Abstract
Humans are highly efficient at recognising familiar faces, independent of whether these faces are known from real life (i.e., personally familiar faces) or via media (e.g., celebrity faces). It is unclear, however, whether recognition in these two cases relies on the same type of facial representation, and some researchers have suggested qualitative differences between media-based and real-life face familiarity. Here, we used event-related brain potentials (ERPs) to test whether early visual recognition (as reflected in the N250 familiarity effect; 200-400 ms after stimulus onset) and later stages of integrating visual with additional person-related information (as reflected in the Sustained Familiarity Effect [SFE]; 400-600 ms) are different for real-life versus media-based familiarity. In two experiments, we presented multiple “ambient” face images (varying naturalistically in lighting, viewing-angles, emotional expressions etc.) of a highly personally familiar person, of the participant’s favourite celebrity, and of an unfamiliar person. We consistently observed clear N250 familiarity effects, with more negative amplitudes for both personally familiar and favourite celebrities relative to unfamiliar faces. A similar pattern was observed in the subsequent SFE time window, which additionally showed larger familiarity effects for personally familiar relative to favourite celebrity faces (Experiment 1). In a third experiment, we compared favourite celebrities with disliked celebrities to test for a potential influence of valence on ERP familiarity effects. Again, clear effects were observed for all familiar face IDs, with a reduced SFE for disliked relative to favourite celebrities. Overall these findings suggest that ERP familiarity effects are not qualitatively different for personal versus media-based familiarity. Instead, increasing familiarity may result in more robust representations that allow recognition from increasingly variable images and irrespective of the valence of one’s personal feelings.