Abstract
Biederman & Kalocsai (1997) proposed that differences between face and object recognition exist because the neural representation of faces, but not objects, retains low-level spatial frequency (SF) information extracted by early visual areas. They showed that individuals are better matching identical pairs of faces relative to complementary pairs (with opposite SF and orientation information) but perform equally well with identical and complementary pairs of chairs. Yue, Tjan & Biederman (2006) suggested that this face-specific effect is not due to expertise, since they failed to find increased sensitivity to SF content following training with novel objects. Because their training protocol may be insufficient in establishing any face-like effect, we used naturally occurring car expertise to reexamine this question. In E1, car experts and car novices matched sequentially presented faces and cars. Relative to the study image, the probe could be the same identity with either identical or complementary SF information, or a different identity. All observers showed sensitivity to SF manipulations for both faces and cars, suggesting that such sensitivity can be found with common objects. This effect was not modulated by car expertise. E2 used upright and inverted images to explore whether configural processing (stronger for upright than inverted faces) was the source of the effect. Car novices showed sensitivity to changes in SF content for upright and inverted faces, and cars. However, only faces showed a larger cost of changing SF content for upright images. Expertise does not appear to increase sensitivity to SF content, and configural processing is not sufficient to account for this effect. Indeed, even novice perception of inverted cars was sensitive to SF content. In addition, we conclude that short-term memory for common objects in novices does not solely depend on representations where SF information is absent, such as edge or volume-based descriptions.
This research was funded by grants from the James S. McDonnel Foundation and the NSF.