Eye movements Affect Heading Perception from Optic Flow

Poster Presentation: Saturday, May 17, 2025, 2:45 – 6:45 pm, Pavilion
Session: Motion: Biological, self-motion

Matthew Anderson1 (), Emily Cooper1, Jorge Otero-Millan1; 1University of California, Berkeley

Self motion creates a global pattern of retinal motion signals called optic flow. The direction of self motion (i.e., heading) influences the focus of expansion of the flow, which is the retinal location from which motion vectors radiate. Previous work has reliably shown that humans use the focus of expansion as a cue for estimating heading. However, as we move through the visual world our eyes are continuously moving in order to monitor the environment (e.g., for hazards and navigational landmarks). These eye movements modulate the pattern of optic flow on the retina, and therefore the location of the focus of expansion. Although previous work has shown that heading perception is robust to some eye movements, the range of interactions between eye movements and heading perception are poorly understood. We manipulated gaze variables such as fixation position and smooth pursuit speed/direction to examine how they affect heading perception. Human observers viewed optic flow stimuli that simulated locomotion across a flat ground-plane. Meanwhile, observers pursued a dot that varied in position, speed, and motion direction relative to the optic flow’s focus of expansion. We tracked observers’ gaze with a video-based eye-tracker. After each stimulus was presented (for 1.5 secs), observers judged the heading direction. We found that heading judgements were strongly affected by fixation position, but weakly affected by the speed/direction of the fixated dot. As horizontal (azimuth) distance of fixation from the focus of expansion increased, heading accuracy significantly decreased, yet accuracy did not significantly vary with vertical (elevation) distance. We model these behavioral results with a Bayesian ideal observer that incorporates priors in heading perception and noise in both retinal cues (visual motion sensitivity) and extraretinal cues (efference copies of eye position and eye velocity).

Acknowledgements: This work was partially supported by Alcon.