Vision, High-level Theory of

Abstract

High-level vision concerns those aspects of visual perception that involve stored information, and it is usually distinguished from low-level vision, thought of as driven primarily by visual input. Intermediate-level vision bridges the gap between low- and high-level vision by, for example, explicitly representing the surfaces (derived from low-level features such as edges and texture elements), which are needed to describe the three-dimensional (3-D) objects in our world (high-level vision) visually. About 60 percent of the neurons in the primate cortex appear to be responding to visual stimuli. A large proportion of them are in the extrastriate cortex, in areas involved in high-level vision tasks, supporting diverse functions such as object recognition, face perception, mental imagery, picture naming, visual categorization, scene perception, and so on. Considerable progress has been made in our understanding of these processes thanks to contributions from vision scientists and cognitive neuroscientists from all sorts of backgrounds (e.g., psychophysicists, experimental psychologists, cognitive psychologists, cognitive neuropsychologists, neurophysiologists, computer scientists). Theoretical developments in this area are the focus of this article.