CONNECTEDNESS AND PART-RELATION INTEGRATION IN SHAPE CATEGORY LEARNING

We investigated the role of connectedness in the use of part-relation conjunctions for object category learning. Participants learned catego ries of two-part objects defined by the shape of one part and its loca tion relative to the other (part-relation conjunctions). The topologic al relationship between the parts (connected, separated, or embedded) varied between participants but was invariant for any given participan t. In Experiment 1, category learning was faster and more accurate whe n an object's parts were connected than when they were either separate d or embedded. Subsequent experiments showed that this effect is not d ue to conscious strategies, differences in the salience of the individ ual attributes, or differences in the integrality/separability of dime nsions across stimuli. The results suggest that connectedness affects the integration of parts with their relations in object category learn ing.