To the Editor: We are pleased that the DSM-5 Personality and Personality Disorder Work Group is considering feedback from the larger mental health community and invites continuing dialogue. We look forward to seeing the revised personality disorder proposal to which Dr. Skodol refers.

We reiterate our conviction that while dimensional trait models are useful as research tools, they are of limited relevance and utility in clinical practice. They should not be offered in lieu of clinically coherent personality prototypes, nor offered in a way that competes with or complicates the use of a prototype-based diagnostic system.

A good diagnostic system is like a good map in that it must accurately depict the territory. However, sometimes one requires a road map, sometimes a topographical map, and sometimes a political map. A mountaineer in a wilderness region will have little use for a highway map, regardless of its accuracy.

We do not consider the trait model map inherently better or worse than the personality prototype map. Rather, we believe it is the wrong map for clinical purposes. The two kinds of maps address completely different questions. Academic personality researchers ask questions about the relationships among variables in a general population. Dimensional trait models help answer such questions. Clinical practitioners need to understand the interrelation of psychological processes in an individual patient. Personality prototype models facilitate such understanding. Neither approach is more scientific than the other; it is a matter of what questions one is trying to answer. The primary purpose of DSM is the clinical diagnosis of patients. We therefore believe that the DSM-5 diagnostic system for personality disorders should be based on personality prototypes, not trait dimensions.

We strongly agree with Dr. Skodol's wish to select DSM-5 prototypes on "good scientific rationales." The prototypes used in DSM-5 should be those that have emerged empirically in research specifically conducted to develop personality prototypes, that have demonstrated their clinical value-not borrowed from DSM-IV to "ease the transition" to DSM-5 or to an eventual predetermined dimensional trait model for DSM-6.

The empirical literature on personality trait models developed independently of the clinical practice literature and with little input from clinical practitioners. It reflects mainly the methods and concepts valued by academic researchers who do not interact with patients. It would be unfortunate if the official psychiatric diagnostic manual mirrored this bias, to the detriment of patient care.