Motion-in-depth uncertainty is not much worse than lateral uncertainty: evidence from continuous psychophysics

Motion-in-depth (MID) is harder to discriminate than lateral or fronto-parallel motion, probably because the use of monocular and binocular cues in MID needs the support of extra-retinal signals. On the other hand, MID is ubiquitous in many perception-action tasks in which people perform quite well (e.g. catching a fly-ball). A recent framework, continuous psychophysics (CP), allows to measure sensory uncertainties of visual attributes (e.g. position) from tracking data. Using CP, one study has shown that MID was impaired based on spatio-temporal properties of the tracking behaviour when compared to lateral motion. However, motion consisted of a random-walk, so the use of velocity was not possible. Here we ask participants (N=7) to cancel any perceived motion, by tracking the trajectory of a set of dots with the eye-head system within a VR setting (HTC VivePro @90Hz). The dots underwent an elliptical motion with lateral and depth components. The temporal frequency (W) of the movement was the same for both components and during a given trial (21 seconds) W could be: 0.10, 0.14, 0.20, 0.29, 0.41, 0.58 or 0.82 Hz. Due to biomechanical constraints, the average lateral amplitude (0.12 m) was about half the depth amplitude (0.27 m). In addition to cancel the motion, participants reported whether they were actually perceiving any motion by continuously pressing a key. Results indicate that they succeed in cancelling the motion for about 15 seconds/trial. We estimated the positional and speed measurement uncertainties by using a Kalman filter defined with the stimulus dynamics. Positional uncertainty was larger in depth (15 cm vs 10 cm) and increased more with W than the lateral. Although a similar trend was observed for speed uncertainty (means of 14 vs 19 cm/s), the differences are smaller than the ones previously reported, rendering CP a promising framework for measuring motion uncertainties.