Visual experience is necessary for selectivity of faces over language in the fusiform

The contribution of innate constraints and experience to the development of face selectivity in human cortex is hotly debated. Unlike the sighted, people born blind report little interaction with faces in any modality. A recent study with blind adults found preferential responses to tactile faces over scenes and objects in the location of the fusiform face area (FFA) (Ratan Murty et al., 2020, PNAS). Blind adults also show responses to spoken and written language in a similar location (Kanjlia et al., 2016, PNAS; Tian et al., 2022, Cerebral Cortex). Does the FFA develop equivalently regardless of visual experience or does it respond to both faces and language in blind people? Congenitally blind adults (n=8) performed both ‘face touching’ and language tasks during fMRI. Participants felt 3D-printed models of faces, one at a time, and pressed a button if the same face repeated. In a control condition, participants touched 3D-printed scenes with different spatial layouts. Participants read tactile Braille words and felt control shapes comprised of raised dots and listened to spoken words or backwards speech (control) while performing delayed-match-to-sample. A sighted control group performed analogous tasks with visual faces/objects, visual words/false fonts and auditory words/backwards speech. We replicate the preference for face over scene touching in the anatomical location of the FFA in congenitally blind people. In the blind group, responses to faces were stronger in the left hemisphere, and these face-responsive regions also showed robust preferences to tactile Braille and auditory language relative to control conditions. By contrast, in the sighted, face-responsive regions did not respond to spoken or written language more strongly than control conditions in either hemisphere. We conclude visual experience is necessary for development of a selective response to faces over language in the fusiform. Innate constraints may bias this part of cortex for agent-relevant information.