Modality-specific encoding of conceptual person identity in the Fusiform Face Area

Abstract

I employed multi-voxel pattern analysis to fMRI data from two functionally defined clusters in the Fusiform Face Area (FFA); mFus and pFus. Stimulus information was decoded from hemodynamic response patterns evoked by three fictional identities. These were comprised of sets of semantic person information and corresponding faces. By presenting identities with overlapping visual and semantic features we could attribute successful decoding to the differentiating feature. Our data suggest a functional differentiation between pFus and mFus. Analysis of response patterns in pFus indicated that this cluster is engaged in perceptual analysis during face presentation and in retrieval of the corresponding face representation during presentation of semantic stimuli. In mFus semantic items appear to evoke a single corresponding face representation whereas during face perception this cluster is likely to be engaged in retrieval of semantic features by activation of a collection of corresponding visual representations. I propose that mFus supports conceptual knowledge of people by activating visual features that correspond to semantic categories.