Alignment in Anticipation Drives Successful Coordination in Dynamic Interactive Tasks

The ability to coordinate with others depends critically on anticipating and adapting to each other’s actions. While predictive mechanisms are central in theory, empirical studies often rely on simplified tasks that overlook the sensory, motor, and social complexity of real-world interaction. We introduce a quantitative framework that integrates multimodal tracking of eye and body movements and heart rate with machine learning and Bayesian models to identify predictors of coordination in a fast-paced ball-hitting task. Coordination success depended primarily on similarity in partners’ predictive styles rather than on general anticipatory ability. Well-coordinated pairs showed similar anticipatory saccades in individual tasks, enabling more efficient prediction of self and partner actions during joint tasks. By contrast, demographic variables, overall physiology, motor activity, and self-assessments were minimally linked to success. By reverse-engineering behavioral signatures, our framework identifies alignment in action anticipation as a key mechanism of coordination in both human–human and human–robot interaction.