ERP signatures of face redundancy and identity variation in multi-face arrays

Understanding how the brain distinguishes between visually similar faces presented simultaneously is essential for models of face perception and social cognition. Using event-related potentials (ERPs), we investigated how neural responses vary as a function of image-level repetition and identity-level variability within multi-face arrays. Across two experiments, participants viewed 2 × 2 face grids that systematically varied in whether they contained identical images, different images of the same individual, or different individuals; Experiment 1 additionally includes a single large-face condition. In Experiment 1, which used unfamiliar faces and a one-back task emphasizing perceptual configuration, early ERP components were primarily sensitive to stimulus layout and face multiplicity, with layout-dependent modulation of the N170 and P200 across single-face and multi-face grid conditions. In Experiment 2, which used familiar celebrity faces and an explicit identity judgment task, variability in facial appearance and identity robustly modulated the N170 and P200, and elicited a pronounced late posterior negativity (LPN) in the 250-350 ms time window, with larger amplitudes for arrays containing greater identity diversity. Post hoc image analyses confirmed that these effects were not attributable to low-level visual differences across conditions. By dissociating physical image repetition from identity-level variability within simultaneously presented face arrays, the present findings clarify how distinct stages of face processing contribute to perceptual individuation and integrative processing in complex social scenes.