Neural representations of naturalistic person identities while watching a feature film

Abstract Recognising other people in naturalistic settings relies on differentiating between individuals (“telling apart”), as well as generalising across within-person variability (“telling together”; Burton, 2013; Lavan, Burston, & Garrido, 2019; Lavan, Burton, et al., 2019). However, previous neuroscientific investigations of face and voice recognition have tended to measure identity-related responses and representations using tightly controlled stimuli, thus under sampling the naturalistic variability encountered in everyday life. In this study, we tested whether cortical regions previously implicated in processing faces and voices represent identities during naturalistic and task-free stimulation. Representational similarity analyses were conducted on functional MRI datasets collected while human participants watched feature-length movies. Identity representations—defined as similar response patterns to variable instances of the same person (“telling together”), and dissimilar patterns in response to different people (“telling apart”)—were observed in established face and voice processing areas, across two independent participant groups viewing different sets of identities. We also explored contributions of face versus voice information to identity representations, finding more widespread preferential sensitivity to faces. We thus characterise how the brain represents identities in the real world, for the first-time accounting for both “telling people together” and “telling people apart.” Despite substantial differences to previous experimental research, our findings align with previous work, showing that similar brain areas are engaged in the representation of identities under experimental and naturalistic exposure.