Expert systems as a mindtool to facilitate mental model learning

This exploratory study investigated the question, Does the process of constructing an expert system model of a hydraulic drum braking system promote the formation of expert-like mental models? Thirty-three participants in three countries, who reported no knowledge of the subject domain in which the expert system was to be created, read encyclopedia extracts, viewed graphics, and then created small expert systems. The study was conducted via mainframe-mediated network using a variety of communication protocols. Participants' mental models were assessed using Pathfinder concept networks technique, troubleshooting tests, and prediction (change of state) tests. Each of the three measures was administered three times, as pretests, midtests, and posttests. Scores on all three measures suggested that there were small increases from pretest to midtest after participants studied the brake text and graphics. There were substantially larger increases in all scores from midtest to posttest when the expert system was constructed. In poststudy surveys and interviews participants responded that their knowledge in the subject domain had increased, that the creation of the expert system had been very helpful in learning the material, and that they had had fun doing it. When used as a mindtool, an expert system becomes a process, not the end product. It can facilitate a learning gain by engaging students, fostering their concentration, and assisting them in organizing systemic information.