Detection of Motion under Target Uncertainty and Peripheral Presentation

We have developed a model to account for the loss in visibility of moving targets that occurs when an observer is uncertain about the target's direction of motion. The model's key features are an array of directionally selective visual mechanisms and a rule governing the mechanisms from which an observer will derive sensory data. In response to uncertainty about two possible directions of motion, the observer is assumed to use a mechanism whose peak sensitivity is to a direction midway between the two possible directions. Seven experiments, using both reaction time and forced-choice data, demonstrate the predictive advantages of this midway model over competing single-band and multiple-band models. Additionally, the experiments reveal several new properties of human motion perception: Direction and velocity information have orthogonal representations in the visual system; although motion sensitivity does not vary with direction, the precision with which small changes in direction can be recognized does, reflecting differential breadth of tuning for directionally selective mechanisms sensitivity to various directions; and motion-analyzing mechanisms are broadly tuned for direction as well as speed. (Author)