Change Detection in Visual Short-Term Memory

Abstract

Numerous kinds of visual events challenge our ability to keep track of objects that populate our visual environment from moment to moment. These include blinks, occlusion, shifting visual attention and changes to object's visual and spatial properties over time. These visual events may lead to objects falling out of our visual awareness but can also lead to unnoticed changes, such as undetected object replacements and positional exchanges. Current visual memory models do not predict which visual changes are likely to be the most difficult to detect. We examined the accuracy with which switches (where two objects exchange locations) and substitutions (where one or two objects are replaced) are detected. Inferior performance for one-object substitutions vs. two-objects switches, along with superior performance for two-object substitutions vs. two-object switches was found. Our results are interpreted in terms of object file theory, trade-offs between diffused and localized attention, and net visual modification.