Abstract
Perception of motion transparency arises when two or more surfaces move over each other with different directions or different speeds. In all cases, a clear depth percept accompanies the segmentation and integration processes, even though depth is left ambiguous in the display. We are interested here in the way depth is assigned to each surface. Stimuli consisted of random dot kinematograms representing two surfaces moving at the same speed but in opposite directions. The stimuli were sparse (2% density), restricted to a circular aperture (8 deg diameter) and the common speed was set to 2 deg/sec. The orientation of the dot trajectories was randomly varied in 15 deg steps. One surface was arbitrarily displayed with black dots while the other one used white dots. Observers had to report the polarity (black vs. white) of the surface that appeared in front. While depth is completely ambiguous in the display, we found systematic biases in depth assignment. Observers consistently chose the front surface as the one that moved to the left and down. These results can be related to previous reports of a leftward bias for simple translation (Morikawa & McBeath, 1992, Vision Research, 6, 1137–1141). Our results suggest that such biases participate to depth ordering. We discuss the implication of this biased depth assignment for the mechanism underlying motion transparency.