The first purpose of this study was to demonstrate the durability of spacing effects produced by the Modified Low-First Method. The Modified Low-First Method is an optimal spaced learning method which was derived from a reactivation theory of spacing effects and was designed to be effective by setting as advantageous spaces as possible for all items and for any learners with various working memory capacities. It consists of three principles; the first is to sort all items by their probabilities of recall in ascending order at the end of each learning session for the subsequent session, and the second is to omit items whose probabilities of recall have reached a certain level, and the third is to transit to a new learning session when the number of unrecalled items in a session have reached a certain number. The second purpose of this study was to extend a reactivation model to be able to predict probabilities of delayed recall by incorporating a forgetting function into the previous model. In Experiment, the probabilities of delayed recall in the control condition with simple repetition method and in the experimental condition with the Modified Low-First Method were compared. The result indicated that the Modified Low-First Method was constantly more effective than the former for as many as 14 days. In Simulation, a forgetting power function derived from the experimental data was incorporated into the existent model, and the experiment was simulated with it. The result showed that the estimated probabilities of delayed recall approximated closely to the experimental data, indicating the adequacy of the extended reactivation model.
