Dr Kendell (Reference Kendell2002) offers a number of explanations for the reluctance of British psychiatrists to treat patients with personality disorders. He also makes it clear that, whether personality disorder is regarded as an illness or not, it is usually associated with a range of other diagnoses and with a poor response to treatment. This indicates that psychiatrists need to understand them, but whether lack of knowledge of the ‘underlying cerebral mechanisms’ of these patients (or of the psychiatrists whom they irritate) is the problem is, in my view, dubious; the need is rather for an understanding of persons.

While it may have been true in the past that few links were made between the concept of personality disorder and the psychological literature on personality structure and development, the situation has changed considerably in recent years (see Reference LivesleyLivesley, 2001). One such link is offered by the model of borderline personality disorder developed within cognitive—analytic therapy — the ‘multiple self states model’ (Reference RyleRyle, 1997). This model is based on an understanding of development which emphasises the key role of the intense interactions between infants and their caretakers in shaping personality and patterns of interaction (Reference TrevarthenTrevarthen, 2001). These patterns (called ‘reciprocal role procedures’ in cognitive—analytic therapy) determine subsequent ways of relating to others and of managing the self. In the case of people with borderline personality disorder, reciprocal role patterns of abusing—abused and neglecting—deprived are commonly acquired in childhood and these patients continue to expect and accept abuse from others and to inflict it on others and on themselves. Faced with perceived repetitions of abuse or neglect they commonly switch to partially dissociated, more manageable states, responding, for example, with pseudo-compliance, by seeking ideal care from idealised others or by maintaining emotional distance (with or without the use of drugs). Switching between states is often abrupt and evidently unprovoked and is confusing to the self and to others; it also disrupts what capacity patients have for self-reflection and learning from experience.

Clinical staff, whether offering nursing care or occupational, cognitive—behavioural or pharmacological treatments, are always liable to be perceived in terms of the patient's patterns and will often be induced or provoked to reciprocate with, for example, counter-hostility, withdrawal of care and attention, or unrealistic offers of help. They are also liable to be confused and de-skilled by the switches. It is here that management based on the multiple self states model can be valuable.

The model was developed in the context of individual psychotherapy. In practice it involves working with the patient to create diagrammatic descriptions of the different states and of the shifts between them. Such descriptions serve to increase the patient's capacity for self-reflection and control and the clinician's ability to avoid or correct responses likely to reinforce the damaging patterns. More recently, and of particular relevance to psychiatry, diagrammatic reformulation has proved valuable as a basis for care planning and staff supervision; applications in community mental health services and in forensic settings are reviewed in Ryle & Kerr (Reference Ryle and Kerr2002). Working with patients with damaging personality disorders using this approach allows clinicians to respond appropriately, consistently and non-collusively, rather than to react. It can, I believe, be both more effective and professionally more rewarding and could overcome the reluctance of psychiatrists to take responsibility for these neglected patients.