Figure 1

Some of the visual processing areas that respond to faces. Previous image enhancement techniques for improving face identity perception in AMD have targeted low level vision in early visual areas. Our caricaturing method is designed to tap potential for additional benefits from improving coding of face-shape information in mid- and high-level processing regions^17,18,19,20. Note the precise origin of caricature benefits within mid-to-high level regions is not known (although face adaptation aftereffects suggest face perception is influenced by a mix of high-level face-specific coding and more generic opponent shape coding⁵¹). Image based on Irons et al.³³.