Abstract
We studied a novel phenomenon indicating that the integration of extra-retinal information and the optic flow is a necessary but not sufficient condition for slant constancy during active head motion; congruency between monocular and binocular cues to depth is also required.
In two experiments, we measured observers' performance in a rotation-detection task during active vision. Two viewing conditions were compared: binocular vision with null disparity (same image projected to the two eyes) and monocular vision. Static or rotating slanted planar surfaces were simulated with 90/270-deg tilt (Experiment 1) and 0/180-deg tilt (Experiment 2). Observers produced oscillatory lateral head movements which were recorded by an Optotrack Certus system. The position of random dots on a CRT monitor were updated in real time, simulating the correct projection of a random-dot planar surface on the image screen by taking into account the head position of the observer and the rotation of the surface.
Full perceptual constancy was found for simulated stationary surfaces under monocular vision, but not under binocular vision with null disparity. A stationary surface appeared to be stationary when it was viewed monocularly; however, it appeared to be rotating when it was viewed binoculary with null disparity. Moreover, in both experiments, rotation sensitivity was larger for monocular than for binocular vision. Response bias was shifted in opposite directions, with an overall tendency to judge surfaces as being stationary in monocular vs. rotating in binocular vision.
Monocular vision supports full slant constancy. Binocular vision with null disparity (1) disrupts the effect of extra-retinal signals produced by head movement, and (2) induces the perception of an apparent rotation counter to the heading direction.