Abstract
The ability to see 3D motion utilizes two forms of dynamic binocular information. As most vision scientists would expect, the visual system can exploit temporal changes in conventional binocular disparities to extract 3D direction. But the visual system can also use a velocity-based cue (the ‘inter-ocular velocity difference’) by exploiting the different velocities projected upon the retinae when an object moves through depth. In this talk, I will describe a series of psychophysical experiments demonstrating that this velocity-based cue not only makes a distinct contribution to 3D motion perception, but that it serves as the primary binocular source of information for a wide range of ecologically-important motions through depth. A special stimulus (the ‘dichoptic pseudoplaid’) then shows that this inter-ocular comparison of eye- specific motions operates upon later stages of motion integration (pattern motion), instead of early stages (component motion). Results from neuroimaging suggest that these 3D motion computations occur in human MT (and MST). I will finish by discussing how 3D motion signals could be present in MT, despite a longstanding tacit assumption that MT neurons encode only frontoparallel motion direction and static disparities.