Abstract
Visual motion interacts with spatial location to produce several phenomena in which static or moving objects are mislocalized in space. Motion is processed in several stages from local estimates to global percepts. Previous studies have used drifting gratings or Gabors (drifting sinusoids windowed by a stationary Gaussian) in which the local and global velocities are necessarily the same. It is therefore not clear at which stage motion information affects perceived position. If the spatial shift is a property of cortical simple cells in primary visual cortex (Fu et al. Journal of Neuroscience, 2004. 24(9): 2165–2171), where only local motion information is represented, we would expect motion shifts to be determined by local motion signals. Rectangular grids of Gabor patches were generated with either vertical gratings and equal drift speeds (3 °/sec) or randomly oriented gratings with drift speeds that vary with the sine of the angle between carrier orientation and the global motion direction. Patches above fixation drifted leftward and below fixation rightward, or vice versa. The rectangular grids (7 high; 7, 3 or 1 element wide) were offset horizontally and subjects reported whether the top array was shifted to the right relative to the bottom array. The offset was varied systematically from trial to trial to generate a psychometric function. The 50% point provided a measure of the perceived spatial shift. We found no significant difference in the magnitude of the shift for the parallel and random Gabor arrays that might indicate a direct influence of the local velocities. We also found that a larger area of motion tends to induce a larger shift in position. Our results indicate that motion information affects perceived position only after motion integration has taken place.