Attention in Depth: Disparity and Occlusion Cues Facilitate Multi-Element Visual Tracking

Abstract

Human observers can track up to five moving targets in a display with ten identical elements (Pylyshyn and Storm, 1988; Yantis, 1992). Previous experiments manipulated element trajectories to prevent intersections of element boundaries, evidently in the belief that transient overlaps among homogeneous elements make the task too hard. We examine whether depth cues such as occlusion (T-junctions) and disparity affect performance in a tracking task when element boundaries, as projected onto the two-dimensional plane of the monitor screen, are allowed to intersect. Elements move smoothly in depth, as well as in horizontal and vertical position, throughout a 7-second tracking period. A probe is then flashed, and subjects report whether the flash occurred on a target or on a non-target. Overlapping circular objects form T-junctions when shaded to appear like spheres or figure eight regions when rendered as disks. Two factors, disparity and T-junctions, are considered. Results from eight naive observers show that performance improves for displays with depth information (T-junctions or disparity), suggesting that depth cues are useful for multi-element tracking.