Abstract
Purpose: The rotation of a kinetic depth (KD) figure in parallel projection is perceptually ambiguous. Additional information, i.e., binocular disparity, can disambiguate the perceived rotation. As recent work suggests eye movements have a role in depth assignment in motion parallax, we wondered if eye movements might also disambiguate the perceived rotation of KD figures. That is, for motion parallax, retinal motion in the same direction as the eye movement is assigned to near depth while motion in the opposite direction is perceived as farther than fixation. If this were to apply to KD, then a KD figure translating from left to right should be perceived as rotating “front to right” and translation to the left generating rotation “front to left”.
Method: Stimuli were rotating, random-dot KD figures. The task was to report the figure's perceived direction of rotation. Direction of rotation (front to right/left) and direction of translation (left/right) varied between trials. In the first and third conditions (C1 & C3) observers used LCD glasses and frame sequential presentation to view “stereo” KD figures. In conditions two and four (C2 & C4) observers viewed KD figures monocularly. Observers fixated a stationary spot just below the figure's path in C1 and C2, and fixated a point translating along with the figure in C3 and C4. Trials from C3 and C4 were randomly interleaved in an attempt to obscure to the observers the covariation of translation and rotation predicted in the KD trials.
Result: Binocular disparity was effective for disambiguating KD rotation: 91% with stationary eyes (C1) and 96% with eye movements (C3). Eye movements were equally effective in disambiguating KD rotation: 98% in C4. It was unexpected that translation of the KD figure, without eye movements, was almost as effective: 90% in C2.
Conclusion: Eye movements and lateral translation disambiguate KD rotation as well as binocular disparity. The actual effect of each remains to be determined.
Support: NIH grant EY12541