Categorization and sensitivity to correlation.

Three categorization experiments were run in which participants saw a set of stimuli that varied on 4 continuous dimensions. Participants first categorized the stimuli and tried to predict some of the dimensions, given the values of others. Experiment 1 used iris-like stimuli based on the descriptions of R. A. Fisher's (1936) taxonomic descriptions. It showed that having participants categorize the stimuli was essential to being able to perform the prediction task and that merely observing the stimuli was not sufficient. It also indicated that participants could use within-category as well as between-category correlations for predictions. Experiments 2 and 3 used stimuli with artificial variations of values. Participants processed categories that had different within-category correlations. Participants' behavior could be predicted as a combination of sensitivity to within-category correlation and bias about the sign of the correlations. These results were fit to the rational model of categorization (J. R. Anderson, 1991) and to an exemplar model (R. M. Nosofsky, 1988).