Variation in Working Memory Capacity and Episodic Recall: The Contributions of Strategic Encoding and Contextual Retrieval
Unsworth, Nash, Spillers, Gregory J., Psychonomic Bulletin & Review
The present study examined the extent to which differences in strategic encoding and contextual retrieval account for the relation between individual differences in working memory capacity (WMC) and variation in episodic recall. Participants performed a continual distractor task under either incidental- or intentional-encoding conditions. High-WMC individuals outperformed low-WMC individuals across both encoding conditions and, notably, to a greater degree in the intentional-encoding condition. These results suggest that WMC differences in episodic recall are likely due to a combination of differences in both contextual-retrieval and strategic-encoding processes. These findings are consistent with prior work showing that high-WMC individuals are better at engaging in strategic-encoding processes during the presentation of items than are low-WMC individuals and are better at using contextual cues to focus the search on correct items during retrieval.
A great deal of prior work has demonstrated that working memory capacity (WMC) is an important predictor of performance on a number of higher and lower order cognitive tasks (Engle & Kane, 2004). Recent work has suggested that individual differences in WMC (as measured by complex span tasks) are related to differences in episodic retrieval in tasks such as free recall (Bailey, Dunlosky, & Kane, 2008; Kane & Engle, 2000; Unsworth, 2007). However, the reasons for this relation are still not well understood. It is possible that these differences arise primarily from differences in strategic retrieval processes, differences in strategic-encoding processes, or some combination of both. That is, variation in WMC likely reflects variation in control processes (Engle & Kane, 2004; Unsworth & Engle, 2007), and these control processes likely operate both at encoding, in the form of differences in encoding strategies, and at retrieval, in the form of differences in retrieval strategies and the selection of appropriate probes or cues (Atkinson & Shiffrin, 1971).
In previous work, we (Unsworth & Engle, 2007) have suggested that the relationship between WMC and episodic recall is due, in part, to differences in contextual-retrieval abilities. In line with Glenberg and colleagues (Glenberg et al., 1980; Glenberg & Swanson, 1986), we suggested that, at encoding, items are associated with multiple levels of context (e.g., global, list-specific, and word-specific contexts). At retrieval, these contextual features are used as cues/probes to constrain the search to the items from the most recently presented list. In terms of individual differences in WMC, we suggested that low-WMC individuals are less efficient than high-WMC individuals at selecting appropriate retrieval strategies and retrieval cues/probes (particularly, contextual cues), which leads to the inclusion of more irrelevant items in their search sets. The inclusion of more irrelevant items in the search sets of low-WMC individuals increases the overall size of their search sets (cue overload), leading to a less accurate and slower search.
According to this contextual-retrieval hypothesis, low- WMC individuals demonstrate poorer episodic recall than do high-WMC individuals because they use less efficient contextual cues to effectively constrain the search to only the relevant items. Evidence consistent with this hypothesis has come from studies in which high- and low-WMC differences in free recall tasks have been examined, where low-WMC individuals not only recalled fewer target items than did high-WMC individuals, but also were slower to recall items and recalled more intrusions than did high- WMC individuals (Unsworth, 2007). Collectively, these results suggest that low-WMC individuals search through a larger set of items than do high-WMC individuals, due to the fact that low-WMC individuals are less able to use contextual cues to focus the search. Furthermore, simulations that tested various explanations for differences between high- and low-WMC individuals in episodic recall (e. …