Proceedings of the Annual Meeting of the Cognitive Science Society

After being taught by teachers, learners often need to work independently in new situations. However, a teaching strategy that most efficiently fosters independent learning remains elusive. In this study, we developed a novel experimental paradigm to compare various teaching strategies. In addition, we formalized autonomy-supportive teaching and constructed an autonomy-support algorithm that estimates learners' mental states and aims to enhance both learners’ competence and autonomy. In the experiment, participants were taught through different teaching algorithms depending on the experimental conditions, after which they independently worked on a new set of tasks. Our results demonstrate that compared to the all- and no-teach algorithms, the autonomy-support algorithm enhances learners' engagement while being taught and enhances performance when learners independently work on a new set of tasks. Our findings contribute to the existing observational and interventional research on education by providing rigorous evidence in an experimentally controlled setting.