Abstract
Face recognition is thought to rely on neural mechanisms in the right hemisphere and a corresponding left visual field (LVF) processing advantage. Here we report results from several experiments which lead us to conclude that the LVF advantage for face recognition reflects the splitting of identities between the retinal hemifields for centrally viewed faces. In one experiment we verified the existence of the LVF half-face bias using a two-alternative forced choice (2AFC) task. Observers viewed unfamiliar faces and subsequently indicated which of two symmetric faces, created by mirroring each half of the previously viewed face, best matched the initially viewed face. Consistent with previous reports, observers were more likely to match the initial face to a symmetric comprised of the mirrored LVF half. In another experiment, using a similar method, we found that the LVF half-face bias occurs similarly for chimeric faces comprised of two easily discriminable half-faces (i.e., half-faces from different individuals, merged at the vertical meridian). In another experiment we used an adaptive (staircase) method to measure the LVF half-face bias for chimeric faces comprised of famous face pairs morphed to different degrees in each hemifield. In a final experiment, we found that the LVF bias does not occur for half-faces viewed in isolation (i.e., halves of occluded faces viewed in either hemifield). Taken together, our results show that the LVF half-face bias only occurs when a centrally viewed face is visually split between the two hemifields. This bias implies that each half of a face is perceived to have a different identity, with the identity of the half-face in the LVF dominating the other half. We refer to this phenomenon as the visual splitting of identities during face recognition.