Abstract
Recent research shows that gaze following relies on both the computation of a gazer’s line-of-sight (gaze directionality) and an understanding of their mental state (adopting their visual perspective). Here we examined how these two mechanisms operated when participants responded from the gazer’s mental perspective. We devised a task in which a central avatar gazed at a peripheral target or distractor before returning its gaze to the center. Response targets invoked either a combined (i.e., both the observer and the avatar see an 8) or a dissociated mental representation (i.e., observer sees an E while the avatar sees a 3). Participants were asked to localize the target from the avatar’s perspective. Crossing the line-of-sight and mental perspective conditions provided the test cases to examine target performance when the line-of-sight and mental perspective are spatially combined and when they are spatially dissociated. Participants overall performed with high accuracy. They were faster to localize targets in conditions in which both the line-of-sight and avatar’s mental perspective cued the target relative to conditions in which one of the two signals were inconsistent with the target’s location. Inconsistent visual perspective between the observer and the avatar invoked a larger detriment on target performance when the avatar’s line of sight was spatially incongruent with the target relative to when it was spatially congruent with the target. Thus, humans can follow gaze from another gazer’s perspective, with the computations of the gazer’s line-of-sight and their mental state operating similarly as in typical gaze following measured from the observer’s perspective.