Intuitive and empirical prototypes in childhood psychopathology.

Abstract

This study examines the validity of intuitive classifications of experienced diagnosticians, that is, the empirical foundation of their intuitive prototypes. Intuitive prototypes of I 0 experienced diagnosti­ cians were collected in two tasks and were compared with the empirically based assessment system ofT. M. Achenbach ( T. M. Achenbach, K. C. Conners, H. C. Quay, F. C. Verhulst, & C. T. Howell, 1989). Results show that the intuitive prototypes correspond very poorly with the empirical core syndromes, instead consisting of different groups of symptoms from various core syndromes. The study also showed a low similarity between behaviors that clinicians judge as co-occurring with an incompletely specified syndrome and behaviors that empirically co-occur with that syndrome. Reasons for the difference between intuitive prototypes and empirically derived classification struc­ tures are discussed, as are consequences for diagnostic practice. Research on intuitive prototypes in psychopathology has demonstrated that the prototype view corresponds to how cli­ nicians actually think about and use diagnostic categories (Cantor, Smith, French, & Mezzich, 1980). Prototypes can be regarded as classes or categories in which features are orga­ nized. In the prototype view of classification, category member­ ship is a matter of degree. In the domain of child psychopathol­ og) this means that problem behaviors can be more or less typ­ ical for a particular prototype. According to Chan and Jackson (1982), intuitive prototypes or implicit personality theories arise from the observation of co-occurrences of behaviors. Therefore, they reflect the distribution of traits in people and provide a relatively valid foundation for the judgement of spe­ cific targets. The issue of whether these intuitive prototypes reflect reality has been raised (Clark, McEwen, Collard, & Hickok, 1993; Ho­ rowitz, Wright, Lowenstein, & Parad 1981; Rosch, Mervis, Gray, Johnson, & Boyes-Bream, 1976). However, we are not aware of any study that actually compares the intuitive proto­ types of diagnosticians with classes of dysfunctional behavior or syndromes derived from multivariate analyses of systematically collected problem behaviors. The expectation that these intu­ itive prototypes do not reflect reality is based on various argu­ mt:nts, such as the illusory correlation bias (i.e., seeing a corre­ lation between two events as being stronger than it actually is;