Phenomenological Classification Systems: The Case of DSM-III

PHENOMENOLOGICAL CLASSIFICATION SYSTEMS: THE CASE OF DSM-III MICHAEL T. McGUIRE* Introduction A fundamental aim of taxonomic endeavors is to divide populations into meaningful subgroups. In most instances, achieving this aim is more difficult than imagined, particularly when taxonomists use self-report statements as their primary data. Such statements are subject to numerous influences (e.g., cultural, memory, motivational, subjects' capacity to comprehend) that are likely to differ across persons. As an example, take a recent publication that includes self-report "reasons" for committing crimes. A sampling of categories that respondents could use includes : losing your job, heavy debts, gang involvement, revenge or anger , excitement or kicks, friends' ideas, to get money for drugs, andjust felt nervous and tense [1, p. 94]. Several types of classification and interpretative problems develop when categories like those listed above are used. Obvious ones include: categories may overlap, for example, they are not logically or behaviorally independent; categories are assumed to be related to behavior, although they may not be; the importance of categories may not be equivalent; and respondents differ in their ability to recall events as well as relate events to subsequent behavior. Such problems are well recognized by taxonomists, and most established phenomenologically based systems have gone through a series of reliability and validity evaluations prior to actual use. Seldom, however, are the possible reasons for the problems examined in detail. Moreover, The author thanks K. M. Colby, E. Liston, L. Jarvik, J. Marmor, D. McGuire, J. Spar, and R. Stoller for comments on earlier drafts. The responsibility for what is said rests with the author. *Department of Psychiatry and Biobehavioral Sciences, Neuropsychiatrie Institute, School of Medicine, University of California at Los Angeles, Los Angeles, California 90024, and Nonhuman Primate Laboratory, Sepulveda Veterans Administration Medical Center, Sepulveda, California 91343.© 1986 by The University of Chicago. All rights reserved. 003 1 -5982/87/300 1-05 1 6$0 1 .00 Perspectives in Biology and Medicine, 30, 1 ¦ Autumn 1986 \ 135 if taxonomists had their way, most would probably prefer to classify events of interest using other than phenomenological data. This option is not always available, however. Such is the case with DSM-III, the official diagnostic system for American psychiatry. The DSM-III poses an interesting set of taxonomic problems, many of which must be addressed in any phenomenological taxonomy. In what follows, I shall argue that many of the problems encountered in information gathering and organization in developing and revising DSM-III are intrinsic to phenomenological taxonomic endeavors dealing with complex systems; others are specific to psychiatric taxonomies; some are a consequence of the populations which psychiatrists encounter ; and some are a consequence of situations in which humans interact to gain taxonomic-relevant information. In the latter group, evolutionary biological concepts may provide valuable insights. Taken together, the points developed in the paper suggest that we should be only cautiously optimistic about taxonomic systems that rely on selfreport data. The DSM-III [2] represents an elaborate partitioning of mental illness . Several hundred disorders are identified. Ideally, a psychiatric taxonomy would do the following: (a) characterize groups of persons suffering from specific types of disorders (e.g., bipolar illness) and their subgroups; (b) develop criteria for identifying group and subgroup members; and (c) construct subgroups that are associated with specific forecasts about the future clinical course of subgroup members (e.g., the "natural history" of subgroup members). Hoped-for by-products of such a taxonomy would include identification of subgroups in which members have similar etiologies and in which clinical forecasts are associated with predictable behavioral change in response to specific interventions . In any taxonomic endeavor dealing with data representative of physiological-psychological-behavioral interactions, no one expects to achieve the ideal and, depending on which DSM-III subgroup one is considering, the ideal is approximated in differing degrees. There were many reasons for developing DSM-III. They will not be reviewed here except to note that many psychiatrists considered its predecessor , DSM-II, far from ideal. During the period in which DSM-III was being developed as well as after its publication, concerns were expressed about its form and its categories: it was superficial [3]; it attempted to sacrifice validity...