In modeling spatial vision, one must begin with the fact that vision performs two distinct tasks. The first is to support visual cognition or perception, our visually informed knowledge about the identities and relative locations of objects and surfaces in the world. A second, sensorimotor, function is to control visually guided behavior. The two functions process somewhat different kinds of information. The functions are sometimes described as 'what' and 'where' systems, though the what/where distinction does not capture the cognitive/sensorimotor contrast because 'where' questions can be asked of either the cognitive or the sensorimotor systems. The difference refers more to their outputs, the cognitive system governing perception (including perception of motion and position) while the sensorimotor system handles behavior.