Folk Taxonomies Versus Official Taxonomies

Flanagan and Blashfield’s paper continues a highly original program of research on clinicians’ understandings of psychopathology. This work is unique in bringing concepts and methods from cognitive anthropology to bear on psychiatric classification. At first blush, it might seem questionable to treat clinicians’ beliefs about psychiatric disorders as folk taxonomies, no different in kind from classifications of birds produced by members of a forest-dwelling tribe. Unlike birds and other natural objects of folk taxonomies, mental disorders are not concrete entities and they cannot be distinguished on perceptual grounds. Unlike tribespeople, clinicians are not simply “folk” whose naïve classifications can be compared with those of scientific authorities. They are, in fact, heavily exposed to the reigning official taxonomy during their professional training, and their thinking can hardly remain unaffected by it. It is a testament to Flanagan and Blashfield’s work that despite these apparent disanalogies, their folk-taxonomic approach manages to illuminate how clinicians carve up the psychiatric domain.

This paper reports some careful, creative, and labor-intensive studies to make a few fundamental claims about how clinicians classify. First, the authors argue that clinicians hold simplified taxonomies compared with the official DSM-IV taxonomy: They encompass fewer disorders and classify them in a shallower hierarchy. Second, they claim that clinicians substantially agree in how they classify disorders, and that a consensus classification can therefore be extracted from their individual taxonomies. Third, they argue that clinicians’ taxonomies show only a modest consistency with DSM-IV, and that the discrepancies may be systematic and important. Notably, for example, clinicians did not preserve the fundamental divide between the DSM’s Axes I and II. My comments address the second and third claims.

Is There a “Consensus” among the Clinicians?

One of the main achievements of Flanagan and Blashfield’s paper is to develop a “consensual taxonomy,” which is laid out in their Table 2a. This taxonomy clearly distils how their clinicians tended to judge the relations among diagnoses. However, it is an exaggeration to present it as a consensus. If, by Romney, Weller, and Batchelder’s (1986) third criterion, “cultural consensus” requires that the primary factor in a principal components analysis accounts for “most” of the variance in clinicians’ judgments, then Flanagan and Blashfield’s analysis fails to show consensus. Their first factor accounts for only twenty-five percent of the variance, well short of the required majority. This finding implies that there is three times as much inconsistency or variability among clinicians’ taxonomic judgments than there is consistency and commonality. The observed factor-analytic findings are compatible with the average correlation between pairs of clinicians being only about 0.2, a very weak level of agreement.

This is not to say that there is no consistency among the clinicians, or that there is no value in summarizing that consistency, but it does indicate that there is a great deal of diversity as well. Clinicians’ folk taxonomies seem to be highly idiosyncratic, more a Babel of distinctive classifications than a chorus of agreement. It is therefore risky to draw strong generalizations about how clinicians classify, and to imply that there is a single consensual taxonomy.

Several things are very interesting about this diversity of classifications. First, Flanagan and Blashfield show that it cannot be explained by the clinicians’ differing levels of experience or by the different methodologies used to elicit their taxonomies. The variability among clinicians must therefore have some other basis, which awaits further investigation. Second, the level of agreement (“consensus”) obtained here is significantly lower than what has been found in more orthodox studies of folk taxonomy. By implication, the classification of psychiatric disorders is a slipperier enterprise than the classification of animals and plants. People who have extensive professional experience and training in the field of psychopathology—vocational training that includes the standardizing forces of accrediting bodies and a dominant psychiatric nosology—nevertheless tend not to agree substantially about how to map that territory. One guess about why this might be is that clinicians’ folk taxonomies are not merely descriptive maps of the interrelations among diagnostic entities, but are...