Resource flexibility and strategy in multiple object tracking
Date
2017
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
University of Delaware
Abstract
Tracking capacity is typically defined by a limitation of 3-4 objects, suggesting
that there may be a structural constraint on attentional resources for tracking
(Pylyshyn & Storm, 1988). In this view, tracking capacity is item limited and the
amount of resources allocated to each object is fixed. However, studies have found
that tracking limits can be extended to 8 or more objects with appropriate reductions in
velocity (Alvarez and Franconeri, 2007). Contrary to the discrete resource account,
these findings suggest that tracking may depend on a continuous limited-capacity
system in which a finite resource is divided among tracked objects in varying
amounts, depending on tracking difficulty. The present set of experiments examined
the nature of the resource limitations observed in multiple object tracking. ☐ The continuous resource model posits that participants can only increase the
number of objects tracked by reducing the amount of resource allocated to each
tracked object with accompanying reductions in the spatial precision of tracking. This
prediction was evaluated using a novel version of the MOT task in conjunction with a
mixture model to obtain separate estimates of the number of objects tracked and
tracking precision as the number of objects to-be-tracked was varied between 1 and 6.
The results showed marked individual differences at high tracking loads, where half of
the participants tracked only a few targets with high precision while the remaining half
tracked many targets with low precision. Experiment 2 explored whether these
individual differences reflected fixed differences in the underlying tracking
architecture or different strategies that could be freely adopted by any observer.
Participants were instructed and incentivized to maximize the number of objects
tracked at the expense of precision to determine whether tracking performance is
flexible to strategy. Many but not all observers were able to flexibly trade off precision
in favor of a larger number of tracked targets. These results are consistent with models
that point to a limited resource whose allocation is flexibly determined by tracking
demands. ☐ The flexibility of attention allocation was further tested in Experiment 3 to
examine whether resources can be endogenously varied between targets. High priority
targets were tracked with greater precision than low priority targets, indicating that
differential amounts of attention could be distributed between tracked objects.
Experiment 4 examined whether tracking resources are hemifield specific or if a
common pool is shared across visual fields. Although tracking two objects was just as
precise as tracking a single object when they were presented bilaterally, distractors were
more likely to be mistaken for tracked objects with increased set size regardless of
spatial arrangement. The results support a bilateral advantage in tracking, but not
independence. Experiment 5 investigated whether people attend to all tracked objects
simultaneously or if each object is attended serially. ERPs were used to measure online
processing of task-relevant probes on moving objects. Consistent with a parallel
account, probes occurring on tracked objects were processed with no delay regardless
of the number of objects tracked, while probes on distractor objects were processed with
significant delays. ☐ In summary, these results are consistent with continuous resource models that
assume that tracking additional objects is accompanied by reductions in tracking
precision. Allocation of this resource is subject to individual differences that depend
partly on strategy but may also reflect differences in total capacity. Furthermore,
people can flexibly allocate variable amount of tracking resources between targets.
This resource appears to be largely hemifield specific and allocated in parallel to
tracked objects.
Description
Keywords
Psychology, Attention, Event related potential, Mixture model, Multiple object tracking