The invisible copilot? Assessing task engagement and team dynamics in a virtual flight environment using eye-tracking metrics

Gaze behavior represents a vital predictor of flight performance in pilots operating an aircraft. However, its role in collaborative settings with a copilot has been largely disregarded. Our study addressed this gap by simulating a flight environment with role division into a pilot flying (PF) and pilot monitoring (PM). Analyses of linear mixed-effects models revealed that different eye-tracking metrics significantly differed between the PF and the PM. Additionally, classification analysis revealed that these eye-tracking metrices were suitable for discerning between PF and PM with high predictive accuracy (97%) and precision (97%). We discuss implications for a role-based user state diagnosis for adaptive assistance system design in the cockpit, thereby advancing aviation safety.