Abstract
Purpose: We ran 4 experiments to investigate whether the human visual system uses kinesthetic information to disambiguate visual motion perception.
Methods: Subjects moved a cube, tracked by an Optotrak system, behind an occluder. Randomly oriented grating patterns were rendered moving with the cube and co-aligning with its top surface. A 6-degree round aperture was rendered 10mm above the gratings. Subjects moved the cube along a randomly chosen direction for 3500ms and reported the grating motion direction by turning a dial. The hand movement direction was obtained by fitting a line to the cube positions. Experiment 1 and 2 were rendered stereoscopically and 3 monocularly. Fixation was required in Experiment 1 but not the others. Experiment 4 was similar to 2, with 200ms delay added between the hand movement and the grating motion. Four subjects participated Experiment 1, 2 and 4, respectively and 6 participated Experiment 3.
Results: Regression of the reported direction over the hand movement and grating normal direction showed that all subjects in Experiment 1, 2 and 3 perceived the grating motion being consistent with the hand movement direction. In Experiment 4, 3 subjects' perception agreed with the grating normal direction. The other subject's perception coincided with the hand movement. The results were statistically significant.
Discussions: The experiments provide strong evidence that kinesthetic information disambiguates motion direction perception, regardless which terminators the subjects attended to (fixating or not) or the nature of the terminators (extrinsic or intrinsic). Experiment 4 controlled for the possibility that the subjects simply reported the hand movement direction. We also tried actively moving a diamond shape behind 3 invisible bars and spinning ellipses with high aspect ratios. That we didn't see coherent rigid motion in both cases suggests that active manipulation doesn't help integrating local motion signals to global motion perception.