Termination in Psychoanalytic Psychotherapy: An Attachment Perspective


Attachment and separation are inseparable. To become attached is to suffer the possibility of loss. In this brief, rather discursive, and personal paper I consider different patterns and styles of termination in relation to the characteristic ways in which the secure and insecurely attached make, and therefore break their relationships, and how understanding those patterns can be put to therapeutic use. The avoidantly insecure deny the affective aspects of stopping therapy, and need to be helped to see how anger, fear and sadness are denied but still influence their behaviour; the ambivalently attached cling to therapy and have to be prized away, building a sense of autonomy and confidence that, via ‘reinstatement of the lost object’, they can survive termination secure in the comfort of an inner therapist to turn to at times of need. For the disorganisedly attached, termination is a major threat, and attenuated endings are probably the best arrangement. These themes are discussed with clinical, personal and cinematic examples.


A good starting point for discussing termination is Psychoanalysis Terminable and Interminable (Freud 1937), written two years before the author’s death at the age of 82. But as Pedder (1988) points out, the English title could more accurately have been translated as Psychoanalysis Finite or Infinite. The very different linguistic harmonics of that road not taken might have steered therapists away from the abortive or guillotine-like implications of termination, and the irritable connotation of interminability, suggesting instead themes of separation, death, a timeless unconscious, and the infinity of irreversible loss.

The questions surrounding termination are fairly simple, even if the answers less so. When should one end – is it up to the analyst, the patient, or when an agreed fixed term is ‘up’? How should one end – abruptly, or with a gradual winding down of frequency of sessions? Are follow-up, and ‘top-ups’ allowable? Why should one end – what is the theoretical justification for an ending, how does one know that the job is done, and how does a decision to end emerge? In what way can one discern if an ending is good enough (analogous to a ‘good death’ in the hospice literature), premature (as in the Dora case, Freud 1905) or overdue (as with the Wolf Man, 1918)?

While the questions, theoretical and practical, surrounding termination are clear, answers are less certain. Novick (1997, p. 145) convincingly argues that, with honourable exceptions (Balint 1968) ‘neither Freud nor his followers paid much attention to termination as a phase of treatment’ and that ‘for almost 75 years psychoanalysts have been unable to conceive of the idea of a terminal phase…’.

Three possible reasons for this dearth seem relevant. First, ending therapy, as with embarking on one, is a real event, an ‘enactment’ going beyond the bounds of transference and the imagination. The departing patient is not just deconstructing a transference, she or he is disengaging from a fellow human being with whom many hours of close proximity and intimate affect-laden conversation have been passed (c.f. Rycroft 1985). Psychoanalysis struggles to theorize the real relationship – while this is home base for attachment theory (Holmes 2009). Psychoanalysis struggles to theorize the ‘real relationship’. Attachment theory assumes that the therapist and the therapeutic setting provide a real secure base, whose function is to enable the client then playfully to explore the ‘unreality’, yet validity, of her transferential and other imaginings.

Second, a confounding issue for psychoanalysts is that they themselves never fully undergo the process of disengagement which awaits the average analysand. The analyst perforce retains her fundamental belief in the potency and importance of psychoanalysis; is likely to have continuing contact with the analytic world, including her own analyst, through her professional society; and not infrequently undergoes second or even third analyses. If ending analysis is an analogue of leaving home, an analyst continues to retain a foothold in the parental mansion.

Third, the question of termination overlaps, sometimes in confusing ways, with the issue of the aims and objectives of analytic therapy and what a ‘good outcome’ might be. Removal of symptoms, diminished splitting and greater integration of the personality, strengthening of the ego, overcoming of ambivalence towards the breast, ‘genital primacy’, formed the mantra of the early literature. More recently, as character disorder rather than neurosis have come to form the bulk of analytic practice, and subtle research methods for studying outcome have become available, earlier idealised views of analytic outcome have been tempered with reality. A more nuanced view of what can and cannot be achieved in analysis is beginning to emerge, in which the prime aim of therapy is to equip patients with new interpersonal and intra-psychic skills, and to help push psychic equilibrium in a more positive direction. The analyst needs to know when ‘enough is enough’, and to guard against imposing his or her own narcissistic wishes, or colluding with those of the client, for a perfect outcome.

The bulk of this chapter is an attempt to use an Attachment perspective to develop a more comprehensive theorisation of termination. There follows a resonantly brief consideration of time-limited dynamic therapies, concluding with some broader psychoanalytically-inspired reflections on loss, absence and ending.


Ending and Attachment

I discuss the contribution of attachment to thinking about termination under four headings: theorising loss in relation to secure and insecure attachments; termination as co-construction; disillusion and dissolution of the secure base; and mentalising as a core psychotherapeutic construct and its relation to endings.


Theorising Separation and Loss

Bowlby (1988) saw the mutative potential of psychotherapy as arising out of a therapist-client relationship which assumes some of the characteristics of a Secure Base. A patient is a person in distress; distress triggers attachment behaviour – other concerns are shelved and a secure base is sought. When no such attachment figure is available, or seen to be available in the patient’s outside world, a professional who can alleviate distress through proximity, intensity and sensitivity is required. This is the Attachment formulation of the ‘real’ relationship, the therapeutic alliance.

This reality provides the figured base against which transferential distortions, misguided expectations, unconscious wishes and impulses, can be observed and made meaningful. Therapists’ consistency, regularity, and responsiveness have the potential to alleviate distress, and activate clients’ capacity to explore their feelings and their relationships, including that with the therapist. The client’s contribution to impediments with that process is ‘the transference’, the therapists’s, ‘counter-transference’ in the classical sense.

Setting up a therapeutic relationship is an inescapable ‘enactment’ on the part of therapist and client: an action that is ‘real’, observable, performed rather than merely imagined, phantasised about, or desired. The latter comes into play as the meaning of actions and their psychological reverberations become grist for exploration, but a vital pre-condition is the alleviation of attachment insecurity.

But if the reality of establishing an attachment relationship is central, so too is its ending. For Bowlby (1973), separation was the flip-side of attachment: the very purpose of attachment behaviours, on the part of both care-seeker and care-giver, is to mitigate loss. Crying, proximity-seeking, responsiveness and soothing all work to ensure that an individual when vulnerable – whether through physical immaturity, illness, or trauma – gains and maintains access to protection and succour.

When separation is irreversible – i.e. at an ending – Bowlby and his followers such as Parkes (2006) identified the now familiar constellation of reactions and feelings: denial, angry protest, searching, despair, and recovery leading to the establishment of new attachments. Subsequent research on grief and mourning – both normal and pathological (Shaver & Fraley 2008) – have in a number of ways fleshed out, and to some extent modified, Bowlby’s original formulations on separation and loss.

First, a key issue in reactions to separations is not so much the physical presence, but the continuing availability when needed of the attachment figure (Klass et al. 1996). As physical proximity, especially in older children, becomes less salient, what matters is knowing that a helper will be there when called upon. This ‘sense of availability’ can transcend the total separation implicit in a death and makes grieving bearable. Sources of comfort helping with bereavement include: thinking what the lost loved one would have done in a given situation; conferring with photographs or letters; imagining or even hallucinatorily hearing the dead one’s voice; Proustian remembrance of good times past.

Second, as might be expected, attachment styles have a significant bearing on reactions to loss. There are two main patterns of pathological mourning: denial and chronic depression of mood on the one hand; and inconsolable preoccupation with the lost loved one on the other (Parkes 2006). These map well onto the two principal patterns/styles of insecure attachment, deactivation of separation of protest and hyperactivation and inconsolability (Mikulincer & Shaver 2006). In the former there is denial that the absence of the lost one ‘matters’, while physiological and psychological explorations reveal otherwise. In the other there is a doomed and unassuagable effort to recover the lost loved one.

Third, Bowlby’s somewhat pessimistic perspective on reactions to loss have been modified in the light of the findings that under favourable conditions mourning can be negotiated successfully, and that persistent despair is relatively uncommon. The ‘transactional model of attachment’ (Sroufe 2005) suggests a dynamic interplay between attachment style and current relationships that accounts for variable outcomes in loss. A supportive context – whether this be through family, friends, belief system, social group, church, or therapist – ease the passage from grief to recovery; their absence adds to the burden of loss.

Finally, contemporary views on bereavement (Klass et al. 1996) emphasise the role of post-loss ‘continuing bonds’, as already implied. Bowlby was critical of the idea of maturation as a process of increasing distancing from the primary object, in which an atomised autonomy replaced adherence and dependency. He conceptualised instead a move from immature to ‘mature dependency’. In relation to bereavement his view was that (Bowlby 1980: 399) ‘the resolution of grief is not to sever bonds but to establish a changed bond with the dead person’.

We can now apply these ideas and findings to therapy termination as a bereavement analogue. Separation and loss are intrinsic to the process of psychotherapy, which is punctuated by repeated separations, mostly planned and expectable, but also by occasional traumatic interruptions. The former include the end of each analytic hour, weekend and holiday breaks; the latter therapist and client illness, and enactments on the part of therapist or client such as changing or forgetting sessions, double booking, muddles over times, turning up on the wrong day, etc. All of these are potential grist to the mentalising mill.

Ending therapy is a real loss: a significant segment of the client’s life is no longer there. A secure space and time where distressing events and feelings can be digested is now empty. A person who focuses her attention and sensitivity on one’s inner world is now absent. One is on one’s own with one’s story, feelings, and life-history. But, like every aspect of psychotherapy, an ending is ‘polysemic’ (Tuckett et al. 2008). Depending on mood and perspective, the meaning of an ending can be a death, a bereavement, a completion, a liberation, a funeral (with or without a tearfully convivial wake) or a joyful moment of maturation and ‘leaving home’.

Ending brings gain as well as loss: the time and money invested in therapy is now available for other projects; the client no longer feels so ‘dependent’; autonomy and maturity are reinforced; he or she feels more psychologically robust, more able to provide security for others and less in need of it oneself. Just as the bereaved are sometimes said to have ‘earned’ their widow – or widower – hood, the discharged therapy client likewise may feel she has earned her liberation from the obligations, mysteries and miseries of therapy, while still mourning its now-absent comforts and gifts. One way of seeing the point at which ending begins to enter the therapeutic frame is when for the client the balance-sheet of benefit and investment shifts away from the former towards the latter, the effort beginning to outweigh the gains.

These attachment-informed perspectives have a number of clinical implications. First, therapists should bear in mind the client’s predominant attachment style. Deactivating clients may well appear to take an ending in their stride, apparently seeing it as inevitable, natural and appropriate, presenting themselves as eager to move onto the challenges of ‘real life’, now that their symptoms have diminished and they feel stronger. Regret, doubt, anger, and disappointment may be conspicuous by their absence, gratitude superficial and conventional rather than deep-rooted. The therapist will direct the client’s attention to what is missing as manifest in dreams, failed appointments, seeking others forms of treatment, or in manic cheerfulness, fulsome gratitude, or pollyannerishness papering over grief-sprung cracks in the personality.

Clinical folklore holds that as the end of therapy approaches the client’s symptoms, even if alleviated during the course of therapy, may re-appear. This is perhaps particularly likely for hyperactivating clients who may overestimate the negative impact of ending. The therapist may be tempted into a premature proffering of extensions or suggesting an alternative therapist or therapy (such as a group), in ways that are driven by counter-transference-induced guilt rather than clinical need. Such post-therapy arrangements may well be appropriate, but should not be allowed to divert therapeutic focus from first working through the ending.

Second, the client’s social context should be taken into account when deciding on either offering time-limited therapy, or finding an appropriate moment to conclude open-ended treatment. Time limited therapy is much more likely to succeed when the client has a good social and emotional network to which they can ‘return’ once therapy is over. For more disturbed clients in long-term therapy, if treatment has not managed to facilitate the capacity to generate outside attachments, post-therapy relapse is likely. In partial contradiction of the point made above therefore, such clients may need further therapeutic arrangements such as group therapy or key-worker support, and the reality of this needs to be discussed as a period of intensive individual analytic therapy draws to its conclusion.

Third, the therapist needs to consider the meaning of ‘availability’ and ‘continuing bonds’ as conditions for secure attachment – the latter being a key outcome goal for therapy. This may well already have arisen during the course of therapy in relation to separations. One client, who had experienced traumatic separation from both parents at the age of 8 when he was in hospital for a year with tuberculous osteomyelitis, asked at the start of once-weekly therapy – ‘can I email you between sessions if there are things that crop up during the week?’. My rather rigid response was to say: ‘On the whole I would prefer that we contain issues within the sessions, and that email is used for practical things like changes of time’. It was only after some months of therapy that he felt safe enough to reveal how put down, rejected and angry he had felt by my response, and to be able to explore how this had evoked echoes of his childhood feelings of emptiness and terror when cut off from his parents when in hospital.

The same client was keen that we should have a follow-up session six months after our one-year period of therapy came to an end. For him, such an actualising manifestation of availability seemed needed, and it would have been churlish to refuse. Other clients are able to tolerate complete separation from therapy, continuing to draw on its benefits through when needed, imagining what their therapist might have said, or having fully internalised the mentalising function that (see below) is the essence of the developmental help offered by analytic therapy. Attenuated therapy (winding down from intensive work to fortnightly or monthly sessions for a while), or offering an occasional limited series of sessions if a crisis arises in the client’s life, are other examples of helping the client to maintain a live sense of an available attachment figure.

Responses to ending can be theorised bringing together the Bowlbian perspective with Kleinian ideas of working through loss (1940). Klein’s starting point is Freud’s paper ‘Mourning and Melancholia’ (Freud 1917), which is usually seen as the germ from which the field of Object Relations grew. Freud describes the ego as a ‘precipitate of abandoned cathexes’ – i.e. the developmental process involves internalising what were previously ‘external’ relationships with significant Others. For this to happen the bereavement process has, at each stage of development, to run its course. This means, especially from a Kleinian viewpoint, coming to terms with ambivalent feelings towards an object on which one is dependent – and therefore which has the potential to abandon one. Odi et amo: love and hate inextricably co-exist, and it is only once ambivalence is transcended that full ‘reinstatement of the lost object’ in the ego is possible. Only when that mature state is reached is gratitude possible.

Therapists, especially when working in a time-limited way, need to be aware of how this inevitable ambivalence will colour reactions to ending. I suspect that my client who asked for a follow-up wanted to be reassured that his hatred of me and my relative unavailability would not have killed me off in his absence. This is not, as I see it, an argument against various forms of attenuated ending, but more a reminder that the meaning of such arrangements must always be thought about and discussed in therapy – in other words, mentalised.


Termination as Co-construction

The relational approach takes it as axiomatic that the clinician’s as well as the client’s states of mind need to be taken into account if clinical phenomena are to be fully explored and understood. At first glance this viewpoint seems to equate to Object Relations theory (ORT), which moved beyond Freud’s original intra-psychic account to an inter-psychic one in which the therapist’s emotional responses to the client were, via projective identification, included in the therapeutic mix. But here the clinician’s own projects and personality remain in the background; her main role is as a reflexive receptacle for the client’s projections. Relational and attachment approaches go two steps further. First, by reviving Freud’s (Bollas 2009) throwaway remark that analysis at its best involves the direct communication of one unconscious (the patient’s) with another (the analyst’s), the role of the analyst’s implicit character and belief system is acknowledged. Second, and flowing from this, comes the idea of the ‘analytic third’ (Ogden 1987; Benjamin 2004 ), the unique relational structure of any given therapy, built from the differing contributions of clinician and client, but directly derivable from neither.

From this the clinically obvious point it follows that any given client will have a different therapeutic experience with different therapists, and that a given therapist will establish very different therapeutic relationships with different clients. It should also be noted however, in contradiction of an absolutist relational viewpoint, that ‘difficult’ clients tend to do badly by whomsoever they are treated, and that excellent clinicians tend to make most of their clients better (Beutler et al. 2004); in the latter case it may be the very flexibility and capacity to accept differing ‘analytic thirds’ that contributes to these ‘super-therapists’ success.

Attachment research has contributed some empirical data in support of these general considerations. Dozier and colleagues (Dozier et al. 2008) measured clinicians’ as well as their clients’ attachment styles, looking specifically at the interactions between them. They employed two binary classifications for clinician and client: secure/insecure and deactivating/hyperactivating (using the earlier terminology avoidant/ambivalent), and then related these to process analysis.

Their findings suggested that therapeutic process differed markedly for secure and insecure clinicians. The latter tended to reinforce and amplify their clients’ patterns of insecurity; the former to ‘redress the balance’, pushing against the client’s insecure attachment strategies. Thus with deactivating clients who tend to play down feelings, drop out early, and miss sessions, insecure clinicians failed to ‘chase up’ such clients, cut sessions short, and went along with superficial reassurances from their clients that they were feeling better. Secure clinicians questioned all these, pushing for buried feelings. Conversely with hyperactivating clients, insecure clinicians tended to become embroiled in escalating demands for more and more help, while secure clinicians were better able to maintain boundaries and offer a secure therapeutic frame.

It should be noted that the clinicians in this study were not trained psychotherapists but mental health workers with different professional backgrounds, social work, nursing, psychology, etc. Nevertheless there is no reason to suppose that similar considerations do not apply to psychoanalytic therapies. Based on the Dozier et al. findings therefore, the ‘too early/too late’ dilemma in psychotherapy can be understood in terms of the ‘fit’ between patient and therapist (Holmes 2001). With an avoidant/deactivating patient and an analyst whose attachment style leads her to over-emphasise interpretations and intellectual formulation, the ending might be ‘too early’. Conversely with a hyperactivating client and an analyst with a tendency to rely excessively on support and affective resonance, the therapy might become protracted and the ending ‘too late’.

The moral is a mentalising one (see below): ‘clinician know thyself’ – therapists need to be aware of and allow for their own attachment styles if they are to offer mutative rather than quasi-collusive treatments. Each analyst will have her or his unique ‘termination style’, evoked to some extent by any given patient, but also manifesting his or her own attachment history and predilections. The lineaments of an ending need to be thought about as co-constructed. The task is not so much to get it ‘right’, as to use the ending as a powerful exemplar from which the client can learn about the ways his unconscious shapes the way he handles, and has handled, loss and separation. In facilitating this, the therapist must abstract her own attachment style from the therapeutic equation in order to see the client’s for what it is.


Dissolution or Disillusion

Why should the ending of a therapy ‘matter’? After all, the ending of other professional relationships – a builder whose job is finally done, a banking or legal relationship concluded – is usually a relief. By contrast, the therapist and therapeutic relationship are invested – ‘cathected’ – in a way that makes them affectively salient. The therapist has become an ‘attachment figure’, a person with the properties of a secure base, the loss of whom evokes the attachment constellation of pain, protest, despair and recovery already described.

The question of how an attachment relationship, as opposed to other connections such as friendship, colleagueship, and professionalism, is established is not entirely clear. Bowlby (1956 quoted Cassidy 2008, p. 12) put it well: ‘To complain because a child does not welcome being comforted by a kind but strange woman is as foolish as to complain that a young man deeply in love is not enthusiastic about some other good looking girl’. An attachment relationship is one that permeates or ‘penetrates’ (Hinde 1979) every aspect of a person’s life in ways that mark it off from others. The more that this is true for a therapeutic relationship, the greater the significance of its ending.

In classical psychoanalysis, negotiating the Oedipal situation entails renouncing the breast, coming to accept the inevitable discrepancy between wish and reality. In the neo-Kleinian model of Oedipus, the child who can tolerate parental intercourse and his own ambivalent feelings is liberated – able to think for himself, and to identify with, or turn towards, the father and through him the outer world, as he or she moves away from maternal dependency (Britton et al. 1989).

Winnicott’s transitional space model (Winnicott 1971) introduces a third term between the nirvana-like world of unbridled need and wish, and the harsh, brutish brevity of reality. In transitional space, wish and reality overlap so that the baby’s hallucinatory illusion of the breast is matched by the mother’s actual provision of it. This real, albeit short-lived, blissful ‘fit’ becomes the basis for later play, creativity and hope. In the Winnicott model there are also repeated failures of fit – a mother is, can, and should only be ‘good enough’. There is a necessary ‘dis-illusionment’ with the breast if the child is to move towards independence and new attachments, and to avoid the narcissism which finds intolerable the inevitable discrepancy between wish and reality.

Resistance to termination can be seen as impediments to these developmental processes. The therapist and therapy are invested with indispensability, an illusory and anachronistic carry-over of infantile needs and wishes into the present. The therapist fails to meet the client’s overweening need and so cannot be relinquished. Or the therapist provides only the maternal half of the parental imago, and so cannot point the client towards independence. Hatred and need are so stark that they cannot be brought together into the depressive position. ‘Failure’ (in the sense of ‘good-enoughness’) is only bearable if balanced by a sufficient bank of success. As Novick (1988: 362) puts it:

Seldom mentioned in the literature is the necessity for disillusionment in order to begin the process of giving up and mourning the omnipotent mother-child dyad. To a certain extent, the analyst must be experienced as a failure for the patient to respond fully to the treatment as a success.

An attachment relationship is one in which needs are actually met by the Other – to a greater (in secure attachment) or lesser (insecure attachment) extent. It seems likely that an effective therapist offers analogous responsiveness, sensitivity and attunement to that of the security-producing care-giver. There will be occasional ‘moments of meeting’ (Stern 2004), where the therapist’s understanding matches the client’s affective state in ways that parallel the advent of the breast at the moment of its hallucination.

But as well as ‘being there’ for the patient, the therapist is also, albeit in a regular and predictable way, not there. Indeed it is possible that it is precisely the nature of this absence that marks out someone with secure base properties from, to use Bowlby’s phrases, a ‘kind but strange’ or some ‘other good-looking’ person. During separations, a secure base figure holds the care-seeker in mind, and stays in the mind of the care-seeker. A client similarly has the right to expect her therapist to hold her in mind between sessions, and to refer back to things said and felt in previous sessions. As the salience of therapy becomes established, the sessions and the person of the therapist enters the patient’s stream of consciousness and unconsciousness (dreams of the therapist, slips about the therapist’s name, intrusions of therapeutic vocabulary into the client’s ‘ideolect’, etc.).

Weekly therapy patients often report in the early stages of therapy: ‘what we were discussing last week stayed in my mind for a couple of days afterwards and then seemed to fade’. Indeed the frequency of sessions could almost be dictated by the time it takes for these memories to fade; the shorter the time, the more frequent sessions are needed. This affective object constancy, I suggest, is the basis for the salience of the therapeutic relationship, and what perforce attenuates when therapy come to an end.

The psychoanalytic frame is ideally suited for the investigation of these issues. By apparently offering ‘nothing’ other than predictability availability and responsiveness, the analyst enables the wish/reality discrepancies to be explored; every ending and break is a rupture in which absence can make the heart grow fonder – or more enraged; termination becomes a mini-mourning in which separation solidifies into irreversible loss.

Reich (1950) makes a very clear comparison of the ending of analysis with mourning from a patient who came to her for a second training analysis, several years after the first analysis with another analyst:

His description of his reaction to the termination of his first analysis was quite revealing: ‘I felt as if I was suddenly left alone in the world. It was like the feeling that I had after the death of my mother … I tried with effort to find somebody to love, something to be interested in. For months I longed for the analyst and wished to tell him about whatever happened to me. Then slowly, without noticing how it happened, I forgot about him. About two years later, I happened, to meet him at a party and thought he was just a nice elderly gentleman and in no way interesting’ (p. 182).

Seen this way, ‘transference’ becomes more than merely a repetition of past relationships. It is an investment of the therapist with properties of a secure base that reflect not just the wish, say, for an ever-available attuned primary Object, but also the real responsiveness of a fellow-human. The ending of a relationship, including a therapeutic one, entails real dissolution as well as disillusionment, and real gratitude for the (albeit professional, and professionally rewarded) love and attention which the analyst has provided.

The work of mourning is no more and no less than the dissolution of this investment. The conscious awareness of someone who was once ‘everything’ begins to fade into the background; eventually all that is left is a scar which, like a healed physical wound, imposing its restrictions, great or small. A lost parent, partner, or worst of all a child, inhabits the psyche forever; but as the pain of loss gradually lessens, new investments become possible. When this process is incomplete there may be an inconsolable effort to replace like with like, eternally in search of what is irretrievably lost, so condemned to everlasting disappointment. Only when this ‘transference’ is dissolved are new beginnings possible.

In psychotherapy ‘coming to terms with loss’ starts with the establishment of a professional relationship; moves into the all-important transferential investment reactivating past attachments and losses; and ends with acceptance of separation, loss and the fading of the transference.

When I retired from my full-time practice as a psychiatrist I decided, among the other rite-de-passages of separation from a job I had undertaken for 30 years or so, to have a leaving tea party for ‘my’ patients. I took the list of all my ‘clinic’ patients and sent each of them an invitation.

These were not, strictly speaking, ‘psychotherapy’ patients, but, rather, people with severe mental illnesses whom I had got know quite intimately over the years. We had, together, gone through the vicissitudes of psychiatric practice: most had been in-patients, some involuntarily so, incarcerated at the stroke of my pen. I had prescribed medication for them, written letters about them, helped them to find accommodation, and had seen them regularly in my supportive clinic (typically for half an hour or so every 8 weeks), so that I knew a lot about their lives (as well, no doubt, as much not known). They too had got used to me as ‘their’ psychiatrist’, tolerating me, more or less, my good and bad points.

As the day of the party drew near I became more and more anxious. Would anyone turn up? Would my guests have anything to say to each other, or would the whole thing be conducted in a funereal silence? I wanted to give my patients a good-bye present. The time was near Easter, so a chocolate egg seemed suitable. I had, à la Winnicott, originally thought of a teddy-bear (i.e. prototypical transitional object) for everyone, but the cost was prohibitive, so I settled for a post card depicting a teddy-bear. On the back I wrote a note of gratitude – without his patients how empty a health worker’s life would be. (Thanks to the UK National Health Service, we are paid so that our patients can be seen ‘free at the point of need’).

As an aside, the teddy bear postcard was something of a joke to myself. I once attended a debate between the proponents of psychoanalysis and those who advocated Cognitive Analytic Therapy, which uses written communications and instructions to patients (Ryle 1999). Although Cognitive Therapy can be useful, I had argued that giving very disturbed patients written communications was comparable to offering a hungry crying baby a piece of paper with the word ‘milk’ written on it and expecting that to assuage the distress.

As it transpired my fears were confounded. The turnout was good, the atmosphere sociable and festive. Music played, food was scoffed, jokes were told and even one or two games were played (I had prepared a box of chocolates wrapped up in ‘pass the parcel’ as an ice-breaker). I enjoyed myself and so, it seemed, did everyone else. The event was indistinguishable from the outside from any tea party with a group of people who knew each other quite well – there is a ‘subculture’ of psychiatric patients which means that one way or another they help and interact with each, possibly far more than professionals give them credit for.

The whole event reminded me of a scene from the film version of that painful (for a psychiatrist) masterpiece One Flew Over the Cuckoo’s Nest in which the ‘inmates’ on the mental hospital escape for a day and, led by the Jack Nicholson character, hire a boat for a trip on the river. Just as they are setting out they are challenged by the boat owner – ‘Who the hell are you, escaped nuts or something?’. There is a terrible moment when the viewer thinks all will be revealed and they will miss their pleasure trip, but in a moment of inspiration Nicholson introduces each one as distinguished Professors. As the camera pans across the familiar faces the inmates are miraculously transformed from the emiserated inhabitants of an impoverished and degraded mental hospital world into the distinguished faces of freedom and respectability. Oddness becomes loveable eccentricity and genius. Context and expectation is all.

So too at the party my patients appeared utterly normal and behaved accordingly. What is more they treated me as though I was one of them: kissing, hugging, gossiping, teasing, enquiring as they might with a friend, or colleague.

As I reflected on this moving event afterwards, I realised that the reason I decided the party was needed both for me and my patients, was to help with the ‘dissolution of the transference’ (Sarra, personal communication, 2003). My patients needed to ‘disinvest’ me with the power of good and evil, and to see me for what I was – a person like them, nearing the end of his working life, frail, flawed, slightly lost without his role and his job, wanting to say goodbye. I needed to be diminished, made vulnerable and ordinary in their eyes, so that they could begin to move on and to invest my successor with the transference that I had carried for all the years we had been working together. Reciprocally, I needed to ‘forget’ their patient-hood and dependency (theirs on me; mine on them), their vulnerability, and to see them, like everyone else, as equals, with their strengths and frailties.

All this was sad, and somehow humbling, but also reassuring. Attachment and separation, investment and dis-investment are part of the flux of life, two poles whose psycho-magnetic field we inhabit, orienting ourselves endlessly between the affective demands of fantasy, and triangulation of reality.


Mentalising Termination

Mentalising is meta-ratiocination, thinking about thinking. It starts from the Kantian perspective that absolute truth is ungraspable, and that reality is always filtered through a mind (Allen & Fonagy 2006). However the combination of two minds looking at the same phenomenon means that, via ‘triangulation’ (i.e. two – the patient’s and therapist’s – perspectives on the same fixed entity), reality can be more or less approximated (Cavell 2006).

This chapter has been informed by two perhaps paradoxical principles. First, a ‘perfect’ ending is both impossible and undesirable. There will always be themes and issues left unexplored in any given therapy. Interviewing analytic patients five years post-termination showed that although most were much ‘better’, the presenting conflicts and themes had not gone away, merely become less dominant and overwhelming (Bachrach et al. 1991). While the evidence suggests that for clients with complex disorders longer therapies have better outcomes, there will always be for one reason or another – money, time, geography – a point when therapy perforce comes to an end. Improvement in therapy takes the form of a negative logarithmic curve (Orlinsky et al. 2004) which means that the ‘law of diminishing returns’ operates, and that it takes more and more time to produce a smaller and smaller benefit. The search for perfection, on the part of either patient or therapist is, as already mentioned, a narcissistic delusion which needs to be examined, mentalised, and discussed, rather than ‘acted on’ – by inaction in relation to ending.

The second implicit point is that an ending cannot be other than an enactment. A decision is made: we will end on such and such a day, after so many sessions, with this or that follow-up arrangements, or none. Since the aim of therapy is to replace action with thought, ending is in this sense always counter-therapeutic. But never-ending therapy is ultimately equally unhelpful. The resolution of this paradox lies in the concept of mentalising. If the main therapeutic leverage in psychoanalytic therapies is that they instil the capacity to think about thinking, and therefore better to know oneself and others, and Self-Other interactions, then it is not so much ending as such that matters, as the capacity to think about termination, the feelings it engenders, its meaning, antecedents and sequelae.

Here is an example illustrating these points. John, in his early 40s, was in once-weekly psychoanalytic psychotherapy. His presenting wish was a vaguely expressed desire to ‘gather my strengths’. He had had a varied life-course, including living in a Tibetan monastery for a while in his 20s, but had settled down, working part-time as a part-time teacher, and devoting himself to his family of three children. He saw his wife, a lawyer, as more of a ‘high-flyer’ than himself, as was his father, a headmaster of a large secondary school.

Sent to high-pressure academic boarding school at 8, he had felt undervalued by and estranged from both his parents, typified, he said, by finding a letter home he had written from school signed with both his first and second name, Jon Smith, as though he couldn’t otherwise be certain that his mother (whom he perceived as wrapped up in his baby sister) would know who it came from.

About a year into therapy he began his session by saying how much better he was now feeling, less compelled to control his wife and children, more ready to lead his own life rather than seek out ‘wise men’ (including, by implication, his therapist) who he had thought were in possession of the answers he was looking for.

Despite this apparent vote of confidence, I sensed there was an implicit attack in this announcement and that he was somehow angry with, or disappointed in, me. While mulling this over I noticed that a potted plant on my window-sill looked neglected and half-dead for want of watering, thinking to myself that ‘I must do something about that before my next patient comes’.

I said ‘I wonder if you are trying out in your mind the idea of leaving therapy’. He said ‘yes’, hurriedly adding ‘not immediately of course…’.

I asked when the thought had first arisen. ‘I think it was when I was in your toilet after my last session’, he replied, ‘it seemed so neglected, so full of cobwebs. It reminded me of my Dad, with his stellar career – yet he neglected all the other parts of his life’.

 ‘You included?’ I asked. ‘I can hear a story of disillusionment, or disappointment here…’.

 ‘No I don’t feel disappointed, sad perhaps. I realise I’ve made my own choices; what matters to me is my wife and family, the everyday things of life. I feel happy to live by my own lights now, not following an impossible dream of my Dad’s’.

 ‘And you seem to be feeling that the so-called ‘wise men’, including me, are an illusion, they neglect what really matters to you; the answers lie within yourself’, I suggested, adding: ‘I have to confess that I was thinking about that plant over there; it looks, like you, as if could have done with some tender loving care’.

 ‘Well, I suppose I do feel angry with you for not transforming me into the perfect person I thought I wanted to be, but also grateful at the same time for the attention and validation you have offered me’, he said.

As he left he said jokingly ‘I don’t need to go to the toilet today!’.

I replied in kind: ‘But it’s pristine, all the cobwebs have been cleared away!’. We both laughed; the session seemed to end with a good feeling on both sides.

John was deciding to end therapy, prematurely perhaps, but perhaps in a more creative and balanced way than the little boy who had to remind himself of his father’s name when writing home. He now knows who he is, and is not. He is more autonomous. He can turn to his ‘high-flying’ wife with his own manhood more firmly established, less needing to be controlling, or to borrow an idealised masculine identity from his ‘wise men’. His feeling of having being ‘unwatered’ as a child is confirmed, via triangulation, with my sense of having neglected my plants. He can see his feelings for what they are – real but not necessarily appropriate to the context he finds himself in. The ‘rupture’ of the dirty toilet, perhaps a receptacle for his shitty feelings of rage at neglect and lack of care as a child, became a validating moment, moving him from immature dependency (the ‘wise men’, the high flying wife) to mature dependency (leading his own life, caring for his family). By acknowledging that the toilet was dirty, while at the same time exploring what a ‘dirty toilet’ might represent in his inner world, transference and reality were beginning to be differentiated. This example of rupture repaired gave him a sense of validatory empowerment and enabled him to decide to leave therapy at a moment which, on balance, felt right to him.


Brief Dynamic Therapy: Foregrounding Termination

 The rationale informing Brief Dynamic Therapies (e.g. Gustafson 1986) begins and starts with termination: ‘In my beginning is my end’ (Eliot 1986). A time limit is implicit from the first moment of therapy. The therapist will ‘count-down’, usually starting each session by announcing ‘this is our seventh session’ or, ‘we’ve another three sessions to go’ or something similar. The termination hangs above the therapy from the start – conspicuous either by its absence (patient ‘pretending’ it does not exist, sometimes collusively with the therapist); or by its inhibitory presence (‘what’s the point of going into all this, I’m only going to be seeing you for another six times’); but always grist to the mentalising mill (e.g. ‘I wonder if the fact that you know you are going to lose me means that you cannot fully make us of me, rather as you never really let your weekends-only Dad know how angry you were with him for leaving your Mum’).

Different varieties of brief dynamic therapy handle the ending in different ways. Balint (1968), who realised that for psychoanalytic psychotherapy to reach out from the ivory couches of Hampstead to the masses, it must perforce abbreviate itself, and who had himself suffered major discontinuities in his life (leaving Budapest for the UK to escape the Nazis, the premature death of his first wife), suggested that at the end of therapy the patient should feel both very much better and very much worse, and that what mattered was that this could be acknowledged (i.e. mentalised). Mann’s 12-session take-it-or-leave-it approach (Mann 1973) is justified as an analogue of the existential irreversibility of death. If the pain of loss can be experienced it can be transcended; follow-ups and interminable therapies are simply attempts to evade the reality of irreversible separation. Ryle’s (1990) CAT, already mentioned, offers, after the prescribed 16 sessions come to an end, a tangible good-bye letter, a memento that can mitigate absence, and trigger the activation of an internalised good object that effective therapy can instate. Intensive Short-term Dynamic Psychotherapy, has a developing evidence-base (Malan & Della Selva 2006) is unabashedly time-limited, but pays less attention to termination than other short-term therapies. Its emphasis is on trauma resolution through direct emotional confrontation of avoided feelings – fear, rage, pain and yearning. This approach is perhaps theoretically closest to Freud’s formulations, and picks up on his observation (Freud 1937) that it is perhaps only traumatic cases are capable of full resolution.



 Yallom (2008) makes a convincing case that death anxiety, a fundamental existential issue, tends to be avoided by patients and therapists alike. Addressing the full implications of termination brings one face-to-face with the transience of life, the distorting impact of trauma on development, the limitations of therapy and the inevitability of suffering. Schopenhauer (1984), the supreme yet unbowed pessimist, introduced to Western philosophy the Buddhist precept that suffering is where we start from, and that embracing suffering is the first step towards transcending it.

Intrinsic to this is the distinction in Schoperhaurian philosophy between the ‘world as will’, and ‘world as idea’, the former corresponding roughly with Freud’s notion of the unconscious, the latter with the conscious mind. The Will, like the unconscious, is infinite and timeless, driven by intrinsic energetic forces that predate human existence and will continue once human life has passed from the universe. The world as idea is the familiar world of experience, in which time’s winged arrow is always felt at one’s back – in Marvell’s ‘had we but world enough and time’ view a good way for a man to persuade a woman to yield to her libido before the death instinct once more holds sway! A balanced combination of acceptance of suffering and carpe diem, living in the present moment, are the antidotes to the despair that might seem to be the inevitable consequence of ‘full catastrophe living’ (Kabat-Zinn 1990).

Against this synchronic/diachronic dichotomy of time one can set the biologically-informed attachment view of a life cycle with its nodal points. These include in the up-swing: conception, birth, weaning, walking, talking, school-entry, friendship, adolescence, sexual experimentation, leaving home, finding a sexual partner, occupation, procreation, parenthood. Then follow from the zenith the beginning of the slow pathway to involution: children leaving home, declining powers spiritual and temporal; the mitigating pleasures of grandparenthood, diminishing responsibilities and returning freedom to play and look back on a life’s troughs and peaks. Each of these, especially if interrupted or perverted by trauma, will play themselves out in their positive and negative aspects, in the metaphor of the therapeutic relationship (Waddell 1998). The task of the therapist is to tune into the pulse of this underlying biological trajectory and to bring them into consciousness so that the patient can better understand where she is on life’s journey.

To conclude, illustrate this, and as an antidote to Schopenhaurian gloom, consider the stage and screen hit Mamma Mia!, adapted from the 1968 film Buona Sera, Mrs Campbell. The success of the piece depends largely on its sing-along use of music from the ABBA, a palindromic Swedish 1970s two-couple pop-group, now disbanded, each man-woman partner now separated.

The setting is a Greek Island. Sophie, a teenage girl, brought up by a single parent, Donna, who runs a hotel that has seen better times, is about to get married. Her fiancée, Sky, is a reluctant bridegroom and feels they’d be better off exploring the world (and perhaps each other) before deciding on marriage. Sophie does not know who her father is; her mother’s diary suggests three possible candidates; unbeknownst to Donna, Sophie invites all three to the wedding. Who is the ‘real’ father? Sophie assumes that she will be able instantly to select the right one when she meets them, but to her dismay she discovers the pre-DNA truism that no one can be absolutely sure who their father is, and that they are all possibilities. Who is to give her away? In her confusion she asks each one of them. The wedding ceremony begins. The naively presiding Greek-orthodox priest invites the father to give the bride away. All three rise to assume the honour. Sophie graciously accepts their blessing, is happy to waive the DNA test and accept all three as her fathers, but suddenly announces that she is not ready for marriage and that the wedding is off. At this crisis point one of the three fathers, Sam, steps into the breach: ‘why waste a good wedding?’, he says, and proposes to the love of his life, Donna. She accepts. Sophie and Sky are delighted and relieved and announce that they will embark on a round-the-world trip, and the movie ends happily with Greek feasting and dancing.

The attachment implication is that one can only leave home if there is a secure base to return to. Now her mother has a man, Sophie can now look after herself rather than play the role of the parentified child looking after her mother (a common pattern in children with disorganised attachment styles). With a secure base now in place and available when needed, she is free to explore the world.

The psychoanalytic implication is that once an internalised good ‘combined parent’ is instated, one is liberated to explore one’s own emotional and sexual life. Renunciation of Oedipal longings to possess the parent, attendant feelings of sadness and envy overcome, is a necessary developmental step towards psychosexual maturity. Finding a good internal combined (‘primal scene’) parent, accepting and transcending one’s envy and feelings of exclusion, and/or desire to control and possess the primary object, and embracing the independence and freedom of movement (literal and metaphorical) that implies, are the parallel psychoanalytic conditions for termination. Successful termination of therapy implies the establishment of a more-or-less secure attachment dynamic, with internal feelings of security matched by external relationships, including if needs be a continuing relationship with a therapist.

In both perspectives coming to terms with loss is a central theme. Sophie can leave home and move onto new attachments (from Donna to Sky) secure in the knowledge that Donna, herself at last firmly attached to Sam, will be available to her when needed. Donna is securely instated in Sophie’s inner world. Sophie no longer needs to push Donna’s neediness away, thereby evading her own vulnerability, nor cling adhesively to Sky in ways that inhibit her exploration. Her inner world is intact, threatened neither by her own aggression, nor needing a rigid external scaffold to support it. The listener, bathed in nostalgia, is reassured that despite the vicissitudes of life – the passing of time, loss, separation (including the dissolution of ABBA as a group and of its members as couples), ‘failure’ (the marriage that never happened; Donna’s single parenthood, Sam’s divorce) – through the healing power of music continuity and reparation are possible.

Similar principles apply, all being well, to the therapeutic powers of psychotherapy – including its termination.


Allen, J. & P. Fonagy, P., eds. (2006) Handbook of Mentalisation-based Treatment (Chichester: Wiley).


Bachrach , H., Galatzer-Levy, R., Skonikoff, A. & Waldron, S. (1991) ‘On the Efficacy of Psychoanalysis’, J. Amer. Psychoanal. Assoc. 39: 871–911.


Balint, M. (1968) The Basic Fault (London: Tavistock).


Benjamin, J. (2004) ‘Beyond Doer and Done to: an Intersubjective View of Thirdness’, Psychoanalytic Quarterly, 73: 5–46.


Beutler, L., Malik, M., Alimohamed, S. et al. (2004) ‘Therapist Variables’, in M. Lambert, ed. Bergin and Garfield’s Handbook of Psychotherapy and Behaviour Change  (Chichester: Wiley), pp. 227–306.


Bollas, C. (2009) The Infinite Question (London: Routledge).


Bowlby, J.:

– (1973) Separation (London: Penguin).

– (1980) Loss (London: Penguin).

– (1988) A Secure Base (London: Routledge).


Britton, R., Feldman, M. & O’Shaughnessy, E. (1989) The Oedipus Complex Today (London: Karnac).


Cassidy, J. (2008) ‘The Nature of the Child’s Ties’, in: J. Cassidy & P. Shaver, eds, Handbook of Attachment (New York: Guildford).


Cavell, M. (2006) Becoming a Subject (Oxford: Oxford University Press).


Dozier, M., Stovall, K. & Albus, K. (2008) ‘Attachment and Psychopathology in Adulthood’, in J. Cassidy & P. Shaver, eds, Handbook of Attachment, 2nd Ed. (New York: Guilford).


Eliot, T. S. (1986) Collected Poems (London: Faber).


Freud, S.:

–  (1905) ‘Fragment of an Analysis of a Case of Hysteria’, SE, 7.

– (1918). ‘From the History of an Infantile Neurosis’, SE, 17.

– (1937). ‘Analysis Terminable and Interminable’, SE, 23.


Gustafson, J. (1986) The Complex Secret of Brief Psychotherapy (New York: Norton).


Hinde, R. (1979) Towards Understanding Relationships (London: Academic Press).


Holmes, J.:

– (1997) Attachment, Intimacy, Autonomy (New York: Jason Aronson).

– (2009) Exploring in Security: Towards an Attachment-informed Psychoanalytic Psychotherapy (London: Routledge).


Kabat-Zinn, J. (1990) Full Catastrophe Living (New York: Delta).


Klass, D., Silverman P. & Nickman, S., eds. (1996) Continuing Bonds: New Understandings of Grief (Washington DC: Taylor & Francis).


Klein, M. (1940) ‘Mourning and Its Relation to Manic Depressive States’, in The Writings of Melanie Klein, Vol. 1 Love, Guilt, and Reparation (London: Hogarth), pp. 262–89.


Malan, D. & Della Selva, P. (2006) Lives Transformed: a Revolutionary Method of Dynamic Psychotherapy (London: Karnac).


Mann, J. (1973) Time-limited Psychotherapy (Cambridge Mass: Harvard University Press).


Mikulincer, M. & Shaver, P. (2008) ‘Adult Attachment and Affect Regulation’, in J. Cassidy & P. Shaver, eds., Handbook of Attachment (New York: Guilford).


Novick, J. (1988) ‘The Timing of Termination’. Int. J. Psycho-Anal., 15: 307–318.


Ogden, T. (1987). The Matrix of the Mind (Northvale NJ: Aronson).


Orlinsky, D., Grawe, K. & Parks, B. (2004) ‘Process and Outcome in Psychotherapy’, in M. Lambert, ed., Bergin and Garfield’s Handbook of Psychotherapy and Behaviour Change(Chichester: Wiley), pp. 227–306.


Parkes, C. (2006) Love and Loss: the Roots of Grief and Its Complications (Washington DC: Taylor & Francis).


Pedder, J. (1988) ‘Termination Reconsidered’, Int. J. Psycho-Anal, 69: 495–505.


Reich, A. (1950) ‘On the Termination of Analysis’, Int. J. Psycho-Anal, 31: 179–183.


Rycroft, C. (1985) Psychoanalysis and Beyond (London: Chatto).


Ryle, A. (1990) Cognitive Analytic Therapy (Chichester: Wiley).


Schopenhauer, A. (1984) Essays and Aphorisms (London: Penguin).


Shaver, P. & Fraley, R. (2008) ‘Attachment Loss & Grief’, in J. Cassidy & P. Shaver, eds., Handbook of Attachment (New York: Guilford).


Sroufe, L. (2005) ‘Attachment and Development: A Prospective, Longitudinal Study from Birth to Adulthood’, Attachment and Human Development, 7: 349–367.


Stern, D. (2004) The Present Moment in Psychotherapy and Everyday Life (New York: Norton).


Tuckett, D., Basile, R., Birkstead-Breen, D. et al. (2008) Psychoanalysis Comparable & Incomparable (London: Routledge).


Waddell, M. (1998) Inside Lives (London: Karnac).


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


Yallom, I. (2008) Staring at the Sun (London: Piatkus).


Jeremy Holmes, MD, FRCPsych, BPC, worked for 35 years as Consultant Psychiatrist and Psychotherapist in the NHS, first at University College London, and then providing a district psychotherapy service in North Devon, focussing especially on people with Borderline Personality Disorder. He was Chair of the Psychotherapy Faculty of the Royal College of Psychiatrists 1998–2002. He now has a part-time private practice; teaches on a Masters and Doctoral psychoanalytic psychotherapy training and research programme at Exeter University, where he is visiting Professor; and lectures nationally and internationally. He has written more than 150 peer reviewed papers and chapters in the field of ‘attachment theory’ and ‘psychoanalytic psychotherapy’. His many books, translated into 9 languages, include The Oxford Textbook of Psychotherapy (2005, co-editors Glen Gabbard and Judy Beck), Storr’s The Art of Psychotherapy (Taylor & Francis 2012) and Exploring In Security: Towards an Attachment-informed Psychoanalytic Psychotherapy (Routledge) which won the 2010 Canadian Psychological Association Goethe Award. With Arietta Slade he is currently preparing a 6-volume compendium of the most important papers in Attachment (Benchmarks in Psychology: Attachment Theory, SAGE). Literature and the Therapeutic Imagination, and John Bowlby and Attachment Theory 2nd Edition (both Routledge) are due 2013. He was recipient of the 2009 New York Attachment Consortium Bowlby-Ainsworth Founders Award.

Publication Date:

January 18, 2014

Share This Article


European Journal of Psychoanalysis