Abstract:

The automatic grouping effect is the finding that inter-target grouping improves performance, while target-distractor grouping impairs performance during multiple object tracking (MOT). Many features can induce the automatic grouping effect, such as color, shape, interpolation, and symmetry. However, none of these features are spatial-temporal. Motion information is the principal spatial-temporal feature, and it plays an important role in multiple object tracking (MOT). Using motion information stored in visual spatial working memory (VSWM), people can not only keep track of multiple moving targets among identical distractors, but also periodically extrapolate the motion trajectory of targets. We investigated whether or not participants can use motion information as the basis of the automatic grouping effect just as they can with other feature (e.g. color, shape) during attention tracking.

To test this question, we used classic random motion as the baseline and added two new types of chasing motion in MOT tasks. One chasing motion condition was target-chase-target; the other chasing motion condition was distractor-chase-target. Given that surface features can influence the grouping effect based on motion information, we next tested the motion-based grouping effect in a condition where targets and distractors are assigned different color combinations (Experiment 2) and different shape combinations (Experiment 3). The different color combinations included identical color between targets and distractors, distinct colors between targets and distractors, and mixed colors between targets and distractors. The shape combinations were analogous to the color combinations.

Results of Experiment 1 showed that, compared to the random motion condition, tracking performance was better when targets chased targets and worse when distractors chased targets. In other words, motion information can be used as an automatic grouping cue. Results of Experiment 2 indicated that the tracking performance in the target-chase-target condition was always better than the distractor-chase-target condition when colors were identical, distinct, or mixed. However, compared with the identical color condition, the difference between chasing conditions (target-chase-target condition minus distractor-chase-target condition) was smaller in both the distinct color condition and the mixed color condition. When Experiment 3 varied the surface feature from color to shape, the results were the same as Experiment 2. The surface feature (i.e., color and shape) did not eliminate the motion-based grouping effect, but it did reduce the difference between the target-chase-target condition and distractor-chase-target condition.

From these results, we can conclude that motion information can be used in multiple object tracking and may be an automatic grouping feature. This suggests that spatial-temporal features can also induce automatic grouping, regardless of the effect of surface feature in MOT tasks. It may also imply that, although motion information is encoded automatically and in parallel during tracking, it may share visual spatial working memory resources with surface feature. Finally, motion-based grouping effect in MOT showed that the gestalt law was embodied in attentive tracking.

Key words: multiple object tracking, common fate, motion information, grouping effect