A Model-Building Learning Environment with Error-based Simulation

An important purpose of science education is to help students acquire the ability to build appropriate model of systems. For facilitating such learning activity, modelbuilding learning environments (MBE) have been developed. Recent MBEs assist students in building conceptual models of systems and simulating their qualitative behavior. However, their feedback is insufficient when an erroneous model is built because they can’t explain how its behavior is unnatural nor how to correct it. Additionally, when the model includes inconsistent constraints, they can't create simulation itself. In this paper, we introduce a MBE which overcomes these problems using two techniques we developed: a robust simulator and semantics of constraints. The former analyzes the consistency of a model and relaxes some constraints if necessary. The latter is a systematic description of physical meanings of constraints in models and provides heuristics for explaining unnatural behaviors. A prototype was shown to be useful for learning physics in a preliminary test.