Abstract
Purpose. When an ellipse is rotated in the frontal plane, it can be perceived as rigid 2D, non-rigid 2D or rigid 3D. The 3D percept, showing the stereokinetic effect, is a circular disk, tilted in depth. We assume that motion minimization plays an important role in the stereokinetic effect. Motion slowness predicts a rotation around the contour should be perceived that partially negates the rotation in the image plane. In the limit, motion slowness predicts a circle rotating off-center in the image plane is perceived not to change its orientation. Methods. 8 subjects viewed ellipses of 3 aspect ratios (0.5, 0.7, and 0.9) rotated at different speeds in the frontal plane. For the purpose of the experiment the ellipses were displayed with a dashed contour. In this case, the perceived motion of the 3D percept can be broken down in two components. The first component is a rotation of the surface normal of the circle around the viewing direction. Secondly, there is a rotation of the circle around its surface normal. Since the 3D percept is circular, this motion is invisible when the contour of the ellipse is solid. Subjects manipulated the rotation along the dashed-line contour until no added motion was perceived relative to the rotation of the stimulus. We thus obtained the subject's perceived magnitude of the second component. Results. Within subject agreement is high. Moreover, all subjects report near identical perceived rotations (std.err. 0.05 rad/s). As predicted by slowest motion, added rotation was both a function of stimulus rotational speed and aspect ratio. Conclusion. These results suggest that for stereokinetic stimuli subjects establish correspondence by a slowest motion principle. We computed a number of slowest motion predictions, using either 2D projected or 3D motion. The observed data are closely bounded by these theoretical predictions. These findings help explain the majority of the stereokinetic illusions using the slowest motion principle.