This paper presents a learning support system, MIPP (Modeling Interface for Pulley Problems) that supports learners to understand pattern matching implemented in production systems. Although many cognitive scientists acknowledge importance of pattern matching for modeling human cognition, it is difficult for beginners to understand how this works. To support understanding of pattern matching, MIPP provides three features to learners; visualization of WM (Working Memory), description of rules that leads unexpected rule firings and visualization of pattern matching. The authors expected that learners who observed unexpected rule firings would notice characteristics of production systems and human cognitive system. This paper reports five classes. MIPP have been improved through these classes. As a result, the scores of the exam had risen through the five tests, indicating that the learners acquired skills of controlling pattern matching.

抄録全体を表示