Are stereo and motion information to 3-D structure linearly combined?

In this study we investigated whether motion and stereo information to 3-D structure are combined in a linear fashion as predicted by most of the theories of cue combination. Observers binoc- ularly viewed a motion and stereo specified random-dot planar surface. The observers' task was to place a probe dot on the perceived surface. The perceived position of the probe dot could be adjusted by modifying (i) its 2-D velocity, (ii) its binocular disparity, and (iii) a combination of the two, with respect to the planar surface. In different conditions we varied the relative intensity of the velocity and binocular disparity of the probe dot. If the hypothesis of linear combina- tion of motion and disparity information is correct, then settings derived from different relative intensities of motion and disparity information must lie on a straight line in the velocity - disparity space. The results, however, show that this is not the case. This suggests that motion and stereo modules of 3-D shape reconstruction are not independent.