LAUSR

Krueger et al provide a novel and challenging perspective on the perennial divide between the categorical and dimensional approaches to the conceptualization and classification of psychopathological phenomena. Writing on behalf of the recently established Hierarchical Taxonomy of Psychopathology (HiTOP) Consortium, they address critically the “official nosology”, especially as exemplified in the DSM-5. The latter manual is criticized for being “authoritative”, guided by psychiatrists, and not immune against “socio-political” considerations in preserving and presenting an ex cathedra view of psychopathology as consisting of discrete nomothetic entities or taxa. In contrast, the authors highlight the likely empirical advantages of adopting the alternative position on psychopathology as a continuum of quantitative variation that can be organized hierarchically into several higher-order spectra and dimensions. Krueger et al claim that recent research, methodologically stronger than its predecessors, overwhelmingly supports the quantitative-dimensional model of psychopathology, and believe that the latter is now fit to be ubiquitously translated into clinical practice. They advocate placing this model of classifying psychopathology on “an empirical playingfield” insteadof perpetuating the “traditional” nosology, exemplified by the DSMrevision process. Much of the evidence in support of these proposals stems from the comprehensive quantitative review of published taxometric research by Haslam et al. This review was based on a detailed examination and secondary analysis of 177 articles which, when combined, reported a total of 39.9% taxonic results. However, the authors concluded that, after statistically controlling for confounds, the “true” prevalence of taxonic findings was only 14%, mostly involving the domains of schizotypy, autism and substance use disorders. They contended that historical improvements in themethodological quality of taxometric studies, especially the use of simulated comparison data and the linked comparison curve fit index, have contributed to the marked decline of taxonic findings. There are two possible caveats to this line of reasoning. First, the purely statistical analysis and interpretation of the data is no substitute for a well-designed, real-life comparative studyof clinical populations assessed according to both the hierarchical dimensional model illustrated by Krueger et al and the “traditional” categorical nosology of ICD-10 or DSM-5. The outcome criteria in such a hypothetical study should include choice of treatment, prognosis and functional status of the participants. As far as I am aware, no such study has yet been designed or conducted. My second caveat concerns the applicability of the quantitative dimensional scheme to the bulk of psychotic disorders (marginally mentioned in Krueger et al’s paper). Historically, the evolution of the classification of these disorders has taken a path in reverse to that of the common non-psychotic disorders. The theory of the “unitary psychosis” has been dominant in European psychiatry around the middle of the 19th century, being associated with the names of its first proponent A. Zeller and its first critics W. Griesinger and K. Kahlbaum. It postulated a continuum of different stages within a unitary morbid process, terminating ultimately in a complete disintegration of mental life. It was against this background that E. Kraepelin synthesized the three pre-existing entities of hebephrenia, catatonia and paranoid dementia into a single concept, and proposed in 1896 the dichotomy of the unitary spectrum into the discrete entities of dementia praecox and manic-depressive insanity. Renamed as schizophrenia by E. Bleuler in 1908, the former entity was further described as “the group of schizophrenias”, to be split further by K. Leonhard into systematic and unsystematic forms, each containingmany discrete subtypes. Notably, there has been a recent revival of the continuum model of psychotic disorders, which in its turn has been criticized as “scientifically unproven and clinically impractical”. At this point, I shall add my own take on the problem: can a classification of mental disorders be biologically anchored? This is doubtful, at least in the foreseeable future, because: a) the objects classified in psychiatry are explanatory concepts, i.e. abstract entities rather than physical organisms; b) the taxonomic units of “disorders” in DSM-IV, DSM-5 and ICD-10 do not form hierarchies; c) the current psychiatric classifications contain no supraordinate, higher-level organizing concepts. Leaving aside the vexing issue of validity of the categories, the criteria for evaluating psychiatric classifications should at present focus pragmatically on their clinical relevance and utility: capacity of discriminating between syndromes and between degrees of their expression in individual patients; adaptability to different populations and cultural environments; reliability; cognitive ease of use; and reducing stigma.My prediction is that the quantitative/dimensional and the taxonic/discrete approaches to the classification of mental disorders will remain dialectically interconnected as the “yin” and“yang”. A methodological tool eminently suited for empirical research is the grade of membership (GoM) latent structure analysis, which enables the aggregation of clinical and/or neurocognitive measures into a parsimonious number of “pure types” (taxons) which represent fuzzy sets, rather than discrete categories, and assigns to each individual a quantitative affinity score indexing the degree to which he/she resembles each one of the taxons. My research group has been using the GoM to split a large cohort of schizophrenia patients into subtypes based on neurocognitive measures and to specify each patient’s affinity to any one of the taxons. I am reminded of the Hegelian postulate about the transition (“phase shift”) of the accumulation of quantitative changes into a new quality. This sums