Abstract

Is gaze a window into the cognitive landscape of a viewer? While it is well established that visual features of the external environment guide attention, the extent to which gaze reflects our endogenous cognitive priorities remains less clear. Naturalistic contexts are ideal for studying these endogenous factors, as individuals must draw on their own conceptual knowledge bases and priorities during active visual selection. In this talk, I will present two ongoing eye-tracking studies that leverage computational language models to uncover the cognitive priorities guiding gaze in individuals with and without autism in naturalistic contexts. First, in immersive, real-world environments, eye-tracking reveals idiosyncratic, trait-like “gaze fingerprints” in individuals with and without autism, which are partly explained by the conceptual feature space of a large language model, capturing unique variance beyond spatial and vision-based models. Second, in dyadic conversations, mobile eye-tracking shows that gaze to the conversation partner’s face is modulated by the ongoing semantic context of the exchange, including linguistic surprisal – with differences between individuals with and without autism. Together, these findings position gaze as a window into the semantic and predictive processes that shape attention, providing new leverage for modeling individual differences in natural contexts.