Abstract
Almost all prior research on where people look on faces has been conducted in the lab, using fixed-head eye trackers while people view static images or movies of faces. Yet this situation is radically different from real-world face perception, where eye movements are likely to be influenced by the fact that the face you are looking at may be looking back at you. Here, we asked if the pattern of stable individual differences in face-looking behavior observed in the lab (Peterson & Eckstein, 2013; Mehoudar et al., 2014) generalizes to real-world vision. To measure in-lab behavior, subjects’ eye movements were recorded with a fixed-head eye tracker while they performed a speeded (500ms) famous-face identification task using static images. Preferred face-fixation behavior was defined as the mean initial into-face fixation location. Real-world face-looking behavior was measured with a mobile eye tracker while subjects socially interacted with others, and walked around the MIT campus, passing people in hallways and open areas. Consistent with previous work, we found large individual differences in the preferred location of laboratory-measured face-fixations, ranging from just above the brows to the mouth. Preliminary data suggests that this pattern of behavior is conserved in real-world viewing conditions. These results suggest that, at least for the critical initial into-face fixations that support identification, gaze measured in the lab accurately reflects real-world looking behavior.
Meeting abstract presented at VSS 2015