Abstract
The perception of depth motion is critical for interacting with a dynamic three-dimensional (3D) world. Surprisingly, when only binocular cues are available, continuous tracking of depth motion is impaired relative to frontoparallel tracking (Bonnen et al., 2017). This impairment was thought to be a result of both geometric constraints and the sluggish nature of disparity processing. Here, we used Bonnen et al.’s continuous 3D tracking task in a rich VR environment to test whether depth tracking deficit persisted in the presence of additional cues to 3D motion. A target object moved in a Brownian random walk through a 3D VR environment (VIVE Pro Eye Headset, 1,440 x 1,600 pixels/eye, 110-degree FOV, 90 Hz refresh rate). Participants continuously tracked this target with a “cursor” object using a hand controller (Valve Index). In the “sparse” environment (a replication of Bonnen et al.), only the target and cursor were visible and only binocular depth cues were available. The “rich” environment contained textured surfaces and objects as depth references. We tested the relative contributions of various cues to 3D motion (motion parallax, retinal size change, disparity, and shadows) in a fully crossed design. Performance was quantified using cross-correlograms between target and response velocities (Bonnen et al., 2015). In the sparse condition, tracking for the depth component was relatively sluggish, replicating Bonnen et al. (2017). With all cues present, depth tracking improved slightly in some cases, but was always worse than frontoparallel tracking. The time lags at peak correlation for the depth CCGs were typically 500 ms or more, while frontoparallel lags were about half that. In the monocular conditions, depth tracking was abysmal with the peak correlations close to zero. These results indicate that while monocular cues can perhaps slightly improve the perception of depth motion, binocular cues are vital.