Abstract
Adults appear to be more sensitive to configural information, including the spacing of features, in faces than in other objects (reviewed in Maurer, Le Grand, & Mondloch, 2002). This difference arises even when adults are simply primed to perceive 4 blobs (placed in the position of two eyes, nose, and mouth) as facial features rather than points of the letter Y (Nishimura, Maurer, & Mondloch, 2004). The difference may arise because spacing information plays a greater role in learning to identify individual exemplars of an object category (e.g. faces: Bob vs. John) than in learning to identify objects at the basic level of categorization (e.g. table vs. chair; Gauthier & Tarr, 1997). We simulated this learning difference by having two groups view the same stimuli but learn to label them only at the categorical level or at both the categorical and individual levels. One group (n=9) was trained to label three categories of ambiguous stimuli: bobos formed from 4 blobs, tikas formed from 6 blobs, and pelis formed from 7 blobs. The other group (n=9) was trained, in addition, to use different labels for the three individual bobos, each of which has a slightly different spacing of its constituent blobs. The two groups were matched based on a pre-test of sensitivity to spacing differences in a different set of bobos. On a post-test with novel bobos, the group trained to label individual bobos was significantly more accurate (M=71.1%) in detecting changes in the spacing of the constituent blobs than the group that learned only the category labels (M=64.8%; p = .03, one-tailed). The spacing changes were of the magnitude that naturally exists among human faces. The findings are consistent with the hypothesis that we become more sensitive to the spacing of features in faces than in other objects because we have more experience identifying individual faces than identifying individual members of non-face categories.