Vision gives us enormous adaptive advantage because it allows us to see and react to events before they reach us. This predictive advantage is most obvious for an object in motion where the current trajectory predicts future location. It is equally important when the object’s motion results from the movement of our own eyes. Here our visual system uses knowledge about the upcoming motion of the eyes to predict the future location of the object. In this proposal, we will analyze both instances as two sides of a common predictive process that operates on the maps that control eye movements and spatial attention. We propose that this predictive positioning is involved not only in the guidance of eye movements and the deployment of attention to expected target locations but also for the location at which the target is perceived, including when a target is seen at a predicted location even though it is never there. This framework of a “master map” of target locations for perception as well as overt and covert orienting is a radical departure from the standard “labeled-line” model in which active neurons throughout the visual system specify the position of a target by virtue of their receptive field locations. The predictive shifts of location for moving stimuli and moving eyes, often deviating far from the retinal input, provide a powerful means for evaluating this proposal on many fronts. We will test these predictions with behavioral, fMRI and TMS techniques in healthy and neurological patients, and neurophysiological techniques in non-human primates. According to our working hypothesis, predictive position coding is a core function of the eye movement control system and its companion spatial attention system. The results of the proposed experiments will have the potential to show that it is the properties of action that determine the perception of position, reversing the common assumption that perception guides action.