Abstract
Multiple object tracking (MOT) has been used to great effect in delimiting what perceptible features and behaviors are intrinsic to the representation of objects. We considered the ecological condition in which objects are conceptually instantiated and tracked prior to – or entirely without – becoming visible. For example, hidden animals are readily inferred from leaves and grasses that they brush in passing. Here, we present a novel MOT display of 'virtual' targets and distractors that are not themselves visible, but whose locations are haphazardly revealed by their displacement of numerous squares filling the display. In Experiment 1, we evaluated viewers' abilities to track from 1 up to 5 targets among ten identically-appearing objects. Viewers were able to track multiple independently-moving objects – albeit less capably than found under the conditions characteristic of previous research. The second experiment ruled out that this lower performance for tracking virtual objects resulted from the visually complex background. Under the same viewing conditions, tracking accuracy for visible targets was comparable to that in previous research and significantly greater than for virtual targets. In a third experiment, we evaluated tracking for objects visible at the outset which then all remained visible, or each briefly and completely disappeared once, or each briefly disappeared but persisted 'virtually'. Unsurprisingly, completely disappearing objects were tracked less capably than either continually visible or briefly virtual objects. However, objects that underwent virtual disappearances were tracked as well – and perhaps even better – than those remaining visible throughout the trial. Although inherently more challenging, viewers can infer and track multiple independently-moving objects from indirect evidence of their location – and absent any visible features of the objects themselves. Finally, virtual object tracking is a novel method that may be uniquely well-suited for studies otherwise at risk of confounding by lower-level visual saliency.
Meeting abstract presented at VSS 2014