Abstract
Background: Modern theorists conceptualize visual selective attention as a competition between objects for the control of extrastriate responses (e.g., Duncan & Desimone, 1995). Implicit in the notion of attention as a competition for spatially selective neural resources is the hypothesis that objects within a limited region of the visual field should be difficult to represent simultaneously, but that spatially separated objects might be attended and represented in parallel and without interference. Here, we used analysis of response time distributions (Townsend & Nozawa, 1995) in a redundant-targets shape-identification task to test this prediction. Method: Subjects were asked to identify the shape of (a) a single target object or (b) a pair of identical target objects. The spatial separation between redundant targets varied. Targets appeared embedded within a ring of task-irrelevant distractors, such that the number and density of stimuli was held constant across conditions. The only systematic differences between displays were in the number and position of attended objects. Results: When redundant target objects were adjacent to one another, performance matched the predictions of a serial or fixed-capacity parallel model. When targets were spatially separated, performance showed evidence of parallel processing with some excess capacity. Even at the largest target separations tested, however, processing capacity remained limited. Conclusions: Results indicate that attention-mediated capacity limits constrain perceptual processing within small regions of the visual field, but suggest that higher-level processing limits may also exist.