Optimizing practice through self-testing

Memory benefits from retrieval. This fact has motivated an entire literature on the testing effect, which demonstrates that retrieval practice benefits memory more than additional restudy opportunities. The overall robustness of this effect masks a surprising variability in just how advantageous (or not) retrieval practice actually is in practice, particularly for items that are difficult to retrieve or for learners who are highly unskilled or untrained. In a series of experiments, this dissertation examines effects of self-testing across a variety of levels of difficulty. The goal is to find techniques that allow precision in determining at what level of mastery the risks and benefits of self-testing outweigh the certain but modest benefits of restudy. For learners, an optimization algorithm would be most useful if it can translate to a practice schedule that adapts to them, based on their current knowledge level. The first four experiments attempt to determine where testing is most and least effective, based on subjects’ own judgments of learning during study, and whether restudy events can be profitably reintroduced to practice. Experiments 5 and 6 allow participants themselves to choose practice, to test whether these choices can be modified to improve overall memory. Across these experiments, retrievability and the number of practice sessions both modulate the magnitude of the testing effect. However, modifying practice on the basis of these variables in Experiments 5 and 6 did not reliably improve memory