Spatiotemporal dynamics of similarity-based neural representations of facial identity

Significance Humans can rapidly discriminate among many highly similar facial identities across identity-preserving image transformations (e.g., changes in facial expression), an ability that requires the system to rapidly transform image-based inputs into a more abstract, identity-based representation. We used magnetoencephalography to provide a temporally precise description of this transformation within human face-selective cortical regions. We observed a transition from an image-based representation toward an identity-based representation after ∼200 ms, a result suggesting that, rather than computing a single representation, a given face-selective region may represent multiple distinct types of information about face identity at different times. Our results advance our understanding of the microgenesis of fine-grained, high-level neural representations of object identity, a process critical to human visual expertise.