Abstract
Processing of motion parallax relies on the ability of the visual system to segregate locally different motion signals and integrate them over space and time in order to reconstruct a 3-D structure. In a series of experiments, we evaluated the thresholds for differential motion detection and their dependences upon the spatial structure of the optic-flow field.
Visual stimuli were large moving random-dot displays (80 deg × 60 deg). Each dot was randomly attributed one of two velocities whose difference ranged between 0% and 80% of their average velocity, which ranged between 2 and 64 deg s−1. Stimulus duration was from 130 to 1040 ms. Subjects were instructed to stare at a central fixation mark and had to decide whether the display specified one or two surfaces. At a duration of 130 ms, subjects needed more than 40% difference between the two velocities to reach a 75%-correct detection criterion. Thresholds reached a minimum value of about 20% at a duration of 500 ms. Thresholds always increased with the average velocity and were not affected when a static form segregation cue was added to the displays. They were always larger than thresholds for velocity discrimination in successive displays (about 10% at 260 ms). We finally investigated the spatial properties of differential velocity detection. The visual field was divided into horizontal stripes of equal widths. Adjacent bars alternately contained random dots moving with one of two velocities. Thresholds for differential velocity detection increased monotonically when the bandwidth decreased, up to an asymptotic value, equal to those observed with a transparent display.
Get full access to this article
View all access options for this article.