Brett Fajen¹ (), AJ Jansen¹; ¹Rensselaer Polytechnic Institute

The ability to guide high-speed steering along a winding road depends on the skillful deployment of gaze strategies. Recent research on steering behavior concludes that gaze strategies are driven by model-based predictive mechanisms (Lappi and Mole, 2018). By this account, internal models are necessary to identify locations (waypoints) along the desired future path through which the driver intends to travel. The strongest evidence for this hypothesis comes from studies in which subjects drove through a series of waypoints positioned along an S-shaped path in a virtual environment (Tuhkanen et al., 2019, 2021). Occasionally, one of the waypoints was not rendered. Nevertheless, subjects made fixations to the location where the missing waypoint would have been. The authors concluded that gaze is driven by internal models that predict locations of upcoming steering targets based on prior knowledge of the path. First, we explore alternative model-free accounts to capturing the observed gaze behavior based on simple heuristics that predict the positions of future waypoints in the context of the highly repetitive Tuhkanen steering task. Second, the previous studies aimed to demonstrate an influence of prior knowledge under conditions in which the upcoming waypoint was missing (i.e., information was absent). This leaves open the question of whether such knowledge also plays a role when waypoint information is available. To address this question, we replicated the Tuhkanen study with one major difference. Rather than introducing occasional gaps between waypoints, we occasionally shifted waypoints off the path by a small amount. Analyses focused on the accuracy and timing of fixations to incongruent versus congruent waypoints, which would be expected to differ if gaze is driven by model-based predictions but not if they are based on currently available information. We discuss the implications of these findings for theoretical accounts of gaze behavior during steering.

Acknowledgements: NSF 2218220