Abstract
Brief presentations of feature-deleted images of line drawings of objects, in which every other edge and vertex are deleted from each part, prime their complements (comprised of the deleted features) as well as they do themselves (Biederman & Cooper, 1991). This result suggests that the representation mediating visual priming—the facilitation in the speed and accuracy in the naming of the identical image over a same name, different-shaped exemplar (presented 7 min later)—does not specify local features but, instead, simple object parts. Presumably once the representation of these parts is activated (by either member of a complementary pair) the memorial representation is of the parts rather than the local features. We assessed whether these priming results would be reflected in fMRIa (a for adaptation), in which the second presentation of a repeated stimulus 400 ms later, results in a reduced BOLD response compared to the presentation of a different object (e.g., Grill-Spector, Kurtzi, & Kanwisher, 2001). A contour-deleted image of an object was followed 400 msec later by the Identical image, its Complement, or a same-name Different-shaped Exemplar. The second presentation was always a mirror reversal of the first to reduce the potential role of local spatial or completion processes. Subjects passively viewed the sequences but were instructed to identify/silently name the objects to themselves. If fMRIa in area LO provides an index of priming, a reduced BOLD response for Identical compared to Different Exemplar images would be expected in that area. Preliminary results are consistent with such an effect, suggesting that the reduction in the BOLD response cannot be attributed to reduced activation from repetition of the name or basic-level concept of the object. Moreover, the BOLD response for Complementary images is equivalent to that for Identical images, a result parallel to that found in the behavioral experiments.
NSF 04207994 and 0426415; Human Frontiers Science Program 99-53; Bosco Tjan.