Abstract
Magnetoencephalography (MEG) is a neurophysiological technique with high temporal resolution that has the potential to differentiate processing stages occurring tens of milliseconds apart. Recent research has characterized two early stages of face processing at 100 and 170 ms after stimulus onset, known respectively as the M100 and M170. While the M100 amplitude is correlated with successful categorization of a stimulus as a face, only the M170 amplitude is correlated with successful face identification (Liu et al., 2002). Here we examined the M100 and M170 responses in two developmental prosopagnosics, individuals who have a normal ability to categorize stimuli as faces but impaired identification of particular individuals. Given their behavioral deficits, we predicted that we would find a normal M100 but a reduced or absent M170. In keeping with this hypothesis, both subjects showed a higher M100 response to faces than to houses, but no difference in the M170 amplitude for faces and houses. Further testing in one subject revealed that in terms of face selectivity the sensors of interest defined in the previous experiment were within the normal range at the M100, but the face selectivity of his M170 was below that of normal subjects. Moreover, stimuli manipulated to contain only face parts or only face configurations elicited a normal pattern of response at both the M100 and the M170 in this subject, suggesting that his problem may be one of integration of these independent processing streams. In sum, our findings suggest that in some instances, MEG may prove a useful method for determining neural correlates of the behavioral impairments observed in developmental prosopagnosia.