The burden of patients with borderline personality disorder on mental health services is now recognised to be considerable (e.g., Oxfordshire Mental Healthcare NHS Trust, 1998), so that the importance of developing a potentially effective brief therapy for this difficult-to-treat population can hardly be overstated. A recent cohort study of 27 patients with the disorder treated with cognitive–analytic therapy (CAT) showed improvement at 18 months for 14 patients (Reference Ryle and GolynkinaRyle & Golynkina, 2000). Five-year follow-ups are not yet available nor, crucially, has there yet been a randomised control trial of CAT (Reference MargisonMargison, 2000). Ideally, a comprehensive service for patients with borderline personality disorder should include facilities for brief hospitalisation (to manage suicidal crises), for partial hospitalisation (for short-term containment of dangerous or very disturbed behaviour), for brief therapies of up to 6 months' duration (to stabilise impulsive behaviours and increase psychological mindedness) and for longer-term therapies (of 2 to 5 years, to allow personality change and growth) (Reference Gunderson, Davis, Youngren, Rosenbluth and YalomGunderson et al, 1997). CAT provides an option for brief therapy, while most long-term therapies are psychodynamically based.

Coming from a psychodynamic background, the creator of CAT, Anthony Ryle, has used psychodynamic concepts to underpin it, so that most analytically trained therapists will feel familiar with CAT theory. For instance, the important CAT concept of role reciprocity is a clear exposition of projective identification (Reference Klein and Money-KyrleKlein, 1946) and role responsiveness (Reference SandlerSandler, 1976). However, the omission of the idea of defence against intrapsychic conflict (including both repression and splitting), the minimisation of destructive attacks by the self on knowledge and linking and the lack of place afforded the unconscious would place CAT outside the usual arena of psychodynamic theory. Along a continuum of psychological views about the importance of drives v. cumulative trauma, Ryle would perhaps sit at the extreme traumagenic end of the spectrum. However, his attacks on the drive end of the spectrum (Reference RyleRyle, 1993, Reference Ryle1995a ), as represented by Kleinian theory and its drive-related formulation of aggression, echo views expressed by Sutherland (1983).

Taking into account the mental set and the practical procedures of CAT, most analytically trained therapists would not see it as belonging to the spectrum of analytical therapies (Reference Ryle and RyleRyle, 1995b ), and perhaps there are a number of reasons for this.

First, although both CAT and analytical psychotherapy would aim to increase insight, CAT is also aiming at specific symptom change. Concentration on symptom change would, in the view of many analytical patients, take the therapist out of a state of analytical neutrality. This would in turn reduce the ability of the therapist to make an accurate transference interpretation, which would be thought to be the main mutative agent in analytical therapy. The mutative agent in CAT seems to be an increase in self-reflection; this is a cognitive function, which contrasts with the transference interpretation as an experience. Transference interpretations may occur at crucial times in CAT, but they are not a primary feature of the therapy. It should be noted that analytical neutrality does not equate with a bland, opaque or unresponsive therapist; analytical therapists strive for a neutrality that is equidistant from points of conflict (Reference KernbergKernberg, 1984) in the patient.

Second, analytical therapists operate optimally using two perspectives, described by Bion (1974) as binocular vision, whereby one eye views the patient and his or her material through theory, the other operating “without memory or desire”. Too strong a concentration on theory may interfere with the evenly hovering reverie needed for the state of being without memory and desire, which has been described as follows:

It is clear that in CAT the therapist is very busy forging a therapeutic relationship with the patient, a busyness that reduces the space for reflection. The need of the CAT therapist to spot certain prescribed role relationships may blind the therapist to what is unique about the patient in the room. In a state of viewing without memory or desire, the therapist is thrown back on his or her own internal world, which will include identification with his or her personal therapist and supervisors. For a therapist without a depth analytical training, such a situation may be quite frightening and may lead to counterprojective identification on to the patient. The ‘scaffolding’ of the therapy and activity of the therapist in CAT may allow CAT therapists to function adequately in this model without depth training.

Third, there may be, among analytical therapists, scepticism about what is internalised as a result of a brief therapy that has no guarantee of progress into a longer therapy. A core feature of borderline disorder is patients' intolerance of being alone (Reference GundersonGunderson, 1996), and coming into any therapy they seek a containing relationship. A brief therapy (even if this is explained at the beginning) may be a teasing, frustrating experience for a patient, with the therapist taking on the role of Fairbairn's exciting, rejecting bad object (Reference FairbairnFairbairn, 1952) and the patient feeling (in a mental state that may be split off from an ongoing alliance-seeking state) retraumatised. Successful interpretations of a negative transference of an exciting, rejecting persecutory figure can be made only if the patient actually feels held, psychologically, by the therapist over time. Thus, in the case vignettes cited by Denman (2001, this issue), when Jenny, with a background of being abandoned as a child, explores feelings of disappointment at not being held in a long-term frame by her therapist, is the disappointing therapist internalised? When Paul, at assessment, sets a date to die, is this a retaliatory response to the therapist informing him that the therapy had a date to end? Could his changing states (of rebellion and defiance; then misery and dependence in relation to an uncaring other; then being furious and contemptuous of help offered) be understandable concomitants of the here-and-now recognition of relational disappointment? The transference link is apparently not made by the therapist, so is what is internalised at this point an actual uncaring, unthinking therapist?

Fourth, an analytical therapist might be afraid both of the use of suggestion and of closure in making a written formulation early on in a therapy. Patients with borderline personality disorder may identify with the content of a formulation (Fonaghy, 1995) without identifying with the ability to think about it. This may lead to a concrete closure around the formulation, preventing rather than promoting further psychic growth.

Some of these reservations may be resolved with more precise outcome studies, but it may also be that those most at home as CAT therapists will be those coming straight into the therapy, rather than those with a psychodynamic background.