Abstract
Our recent study showed that a relatively large number of exposures coupled with sufficient number of variation in expression can help created reliable representation of newly learned faces and put them on the trajectory of being transformed into familiar ones (Shyi & He, 2011). Here we examined the underlying neural mechanisms by recording ERP components for faces that were learned with either relatively large number (i.e., 24) or relatively small number of exposures (i.e., 12), while being subject to variation of a single expression or multiple expressions. The ERP results during learning showed that there were no differences in mean amplitude for N170 and N250 regardless whether faces were learned with a single vs. multiple expressions or with different number of exposures (12 vs. 24). However, those learned with 24 exposures exhibited shorter peak latency for both N170 and N250 than those learned with 12 exposures, suggesting differential strength of memory trace as a consequence of exposure frequency. During the subsequent test of recognition and generalization, the results of corrected recognition accuracy (i.e., Hit – FA) largely replicated our previous finding and showed that as exposure frequency increased, faces learned with greater expression variation yielded better recognition and generalization. Furthermore, the ERP results showed (a) a larger P1 component for 24 exposures than for 12 exposures, and (b) an interaction effect on N170 component where the difference in mean amplitude between single and multiple expressions increased as exposure frequency increased from 12 to 24, even though the overall negative deflection in amplitude was greater for the former than for the latter. Taken together, the ERP findings corroborate those obtained with behavioral measures, and unveil the neural correlates of the joint influence of expression variation and exposure frequency on face recognition and generalization.
Meeting abstract presented at VSS 2013