Abstract
Confusion between mirror-symmetric images has been demonstrated in children, adults, mammals and even invertebrates. For example, children frequently confuse mirror-symmetric letters, such as ‘b’ and ‘d’. Mirror confusion has also been reported for head views when subjects failed to remember the direction of Abraham Lincoln's profile on a US penny coin. Several single-unit recording monkey studies in inferior temporal cortex and banks of the superior temporal sulcus reported the possible neural mechanisms of mirror-image confusion of simple shapes, objects and heads. However, to date no evidence for such representation has been shown in the human brain. In a functional MRI study we presented subjects with images of human faces in five different views (left/right profile, left/right half profile and front). We used multi-voxel pattern analysis (MVPA) technique to decode the direction of head view from neural activity. High-level visual areas (Fusiform Face Area and Lateral Occipital Complex) showed worse discrimination between mirror-symmetric head views (e.g., right-left profiles) and best discrimination between profiles and front faces. These findings indicate similar neural representation for mirror-symmetric head views. Notably, this pattern was not observed in early visual cortex, which showed best discrimination of mirror-symmetric profiles that decreased monotonically as a function of the angle difference between head views. These findings complement a recent single-cell recording study of face-selective patches in the macaque, which revealed large proportion of mirror-symmetrically tuned neurons in the anterior lateral face patch. Taken together, these results highlight the importance of such bi-modal tuning to mirror symmetric images as an intermediate step before the generation of an invariant representation. Finally, mirror-confusion for face views was not limited to face-selective areas, consistent with the idea that mirror confusion is a general shape processing phenomenon rather than a face-specific effect.