Retrieval Practice and Context Reinstatement

Retrieval practice is a powerful learning strategy, but the mechanism(s) behind it are not fully understood. One account of this testing effect (i.e., enhanced memory for information that was previously retrieved vs. restudied) is the episodic-context account (Karpicke et al., 2014). According to this theory, successful retrieval requires contextual reinstatement which updates the target memory representation to include features of the current test context along with features of the initial study context. The resulting composite trace provides varied features that are more likely to match those on the final test. However, the specific features necessary for reinstatement remain unclear. The current study was designed to clarify the nature of these features using a three-phase paradigm. In phase 1, participants studied targets paired with different cues (phonemic vs. semantic) and presented in one of two different lists. In phase 2, participants either: (a) restudied the words; (b) restudied and made cue discrimination judgments; (c) restudied and made list discrimination judgments; or (d) freely-recalled the words. In phase 3, a final recognition test assessed item and context memory (i.e., cue type and list number). Unconditionalized analyses of final test performance found no differences between learning conditions. However, when only looking at words successfully retrieved at practice and thus subjected to reinstatement, free recall practice generally outperformed all other conditions on item and context memory, aside from the Restudy + Cue Discrimination condition. These results suggest that semantic, rather than temporal, features may be most relevant to the learning benefits of retrieval practice