Configural and featural processing, and their role in the face network

Abstract

Object recognition can rely on the detection of individual features, as well as the spatial configuration of features, which is especially prominent in the identification of faces. We previously showed that canonical 'face effects' (e.g., inversion, part-whole) can be elicited by non-face stimuli that are identified strictly by configuration, but not in stimuli that are identified by features. Here we asked 1) whether there exist distinct neural bases that support domain-general configural vs. featural processing; and 2) how domain-specific face-areas may respond and interact with these processors. We trained adults to recognize these configural and featural stimuli, and then used fMRI to observe neural responses associated with each type of processing. Configural stimuli drove responses in superior parietal, superior frontal, and lateral occipitotemporal cortices, while featural stimuli primarily elicited responses in the ventral fusiform gyrus. Next, we identified face-areas with a separate fMRI localizer; the Occipital Face Area (previously linked to the identification of face parts) was selectively activated during recognition of featural stimuli, while the Fusiform Face Area was ambivalent, preferring neither configural nor featural stimuli. The posterior Superior Temporal Sulcus face-area, typically involved in the moveable aspects of faces (e.g. facial expressions), selectively responded to configural stimuli over featural stimuli. Finally, we observed that the functional connectivity of these face-areas to configural or featural regions mirrored this functional pattern. These domain-general processors and their differential connectivity with the face network may offer a mechanism by which face-areas use this information to accomplish their distinct roles in face perception.