Failed change detection produces volatile short-term memory

Abstract

Aim: We probably all had experiences where we gave up looking for a lost item and resumed searching for it later. Although the initial search would be a failure in this case, it could still provide information to aid the exact same search later on, provided that some form of memory keeps such information accessible. The motivating question behind our study is: what kind of information, if any, gets encoded into memory from failed visual searches? Furthermore, if some form of memory arises from failed searches, we wish to know whether it is useful for the same future searches. Methods and results: We address these questions using a series of change detection experiments involving flickering displays of randomly arranged polygons. Experiments 1 and 2 present these displays for a predetermined time to limit search success on individual trials, and repeat these trials with novel trials mixed in throughout the experiment. We find that subjects perform on par with or worse than a model that assumes no memory benefits on each repeat of a trial, suggesting that even if memory remains from failed visual searches, it is unable to aid future searches and may cause some people to keep repeating their errant search patterns. Experiments 3 and 4 repeat these time-limited displays more rapidly than experiments 1 and 2 by alternating presentations of those displays with various types of short delays. In contrast to experiments 1 and 2, they suggest that some memory from previously failed searches can aid future searches that occur after delays lasting only a few seconds, but this memory can be disrupted by performing other search tasks during the delays. Conclusion: Taken together, these experiments suggest that although some form of memory develops from failed visual searches, it seems short-lived, appears prone to disruption, and may not always assist searches through the same display in the near future.