Abstract
A dominant idea in the motion perception literature is that 1st-order motion is specified by spatiotemporal changes in raw luminance; i.e., motion energy (ME). The underlying principle is that motion can be perceived without extracting the boundary and surface information that determines a moving object's form. This follows from the dispersal of boundary/surface features across the spatial frequency spectrum, and the detection of motion by spatiotemporal filters responsive to a limited range of spatial frequencies. We propose that these edge and surface features are responsible for perceiving an object's motion and its form (see also the formotion model; Francis & Grossberg, 1996), and further, that there are dual pathways, one for object motion and the other for ME. Support for this proposal comes from a stimulus composed of 6 adjacent rectangles. The rectangles vary in luminance such that ME and changes in edge contrast specify motion in opposite directions. Consistent with Wertheimer's distinction, “objectless” phi motion is perceived in the ME-specified direction, predominantly for brief frame durations (high speeds). Beta motion entailing the perception of a moving edge (as in the line motion illusion) is perceived in the direction specified by changes in edge contrast, predominantly for longer frame durations (slower speeds). Decreasing the width of the rectangles reduces edge-based motion and introducing thin gaps between them completely eliminates it. ME is minimally affected by these changes. Conversely, ME perception is substantially weakened by reducing the number of frame-changes, whereas a single frame change suffices for perceiving edge-based motion.