UC San Diego

Retrieval practice is a learning technique which involves taking practice tests on to be-recalled information. Because it is one of the most potent memory enhancers known to learning science, many cognitive and educational psychologists endorse retrieval practice as a highly effective method for enhancing academic performance. However, retrieval practice’s capacity to enhance transfer of learning, or the generalization of learning to different contexts, has yet to be fully characterized. Given that learning often needs to be flexibly applied, the importance of transfer to education is paramount. In the current research, four studies investigated retrieval practice and transfer to rearranged stimulus-response elements. With this type of transfer, practice and criterial tests target

the same basic information, such as a specific fact or concept, but feature different cues and responses. For example, given the fact “Thomas Jefferson purchased Louisiana from France”, a practice test asks, “Thomas Jefferson purchased what from France?”, whereas a criterial test asks, “Who purchased Louisiana from France?” The chief question was whether learning would transfer—as evidenced by better accuracy—to criterial test questions featuring rearranged stimulus-response elements. Studies 1-4 used triple associate words, Advanced Placement History and Biology course facts, term-definition neuroscience facts, and process-based biology concepts, respectively. Throughout these studies, retrieval practice consistently improved accuracy when cues and responses were consistent on practice and criterial tests, relative to a non-retrieval practice reexposure control condition. However, substantially reduced improvements were observed when the criterial test featured rearranged stimulus-response elements. This generally occurred regardless of the amount of, and the format of, retrieval practice. Only the addition of detailed and extensively processed feedback, as was investigated in Study 4, yielded exceptions to the prevailing pattern of minimal-to-no transfer. Overall, the present research yields insights into the transfer properties and mechanisms of retrieval practice, including its often selective enhancement of trained cue-response combinations. It also provides support for dual memory theories of retrieval practice effects. Practically, this research suggests that retrieval practice should be targeted at as many cue-response combinations as possible during training, or should incorporate extensively processed feedback, in order to yield maximum learning benefits.