Abstract
Face recognition models typically assume a basically serial architecture, in which (i) perceptual representations are generated and then compared to (ii) stored long-term face representations, which in turn allow access to (iii) domain-general person representations. However, recent developments seem to question this architecture. Here, we utilised the high temporal resolution of event-related brain potentials (ERP) to examine potentially separable processing stages during face and person recognition. In Experiment 1, we observed a clearly enhanced N170 for contrast negative faces, a manipulation known to disrupt face perception. Importantly, ERP familiarity effects, with more negative amplitudes for personally familiar relative to unfamiliar faces at occipito-temporal channels, were observed in a subsequent time window, starting 200 ms after stimulus onset. In Experiment 2, familiar and unfamiliar target faces were preceded by name primes of either the same or a different person. While familiarity effects were again evident from 200 ms onwards, identity-congruent names increased the effect in a subsequent 300–400 ms time window. Together, these findings demonstrate separate processing stages representing perceptual (N170), facial long-term (app. 200–300 ms), and domain-general (app. 300–400 ms) representations, in line with classic models of face recognition.
Authors
Bojdo, M. M., Zakriev, D., Schipper, M., Ciocan, M., Lidborg, L. H., & Wiese, H.
https://doi.org/10.1016/j.biopsycho.2025.109102