Veridical perception of global motion from disparate component motions.

Although it is in principle possible to determine the direction of motion of an object by combining the motion of its one-dimensional oriented contours (Fennema CL, Thompson WB. Comput. Graph. Image Processing 1979;9:301-315) there is still much debate on whether human observers can do so. The Intersection Of Constraint (IOC) rule proposed by Adelson and Movshon (Adelson EH, Movshon JA. Nature 1982;300:523-525), although compatible with the veridical object's motion, was challenged by recent psychophysical data obtained with type II plaids or lines moving behind apertures: perceived direction of motion is biased toward the vectorial average of the component motions, rather than in the direction predicted by the IOC rule. Since the velocity predicted by the vectorial rule is inconsistent with the physical velocity, its use leads to the puzzling prediction that the perceived position of a moving object becomes inconsistent with its actual position. In the present paper, the perceived path of a figure defined by its one-dimensional contours and moving behind apertures along a circular trajectory is compared with the discrepant predictions of the IOC and of the Vectorial model. The results show that the perceived path is close to veridical with these stimuli, therefore challenging the idea that the visual system uses a vector averaging rule.