How We Recognise Faces in the Real World: Linking Perception and Person Knowledge

Recognising a person from their face is central to social interaction, yet the neural mechanisms that support robust identification across highly variable visual appearances are not resolved. Classical cognitive models propose that recognition is mediated by image-invariant face recognition units formed within the visual brain. However, converging evidence from behavioural and neuroimaging studies evidence increasingly challenges the sufficiency of this account. In this seminar, I will present findings from cognitive psychology, neuroimaging, and neuropsychology to re-examine how familiar face recognition is achieved in the brain. I will first outline the computational challenges posed by within-person and between-person similarity, and review evidence showing that fully image-invariant identity representations are not consistently instantiated within core face-selective regions. I will then consider evidence that robust identity representations instead emerge within an extended face-processing network, in which perceptual information is integrated with semantic, episodic, affective, and social knowledge about individuals. Drawing on recent work using naturalistic viewing paradigms, I will argue that familiar face recognition reflects dynamic interactions between exemplar-based perceptual representations and person knowledge distributed across cortical systems. This framework provides a mechanistically grounded account of how stable identity recognition can be achieved without invoking fully invariant visual codes.