TAME: A Method of Teachable Agent Modeling for Error-Visualization

In the context of learning, the effectiveness of learning-by-teaching (LbT) is widely recognized. LbT is a method through which learners deepen their understanding by teaching others what they have learned. In order to realize LbT, it is necessary to have the behavior of the teachable person, such as the knowledge of other and an attitude towards receiving instruction, as well as the existence of the teachable presence. One technology for approaching this challenge is called a teachable agent. A teachable agent is an agent that behaves as if it is learning what is taught by the learner. Another effective learning method is error-visualization. Error-visualization is a feedback-based approach to simulating the answers by the learner as (abnormal) behavior. However, no modeling methods have been proposed that combine teachable agents with error-visualization. In this paper, we propose teachable agent modeling for error-visualization (TAME) as a teachable agent modeling method that aims to combine teachable agent with error-visualization. As a case study, we realize a teachable agent using the TAME modeling method based on a conventional learning support system called Monsakun.