Multiple-Object Tracking Is Based on Scene, Not Retinal, Coordinates.

Abstract

This study tested whether multiple-object tracking-the ability to visually index objects on the basis of their spatiotemporal history-is scene based or image based. Initial experiments showed equivalent tracking accuracy for objects in 2-D and 3-D motion. Subsequent experiments manipulated the speeds of objects independent of the speed of the scene as a whole. Results showed that tracking accuracy was influenced by object speed but not by scene speed. This held true whether the scene underwent translation, zoom, rotation, or even combinations of all 3 motions. A final series of experiments interfered with observers' ability to see a coherent scene by moving objects at different speeds from one another and by distorting the perception of 3-D space. These reductions in scene coherence led to reduced tracking accuracy, confirming that tracking is accomplished using a scene-based, or allocentric, frame of reference.

Attentional costs in multiple-object tracking

“…This may also be the case in the present experiments. Target losses are also more likely as the speed of the objects increases (our Experiment 3; Liu et al, 2005). Our results show that increases in speed also increase the attentional demands, which suggest that attention may be important for keeping an up to date representation of target locations.…”

“…As the speed of objects increase, the probability that the model will lose the targets also increases because targets move too far from the predicted location to be differentiated reliably from distractors. In practice, Liu et al (2005) have shown that participants do indeed produce more errors in tracking when the speed of objects in a display is increased from 1°/s to 6°/s. Though sensitivity to speed is an attribute of serial mechanisms, it would also be an attribute of parallel mechanisms with limited spatial or temporal capacity.…”

“…While the hunters get into position they must track these targets, which can be an easy task if the herd is sparsely distributed and moving slowly, but may become more difficult as the herd becomes more dense and the animals move amongst each other at a faster rate. Tracking difficulty, as assessed by the multiple-object tracking paradigm in the laboratory, also appears to depend upon the proximity of targets to distractors (Pylyshyn, 2004) and the speed at which the objects are moving (Liu et al, 2005). Though tracking has been hypothesized to be attentionally demanding Scholl, 2001), it is unclear how and when attention is used in this task.…”

Spatial updating of dynamic scenes: Tracking multiple invisible objects across viewpoint changes

“…In conditions with objects visible throughout the entire trial, we expected lower visual tracking performance with abrupt viewpoint changes, as compared with the conditions with smooth and without viewpoint changes (Huff et al, 2009;Liu et al, 2005). In conditions without viewpoint changes, we expected slightly impaired MOT performance if the objects were temporarily invisible (Keane & Pylyshyn, 2006).…”

“…Recently, however, MOT was examined with scenes in motion. Liu et al (2005) showed that smooth scene motion-such as rotation, translation, and zoom-did not impair tracking performance. Continuous visual information about scene motion was sufficient to establish correspondence between targets from one moment to the next.…”