Abstract
A widely held notion is that face and object expertise utilize the same sources of perceptual information. Although much is known about face expertise, the nature of the diagnostic information for object expertise and the stage in the visual processing hierarchy at which it is utilized are still unknown. To address this question, we compared performance and event related potentials (ERPs) of 15 car experts and 15 car novices performing a category verification task. The two groups categorized objects with which they had different levels of expertise (faces, cars, and airplanes) at basic and subordinate levels. To manipulate the information contained in the images we spatially-filtered them comparing performance and ERPs for broadband (BB) images and images that were either high- or low-pass filtered (HSF and LSF, respectively). In both groups, face subordinate categorization relied more on LSFs than on HSFs, while subordinate categorization of airplanes relied more on HSFs. Critically, the experts relied more on HSFs than on LSFs for subordinate car categorization. This pattern of spatial frequency (SF) usage in car expertise contrasts that found in face categorization. The N170, early face-selective ERP was modulated by expertise. In the experts' left hemisphere N170 amplitude in response to cars was equivalent to N170 amplitude to faces, both higher than to airplanes. In novices the N170 in response to cars was equivalent in response to airplanes, both smaller in amplitude than the N170 to faces. However, this effect did not interact with SF or categorization level. Overall, our findings suggest that opposite to some theories of visual expertise, the diagnostic information needed for expert car recognition is qualitatively different from face recognition. Furthermore, expert use of specific SF scales does not occur at the early perceptual stages reflected by the N170 and thus should be related to later post-perceptual processes.