange. Further development of reflective function normally leads to an integration of these partial representations. The hopelessness of this task in the face of a deficit in reflective capacity may be seen as the direct cause of the permanent fragmentation of the internal world of such patients. The recovery of reflective function, and its active use within the analysis, enabled Mr A to integrate his representations, so that his objects could have conflicted motives and yet retain their coherence (see Figure 6).

What brings about representational change?

What aspect of psychoanalysis brings about these changes? It is widely recognised that insight is not enough and that something akin to Loewald’s (1960) real object is necessary to explain change.

The dichotomy between interpretive-insight on the one hand and relationship generated change, however, is probably false. It is based on a confusion of means and ends. Verbal communications by the analyst may change representational structures in the patient’s mind and we may, if we so desire, talk of such changes insight as long as the term is restricted to the mental states that are experienced as motivating behaviour in self and other. However, as my colleagues and I took great pains to point out, in a little-cited and no doubt even less read paper some years ago (Fonagy, Moran, Edgcumbe, Kennedy, & Target, 1993), relationships as much as interpretations readily lead to such changes in representational structures. As we observe the behaviour of another we automatically interpret, infer mental states and restructure and potentially enrich our representations of how we see self and other interacting. The perception of the analyst as having empathy (Emde, 1990) or healing intentions (Stone, 1961) may bring about changes in object representations through the same mechanism of change as interpretations. Thus there is no qualitative difference between the means by which therapeutic change is achieved via interpretation and via a new relationship.

Relational models of psychic change, on their own, lack specificity. For example, Wallerstein (1986) in his detailed study of the 42 Menninger patients, noted a number of cases where change was primarily attributable to the relationship with the therapist. In this category he placed side-by-side the individuals whose symptomatic improvement depended on the ongoing availability of the therapist alongside others who transferred a positive dependency from a therapeutic relationship to another person. The fromer, I would call shifts in representation, as the earliest Mr. A above. In the latter, we may be looking at representational change, a changed set of expectations concerning the reaction of attachment figures. But when we ask about either intrapsychic or interpersonal processes that might determine the direction a particular patient might follow, Wallerstein’s observations yield few clues.

So my brief answer to the question, is it the relationship or is it the interpretation, would be: it is both.

What changes in representational change?

So, what changes in representational change? Schematic models of therapeutic change (1980; Abrams, 1987; 1988; Joffe & Sandler, 1969; Sandler & Joffe, 1966; Sandler & Joffe, 1967; Sandler & Joffe, 1969) attribute therapeutic action to changes in long-term memory which underpins the representational system. However, it seems fair to say that while the aims of psychoanalysis (Sandler & Dreher, 1997) have been greatly elaborated since Freud’s original model of undoing repression and recovering memory into consciousness (Freud & Breuer, 1895), these advances have not brought with them an updating of the role of memory in the therapeutic process.

Memory, we now know, is not a singular mechanism but is made up of a number of different systems. Cognitive science makes a key distinction between the two kinds of memory system both which have important functions in psychoanalytic treatment. Of particular relevance is the distinction between a declarative or explicit memory which is involved with the conscious retrieval of information about the past, and the so-called procedural or implicit memory system from which information may be retrieved without the experience of remembering. The distinction was originally made by Cohen (Cohen, 1984; Cohen & Squire, 1980) to distinguish knowledge characterized by content from the kind of content-independent information which is involved in acquiring skills such as playing the piano (regardless of the specific piece) or driving (independent of destination). Declarative memory relates to remembering events and information. The aspect of this system that we, as analysts, are most concerned with is autobiographical memory: high in self reference and frequently accompanied by personal interpretation (Conway, 1996).

Robert Clyman (1991) was perhaps the first to suggest explicitly adopting the distinction between procedural and declarative memory for use within psychoanalytic models of normal and pathological development. He proposed that the procedural rather than the declarative memory system may be involved in the transmission of early experience into adult personality. Even earlier, Pat Crittenden (1990) suggested that Bowlby’s notion of internal working models may be best understood in terms of procedural knowledge. More recently, a group of psychoanalysts working in Boston including Daniel Stern, Ed Tronick, Karlen Lyons-Ruth, Alexander Morgan and Alexie Harrison have been working intensively to integrate the concept of implicit memory with ideas about the therapeutic process.

In agreement with these authors, I would like to propose that experiences contributing to internal representations of object relationships are not, by and large, stored in declarative memory. The extent to which episodes of interaction with the caregiver may be remembered (encoded and stored in autobiographical memory) may be incidental in the development of internal representations of relationships. What lies at the root of interpersonal problems, the transference relationship and quite possibly all aspects of the personality which we loosely denote with the term unconscious, is a set of procedures or implicit memories of interactional experience. These may be represented as self-other-affect triads, as Otto Kernberg (1988) has suggested, as a network of unconscious expectations, as Bowlby (1988) conceived, or they may simply be emergent properties of the nervous system, abstracting invariant information through the tendency of nerve-cells to survive together if activated together (Edelman, 1987; Stern, 1994).

These are "generic memories" or "scripts" (Nelson, 1993a) where the child’s predominant experience of a relationship is retained. Individual experiences which have contributed to this model may or may not be "stored" elsewhere, but in either case the model is now "autonomous", no longer dependent on the experiences which have contributed to it. The models exist non-consciously as procedures which organize interpersonal behaviour but are not consciously accessible to the individual unless attention is specifically directed to them. The models are not replicas of actual experience but are undoubtedly defensively distorted by wishes and fantasies current at the time of the experience. Thus in no sense can they be thought of as bearing testament to historical truth. Many of these procedures may be highly dysfunctional. For example, infantile grandiosity defensively activated in the face of maltreatment may generate an image of the self as responsible for neglectful or cruel behaviour. Although an illusion of control and predictability may be acquired in this way, the self will also be seen as guilty and deserving of punishment. We then commonly encounter this pattern of role relationships in the transference, the patient’s behaviour may be provocative, based on the powerful expectation that a punitive reaction is to be expected. The person will be unaware of the operation of this implicit model unless attention is specifically directed to it by appropriate interpretation.

So where is therapeutic action -- in implicit or explicit memory? Memories of past experience can no longer be considered relevant to therapeutic action. Psychic change occurs as a function of a shift of emphasis between different mental models of relationship. Change occurs in implicit memory leading to a change of the procedures the person uses in living with himself and with others. With reference to the recent controversy on the recovered memories of trauma (Brenneis, 1997), it seems that memories may or may not be forgotten. They may or may not come to the surface as part of a thorough scrutiny of patterns of internal and external relationships in a psychoanalysis. And the analysis may or may not be successful regardless of the recovery of traumatic personal experience, true or false. The emergence of the memory is part of a separate process. It is not the undoing of repression but a process of active construction, the creation of an experience, true or false, congruent with the pattern of self-other relationships close to awareness.

In psychoanalytic psychotherapy therapeutic action is predominantly in the implicit memory system rather than relatively superficial changes in autobiographical memory. The term superficial is used advisedly here, since inherently experiences constructed in autobiographical memory are more likely to be inaccurate than accurate. Undoubtedly, recovered memories will contain the essence of a wide range of events, distilled into a configuration represented within that mental model. In that sense the memory is true. It is true to the model of a particular experience-of-being-with. However such a model may be profoundly distorted by fantasies and other intrapsychic experiences. Truth in psychotherapy makes sense only in the context of psychic reality.

In reality, the analysand works backwards and pulls together elements of early experience consistent with a freshly discovered perception of himself in relation to the other. It is not surprising then that memories from adolescence and latency will dominate a patient’s material despite our conviction that earlier experiences were the formative ones. The mental model uncovered by the patient in analysis is likely to have been generated by early experiences which antedate the development of autobiographical memory and therefore will never be retrieved.

There is good reason to believe that psychoanalysis works by modifying procedures rather than by creating new ideas. The latter is suggestion which probably cannot by itself sustain change. Of course it should be admitted that just as implicit memory may activate specific elements within the autobiographical or episodic memory system. So might newly-found awareness of particular experience influence the way an individual experiences themselves in relation to others. This process, however, is unpredictable and always vulnerable to the "recovery" of yet further memories. Psychoanalytic attention is appropriately generally focused on the emotion generated in the context of particular self-other relationships. Given the neurophysiological evidence that the emotional charge associated with experiences stored sub-cortically cannot change without cortical involvement, bringing such implicit structures into conscious focus seems not only desirable but arguably critical and a key component of therapeutic action.

To summarise: Based on the psychological and neurophysiological distinction between the declarative (explicit) and procedural (implicit) memory systems, in agreement with numerous other writers, we have proposed that representations of self-object relationships that regulate interpersonal behaviour are encoded as procedures in implicit memory. Representational change is a function of a shift of emphasis between different models of relationship, as these become manifest in the transference, and are sometimes radically modified as the intentionality of the object representations becomes elaborated. Change occurs in implicit memory leading to a change of the procedure the patient uses in living with himself and others. The old-fashioned process aim, the recovery of autobiographical memories, is a consequence rather than the driving force of these changes. The recovery of memories is less to do with the undoing of repression than with a process of active construction, congruent with the pattern of self-other relationship that has come into therapeutic focus. Pathogenic mental models of relationships uncovered in analysis are likely to have been generated by early experiences which antedate the development of autobiographical memory and therefore will never be retrieved. Autobiographical memory may create a matrix of meanings within which such a change in relationship models can take place but it cannot and does not constitute this change.

A final example: change in procedural memory

Mr. A had a very irritating habit, which undoubtedly contributed to my image of him as Teflon man. Even when he came to be able to accept my interpretations, he seemed to be able to simultaneously negate them by shifting the focus to another person or into the past. So, in the end, although he seemed committed to the treatment, I would feel that my actions were meaningless.

In one session, he was complaining about various things. He had argued with his secretary and then his business partner. He felt that his assistant, Neil, did not support him and the whole company was falling apart. When I tried to say something he indicated that I was unlikely to have anything worth listening to, as I clearly knew nothing about business. Through his irritation he managed to get across, however, that he was indeed concerned to hear something from me. He kept talking about his assistant, Neil, who would not or could not help resolve his problems. Eventually, I managed to say that it seemed hard for us to meet up, that he was interrupting me when I tried to say something because he wanted to discard anything helpful that I might say. But this left him feeling that I was totally unsupportive and unhelpful. I also managed to slip in that I thought he was afraid of something very dramatic happening, like a complete collapse, which on the surface he feared in relation to his company. Perhaps he did not know whether my ‘helping’ would make matters worse or better.

He relaxed and seemed more thoughtful but appeared to interpret my comments totally as if I were talking about his assistant. He said how much he was worried about him because he seemed to be under so much pressure and that perhaps I was right (sic) and Neil was close to a breakdown. From thereon in, whatever I said in that session, whether further interpretations or clarifications, were immediately applied to Neil. Repeatedly, I tried to show him what he was doing, but each time the ball would end up in Neil’s court. Eventually, I lost patience and pointed out quite directly what was happening - that he was repeatedly severing the connection between us and not allowing himself to feel that I was communicating to him. He said he felt that I was accusing him of being mad and broke off the conversation. At other times I could describe, he would use my interpretations but displace them to figures in his past, or to someone he knew.

It took some time for me to be able to make him aware of this strategy he had. But when I did, he could see that it was quite pervasive. He became aware of how he was doing this with his mother, and we finally understood it as a way he had of coping with anyone who he felt was intrusive. Interestingly, as we pursued this line, memories of his mother emerged as someone whose curiosity about him could not be contained. She would burst into his bedroom at night without knocking, hoping (he thought) to find him reading dirty magazines. When she asked about what had happened to him at school, he would inevitably tell her a story about what happened to another boy.

The episodic memories helped to give meaning to his way of behaving but I do not believe them to have been its cause. I think much earlier, perhaps in infancy, he had learned to offer a false, self-created structure rather than the true one, to protect it from invasion. These episodes were no longer accessible but this did not hinder his capacity to change.

Bringing it all together

We have talked of three modes of psychic change which tend to occur in psychoanalytic treatment. They may all be differentiated at at least four levels: (1). In terms of their description; (2) In terms of their functions (i.e. the processes they entail); (3) In terms of their impact on symptomatology and (4) In terms of the techniques which are most pertinent to that phase of treatment.

Initially in the treatment of personality disordered individuals we are most likely to encounter representational shifts (see Figure 7). Descriptively, these are defensive strategies which enable the patient to establish coherent self states by externalising alien parts of the self onto the therapist. At the outcome level, symptom reduction is likely to result. From the point of view of process aims, the therapist must permit such externalisations in order for the therapy to be tolerable to the patient and seek moments of movement, which may be short windows of time, when the patient is able to hear yet is still convinced of the success of the externalisation. From a technical point of view this is the hardest phase, where many treatments fail. A simple verbalisation of the process will not suffice. The analyst, his mind colonised by alien ideas, must find sufficient reflectiveness to perceive patients as they are, having distanced himself from an alien self-construction. Perhaps human generosity is the most pertinent therapeutic capacity at this phase.

The second, concurrent, phase entails the disinhibition of mental processes, chiefly but not exclusively, reflective capacity - mental processes that have been defensively inhibited, as part of an attempt at adaptation to sub-optimal internal and external environments (see Figure 8). The process aim here is a move away from the duality of psychic equivalence and pretend modes of functioning, and engagement of the patient’s mind in forms of mental activity which have felt dangerous in the past. A complicating aspect of this phase is the almost inevitable worsening of the patient’s symptoms. The re-vitalisation of mental processes inevitably induces a heightening of conflicts, which bring with them regression and compromise formation. At the technical level, a focus on mental states in the context of an attachment relationship is vital, as are consistency and coherence of approach. These patients experience an intense hunger for understanding the ways that minds function; they learn this, not so much from the specific comments of their analyst, but rather through the observation of the commitment of the analyst to an understanding of the patient as a thinker of ideas and experiencer of feelings.

Finally, representational change is the reorganisation or restructuring of the representational system (see Figure 9). With neurotic patients this may be all that is required. And, as therapists tend to come to the profession with reasonably intact minds, we are often tempted to assume that the patient on our couch also requires no more. From a process point of view, representational changes are in the direction of a fuller and more elaborated representation of the mental states of internal objects and the self. Enhanced reflective capacity allows patients to integrate split-off parts of the self and create object representations with complex thoughts, mixed emotions, and differentiated desires. Symptomatic improvement should be associated with such changes. But from a technical point of view, we have argued that it is important for the analyst to be aware that the changes sought are not changes in the patient’s awareness of past events but rather changes in procedural and implicit memory. Thus recovery of past experience may be helpful but the understanding of current ways of being with the other is the key to change. For this, both self and other representation may need to alter and this can only be done effectively in the here and now.

It is traditional to end psychoanalytic papers by a dramatic qualification. And here comes mine. I have described phases of treatment but this sequence (see Figure 10), whilst theoretically coherent, may have little to do with the order in which these phases emerge in any particular treatment. We have discussed symptomatic improvement as an inevitable consequence of following these procedures. Yet process outcome research clearly demonstrates that such associations are in reality quite rare. Finally, just a reminder that this model is a simple heuristic and a personal one at that.

So what do I think is actually important about therapeutic action? Squeezed into a corner, I would say that the psychotherapist’s mentalistic elaborative stance helps the patient to find himself in the therapist’s mind, and to integrate this image as part of his sense of self. In this process there will be a gradual transformation of a non-reflective mode of experiencing the internal world (which forces an equation of internal and external reality) to one where the internal world is treated with more circumspection and respect, as separate and qualitatively different from physical reality. This, I believe, was Freud’s heritage to us and what we should cherish in our daily work with our patients.

Figures 1, 2, and 3

Figures 4, 5, and 6

Figures 7, 8, and 9


DSPP Home Page

Peter Fonagy, PhD, FBA
Freud Memorial Professor of Psychoanalysis, UCL
Director of Research, The Anna Freud Centre
Co-ordinating Director, Child and Family Center and Center for Outcomes Research and Effectiveness, Menninger Foundation

Address for correspondence:

Sub-Department of Clinical Health Psychology
University College London
Gower Street
London WC1E 6BT

E-mail: p.fonagy@ucl.ac.uk


To comment on this paper, contact Peter Fonagy, PhD, FBA p.fonagy@ucl.ac.uk