Abstract
To interpret visual scenes, visual systems need to selectively segment or integrate multiple moving features into distinct objects or surfaces. Previous studies have found that the perceived direction separation (DS) between two components of transparently moving random-dot stimuli is wider than the veridical DS. This perceptual “direction repulsion” is useful for motion segmentation. Here we investigate the impact of motion coherence on the segmentation and integration of spatially-overlapping moving-stimuli in both humans and rhesus macaques. Two overlapping random-dot patches moved at 5°/sec within a static aperture 7.5° across. Human subjects performed a 2AFC task to compare the DS of the moving stimuli with the angle of two static lines. Using a staircase method, we measured the perceived DS at different motion coherences and direction separations. In one experiment, the DS between two motion components was either 30° or 60°, whereas motion coherence was varied from 100% to 50%. The perceived DS progressively changed from wide to narrow as motion coherence was lowered, showing a switch from direction repulsion to attraction (i.e. smaller than the veridical DS). In the second experiment, the DS was varied from 15° to 105° while the motion coherence was either 100% or 60%. Consistent with previous findings, at 100% coherence the effect of direction repulsion depended on the DS and occurred mostly at a DS less than 90°. At low coherence, however, direction attraction was present across all direction separations used. In addition, we tested one monkey subject using the 2AFC task and the method of constant stimuli. The monkey showed high sensitivity when discriminating between transparently moving stimuli with different DS. Preliminary results indicated that the monkey's perception also changed from direction repulsion to attraction as motion coherence was reduced. These findings support the idea that motion segmentation yields to integration to overcome stimulus ambiguity.
UW-Madison SMPH, Graduate School, and Dept. of Physiology.