The Story of Research into Cognitive Analytic Therapy:  Tony Ryle

Research has always been integral to the CAT approach. Dr Anthony Ryle, the founder of CAT, comments that: “Working with patients to find the best descriptions of what it was we were trying to change as part of outcome research, demonstrated to me the power of collaboratively arriving at such descriptions.” In this sense, research methodology is within every CAT therapy as well as being a way of looking at overall outcomes for patients and refining the model and method.


CAT was formally defined as a distinct model in 1985 but it emerged from my descriptive, epidemiological and repertory grid research and conceptual developments of the previous decades (Ryle, 1967, 1969, 1975 and 1982) and from the general recognition of the common factors underlying change in psychotherapy. I was interested in the whole person perspective of dynamic psychotherapy but was frustrated by the fact that neither its process nor outcome was researched.  As I saw it, such research required working with patients to establish their individual aims in a form which, in contrast to behavioural and the emerging cognitive models, identified underlying psychological processes and not just symptoms, thoughts and behaviours. Working with patients on the early reformulation of their presenting problems in these terms, in order to produce descriptions which would allow the measurement of change, proved so therapeutically powerful that it transformed my practice and became a key feature of what became CAT.

A complete list of published CAT studies, books and journal articles is published on the website in the bibliography and reference list.

Early research contributing to the development of CAT

As in any new development, observation and description of the process and effects of therapy preceded large scale formal studies. Clinical practice, conceptual development and more formal research have continued to be mutually influential. Early research into individuals and small groups was carried out, combining the use of standard questionnaires with the measurement of individually identified features.  Repertory grids offered a means of measuring relevant changes in how patients construed themselves and others, through the prediction of desirable changes in selected measures identified in pre-therapy grids.  This was combined with rating changes in the individual Target Problems and underlying Target Problem Procedures which were arrived at in the reformulation process (Ryle, 1980). This research led to the first elaboration of a general model, the Procedural Sequence Model, later developed into the Procedural Sequence Object Relations Model (Ryle, 1982, 1990).

Studies of outcome

Brockman et al. (1987) reported a small randomised controlled trial comparing the effectiveness of CAT and a brief psychodynamic model, using both standard (nomothetic) and individual (ideographic) measures.  CAT was more effective as measured by the latter, repertory grid-based, measures. Ryle and Golynkina (2000) published a descriptive, evaluated study of a series of cases of clients given a diagnosis of borderline personality disorder (BPD) treated with CAT.  The clinically and statistically significant effect of adding CAT to a well developed service for older adolescents with borderline features was reported by Chanen et al., 2008, 2009 a and b in a large, well-designed RCT. Professor Sue Clarke has recently completed a positive RCT of 24 session CAT vs Treatment As Usual (TAU) for patients with a range of personality disorders that has been submitted for publication.

Many other evaluated but not controlled trials of CAT in a number of patient groups have been completed. For example, Borderline Personality Disorder (Dasoukis, J., et al, 2008, Livanos, A., et al, 2008); depression (Katsigiannopoulos, K., et al, 2008);  Obsessive Compulsive Disorder (Protogerou, C., et al, 2004; Kosti, F., et al, 2008) and diabetes (Fosbury, J., 1996, Fosbury, J.,  et al, 1996, Fosbury J.A., et al, 1997). Early CAT research anticipated the current demand for evidence based practice; the subsequent failure to offer more RCT-based evidence has a number of explanations, notably the fact that CAT has developed rapidly and in a number of geographical locations but has lacked a central academic base. Moreover, the ethical and design problems of RCTs and their limited clinical usefulness in psychotherapy research have been widely discussed. Single case experimental design studies offer more clinically relevant findings; Kellett (2005, 2007) has demonstrated the use of this methodology to study CAT with personality-disordered patients and this approach has been applied to a series of cases of BPD patients (Kellett, Bennett, Ryle and Thake accepted for publication).

Studies of process

A number of investigations of aspects of process have been published.   Bennett and Parry (1998) demonstrated the accuracy of reformulation.  They developed an empirically-based model of good practice using the method of Task Analysis on the basis of which an audiotape-based measure of therapist’s general psychotherapy and specific CAT skills (CCAT) has been developed (Bennett and Parry, 2004). A study of how threats to the therapeutic alliance in the CAT treatment of BPD were dealt with led to the description of an empirically-validated model of CAT practice (Bennett, Parry and Ryle, 2006) providing a well designed adherence measure .  This was applied and validated in a later study (Daly, Llewelyn and McDougall (2010).  Recently Sue Llewelyn has shown correlation between good therapy outcome and the therapist’s ability to work with and overcome threats to the therapeutic alliance (Rupture-Repair model) (Daly, Llewellyn and McDougall, 2010).

The development of the clinical model

CAT grew from an initial attempt to integrate psychoanalytic and cognitive models and the process of differentiating from these sources continued.  CAT-based critiques of psychoanalysis include Ryle, (1996, 2003) and of cognitive models include Ryle (2001, 2010). The incorporation of ideas from Vygotsky and Bakhtin and of evidence from developmental studies is described in Ryle and Kerr (2002).

The Multiple Self State Model (MSSM) of Borderline Personality Disorder identifying structural dissociation as a key feature was proposed (Ryle, 1997 a and b) and a repertory grid study confirmed that patients could identify the characteristics of their dissociated self states (Golynkina and Ryle, 1999).  During the last decade three books were published describing the use of CAT with adult abuse survivors (Pollock, 2001), with older patients (Hepple and Sutton, 2004) and with offenders (Pollock, Stowell-Smith and Gopfert, 2006).  The use of the CAT model within staff groups has been described and evaluated (Kerr, Dent-Brown and Parry, 2007; Thompson, Donnison, et al, 2008).  CAT has also been used with clients in small groups (Duignan and Mitzman 1994, Maple and Simpson 1995, Hepple 2012).

The development of research tools

A questionnaire indicating poor personality integration, the Personality Structure Questionnaire (PSQ) (Pollock, Broadbent, Clarke et al., 2001, Bedford et al., 2009) provides an effective screening tool and its repeated use provides an indication of the timing and extent of change in the degree of structural dissociation in the course of therapy. The States Description Procedure (SDP) (Bennett, Pollock and Ryle, 2005) and the revised version (SDPr) (Ryle, 2007) provide a basis for patients’ guided self-reflection through which the number and characteristics of individual borderline patients’ dissociated self states can be identified.

The current situation

Two important RCTs are in process.  In Sheffield the SPeDi trial (Sheffield Personality Disorders) is comparing CAT with CBT in patients with BPD.  In Oxford, a pilot RCT of CAT vs CBS in an IAPT population of mild anxiety and depression is underway. More CAT-related research is in process and as the model becomes consolidated it is to be hoped that more formal studies will be carried out.  But the close links between clinical experience, conceptual developments and the process and outcome research developments which has characterised CAT are likely to continue to yield more clinically relevant but imperfect studies of process than more formal conventionally-blessed outcome studies. The Research website will maintain an up-to-date list of published work and will aim to report work in press or in process.

Updated May 2012

Follow this link for the list of key references.

News from ACAT

Planned Website Migration Complete We are pleased to announce that the migration planned for today has now been completed....

New ACAT phone number The ACAT phone number has now changed to 01305 263511. Please be aware the old 0844 number will not work after the 12th of August....

Planned ACAT Website Maintenance - Website Unavailable on Friday 17th July from 12pm BST We are going to be migrating the ACAT website to a new server at midday BST on Friday 17th July 2015. This will mean the site will be unavailable while the transfer takes place....

Click to read all news

Our Next 3 Events

04-12-15 ACAT How to Run a Successful Private Practice: All You Need to KnowHow to Run a Successful Private Practice: All You Need to Know One-day CPD workshop presented by Robert Watson Friday, 4th December 2015 - 09:30 for a 10:00 start, to finish at 16:30 The Augustana Centre, 30 Thanet Street, London WC1H 9QH Cost:...

04-12-15 CAT for apparently unusual experiences, offered by CatalyseCAT for Apparently Unusual Experiences This event is now full. Please contact the administrator (Frances Free at Catalyse) if you wish to go on a waiting list. Please also note the venue has changed, and the workshop will now be held at Dalton Ellis...

07-12-15 Youth CAT - CAT toolsThis event is organised around CAT tools appropriate for work with young people and their families. If you have a particular tool that you have found useful, let us know and we will try to allocate you some time on the timetable. Please circulate amo...

Please help us by completing our visitor survey... it will only take 5 minutes. Thanks!