Using Computational Modeling to Assess Use of Cognitive Strategies

Abstract

Although there are many strategies and techniques that can improve memory, cognitive biases generally lead people to choose suboptimal memory strategies. In this study, participants were asked to memorize words while their brain activity was recorded using electroencephalography (EEG). The participants’ memory performance and EEG data revealed that a self-testing (retrieval practice) strategy could improve memory. The majority of the participants did not use self-testing, but computational modeling revealed that a subset of the participants had brain activity that was consistent with this optimal strategy. We developed a model that characterized the brain activity associated with passive study and with explicit memory testing. We used that model to predict which participants adopted a self-testing strategy, and then evaluated the behavioral performance of those participants. This analysis revealed that, as predicted, the participants whose brain activity was consistent with a self-testing strategy had better memory performance at test.