Optimal Learning Under Time Constraints: Empirical and Simulated Trade-offs Between Depth and Breadth of Study

Learners are often constrained by their available study time, typically having to make a trade-off between depth and breadth of learning. Classic experimental paradigms in memory research treat all items as equally important, but this is unlikely the case in reality. Rather, information varies in importance, and people vary in their ability to distinguish what is more or less important. We test the impact of this trade-off in the study of Graduate Record Examination (GRE)-synonym word pairs. In our empirical Study 1, we split our stimuli set, with some items (focal) being afforded more rounds of retrieval practice than other items (non-focal). All conditions had the same total number of trials (i.e., constant study time), but differed in the number of focal words (breadth) and repetitions (depth). The conditions differed significantly in both mean performance and variance on the day-delayed test. Using this empirical data as a base, we then conducted a simulation (Study 2) modeling depth-breadth trade-offs under various conditions of learner forecasting accuracy and test coverage. In Study 2, we found that a medium-depth medium-breadth strategy was appropriate for most of the learning situations covered by our simulation, but that learners with a well-calibrated understanding of importance may benefit from a more targeted high-depth, low-breadth approach. Our results highlight the complexity of navigating the depth-breadth trade-off. Models of learning strategy optimization will need to account for learner forecasting sensitivity, which itself is likely an interaction between relatively stable individual differences and shifting contextual factors.