Abstract
Human observers can distinguish between highly similar faces in a fraction of a second. A network of cortical regions has been proposed to underlie human face processing. Areas in superior temporal (STS), fusiform (FFA), lateral occipital (OFA), as well as anterior temporal cortical regions and the amygdala are activated more strongly when observers view faces than non-face objects. Previous studies have suggested distinct roles for these areas, such as identity discrimination in the FFA and discrimination of emotional expression and/or gaze direction in the STS. However, previous studies have not determined whether these areas are critical to performing the perceptual discriminations, or whether their activation corresponds to ancillary, post-perceptual, processing (for example social processing, or multi-modal integration.) In the current study we use performance-based fMRI analysis to determine which areas show patterns of activation corresponding to successful perceptual discrimination. Subjects' thresholds were determined during a practice session utilizing a staircase procedure in a carefully controlled psychophysical face identity discrimination task. Presenting the stimuli at a known threshold during the fMRI data collection ensured that all changes in activity could be directly attributed to changes in internal state and not stimulus differences. We show that while areas STS, FFA, and OFA all show higher activation for faces than for a non-face blobby object, only the OFA and FFA show significantly greater activation when subjects are accurate. These findings, as well as other preliminary findings from discrimination tasks on face expression and gaze direction, suggest that areas OFA and FFA play primary roles in the basic perceptual processing of face stimuli.