The long-term consequences of retrieval demands during working memory

Abstract

Although it is well known that distraction impairs immediate retrieval of items maintained in working memory (WM; e.g., during complex span tasks), some evidence suggests that these items are more likely to be recalled from episodic memory (EM) compared with items that were studied without any distraction (e.g., during simple span tasks). One account for this delayed advantage of complex span over simple span, or the McCabe effect (McCabe, Journal of Memory and Language, 58[2], 480-494, 2008), is that complex span affords covert retrieval opportunities that facilitate later retrieval from EM by cumulatively reactivating each successively presented item after distraction. This explanation focuses on the processing that occurs during presentation and maintenance of the items, but no work to date has explored whether the differential demands of immediate retrieval between simple and complex span may explain the effect. Accordingly, these experiments examined the impact of immediate retrieval demands on the McCabe effect by comparing typical immediate serial-recall instructions (i.e., recalling the words in their exact order of presentation) to immediate free-recall (Experiments 1-2) and no-recall (Experiments 2 and 3) instructions. The results suggested that the nature of retrieval may constrain the McCabe effect in some situations (Experiments 1-2), but its demands do not drive the McCabe effect given that it was observed in both serial-recall and no-recall conditions (Experiment 3). Instead, activities such as covert retrieval during the processing phase may underlie the McCabe effect, thus further evidencing the importance of processing in WM for the long-term retention of information.

Keywords: Complex span; Episodic memory; Retrieval; Simple span; Working memory.

Fig. 1

Mean proportion of recalled items, scored as accurate according to their original serial order of presentation (top panel), in any order (middle panel), and at delay (bottom panel) in Experiments 1 and 2. Error bars reflect 95% within-subjects confidence intervals

Fig. 2

Mean proportion recalled at the immediate test in terms of serial scoring (top panel) and free scoring (bottom panel) in Experiment 3. Error bars reflect 95% within-subjects confidence intervals

Fig. 3

Mean proportion recalled at the delayed test in terms of serial scoring (top panel) and free scoring (bottom panel) in Experiment 3. Error bars reflect 95% within-subjects confidence intervals