Abstract
In ‘Kanizsa’ figures, vivid subjective shapes are seen in the absence of explicit contours to define them. When two or more such figures are presented sequentially, so that the subjective shape occupies different positions, good apparent motion of the shape is usually reported. This motion percept must be mediated by a high-level process, in which form extraction precedes motion detection. Some spatial and temporal properties of this motion process are investigated. A major finding is that motion is only perceived when the time interval between successive frames falls below about 500 ms, and the duration of each frame exceeds about 80 ms.