This collaborative project will investigate the relationship between conceptual understanding and problem-solving behavior in physics. The investigators will construct computational models of expert and naive- student problem-solving skills (utilizing existing intelligent tutoring system technology) and naive-student conceptions by analyzing protocol data collected from students solving mechanics problems involving force and motion and from individual student interviews about fundamental physical concepts like force. They will then build a mapping between students' naive conceptions and the problem-solving skills which they would acquire by applying those conceptions in various problem contexts. The skill and concept models, and the mapping, will be used to build an automated system capable of conceptual diagnosis of problem-solving behavior. This system will be used to construct individual student models of skills and concepts. The mapping will be validated by comparing the system's conceptual diagnoses with results from accepted diagnostic tests and student interviews. Several outcomes of this research will be: a demonstration of the impact of naive conceptions on physics problem-solving; a test of the validity of using problem-solving behavior to identify naive physical concepts; and an enhancement of existing frameworks of naive conceptions of force and motion. The results would be applicable to other domains of physics. Furthermore, the diagnostic technology would serve as a means of collecting data that is often difficult or impractical (in most classroom situations) to acquire. As a classroom tool for teachers, the system would help demonstrate the need for physics instruction to address students' naive concepts. Finally, the system could serve as the diagnostic core of an intelligent tutoring system that would diagnose and tutor on students' conceptual problems in physics.
