To compute learning target location by form and motion attentive visuomotor control system

摘要

We propose a new attentive visuomotor control system consisting of form (ventral) and spatial motion (dorsal) primate visual streams. Hypothetical neuronal network is also proved to support the system in terms of brain-like fashion, as well as being used to computationally obtain target location, induced by self-motion estimates and the form perception of targeted object. As a consequence, semantics of the two visual pathways (ventral and dorsal), in this paper is attempted to be revealed with Bayesian frontal lobe's model, by taking into account of learning target location by the visuomotor control system comprising of both form and motion.