Abstract
Relatively little is known about the neural computation of 3D motion (compared with that of 2D motion and static disparity). The perception of 3D motion has been shown to rely upon a combination of a disparity-based cue (disparity changes over time), and a velocity-based cue (velocity differences seen by the two eyes). To better understand the mechanisms underlying 3D motion processing, we performed a series of psychophysical experiments to measure the spatial characteristics of 3D and 2D motion perception.
We employed 3D dynamic random dot stereograms (3DRDS) and manipulated both the 3D or 2D motion strength and the spatial and temporal structure (e.g., stimulus area, dot lifetime). 3DRDSs consisted of dots randomly positioned within an annular volume. Brownian dot motion was created by randomly sampling motion orientation from a radial Gaussian distribution in the x-z plane, centered on the motion direction for that particular trial. The magnitude of z-motion for each dot was then derived from the sine of the Gaussian sample. Motion strength was controlled by varying the standard deviation of the Gaussian direction distribution.
We measured accuracy in a motion direction-discrimination task as a function of motion strength and stimulus eccentricity for both 3D and 2D displays. Overall, thresholds were lower for 2D displays than 3D displays, replicating the phenomenon of stereomotion suppression (Tyler, 1971). 2D motion discrimination accuracy did not depend strongly on stimulus eccentricity. However, 3D motion accuracy decreased with eccentricity, but at a rate far less severe than one would expect from eccentricity effects on stereoaccuity. Such dissociations can reveal the relative precedences of the velocity- and disparity-based cues across the visual field. The characterization of spatial profiles of sensitivity provides a means for characterizing how multiple perceptual cues are integrated to compute 3D motion.