Abstract
Changing object category from a cup to a violin produces a large release from adaptation in LOC compared to when the same images are repeated (Grill-Spector et al., 2001). This release from adaptation with a change in object class could be interpreted as a semantic effect (Simons et al., 2003). However, a class change is not only a change in semantics, it is also a change in shape. In the present experiment, subjects viewed a sequence of two object images, S1 and S2, under one of four conditions: a) the two stimuli were identical, b) the stimuli were identical but mirror reversed, c) S1 and S2 were different exemplars from the same basic-level class, (e.g., two different breeds of dogs in different poses), or d) S1 and S2 were from different, but closely related, basic-level classes, (e.g., a cat in a similar pose to a dog shown at S1). S2 was always translated with respect to S1. The subject's task was to respond with one key if the shapes were identical (ignoring the reversal in condition b) and another key if the shapes were different (conditions c and d). The shape similarity of S1 and S2 in conditions c and d were equated using the Gabor-jet model (Lades et al, 1993), which simulates the multi-scale, multi-orientation filtering of the visual field that is characteristic of early visual areas. Within LOC, BOLD responses for conditions b, c, and d, were substantially equivalent, and all higher than that of condition a, suggesting sensitivity to shape rather than class. In frontal and parietal areas, greater BOLD responses were evident for the reflected and different-exemplar conditions—the more difficult conditions (as assessed by RTs and error rates) for their respective responses —suggesting that activity in these areas may be reflecting task difficulty.
This research was supported by NSF BCS 04-20794, 05-31177, 06-17699 to I.B.