Abstract
Recognition of social entities such as faces and bodies is a primary step to represent social events such as interactions. Another important requirement is to process spatial relations between those social entities, which can provide a first assessment of the type of interaction taking place. For example, face-to-face bodies are more likely to trigger representation of interaction than two bodies back-to-back. We have previously shown that regions for face and body perception in the visual cortex also encode spatial relations between multiple full-bodies, distinguishing between face-to-face and back-to-back body dyads. Within a body, however, the direction of face and rest-of-the-body is not always congruent, which raises the possibility that the two might be processed independently in visual areas. We addressed this, by examining whether: a) visual encoding of spatial relations is category-specific in the sense that face-specific areas care about relationships between faces and body-specific areas care about relationships between bodies; and b) whether there are visual areas that encode for whole-person generalizing across bodies’ and faces’ direction. We used functional MRI on healthy male and female subjects, to measure the brain activity in response to dyads of faces and head-blurred bodies presented face-to-face or back-to-back. We found higher activity for facing (vs. nonfacing) head-blurred bodies in functionally localized body-specific visual areas only, and higher activity for facing (vs. nonfacing) faces in functionally localized face-specific visual areas only. Using multivoxel pattern analyses, we identified a region that generalizes across body and face direction in the lateral occipital cortex. This new region might integrate signals from category-specific (i.e., face and rest-of-the-body) visual areas to form a representation of relationship between people.