Charting the Silent Signals of Social Gaze: Automating Eye Contact Assessment in Face-to-Face Conversations

Abstract

Social gaze is a crucial yet often overlooked aspect of nonverbal communication. During conversations, it typically operates subconsciously, following automatic co-regulation patterns. However, deviations from typical patterns, such as avoiding eye contact or excessive gazing, can significantly affect social interactions and perceived relationship quality. The principles and effects of social gaze have intrigued researchers across various fields, including communication science, social psychology, animal biology, and psychiatry. Despite its significance, research in social gaze has been limited by methodological challenges in assessing eye movements and gaze direction during natural social interactions. To address these obstacles, we have developed a new approach combining mobile eye tracking technology with automated analysis tools. In this paper, we introduce, validate, and apply a pipeline for recording and analyzing gaze behavior in dyadic conversations. We present a sample study where dyads engaged in two types of interactions: a get-to-know conversation and a conflictual conversation. Our new analysis pipeline corroborated previous findings, such as people directing more eye gaze while listening than talking, and gaze typically lasting about three seconds before averting. These results demonstrate the potential of our methodology to advance the study of social gaze in natural interactions.