A New Multiple Object Awareness Paradigm Shows that Imperfect Knowledge of Object Location Is Still Knowledge

摘要

Keeping track of multiple moving items is a fundamental perceptual skill. Nevertheless, our capacity in tasks such as multiple object tracking (MOT) [], multiple identity tracking (MIT) [], or change detection [] is very low—typically just two to four items. In standard tracking tasks, observers monitor M out of N items as they move about the screen. At testing time, observers are asked to mark all tracked items or to state if a specific item is a target. If they are wrong about an item, they are considered completely wrong. Similarly, in an identity tracking experiment, observers might be asked, “Where is the horse?” If they mark the neighboring location where the cow is hidden, again, they are marked completely wrong. If an observer knew approximately, but not exactly, where the horse was, she would get no credit for that knowledge. In our multiple object awareness task, observers continue to click on locations until they locate the target for that trial so that any imprecise knowledge can be used in observers’ responses. A model in which location information degrades with time after an item is attended can be used to estimate a new capacity of multiple object awareness—the number of items about which the observer has some location information. This capacity is at least double the usual capacity estimates for identity tracking.