The Impact of Working Memory Testing on Associative Long-term Memory

Abstract

The long-term fate of to-be-remembered information depends in part on the conditions of initial learning, including mental operations engaged via working memory (WM). However, the mechanistic role of WM processes in subsequent episodic memory (EM) remains unclear. Retrieval processes or increased maintenance time are two ways WM processing can improve EM. Yet, it remains uncertain whether benefits from WM processing are greater after an opportunity for retrieval practice compared to repeated exposure (e.g., restudying) to the memoranda. If there is a relative advantage of WM testing versus mere re-exposure, this implies that the testing effect, the robust finding that retrieval practice improves long-term memory (LTM), can emerge for tests administered within the canonical limits of WM. This dissertation has two aims: 1) to determine whether re-exposure during WM choice recognition tests improves subsequent EM for word-pair associations more so than restudy opportunities and 2) to delineate mechanisms that mediate benefits of WM re-exposure and their potential parallels with the LTM testing effect. Across five experiments (N=696), the results indicate that WM recognition testing can lead to superior associative EM minutes later compared to WM restudy. Consistent with the LTM testing effect literature, any testing advantages after short delays was contingent on initial test performance. When WM recognition was successful overall and errors were removed from the analyses, testing advantaged EM minutes later relative to restudy, revealing a positive “WM-based testing effect”. When WM choice recognition was relatively unsuccessful and errors were included in the analyses, no testing advantages emerged. Additionally, the final experiment (N=112) provided novel evidence that WM recognition testing, but not restudying, can benefit EM after longer delays of over twenty-four hours relative to no re-exposure. The outcomes of this dissertation suggest that similar mechanisms may underlie the positive impact of retrieval practice opportunities during WM and LTM processing and underscore the importance of WM processing in supporting EM for associations after short and long delays.