Abstract
In the context of event-related potentials (ERPs) studies, the N170 component is considered to be the signature of face perception. Human faces evoke a negative potential about 170 msec (N170) after onset, whereas animate and inanimate nonface stimuli do not (Bentin et al.,1996). It is also known that the N170 is delayed and enhanced when faces are presented upside-down (Rossion et al., 1999; 2000; Eimer, 2000). The debate is still open as to why this is so (Rossion & Gauthier, in press).
Here, we use a new principled approach developped by Smith, Jentzsch, Gosselin & Schyns (VSS, 2003) to determine the information in a face that triggers the N170 face inversion effect. Twenty 5.72 × 5.72 deg faces (i.e., [5 males + 5 females] * 2 expressions) sparsely presented through a mask punctured by 16 small Gaussian windows (standard deviation = .22 deg) were used. In half the trials, the stimuli were inverted. Our observers (n=2) had to identify the stimuli. Electroencephalograms (EEGs) were recorded. We performed multiple linear regression analyses on the punctured masks and EEGs for the upright and the inverted (sparsely sampled) faces. A complex pattern of differences was found between the two conditions. We discuss the implications of our findings for face processing.