Abstract
The geometry underlying the recovery of depth order from motion is that the angular velocity originating from the rigid translation of objects relative to the observer is inversely proportional to the distance from the observer to the objects. Previous studies revealed that the calculation of angular velocity requires either retinal (Braunstein & Andersen, 1981, P&P, 29, 145–55; Naji & Freeman, 2004, Vision Res, 44, 3025–34) or extraretinal signals (Nawrot, 2003, Vision Res, 43, 1553–62). We explored how the visual system integrates retinal signals with extraretinal ones to determine depth order. We used a stimulus in which four rows of horizontally moving random-dots had a common motion component and a relative motion component. The direction of common motion relative to the moving pursuit point was opposite to that of the pursuit point, and the velocities of the common motion and of the pursuit point were manipulated independently. The relative motion component consisted of the dots in all four rows moving in the same direction but the dots in the first/third rows and the second/fourth rows had different speeds. Observers were required to report which rows appeared in front. Possible cues for scaling the amplitude of eye-movements were eliminated by presenting the stimulus on a black screen in a darkened room so that the frame of the screen was invisible. We tested the prediction that the depth order produced by retinal motion (i.e., dots moving faster on the retina appear closer) can be reversed by an eye-movement velocity signal that exceeds that of the retinal motion (i.e., dots moving slower on the retina now appear in front). The prediction was confirmed with three different velocities of common motion. The result suggests that object-velocity relative to the head is calculated by adding eye-movement velocity to retinal velocity and is processed for determining depth order from object-produced motion parallax.
Supported in part by JSPS, NICT, and NSERC