The Category Use Effect in Clinical Diagnosis

Abstract

Conventionally, the process of clinical diagnosis is seen as distinct from subsequent clinical decisions. In the current study we challenge this distinction by examining a category use effect whereby diagnostic features that can be used to classify instances and make additional predictions are more likely to be used in diagnosis. During training, senior medical students (n = 36) and undergraduates (n = 44) learned to classify cases into one of two artificial disease categories. They were then asked to make an additional prediction about each instance. Some features were informative for both diagnosis and the additional prediction (relevant-use features) whereas others were useful only for diagnosis (irrelevant-use features). At test, all groups classified instances with relevant-use features more accurately and confidently than instances with irrelevant-use features. This effect was stronger in those with clinical training and when there was a plausible connection between diagnosis and prediction. The implications for clinical psychology diagnosis and judgment are discussed.