Motion capture and visual attention: Comments on Culham and Cavanagh (1994)

In 1987 we described a novel illusion involving interactions between motion and chromatic signals in the visual system. A yellow square was displayed on an equiluminous grey background and a cluster of small black dots was superimposed on the square. When the cluster of dots was moved left right, the yellow square also appeared to move with it in the same direction an example of a class of illusion that we have dubbed "motion capture" The illusion does not depend on the presence of chromatic edges and can be seen even with a low contrast light grey square on a dark grey background. Indeed, an especially striking version of the illusion can be produced using a simple 2-frame apparent motion square in which a cluster of 5 spots was presented briefly in Frame 1 followed by the same cluster shifted horizontally in Frame 2. If one of the dots was occluded by a stationary opaque white square in Frame 2, the correspondent dot appeared to move horizontally as though it was being "dragged along" by the surrounding dots-even though it appeared only in Frame 1 and was just blinking on and off. We suggested that the motion signals from the low spatial frequencies associated with the surrounding dots was somehow being spontaneously "attributed" to the single unpaired dot in the middle (Ramachandran, Inada & Kiama, 1986). Our work also suggests that motion capture: (a) can be seen simultaneously in opposite directions and, therefore, cannot be based on eye movements; and (b) is sensitive to certain topological image characteristics such as "inside" vs "outside" and to the distinction between figure and ground (Ramachandran, 1985), e.g. if a jumping illusory square is superimposed on stationary dots, the dots inside the square get captured but not the ones outside. This sensitivity to figure and ground is important for it poses a challenge to computational models of motion capture such as those based on cooperative algorithms or winner-take-all schemes (Bulthoft, Little & Poggio, 1989). Motion capture can also be modulated visual attention. To demonstrate this we used a 2-frame apparent motion display similar to the one described above except that we had two clusters of dots-a red one and a green one-superimposed on each other and jumping simultaneously in opposite directions, with a single unpaired dot of neutral color in the middle that appeared only in Frame 1. In this display, if one paid …