Puneeth N. Chakravarthula¹ (), Ansh K. Soni, Miguel P. Eckstein; ¹University of California, Santa Barbara, ²University of California, Santa Barbara, ³University of California, Santa Barbara

Accurate face recognition relies on discerning subtle facial feature variations. The visual system uses specialized detectors that balance position invariance and specificity to encode facial shapes and positions. Recently, we found that individuals who prefer to initially fixate near the eyes (upper-lookers) have more position-invariant neural face codes than those who fixate lower (lower-lookers). How do these groups achieve efficient face recognition despite neural code differences? To investigate this, we tested 11 upper- and 11 lower-lookers' ability to recognize faces varying subtly in shape (featural faces) or position of facial features (configural faces). Shape and configural feature sets were difficulty-matched. During a 5-alternative forced-choice recognition task, observers maintained their gaze either at the eyes or near the mouth. Further, we also tested their sensitivity to variations in feature shape and position, comparing efficiency at integrating feature variations across the face under separate single-feature conditions (eyes-only, nose-only, mouth-only). Both groups recognized featural faces equally well across fixation locations. However, lower lookers could only recognize configural faces accurately when fixating the mouth, while upper lookers did equally well, irrespective of their gaze position (looker type × fixation location F (1,21) = 19.9, p<0.001). In single-feature conditions, upper-lookers were more efficient at extracting eye position (but not shape) information than lower-lookers (t (18) = 4.3, p < 0.005). No differences in featural or configural processing efficiency were found for the nose or mouth. Finally, using an integration index (a measure of the relative efficiency for processing whole faces vs. parts), we found that upper-lookers efficiently integrate shape and configural information for both fixation locations, while lower-lookers better integrated configural information when fixating the mouth compared to the eyes. Taken together our findings suggest that upper- and lower-lookers rely on distinct mechanisms for configural face processing.