Abstract
Human subjects can simultaneously track up to 5 dots, when presented with an array of dots moving in a random manner (Pylyshyn & Storm, 1988, Spatial Vision, 3, 179–197). This study investigates to what extent positional information is degraded as the number of tracked trajectories increases.
In pilot experiments, on each trial, a single dot in apparent motion, followed a diagonal linear trajectory from the lower-left of the screen to the upper-right. At a horizontal position indicated by fixed vertical markers, the trajectory deviated either clockwise or anticlockwise; the observer reported the perceived direction of deviation. Using a method of constant stimuli, deviation thresholds at the 84% point were estimated. The length of the trajectories and the dot velocities were adjusted to obtain optimal thresholds (typically 2°).
In the main experiments, the actual number of dots moving (N) was varied, each having the optimal parameters determined above. The average orientation of the trajectories was fixed between trials, but within a trial the N trajectories were jittered about the average; deviation from parallelism could not be used as a cue. In one set of experiments each of the N trajectories deviated the same amount, and thresholds were only mildly affected by variation in N (typically 10° for N=10). In another set only one of the N trajectories deviated; here thresholds increased rapidly with N, with thresholds as large as 30° for N = 4, and with most observers performing unreliably when N was 6 or more.
Severe loss of positional information results when multiple trajectories are tracked. This cannot be attributed to inter-trajectory interference, since thresholds were only mildly elevated when all trajectories deviated. While the identity of up to 5 dots can be reliably tracked, memory of their trajectories is severely compromised. Tracking deviations in trajectories of dots is far more demanding on attentional resources than tracking their identities