Oliver Layton¹ (), Daniel Yu¹, Alexander Lyon¹, Maddie Puzon¹, Sam Cohen¹; ¹Colby College

Moving observers experience structured patterns of retinal motion known as optic flow. Humans can accurately judge their linear direction of self-motion (heading) from optic flow, often within 1° for nearly straight-forward movement (Warren et al., 1988). However, error increases for more eccentric headings, reaching 10° or more (Sun et al., 2020). Studies that characterize the accuracy of heading perception over the widest range have focused on the horizontal plane and report systematic biases toward peripheral (left/right) headings (Crane, 2012; Cuturi & MacNeilage, 2013). In the present study, we quantified 3D heading perception across the full range of azimuths and elevations. On each trial, observers viewed optic flow generated by simulated linear self-motion through a 3D environment in a VR headset. After the motion ceased, observers indicated their perceived heading by physically pointing to align a virtual laser, yoked to a handheld motion-tracked controller, with their perceived direction of movement. Consistent with existing work, forward headings were judged more accurately than backward headings. Contrary to horizontal-plane studies, however, we found central rather than peripheral biases: responses were shifted toward the fore–aft axis (0°/180°) in azimuth and toward the equatorial plane (0°) in elevation. To reconcile these divergent results, we conducted a control experiment in which observers judged horizontal headings either by pointing or by adjusting a dial, the method used by Crane (2012) and Cuturi & MacNeilage (2013). Strikingly, we replicated both effects: pointing judgments yielded center bias, whereas dial adjustments yielded peripheral bias. These findings provide a more comprehensive characterization of 3D heading perception and suggest that previously reported directional biases may reflect not only perceptual mechanisms, but also task-dependent response constraints introduced by different psychophysical paradigms.

Acknowledgements: The Haynesville Project