Evidence of distractor forgetting in working memory changes with testing format.

Working memory distractors maintain some context bindings.

Pattern comparison can be used to lengthen a verbal complex span task.

Research on forgetting irrelevant information in working memory has supported two competing models, SOB-CS (removal) and TBRS (decay). This may be due to the fact that different methods have been used to test each model. To rectify this, we developed a modified complex span task that allows for a better comparison of the models. Participants processed a series of words that were to be remembered (targets) or were to be forgotten (distractors). The primary manipulation was the free time available after viewing a distractor, either a short (.2 s) or long amount of time (1.5 s), with total time held constant across conditions. According to the SOB-CS model, distractors are not completely removed in the short condition, thereby increasing the number of intrusions at test relative to the long condition. In contrast, the TBRS model proposes that distractors decay at the same rate in both conditions, thus predicting no free time differences. We found support for removal with a reconstruction test (Experiment 1) and found inconclusive evidence of decay with a serial recall test (Experiment 2). In Experiment 3 and 4, we incorporated methods from recall and reconstruction tests into a modified reconstruction test and found evidence in favor of removal. Overall, the removal account was better supported than the decay account.

Decay

Removal

Working memory

Forgetting distractors

© 2022 Elsevier Inc. All rights reserved.