Ido Zivli¹, Benjamin Wolfe¹; ¹University of Toronto

Saccadic remapping enables stable perception of the world by providing precise information about the object targeted by the impending saccade prior to the start of the eye movement. Work with static stimuli (Wolfe & Whitney, 2015) demonstrated that, compared to covert attention, remapping provides more object-level information. Here, we examined whether remapping helped drivers detect and localize immediate hazards in dynamic road video. Eleven licensed drivers participated in a dashcam video-based hazard localization paradigm, initially fixating 13º below video centre (approximating a dashboard-mounted smartphone) and making a saccade towards the road scene on receipt of an auditory cue. Videos were shown peripherally for 1-3s and were removed from the screen either when the cue was given (covert attention), on saccade initiation (saccadic remapping) or 100 ms after the saccade was completed (first fixation). After removal, the video was replaced with a rectangular frame, and participants used a cursor to report where the hazard was. Click errors for each trial were calculated based on the smallest Euclidean distance from click position to the annotated hazard area. For each condition and participant, we computed the mean error, expressed as a z-score relative to a permuted null (i.e., errors produced by random clicking). We observed a significant main effect of condition (p < .001), with the lowest accuracy in the covert attention condition (mean z = -3.257, SD = 0.98), followed by the saccadic remapping condition (mean z = -3.85, SD = 1.19). As expected, accuracy was highest in the first fixation condition (mean z = -5.375, SD = 0.84). Our results demonstrate that saccadic remapping contributes to perceptual stability even in dynamic, changing scenes, and may be particularly useful for distracted drivers as they look back to the road.

Acknowledgements: This work was supported by an NSERC Discovery Grant to BW