Abstract
Three experiments were conducted to determine the effects of surface markings on the perceived direction of motion in structure-from-motion displays. In Experiment 1, two objects were presented side by side. The standard object was a sphere covered with approximately 82 crosses. One segment of each cross was parallel to the axis of rotation, which was at an angle of −40, −20, 20 or 40 relative to the horizontal. The second segment intersected the first segment at angles of 30, 60, 90, 120 or 150 . The comparison object was a sphere covered with a random dot pattern. The observer's task was to adjust the motion direction of the comparison sphere until it appeared to match that of the standard. We found a consistent bias in the judged motion direction of the standard sphere, with the perceived motion direction shifted towards the bisector of the acute angle between the segments. In Experiment 2, the markings on the standard sphere consisted either of one set of segments from the crosses on one of the standard spheres in Experiment 1, or both sets of segments, but with the segments not intersecting. The bias with one set of segments was similar to the bias with crosses, but the bias with both sets of segments present was much smaller. In Experiment 3 the rotating spheres were replaced with translating planes, with markings and motion directions similar to those in Experiment 1. Perceived motion direction of the planar stimuli was biased towards the bisector of the acute angle of the crosses and was similar to the bias observed with crosses in Experiment 1, indicating that the bias is not specific to curved surfaces. Overall, the perceived motion direction of an object is influenced by the orientation of its surface markings. An explanation based on vector averaging accounts for the overall results, but not for the larger bias found with crosses compared to non-intersecting segments at the same orientations.
Supported by NIH Grants EY-12437 and TW-05753.