Practical, Functional, and Natural Kinds

Keywords: Classification, essentialism, natural kinds, practical kinds.

 

I am grateful to the two commentators for giving my paper their serious attention, and for writing such stimulating, clarifying, and challenging responses. In a brief response I can only begin to discuss a select few issues, although both commentaries could generate a great deal of profitable exchange.

Zachar's commentary elaborates his practical kinds model, discusses several of the graded series of kinds proposed in my taxonomy of psychiatric categories, and disputes some aspects of my treatment of natural kinds. Zachar and I have no quarrel on his first point. I recognize his practical kinds model as a general theory of psychiatric classification rather than simply as a description of one type of category, the "practical kind." It was perhaps bad form to pilfer a term from Zachar's model and give it a more circumscribed role in my own, although by doing so I intended both to acknowledge the importance of the concept and to situate it among some of its alternatives.

All the same, I stand by the account of practical kinds developed in my paper, even if it reflects a narrower sense than Zachar's. In my sense, practical kinds are pragmatically drawn distinctions imposed on continuous dimensions, whereas to Zachar all psychiatric categories are at least to some degree practical. He therefore proposes a broad "continuum" of practical kinds. The difficulty that I have with this proposal is that such a continuum implies a fundamental unity among psychiatric categories: categories are all to be understood first and foremost in pragmatic terms and differences in kind between them in internal structure are to be neglected. In my view, although the pragmatic bases of psychiatric classification are vitally important, and perhaps inadequately formulated in my paper, the neglect of internal structure and of possible breaks in the continuum of kinds are real limitations of the practical kinds model. Although Zachar is partially justified in claiming that his view is pluralistic, allowing as it does for variability in degrees of practicality, by failing to acknowledge distinctly different kinds of categories in the psychiatric domain, his view is less pluralistic than I believe it should be.

One example of what I take to be this neglect of structural differences is a conflation of two sorts of practicality in Zachar's work. On the one hand, certain categories such as hypertension or obesity are defined on pragmatic grounds, a binary distinction being drawn in the absence of an internal discontinuity, because individuals identified as category members are adjudged to be at significantly elevated risk. On the other hand, some categories just happen to have high practical utility. Zachar argues that schizotaxia, for example, is practical because it affords a variety of scientific and clinical inferences.

It seems to me that these sorts of practicality—or perhaps these sorts of relations to practicality—are importantly different. Hypertension is defined as a category because of its practical utility—because a certain blood pressure threshold marks an inflection in actuarial risk of other illnesses—and might therefore be said to have "front-end" or constitutive practicality. Schizotaxia, which taxometric research shows to be a discrete kind (Haslam, 2003), is not defined and identified primarily on the basis of its relation to an external criterion, but on the basis of its internal structure. Its features cohere in a way that indicates a category with a determinate prevalence. It has "back-end" or consequential practicality to the degree that this category enables useful predictions. These two proposed forms of practicality are equally important: both mean that categories will be pragmatically valuable. However, they are also different, being characteristic of different kinds of categories and perhaps even negatively related. Kinds with front-end practicality, that is, may tend to have a more restricted range of practical implications, defined as they are in relation to specific external criteria. Kinds with back-end practicality, in contrast, may enable a wider array of predictions to the extent that...