Over the years I have discovered various illusions of motion. The bicycle-spoke, reversed-phi, crossover, streaming-dot, boogie-woogie and footsteps illusions tell us about early, bottom-up processing of neural signals of motion. The global-motion and chopstick illusions tell us about the top-down processing of these neural signals that recovers moving objects.

Low level, bottom-up processes. The outputs of Reichardt motion detectors are strongly influenced by the luminance and contrast of the motion signals. For instance, in the bicycle-spoke illusion, stationary grey lines flanked by regions of changing luminance appear to move. In crossover motion, a black and a white line exchange luminances; does one see a white line jumping, or a black line jumping, or two lines flickering in place? The answer depends upon the surround: on a light surround it is the black line, and on a dark surround the white line, that appears to move. So the line with the higher contrast is seen as moving. The lines can be embedded in longer, collinear lines (White’s effect), and I find that the luminance of the embedding lines is 3 times as effective as the surround luminance in driving the seen motion. The streaming-dot and boogie-woogie illusions show that the perceived speed of a line that moves orthogonal to its own length can be strongly affected by dots that stream along the length of the line. Further strong influences of contrast are demonstrated by the footsteps illusion, in which a light and a dark square move at constant speeds across stationary black and white stripes. The dark square has a high contrast when it crosses a white stripe, and spears to speed up. It had a low contrast with it crosses a black stripe, and appears to slow down. The opposite is true of the light square.

Higher level, top-down processes determine the global or local organization of certain patterns of moving dots; and also the perceived motion of orthogonal lines that slide across each other in the chopsticks illusion – an illusion related to the aperture problem and to plaid perception. A vertical and a horizontal chopstick overlap and move along counterclockwise paths without rotating. The intersection of the lines actually moves clockwise but it appear to move strongly counterclockwise. Probably, the motion of the line tips (terminators) propagates along each line and is blindly assigned to the central intersection. I combined this chopstick illusion with the flash-lag effect, by flashing up a spot of light on an intersection that actually moved in one direction but appeared to move in the opposite direction. Results: The flash-lag effect was appropriate to the physical, not the perceived direction of motion.

I hope that all these illusions will shed light upon the analysis of movement at different levels in the visual system. I thank my collaborators Patrick Cavanagh, Brian Rogers, David Smith, Pete Thompson and others.