Humans are remarkably accurate at recognising familiar faces, while their ability to recognize, or even match, unfamiliar faces is much poorer. However, previous research has failed to identify neural correlates of this striking behavioural difference. Here we show a clear difference in brain potentials elicited by highly familiar faces versus unfamiliar faces. This effect starts 200 ms after stimulus onset and reaches its maximum at 400-600 ms. This Sustained Familiarity Effect is substantially larger than previous candidates for a neural familiarity marker and is detected in almost all participants, representing a reliable index of high familiarity. While its scalp distribution is consistent with a generator in the ventral visual pathway, its modulation by repetition and degree of familiarity suggests an integration of affective and visual information.

Peer-review status of attached le: eerEreviewed Citation for published item: ieseD rolger nd sngrmD frndon F nd illeyD wegn vF nd ¤ uttenergD imone gF nd furtonD eF wike nd oungD endrew F @PHIWA 9vter ut not erly stges of fmilir fe reognition depend strongly on ttentionl resoures X evidene from eventErelted rin potentilsF9D gortexFD IPH F ppF IRUEISVF

Human observers are highly efficient at recognizing the faces of the people they know (e.g., Bruce & Young, 2012) and accomplish this task many times every day without apparent effort. Yet the complexity of the task becomes evident when comparing familiar to unfamiliar face recognition (Young & Burton, 2018). It is usually easy to see

No abstract