Correlated symptoms and simulated medical classification

Investigated predictions of category learning models in a simulated medical diagnosis task with 158 undergraduates. Ss learned about a fictitious disease or 2 diseases from hypothetical case studies in which some symptoms correlated with each other and others were independent. They then judged new cases. Across 4 experiments, Ss proved to be sensitive to configural information. When choosing between pairs of new cases, Ss tended to choose the case that preserved the correlation over the case that broke the correlation, even when the case with correlated symptoms contained fewer typical symptoms. When judging which disease was present in a single case, Ss' diagnoses were determined primarily by the correlated symptoms. Implications for process models of categorization are discussed. (20 ref) (PsycINFO Database Record (c) 2006 APA, all rights reserved).