Over the past two decades, numerous attempts have been made to address what it is about the psychotherapeutic process that is healing. I have found

[…] Stephen Mitchell's 1988 book “Relational Concepts in Psychoanalysis”, to be the most intellectually satisfying and clinically useful conceptualization of the different modes of therapeutic action. The theory he advances synthesizes three different perspectives: the drive-conflict model, the deficiency-compensation model, and the relational-conflict model.

Drawing upon Mitchell's ideas (as well as the ideas of others), I have developed an integrative model of therapeutic action that takes into consideration many different schools of thought. It is my belief that most psychotherapeutic models boil down to advocating […] either enhancement of knowledge, provision of experience or engagement in relationship - as the primary therapeutic agent. […]

[Pursuing and unfolding Mitchell’s integrative initiative represents a seminal and significant paradigm shift towards an embracing, inclusive, pluralistic perspective. This is a wonderful, brilliant, worthy attempt. But right here, at the beginning, in that first paragraph describing what she is setting out to do, is also contained what I perceive to be one major flaw and shortcoming of the whole formulation/model. It's not very obvious here, because it's just the initial summary, but it is already tangible.

She then continues through most of her introduction describing these three different models and what different kinds of therapeutic relatedness they imply. MS]

The first model is the interpretive model of classical psychoanalysis. Structural conflict is seen as the villain of the piece, and the goal of treatment is thought to be a strengthening of the ego by way of insight. Whether expressed as …

• the rendering conscious of what had once been unconscious (in topographic terms);
• where id was, there shall ego be (in structural terms); or
• uncovering and reconstructing the past (in genetic terms),

in Model 1 it is "the truth" that is thought to set the patient free. Interpretations, particularly of the transference, are considered the means by which self-awareness is expanded. How do interpretations lead to resolution of structural conflict?

As the ego gains insight by way of interpretation, the ego becomes stronger. This increased ego strength enables it to experience less anxiety in relation to the id's sexual and aggressive impulses; the ego's defenses, therefore, become less necessary. As the defenses are gradually ally relinquished, the patient's conflicts about her sexual and aggressive drives are gradually resolved.

The Model 1 therapist sees herself not as a participant in a relationship, but as an objective observer of the patient. Her unit of study is the patient and the patient's internal dynamics. The therapist conceives of her position as outside the therapeutic field and of herself as a blank screen onto which the patient casts shadows that the therapist then interprets.

Model 1 is clearly a one-person psychology.

In some ways it is not surprising that Freud would have been reluctant to recognize the importance of the actual relationship, because Freud never had any "relationship" whatsoever with an analyst. His, of course, was a self-analysis. By way of a meticulous analysis of his dreams, he was able to achieve insight into the internal workings of his mind, thereby strengthening his ego and resolving his intrapsychic conflicts.

But there were those analysts both here and abroad who found themselves selves dissatisfied with a model of the mind that spoke to the importance not of the relationship between patient and therapist but of the relationships among id, ego, and superego.

Both self psychologists in the United States and object relations theorists in Europe began to speak up on behalf of the individual as someone who longed for connection with others.

In fact, Fairbairn (1943), writing in the 1940s, contended that the individual had an innate longing for object relations and that it was the relationship with the object and not the gratification of impulses that was the ultimate aim of libidinal striving. He noted that the libido was "primarily object-seeking, not pleasure-seeking" (p. 60).

Both the self psychologists and the European (particularly the British) object relations theorists were interested not so much in nature (the nature of the child's drives) but in nurture (the quality of maternal care and the mutuality of fit between mother and child).

Whereas Freud and other classical psychoanalysts conceived of the patient's psychopathology as deriving from the patient (in whom there was thought to be an imbalance of forces and, therefore, internal conflict), self psychologists, object relations theorists, and contemporary relational analysts conceived of the patient's psychopathology as deriving riving from the parent (and the parent's failure of the child).

How were such parental failures thought to be internally recorded and structuralized? Interestingly, some theorists (Balint 1968) focused on the price the child paid because of what the parent did not do; in other words, "absence of good" in the parent-child relationship was thought to give rise to structural deficit (or impaired capacity) in the child. But other theorists (Fairbairn 1954) focused on the price the child paid because of what the parent did do; in other words, "presence of bad" in the parent-child relationship was thought to be internally registered in the form of pathogenic introjects or internal bad objects-filters through which the child would then experience her world.

But whether the pathogenic factor was seen as a sin of omission (absence of good) or a sin of commission (presence of bad), the villain of the piece was no longer thought to be the child but the parent - and, accordingly, psychopathology was no longer thought to derive from the child's nature but from the nurture the child had received during her formative years. No longer was the child considered an agent (with unbridled sexual and aggressive drives); now the parent was held accountable - and the child was seen as a passive victim of parental neglect and abuse.

When the etiology shifted from nature to nurture, so, too, the locus of the therapeutic action shifted from insight by way of interpretation to a corrective experience by way of the real relationship (that is, from within the patient to within the relationship between patient and therapist).

No longer was the goal thought to be rendering conscious the unconscious, so that structural conflict could be resolved; now the goal of treatment became filling in structural deficit and consolidating the self by way of the therapist's restitutive [in Clarkson’s terms, we might say: ‘reparative’. MS] provision.

With the transitioning from a one-person to a two-person psychology, sexuality (the libidinal drive) and aggression took a back seat to more relational needs - the need for empathic recognition, the need for validation, the need to be admired, the need for soothing, the need to be held.

The therapist was no longer thought to be primarily a drive object but, rather, a selfobject (used to complete the self by performing those functions the patient was unable to perform on her own) or a good object-good mother (operating in loco parentis).

To repeat, the deficiency-compensation model - embraced by the self psychologists and by those object relations theorists who focused on the internal recording of traumatic parental failure in the form of deficit - conceived of the therapeutic action as involving some kind of corrective experience at the hands of a therapist who was experienced by the patient as a new good (and, therefore, compensatory) object.

In Model 2, then, the patient was seen as suffering not from structural conflict, but from structural deficit - that is, an impaired capacity to be a good parent unto herself [this is a neat, important formulation, fully rooted in an object relations understanding – highly relevant still today. MS]. The deficit was thought to arise in the context of failure in the early-on environmental provision, failure in the early-on relationship between parent and infant.

Now the therapeutic aim was the therapist's provision in the here-and-now of that which was not provided by the parent early-on - such that the patient would have the healing experience of being met and held.

Of note is that some deficiency-compensation theorists (most notably the self psychologists) focused on the patient's experience of the therapist as a new good object; others (the Model 2 object relations theorists) appeared to focus more on the therapist's actual participation as that new good object.

But what all the deficiency-compensation models of therapeutic action had in common was that they posited some form of corrective provision as the primary therapeutic agent.

It was then in the context of the new relationship between patient and therapist that there was thought to be the opportunity for a "new beginning" (Balint 1968) - the opportunity for reparation, the new relationship a corrective for the old one.

But although relationship was involved, it was more an I-It than an I-Thou relationship - more a one-way relationship between someone who gave and someone who took than a two-way relationship involving give-and-take, mutuality, and reciprocity. [this is a very important recognition, central to the equivalent distinction by Clarkson between the ‘reparative’ and the ‘authentic’ relationship modalities, where the ‘reparative’ is understood in exactly the same terms (i.e. as an I-it role relationship) as opposed to an I-Thou authentic relationship (which by definition needs to be free from role behaviour). It is one of the major confusions in the humanistic movement that these two distinct modalities of relationship get conflated, ignoring the fundamental contradictions between them.
But even Carl Rogers was already clear about the difference, right there in the three core conditions which is why he distinguishes ‘unconditional positive regard’ as a different condition from ‘congruence’. What is not often appreciated is that there is an inherent tension between the core conditions: between empathy and unconditional positive regard on the one hand (where the attention is more on the experience of the other person in front of me) and congruence on the other hand (where - if I want to become aware of it - the attention needs to be on my experience); i.e. the self-other tension runs right through the core conditions, reflecting the tension between ‘reparative’ and ‘authentic’ modalities.

Stark uses the thoroughly humanistic term ‘I-Thou’ relationship here, derived, of course, from Martin Buber, but - as we will see later - she fills it with very different meaning and uses it very differently than humanistic therapists would. Some people use the term ‘I-I relationship’ in opposition to ‘I-it’ relationship (e.g. in Gestalt therapy, see Jennifer MacKewn (2009) Developing Gestalt Counselling), bu in my reading of the literature ‘I-I relationship’ is used exchangeably with ‘I-Thou relationship’, although some meaningful distinctions might be made between the terms. MS]

It is for this reason that self psychology, which is a prime example of a deficiency-compensation model, has been described as a one-and-a-half-person-person psychology (Morrison 1994) - it is certainly not a one-person person psychology, but then it is not truly a two-person psychology either. [which is then clarified and implicitly defined precisely as an ‘I-I relationship’ between two subjects - in this respect Stark’s and Clarkson’s distinctions and formulations entirely coincide, although – as we will see- there are major confusions to come as to whether two-person psychology involves two subjects or objects. MS]

Let us return to the issue of what constitutes the therapeutic action. There are an increasing number of contemporary theorists who believe that what heals the patient is neither insight nor a corrective experience. Rather, what heals is interactive engagement with an authentic other [this coincides not only with what most humanistic therapists believe to be the ‘I-Thou relationship’, but importantly they also agree that it is this modality (and no other!) that ultimately produces therapeutic action. However, although she is using a thoroughly humanistic term, what she means by ‘authentic other’ will turn out to be quite different from what humanistic therapists mean by it. MS]; what heals is the therapeutic relationship itself [[to formulate this as “the therapeutic relationship itself” is tricky, and once again points to another flaw in Stark's formulation which I will come back to in much more detail later; is she implying that Model 1 and Model 2 are not actually ‘therapeutic relationships’ at all?!? - Does she see all three models as valid ingredients in therapeutic relating, or does she consider Model 3 to be the 'real' thing? - she gives thoroughly contradictory messages about that at different points. MS].

Relational (or Model 3) theorists who embrace this perspective conceive of patient and therapist as constituting a co-evolving, reciprocally mutual, interactive dyad - each participant initiating and responding. For the relational therapist, the locus of the therapeutic action always involves this mutuality of impact.

[And here comes the crucial formulation that distinguishes the psychoanalytic conception of two-person psychology from the humanistic one: MS]

Unlike Model 2, which pays relatively little attention to the patient's exerting of pressure on the therapist to participate in certain ways [I think that is a highly questionable assertion: Joseph Sandler - very much from within a one-and-a-half-person perspective - used the terms 'role responsiveness' and ‘actualisation’ during the 1970s already precisely to pay attention to the dynamics created by the patient’s unconscious exerting of pressure onto the therapist. I think the recognition of this pressure is very much a Model 2 principle, at least across the psychoanalytic schools in the UK. MS], Model 3 addresses itself specifically to the force field created by the patient in an effort to draw the therapist into participating in ways specifically determined by the patient's early-on history - ways the patient needs the therapist to participate if she (the patient) is ever to have a chance to master her internal demons. [because this pressure between two people is the transference, we are talking here about an understanding of the supposed ‘I-Thou’ interaction between two people that fully includes the recognition of transference pressures. She now goes on to elaborate that understanding: … MS]

In other words, in Model 3 [supposedly defined as an authentic I-Thou relationship. MS], the patient with a history of early-on traumas is seen as having a need to re-find the old bad object - the hope being that perhaps this time there will be different outcome [the patient is only seen as having that need in the psychoanalytic tradition, not in the humanistic formulation of two-person psychology, which tries to find a ‘real’ I-I meeting without and beyond the transference. The relational movement to this day, having latterly grown into an amalgamation of humanistic and psychoanalytic traditions, is thoroughly divided and confused on this issue. The paragraph above makes it clear that Martha Stark not only presumes and takes for granted key principles of the psychoanalytic tradition, but makes them definitional and axiomatic in terms of what she means by ‘relational’; she either by definition excludes the humanistic and existential perspectives from her understanding of the ‘relational’, or appears to be oblivious of how humanistic therapists are - at least originally and traditionally - talking about the same thing from within completely different assumptions, i.e. defining authentic as transference-free – so while Stark by definition seems to exclude them, existential therapists strongly claim the relational principle for themselves, insisting on an “a priori inter-relational grounding from which our unique sense of being arises” (see: Ernesto Spinelli: Practising Existential Psychotherapy: The Relational World). MS].

In order to demonstrate the distinction between a theory that posits unidirectional influence (Model 2) and a theory that posits bidirectional (reciprocal) influence (Model 3), I offer the following:

As we know, self psychology (the epitome of a corrective-provision model) speaks to the importance of the therapist's so-called "inevitable" empathic failures (Kohut 1966). Self psychologists contend that these failures are unavoidable because the therapist is not, and cannot be expected to be, perfect.

How does relational theory (Model 3) conceive of such failures? Many relational theorists believe that a therapist's failures of her patient are not just a story about the therapist (and her lack of perfection) but also a story about the patient and the patient's exerting of interpersonal pressure on the therapist to participate in ways both familial and, therefore, familiar to the patient (Mitchell 1988).

[This is another point where the traditions fundamentally disagree. Stark does not even take into account that the humanistic tradition would have a third perspective here: that the “therapist’s failures of her patient” are to do with the therapists ‘stuff’, like in a really real ‘real’ relationship, like a marriage, where we assume that if things go wrong it's 50-50 between two ‘real’ people. What Stark doesn't fully realise and acknowledge is that here she is confusing Model 3 and Model 1 - the recognition that the patient exerts “interpersonal pressure on the therapist to participate in ways both familial and, therefore, familiar to the patient” is already a much earlier Kleinian insight (because that pressure then easily becomes a projective identification), more readily recognised from a Model 1 stance than from what the humanistic therapist would consider a ‘true’ two-person psychology where any failures are to do with the woundings of the two real people in the room. So when she continues now to speak about ‘relational theory’, she is implying that only two-person psychology is ‘relational’, when the main point she's just using to define the ‘relational’ is actually more a Model 1 ‘one-person psychology’ recognition. One of the perennial confusions and flaws of the model is that she oscillates between defining ‘relational’ as Model 3 two-person psychology only, when at other times she speaks about ‘relational’ as an integrative position, embracing all three models.
Anybody who knows Ken Wilber's writings will have come across a way to solve this kind of confusion: Wilber's idea of second-tier evolution implies that there comes a stage which is not one more new dogmatic model alongside the previous ones, but is of a different order in that it integrates and gives valid recognition to all preceding ones (each stage – in Wilber’s terms - ‘transcends and includes’ the previous one, but only the second-tier stages explicitly validates all of them).
For myself, I am very clear that I want to call ‘relational’ only this integrative notion which embraces all three models as valid in principle (in different situations) and sees them all as relational in their own particular ways, rather than claiming that two-person psychology is ‘the’ true version of relationality and ‘true’ therapeutic relating. The recognition of the equal validity of the different models is much more strongly held in Clarkson's formulation - she does not speak of different ‘models’, but diverse ‘modalities’, implying that they are all valid aspects and versions of therapeutic relating. There are some places where Martha Stark formulates the integrative version, but much more frequently she is biased towards Model 3. MS]

Relational theory believes that the therapist's failures do not simply happen in a vacuum; rather, they occur in the context of an ongoing, continuously evolving relationship between two real people - and speak to the therapist's responsiveness to the patient's (often unconscious) need to be failed so that she can achieve belated mastery of her internalized traumas. [again, this is completely the opposite of what humanistic therapists used to mean by ‘two real people’. MS]

In Model 3, then, the patient is seen as an agent [it is very questionable whether humanistic therapists, e.g. John Rowan, would be happy to use the term ‘agent’ here, because they would be reserving that term for the ‘real self’ - and defining the ‘real self’ as beyond this kind of drivenness by early trauma, and would be working towards a real meeting between client and therapist beyond this wounded need and pressure to use each other as objects (which - if we think about Winnicott’s idea of the ‘uses of an object’ is a thoroughly valid and necessary way of relating, both in therapy and in early development). MS], as proactive, as able to have an impact, as exerting unrelenting pressure on the therapist to participate in ways that will make possible the patient's further growth. The relational therapist [because of all the previous points I made in distinguishing the humanistic conception of two-person psychology and ‘relational’ from the psychoanalytic one, it would be more accurate to say here: “the relational psychoanalyst” …. MS], therefore, attends closely to what the patient delivers of herself into the therapy relationship (in other words, the patient's transferential "activity"); by the same token, ever aware of how telling her own response is to that activity, the therapist also remains very much centered within her own experience (in other words, her countertransferential "reactivity").

[Arguably, this notion of the countertransference is very close to Heimann and Racker’s 1950’s formulation, which ushered in the ‘countertransference revolution’, where transference and countertransference are very much seen as interlocking, but with the countertransference still predominantly ‘reactive’ to and against the transference (whilst now recognising that - rather than consisting just of ‘not completely analysed’ remnants of the therapists on pathology - the countertransference was also a precious therapeutic instrument, giving insight into the client’s inner world – let’s note that this kind of transfer of the client’s experience into the therapist’s countertransference isn't really conceivable without some kind of mechanism like projective identification). So if we stick with Stark's own definitions, these uses of the countertransference should really be categorised as ‘one-and-a-half-person psychology - because she's so aware of and embedded in the psychoanalytic understanding and sensitive to the ubiquity of transferential pressures, she never quite gets to what humanistic or existential therapists would call ‘true’ dialogical two-person relating.

Clarkson tries to address this by talking about ‘reactive’ and ‘proactive’ countertransference, as well as the therapist’s transference onto the client. This conceptual minefield reflects the complexity of what's going on - to contributes some clarification, I have suggested the distinction between ‘habitual’ and ‘situational’ countertransference. What remains unclear are the distinctions between one-and-a-half- and two-person psychologies, as that dividing line looks very different from humanistic and psychoanalytic perspectives. MS]

In fact, relational theory conceptualizes the patient's activity in relation to the therapist as an enactment, the unconscious intent of which is to engage (or to disengage) the therapist in some fashion - either by way of eliciting some kind of response from the therapist or by way of communicating something important to the therapist about the patient's internal world. In fact, the patient may know of no other way to get some piece of her subjective experience understood than by enacting it in the relationship with her therapist.

[here she is repeating the essence of what the British object relations tradition has been taking for granted for decades, in fact since the countertransference revolution (e.g. Patrick Casement would be a very good illustration of an analyst who is acutely aware of these pressures and is trying to resist the temptation to fall into countertransference enactments - it's precisely because he understands that, that he usually insists on ‘holding the boundaries’, thus refusing precisely what ordinary mortals would perceive as a ‘real’ two-person relationship). So if we wanted to categorise anybody working on the basis of that recognition of transferential pressures, like Casement for example, we would have to firmly see them as working from within one-and-a-half-person psychology - but Stark here implies that she is talking about a defining feature of two-person psychology.
Rather than a defining feature of the distinction between one-and-a-half-person and two-person psychology, I think, this is more an argument between two polarities within one-and-a-half-person psychology, as it historically developed: between a reparative branch of object relations in the 1950’s (and later self psychologists of the 70’s) on the one hand and earlier Model 1 Kleinians on the other hand, who - precisely because of their fierce critique of reparative and corrective emotional therapeutic stances – continued to hold out against this, insisting on a robust, non-gratifying stance which doesn't allow itself to be cornered by the regressed ‘tyrant in the high chair’. Really, this is an argument between the therapist in the role of the reparative ‘good object’ versus allowing herself to be constructed as the ‘bad object’ (also a role, transferentially given to the therapist, but accepted, based upon an understanding of the transformative potential of receiving such transferential projections, if they can be contained and ‘worked through’ - in fact, that is the definition of ‘working-through’ according to the British independent tradition).
This argument is crucially relevant to Stark’s discussion of theories of therapeutic action - there is a profound conflict between the reparative stance advocating ‘corrective emotional experience’ as an avenue toward healing versus a stance which insists on working-through by robustly refusing any gratification, precisely because it understands the enormous transferential pressures exerted on the analyst, thus causing countertransferential reactivity, which need to be resisted and contained, rather than acted out. Stark reduces Model 2 to the reparative stance only, whereas I would prefer to define Model 2 as made up by the conflict between the corrective and the non-gratifying stances. In the non-gratifying conception (deriving really from Model 1 Kleinians, and persisting for decades throughout the Model 2 British independent psychoanalysts), healing (i.e. therapeutic action) can only occur if the wounding is accepted and mourned by the client, and the demand for reparation given up. This understanding of therapeutic action is equivalent with what Clarkson defines as the transference/countertransference modality (implicitly drawing upon and taking for granted the British independent tradition of psychoanalysis, and decades of arguments between these two polarities which helped to create enormous clarification of the issues). MS]

I will be using the word provocative to describe the patient's behavior when she is seeking to re-create the old bad object situation (so that she can rework her internal demons), inviting to describe her behavior when she is seeking to create a new good object situation (so that she can begin anew), and entitled to describe her behavior when, confronted with an interpersonal reality that she finds intolerable, she persists nonetheless - relentless in her pursuit of that to which she feels entitled and relentless in her outrage at its being denied. [again, the whole tone of this description, if it reminds us of anything, would probably best be characterised as Kleinian, in its perception and insistence on the baby's relentlessness – and I don't think Stark would hesitate to place Klein into the Model 1 category; so it seems to me there is major confusion between insights derived from Model 1 Kleinians and her description and definition of Model 3 two-person psychology. MS]

If the Model 3 therapist is to be an effective container for (and psychological metabolizer of) the patient's disavowed psychic contents [again, this is very much a one-and-a-half-person psychology formulation (container, metabolizer, derived from Winnicott and Bion), where the therapist allows herself to be constructed in the role of the bad object, recognising that it is a role and precisely not her as a person or a subject], the therapist must be able not only to tolerate being made into the patient's old bad object but also to extricate herself, by recovering her objectivity [again, Stark really should stumble across her own language here and recognise that ‘objectivity’ belongs to Model 1, not Model 3, and that, at best, this is exactly the rationale of traditional one-and-a-half-person psychoanalysis a la Casement. From a two-person psychology focus on intersubjectivity, one has a slightly more sceptical notion of objectivity. MS] and, thereby, her therapeutic effectiveness, once she has allowed herself to be drawn into what has become a mutual enactment [there is nothing much in her formulation to suggest that it is – in fact - mutual ( in the way  humanistic and existential therapists might define it), other than that the therapist is reactively falling into it, and so is part of  the enactment (but is that what we mean by 'mutual'? - unless we want to dilute the meaning of 'mutual' to 'two people reacting to each other and affecting each other', which seems unhelpful to me)?  I'd rather hold out for the humanistic definition of 'mutual' -  if we want to use a yardstick: mutual is what happens in a marriage, isn’t it? MS]. The therapist must have both the wisdom to recognize and the integrity to acknowledge her own participation in the patient's enactments; even if the "problem" lies in the intersubjective space between patient and therapist, with contributions from both, it is crucial that the therapist have the capacity to relent - and to do it first [in terms of surviving and engaging with the enactments, I have no quibbles with the substance of these statements and find them helpful; but in terms of conceptual clarity and definitions, this sound slightly different from before - it seems to me that here we have now the first manifestations not exactly of ‘mutual’, but of the therapist participating, but only inasmuch as she is falling into the “problem” in the intersubjective space, created by the patient’s transferential pressures - is this a two-person psychology formulation, or an evolved and sophisticated form of traditional psychoanalysis in familiar one-and-a-half-person mode? MS]. Patient and therapist can then go on to look at the patient's investment in getting her objects to fail her, her compulsive need to re-create with her contemporary objects the early-on traumatic failure situation [same point: I have no argument at all with her description of these pressures, which I take to be valid, useful and comprehensive, and I very much want to be able to think like that about the interlocking transference-countertransference dynamic - the only argument I'm having is how she categorises these recognitions as Model 3, when I think they are derived from Kleinian Model 1, which then evolved historically towards what I take to be modern object relations Model 2. MS]. If the therapist never allows herself to be drawn into participating with the patient in her enactments, we speak of a failure of engagement. If, however, the therapist allows herself to be drawn into the patient's internal dramas but then gets lost, we speak of a failure of containment - and the patient may be retraumatized. [these are precious recognitions and formulations, which I very much subscribe to, but again, any and all of that could have been said by Casement 30 years ago, and I very much doubt he would see himself as a two-person psychologist. MS]

Although initially the therapist may indeed fail the patient in much the same way that her parent had failed her, ultimately the therapist challenges the patient's projections by lending aspects of her "otherness" or, as Winnicott (1965) would have said, her "externality" to the interaction - such that the patient will have the experience of something that is "other-than-me" and can take that in. What the patient internalizes will be an amalgam, part contributed by the therapist and part contributed by the patient (the original projection). [this is a somewhat dodgy formulation in my perception, conflating the mode of therapeutic action in the face of a patient's projections with a developmental achievement between mother and child (which, I think, would be better formulated in Jessica Benjamin's terms than in Winnicott’s). I think she's talking here about the mother's dilemma in confronting the child's grandiosity, trying to establish some recognition of the mother's subjectivity in the face of the child’s attempt to continue using her as an object. This has similarity with (which is the kernel of validity in Stark’s point), but ultimately is not at all the same as a patient transferring the wounding object onto the therapist (who does allow herself to be constructed as an object), and those projections then eventually needing to be challenged by the therapist. MS]

In other words, because the therapist is not, in fact, as bad as the parent had been, there can be a better outcome. There will be repetition of the original trauma but with a much healthier resolution this time - the repetition leading to modification of the patient's internal world and integration on a higher level. It is in this way that the patient will have a powerfully healing "corrective relational experience" - the experience of bad-become-good.

In the relational model, it is the negotiation of the relationship and its vicissitudes (a relationship that is continuously evolving as patient and therapist act/react/interact) that constitutes the locus of the therapeutic action. It is what transpires in the here-and-now engagement between patient and therapist that is thought to be transformative.

And so this third model of therapeutic action is the relational (or interactive) perspective of contemporary psychoanalytic theory. No longer is the emphasis on the therapist as object-object of the patient's sexual and aggressive drives (Model 1), object of the patient's narcissistic demands (Model 2), or object of the patient's relational need to be met and held (Model 2). In this contemporary relational model, the focus is on the therapist as subject - an authentic subject who uses the self (that is, uses her countertransference) [this is precisely the point where humanistic therapists like John Rowan are likely to tear their hair out, at that conflation between the self and the countertransference. MS] to engage, and to be engaged by, the patient.

Unless the therapist is willing to bring her authentic self into the room, the patient may end up analyzed - but never found.

By way of review: Whereas Model 1 is a one-person psychology and Model 2 is a one-and-a-half-person psychology, Model 3 is truly a two-person psychology. And whereas the Model 1 therapist is seen as a neutral object (whose focus is on the patient's internal process) and the Model 2 therapist is seen as an empathic selfobject or good object-good mother (whose focus is on the patient's moment-by-moment affective experience), the Model 3 therapist is seen as an authentic subject (whose focus is on the intimate edge between them [this phrase ‘intimate edge’ is a reference to a well-known book by relational psychoanalyst Darlene Ehrenberg - The Intimate Edge: Extending the Reach of Psychoanalytic Interaction]. In Model 1, although the short-term goal is enhancement of knowledge, the ultimate goal is resolution of structural conflict. In Model 2, although the immediate goal is provision of (corrective) experience, the long-range goal is filling in structural deficit. In Model 3, although the short-term goal is engagement in relationship (and a deepening of connection between patient and therapist), the ultimate goal is development of capacity for healthy, authentic relatedness.

And, finally, whereas Model 2 is about offering the patient an opportunity to find a new good object, so that there can be restitution, Model 3 is about offering the patient an opportunity to re-find the old bad object, so that the traumatogenic early-on interactions can be worked through in the context of the patient's here-and-now engagement with the therapist.

Along these same lines, Jay Greenberg (1986b) has suggested that if the therapist does not participate as a new good object, the therapy never gets under way; and if she does not participate as the old bad one, the therapy never ends - which captures exquisitely the delicate balance between the therapist's participation as a new good object (so that there can be a new beginning) and the therapist's participation as the old bad object (so that there can be an opportunity to achieve belated mastery of the internalized traumas).

Indeed, psychoanalysis has come a long way since the early days when Freud was emphasizing the importance of sex and aggression. No longer is the spotlight on the patient's drives (and their vicissitudes); now the spotlight is on the patient's relationships (and their vicissitudes).

And where once psychoanalysis focused on the relationship that exists between structures within the psyche of the patient, contemporary psychoanalysis focuses more on the relationship that exists between the patient and her objects - or, more accurately, the intersubjective relationship that exists between the patient and her subjects. In Jessica Benjamin's (1992) words: "where objects were, subjects must be" (p. 44).

[here I see her as stumbling precisely across her own ongoing confusion: having defined – for most of this introduction – two-person therapeutic relating very much in object relations terms, how is she going to cross the bridge towards two subjects and intersubjectivity, which - if two-person psychology means anything - would be a necessary implication, it seems to me. But most of her description of relational two-person psychology have not been about subjects at all, but about the ‘working-through’ of bad objects. There is an important paradox there, inherent in the notion of ‘working-through’ via enactment (i.e. that the healing of the wounding inflicted by the bad object occurs through the enactment of the bad object in therapy and via the therapist) which she is grappling with and grappling towards, but as I have repeated, this is an internal argument between three branches of one-and-a-half-person psychology: the provision of the reparative object, the empathically-reparative failing of the narcissistic self object, and the ‘working-through’ of the bad object - none of that is anywhere near two subjects, although all of it is highly precious and valid in my book - it just isn't two-person psychology. All of it can – and needs to - become a valid ingredient of an integrative stance which embraces and needs to rely upon all three models, but can then afford to be conceptually much clearer about the differences between them and from which ones the integrative stance draws upon for their respective gifts. MS]

In the chapters that follow, I will elaborate on the three models of therapeutic action - the interpretive model, the corrective-provision model, and the relational model. In Part I, entitled "The Therapist's Choices," I will develop the three models in much greater depth, making special note of those features distinguishing each model. In Part II, entitled "Clinical Applications", I will demonstrate (by way of numerous examples and clinical vignettes) translation of the theory into the clinical situation; and I will revisit some of the thornier (and more intriguing) conceptual issues raised in Part I.

A particular issue with which I have been struggling involves the following discrepancy: Although much attention has been paid in the literature to the patient's need to be failed and her active efforts to recreate - with her therapist - the old bad object situation (by way of projective identification), scant attention has been paid in the literature to the patient's equally powerful need to find what never was and her active efforts to create opportunities for such restitution. [which the humanistic movement - in its optimistic recognition of human potential - would not at all be surprised about, but would simply conceive of as the self-actualising tendency, the innate impulse towards health and wholeness (or Jung’s teleological ‘transcendent function’ inherent in the inexorable individuation process); and it would rely on that impulse and its recognition as the foundation - and the ultimate driver and rocket fuel - of the therapeutic process. It's only in the more pessimistic psychoanalytic tradition that we would have to break our heads about the question where the impulse towards wholeness actually comes from, and whether it exists in the first place. An integrative stance would wants to find a synthesis between that humanistic optimism (which frequently escalates into therapeutically naive idealism) on one hand, and the psychoanalytically-informed pessimism - they would think of it as realism, of course - on the other (which arguably is not free from the dualistic zeitgeist of Freud’s times and the kind of feeling one might get if one suffers from cancer of the jaw for years). MS]

As Steven Stern (1994) has astutely observed, "there has been no systematic effort to define a ... counterpart to projective identification, that is, the patient's unconscious efforts to evoke in the therapist specific responses that are different from those of the traumatizing figures of the past" (p. 320). [I think, again, that this is more a problem of American psychoanalysis: British psychoanalysts have recognised the evacuation of the idealised object. Susie Orbach used to tell a story about embodied countertransference, where she had a delicious feeling in her body like a purring cat, which she and her patient later understood as a projective identification on Susie’s part with an evacuated idealised object on the client’s part - unconsciously, the patient wanted (arguably: needed) her therapist to inhabit, demonstrate and model what it is like to feel good in one’s body (including probably the understandable wish that her mother might have felt like that). The more as a therapist one allows oneself to be constructed as a bad object, the more obvious it becomes that all fragments and dissociated aspects of the wounding dynamic, including idealised longings, can be evacuated and be experienced in the countertransference via projective identification. MS]

In Part II, I will be addressing this and other compelling theoretical (as well as clinical) conundrums.

Throughout the book, my intent will be to demonstrate the clinical usefulness for the therapist of thinking in terms of enhancement of knowledge, provision of experience, and engagement in relationship as the three primary agents for therapeutic growth and change. I will hope to show that if the therapist is to be optimally effective, then she must be able to work comfortably within all three models of therapeutic action - sometimes using first one approach, then another, sometimes times using two or three approaches simultaneously.

[So here she finally and unambiguously formulates the kind of integrative position that I think deserves the label ‘relational’. This formulation can be nicely complemented and enhanced by combining it with Clarkson's model of relational modalities. What Stark doesn't quite spell out here, but maybe implies, is that there are continuous tensions and conflicts between these three models which co-exist in just about all difficult moments in the therapeutic relationship. And there are potentialities and creative possibilities for responding from within each in just about every moment. The real conundrum of an integrative position is to understand each model and modality as a possible enactment, each with its therapeutic and counter-therapeutic implications, i.e. we want to be able to be aware of and to think about multiple enactments. The question then is not so much ‘how do I operate within this one particular enactment which I happen to be aware of?’, but how in trying to cope with one enactment then precipitates me into another one. So it's more question of jumping from the frying pan into the fire, then jumping from the fire into a comfortable bath tub. In my simple fashion, I formulate that as: the client's conflict becomes the therapist’s conflict, meaning that we are usually being caught in a catch-22 between to equally unpalatable enactments, with there being occasionally some room for manoeuvre in terms of choosing the lesser evil. The intuition towards this recognition of multiple enactments is implicitly present and hinted at in the following paragraph: … MS]

In any event, at each point in time, the therapist must tolerate the necessary uncertainty that comes with the holding open of different possibilities for the therapeutic action, while avoiding the temptation to jump to premature closure in order to ease the anxiety and the strain - all in the interest of being able to enhance the therapeutic potential of each moment.

[Conclusion: I have tried to indicate, in some cases over and over again, what I perceive to be some of the confusions, shortcomings and unhelpful conceptualisations. But there is one major argument I have with Stark’s formulation which isn't very visible and transparent in this first chapter, and hasn't been addressed by my comments so far.  It is this point which I  want to address now and which I see as having confusing and unhelpful side effects on our thinking and practice. It is this point especially which therefore calls out for - and I would argue: requires - complementary input from Petruska Clarkson’s model of relational modalities.

The combination of the two models creates a powerful synergy and allows us to formulate a more comprehensive model of relational integration, relevant both to a) psychotherapy integration generally, b) the integration between humanistic and psychoanalytic traditions specifically, as well as c) clarifications of what we mean by relationality. My evolving formulation of that integration came to a point in 2005 where I started calling it the ‘diamond model’. That integrative model also takes into account Lavinia Gomez’s 2004 warnings about the difficulties, limitations and inherent pitfalls of integrating what she calls a humanistic ‘alongside’ therapeutic stance with a psychodynamic ‘opposite’ stance.

So what is my major argument?
In trying to describe and define the three models as they have historically evolved, Stark characterises each model as having a particular theory of therapeutic action, a particular technique (i.e. a set of interventions), a particular theory of the human psyche (e.g. drive model), and a particular idea about how to relate to the wounding of that psyche as a therapist (i.e. assumptions about the kind of relational stance that is considered therapeutically useful). We could add to that another layer of abstraction, which is that each model has its own implicit meta-psychology (more philosophical assumptions about the nature of the universe, human reality and values etc).
Her assumption is that each model as she describes it has a coherent set of theory, technique, relational stance, theory of therapeutic action and meta-psychology. Her assumption is that each model works within these coherent correspondencies between theory, technique and relational stance.
However, what has become clear to me throughout the last 25 years of psychotherapy integration - and is explicit in Clarkson’s model of relational modalities - that there is no linear, coherent one-to-one correspondence between a therapist’s relational stance and their therapeutic approach including their belief system, their theory and technique.
The whole point of Clarkson’s model is that the relational modalities do not correspond to the traditional therapeutic approaches: one can pursue just about each and every theory and approach from within the different relational stances. Through the hybridization and eclectic combination of all kinds of therapeutic approaches with each other it becomes apparent that there is no particular consistency (although historically there must have been some implicit coherence when these different models first originated – as every paradigms shift emerges out of the logic of a particular zeitgeist). And that a therapist’s approach and identity has inherent contradictions and paradigm clashes, manifesting fundamental and pervasive inconsistencies.
This point of consistency between theory, technique, relational stance and underlying paradigm is precisely the main argument around and against eclecticism in the field of psychotherapy integration: those who argue against eclecticism insist that it is unhelpful to just pragmatically rip a technique out of its theoretical and meta-psychological context within the approach where it was originated and that it was originally embedded in. The argument is that there is an implicit coherence in how each approach has evolved its techniques out of its philosophical and theoretical assumptions; and that if we are going to be using a technique for eclectic and pragmatic purposes, irrespective of theoretical context and coherence, that there are bound to be clashes and double messages - confusing and uncontaining for the client - between the implications of the technique we are using on the one hand and the implications of the rest of our therapeutic presence (theory,  relational stance etc). It is mainly therapists who function in a predominantly instrumental medical model fashion (i.e. who tend to not pay much attention to the relational space and the congruence of their presence, thinking of these as irrelevant to the therapeutic task), who espouse the virtues of eclectic creativity and freedom. However, any therapist who has a sense that “it's the relationship that matters”, or the quality of relationship, will have differentiated appreciation of how the working alliance is affected by the therapist’s congruence or inconsistencies.
One of the ways in which I recognised this for myself in my early development as a therapist was that I increasingly discovered behind my humanistic philosophy of authentic relating (i.e. my sincere commitment to a ‘wounded healer’ perspective implying two-person psychology - a meeting of equals - a dialogical I-I relationship) pervasive, but hidden one-person psychology ‘medical model’ assumptions which were informing and structuring my therapeutic presence. These were derived from Wilhelm Reich and passed down the generation of Body Psychotherapists - I took them totally for granted as part of the practice I was taught and had seen modelled. The therapeutic tradition I was absorbing at the time was full of techniques and ways of working which implied a thoroughly diagnostic stance, a comprehensively pathologising vocabulary and a pressure to perform and intervene in quasi-medical fashion to produce therapeutic results. What I didn't realise at the time was that in supposedly helping to liberate the client's self-actualising tendency (or Reichian ‘core’, or my idea of Winnicott’s ‘real self’), I first had to identify and diagnose the pathological absence of those worthwhile features, which then defined and gave me my therapeutic task (dread to think where I would be without it). None of this was anywhere near consistent with dialogical relating - without being fully aware of it, I was taking my responsibilities for granted to deliver a successful treatment (the supposed outcomes of which were very much defined in terms of humanistic values and philosophy, but the pressure of the pervasive agendas was constantly part and parcel of my habitual countertransference – I might as well been a doctor who was going to get sued if I didn’t perform properly and got rid of the pathology). I was constantly on a mission, so entirely unavailable for the uncertainty of dialogical being together, without anything having to happen. So my actual relational stance - heavily informed by quasi-‘medical model’ agendas - was completely at odds with my humanistic ideas about my relational stance (I wrote some of this up years later: What Therapeutic Hope for a Subjective Mind in an Objectified Body? (2004)).
When as a supervisor I don't oversimplify the inconsistencies and contradictions in my supervisee's therapeutic identity, it becomes possible to see that any therapist can use just about any therapeutic theory to work from a reparative stance. They can also use just about any therapeutic theory to work from a quasi-medical position. Almost any therapeutic theory can feed into attempts at authentic relating, because it all depends how I use the theory within my therapeutic presence (I'm saying ‘almost’ because some inconsistencies are just too large to bridge, although it is not uncommon these days to meet practitioners who say they happily combine person-centred or existential principles of authentic relating with a fairly medical approach like EMDR or CBT - so apparently it can be done!)
So, for example: just because traditionally humanistic therapists didn't conceptualise and attend to the transference, that doesn't mean that in their actual relating they didn't react to transferential pressures - they did engage in what Clarkson would call the transference/countertransference modality, they just did it unconsciously, without explicit awareness. Conversely, psychoanalytic practitioners who officially subscribe to the idea that the transference never disappears, in actual relational reality often measure the success of their ‘treatment’ by how much mutual authenticity is becoming possible in the therapeutic relationship - implicitly they do actually use their perception of that authenticity as a criterion for whether the client is ready to terminate. So although they may theoretically disavow it, from the client's point of view the authentic modality is very much in operation, and can be experienced as a significant ingredient of the therapy.
The first thing I conclude from this is: my implicit relational stance as a therapist is prior and more fundamental than whatever particular theory, technique or approach I am officially using. And for most of us these elements are far from coherent (and I'm not even implying here that such coherence is a significant aim -  if I am arguing for anything, then it is for full and explicit awareness of incoherence). By the way, this is the needs an profound explanation of what is called the Dodo-bird verdict in counselling and psychotherapy: that when outcome research compares the relative successes between different approaches, no significant difference between them can be found - that's  precisely because theoretical orientation is not very significant to what happens in therapy, which is much more influenced by how the therapist is aware of, can move between and - most significantly as we will see below (crucial aspect of my ‘diamond model’)-  handles the dynamic tensions between various relational stances. Most therapists, who reflect upon themselves and their practice in terms of their approach (theory and technique) rather than in terms of relational modalities, do do all of these things - it is just that they do them unconsciously, implicitly (which, according to the Dodo-bird verdict - works out haphazardly well enough with some marginal success in most cases).
So back to the topic of supposed consistency between theory, technique, relational stance, theory of therapeutic action and meta-psychology. Stark's formulation of the three models is a brilliant exercise in bringing conceptual coherence into what is in reality a very inconsistent and conceptually muddled field, and a very contradictory and inconsistent psychological reality within each therapist. Stark formulates and defines the three models as if theory, technique, relational stance, meta-psychology, theory of therapeutic action were different, equally important, facets of a coherent model - as if all these facets were of the same logical order. I don't think they are. I sometimes summarise this in punchy form as: theories do not integrate, people do. If it happens at all, integration becomes possible on the level of relationships, not on the level of concepts or models.

If we reflect on psychotherapy as a relatively young discipline and its development over the last 100-odd years with an eye to the pervasive and ongoing fragmentation and inconsistency, we might want to think of it as one of the main drivers for development, and why we keep on evolving our theories and our thinking (just as the same is true in what drives us to therapy as clients). It is our implicit recognition - and very real suffering from - these inconsistencies which constitutes one of the ingredients that spurs us on towards further development. Any post-modern perspective, that has grown increasingly suspicious of all grand meta-narratives and unification attempts, can help us embrace pervasive inconsistency as a necessary byproduct of pluralism and ongoing evolution (which we take to be good things, aren't they?).

What I conclude from this is that the diversity and tensions between the relational modalities (or different kinds of therapeutic relatedness, generating different kinds of relational spaces) is more fundamental than the divergences between different theories and techniques. When we look at the actual embodied therapeutic presence of therapists supposedly operating from and within the same theoretical model, we find hugely diverse styles. How can their relational presence, and the relational space around them, be so different when they are officially subscribing to the same theory? What are these very noticeable differences based upon? I think we will find on enquiry that what we experience as a different style/presence is based upon different therapists implicitly emphasising different relational stances/modalities – e.g. some more reparatively unequal, some more authoritatively unequal, some more equal/dialogical or ‘on a level’ - and all within the supposedly same therapeutic approach.
The next thing we can conclude from that is (which is one of the fundamental assumptions of my ‘diamond model’), that all the modalities are necessarily present in different degrees in every therapist and in every therapy, whether they are aware of it or theoretically subscribe to it or not. That is because all the relational modalities are all necessary ingredients, for every therapist, without which therapy cannot work, and without which we probably wouldn't even get a proper alliance.

The problem is that in most of the traditional approaches therapists would get trained into habitual unawareness of some of the relational ingredients of their presence; so a relational modality might then be operating only subliminally, unconsciously, but from the client's perspective would  nevertheless be an important part of the relationship (it reminds me of examples when patients after years of psychoanalysis are asked what made the difference, they often report it as the very innocuous and minor human connection which happened - often accidentally - outside the official therapeutic frame - which, to be fair, may only have become possible and have acquired its significance because of all the years of what happened the rest of the time).
The point of formulating the diamond model as an integrative pluralistic relational model is that we bring our theoretical understanding in line with the actual relational diversity that is always already present, anyway, for everybody. As attuned humans and relationally engaged others, all therapists (whether they are aware of it or not, whether they like it or not) implicitly draw upon the whole wealth of relational modalities and potential. The whole contradictory complexity and the tensions between the relational modalities are all present, to different degrees, all the time, as part of the dynamic human whole, except that through the fragmented and dis-integrated traditions of the psychotherapeutic field we have come to identify in a very restrictive and selective way only with certain aspects of the whole (which we have consciously absorbed as part of our therapeutic identity - in that way formulating it, the parallels between our  integrative evolution as therapists and our integrative evolution as people become glaringly apparent, don’t they?).
But from the client's perspective, our whole complexity as another human is always already in the room, no matter what we officially include and exclude. So from my perspective, the only question is how much I awareness I can bring to the complexity and force field between the relational modalities in a particular here and now moment in the therapeutic relationship, and how much of all that will just happen subliminally, unconsciously.

So whilst Stark’s formulation gives us a helpful map for distinguishing between I-it and various degrees of psychoanalytic I-I relating (and the historical evolution of those positions in the psychoanalytic tradition), it over-simplifies the correspondencies between those relational positions and the therapist’s theory, approach and technique as if they formed a coherent whole and therapeutic identity. The prevalence of all kinds of eclectic, contradictory hybridisations between approaches as part of what therapists these days consider psychotherapy integration shows us that there is no such coherence (and also historically has never been – such coherence is more wishful thinking on the parts of some clear thinkers who have tried to systematise the inherent complexity, and the different degrees of incoherence versus coherence that are continuing to evolve). Therapists have always been in a condition where the left hand doesn't know what the right hand is doing, where we theoretically espouse one thing, and in relationship with our client we do something else, and where one's theory is likely to be in opposition to one's technique, meta-psychology or relational stance - there are ongoing and inherited contradictions between all of these facets. It's precisely because there is such pervasive contradictory dis-integration that the idea of psychotherapy integration becomes attractive.

That critique does not at all invalidate Stark’s description of the historical evolution: from Freud’s taken-for-granted ‘one-person’ psychology assumption of analysis as a form of medical treatment, to the continuing deconstruction of that paradigm, swinging all the way to the other extreme, i.e. the humanistic formulation of what I call the anti-‘medical model’, and the currently confused and confusing situation as to what we really mean by relationality and two-person psychology (especially in regard to the lingering misunderstandings between humanistic and psychoanalytic traditions which use the same terms and apparently same concepts, but often claim these for themselves with very different meanings, often oblivious to how the other tradition is using the same term).

So if we just focus on that evolution of relational stances and paradigms (without loading them with supposed correspondencies to theories and techniques) from medical model treatment towards modern intersubjectivity, we can neatly describe it in terms of three relational revolutions over the last 100 years, which have increasingly put the subjectivities of both client and therapist into the therapeutic space - summarised in this hand-out: The Three Relational Revolutions (2007).

I had written up a version of that historical evolution in my 2006 article “How ‘the wound’ enters the room and the relationship” (Therapy Today, Vol 17 Issue 10; December 2006), focusing on the deconstruction of medical model assumptions over the last hundred years. That article focused on both the crucial dualisms (between ‘doctor and patient’ as well as ‘mind-over-body’) which continue to haunt and plague our discipline as long as they remain un-integrated.]