Temporal dynamics of gaze behavior around system activation: The role of trust, mental model, and prior experience in automated driving

Transitions in automated driving require drivers to dynamically recalibrate attention and trust. In this paper, we investigate how gaze behavior evolves around system activation and examine the influence of different trust layers, users´ mental models, and prior experience with automated driving systems (ADS). We conducted a longitudinal naturalistic study with twenty-two participants (10 with prior ADS experience), each completing three automated driving sessions. Across a total of 289 system activations with drives longer than 1:45 min, we captured eye-tracking data from 90 s before to 90 s after each activation. Situational trust was assessed through self-report immediately following system engagement, while mental model and dynamic learned trust were measured post-drive. Our findings reveal significant temporal shifts in gaze allocation: system-focused attention immediately following activation decreased steadily as drivers redirected attention toward non-monitoring regions. This temporal adjustment was correlated with both trust and mental model accuracy. Prior ADS experience further moderated these effects. The findings advance the understanding of how trust, mental models, and gaze behavior co-evolve during automated driving and offer valuable implications for developing objective, continuous trust measurement methods.