CN530 Final Report

How a moving input stimulus becomes a direction percept via processing in the human visual system is a complex and unresolved issue which involves numerous areas of the brain. One of the principal issues in determining direction percept entails decoding of the numerous signals are honing these signals down into one percept. More specifically, how do numerous orientation, disparity, and speed signals from V1 and V2 reduce to form a motion percept in MT, presuming MT is the pooling locus of motion activity. I then give four strategies as an overview in this field for explaining various motion percept tasks: winner-take-all, vector averaging, Bayesian, and a sigmoid feedback function. I conclude by elaborating upon a model by Van Wezel and Britten which details several of these percept strategies, offers a downstream mechanism to account for behavioral visuomotor outputs, and may also support the idea of a sigmoid feedback function being implemented in direction percept and readout of MT. Introduction The ability to detect motion in the human visual system is an evolutionary boon, enhancing our species’ ability to evade predators and, in turn, visually track moving prey when hunting. So in what evolved brain area, or areas, do we code for movement across a visual scene with complex textures and objects moving in numerous directions at different depths? The middle temporal visual area (MT), which is also called V5, has been studied extensively over the past several decades and has since proven to be a key component of motion perception