Motion Distortion Demonstration
The display shows two pairs of linear gratings that translate in opposite directions. Try fixating on the center fixation point (center of the image, though any of the three fixation points will work). Notice that there are two briefly presented flashed lines near the top of the image that straddle the gratings. Most people (>90%, hundreds have observed at conferences) perceive the flashes misaligned: the right flash appears below the left flash. Also notice the two flashed lines toward the bottom of the image. Most people report that the flash on the right is above the flash on the left. Both pairs of flashes are actually physically aligned. You might also try fixating on the upper or lower fixation points, which are in fact precisely aligned with the two pairs of flashes, respectively.
The flashes always appear misaligned in a direction consistent with the nearest direction of motion. Yet, all of the flashes are physically separated from the motion and are stationary (both perceptually and physically). In several experiments, we established that motion over 50 degrees away from stationary flashes cause the flashes to appear displaced in position. Why would motion in one region of visual space shift the positions of remote stationary objects? And how could a physically and perceptually stationary object be displaced in position without appearing to move?
To explain this phenomenon, we have suggested that motion signals distort the representation of visual space. Before positions are assigned to stimuli, motion is integrated over large areas of the visual field and directly influences the position coding of the stimuli. This may explain why motion shifts the positions of both stationary and moving stimuli.
To make the motion as smooth as possible, this demo is extremely small. Slow connections or inadequate memory devoted to your web browser may cause some choppiness, and may reduce the size of the effect. If the motion is not smooth, try quitting other applications, or try using internet explorer or AOL. For reprints or questions, please e-mail David Whitney at whitney@wjh.harvard.edu.
Links: