Abstract
Why can we track only so many objects? We addressed this question by asking when and how tracking errors emerge. To test the hypothesis that tracking errors are predominantly target/non-target confusions emerging from close encounters, we compared standard multiple object tracking trials with trials wherein a non-target turned a random color whenever it approached within 4° of a target. This manipulation significantly improved performance by alleviating the correspondence challenge of a close encounter. In a second experiment, we systematically varied the minimum distance for color changes, and surprisingly, we found no benefit for distances smaller than 3.5°, and similarly, no additional benefit at larger distances. Providing color information below a critical distance could do nothing to alleviate errors already accrued; and providing color information beyond that distance tried to help where help was not needed. Indeed, in a third experiment, inducing color changes whenever non-targets were between 4° and 5.5° of a target, but not below, also produced no advantage. In a fourth experiment, parametrically restricting the minimum distance at which items could approach resulted in all improvements gained at about 4°, providing independent evidence for a critical confusion distance. A fifth experiment compared tracking of four and five targets, varying the number of non-targets so that the total number of close encounters was nearly equal. This produced equal tracking performance, implying that the number of close encounters, not targets, constrain subsequently measured performance. Further experiments explored the effects of speed and target load on the critical close encounter distance. Taken together, these experiments suggest that uncertainty about target location imposes the primary constraint on tracking by leading to confusions with non-targets. But the degree of this uncertainty is fixed, and not dependent on display parameters or tracking load.