Individual differences in the selection and use of features for classifying visual stimuli

Two experiments were performed to investigate the classification of pot-like outlines by human judges. In experiment 1, seventy-two pot-like shapes, drawn by using all possible combinations of values of four pot ratios, were classified by twenty subjects and by a computer program. The shapes varied only in quantitative features and possessed no all-or-none characteristics. In experiment 2,256 shapes traced from drawings of existing pots were classified by fifteen judges. The pots varied in both quantitative and all-or-none features. The results showed that there were differences between judges in the weightings they assigned to different features, and the judges themselves could be classified according to the weightings they gave the features. There were even differences in the way different judges used all-or-none features for classifying. Possible mechanisms are suggested for the basis of these differences.