An intelligent tutoring system for teaching fundamental physics concepts

Students in traditional elementary mechanics classes can master problem solving of a quantitative nature but not those of a qualitative type. Moreover, students' naive conceptions of physics remain unchanged after completing their class. A few approaches have been implemented to improve this situation however none have met with great success. Since elementary mechanics is the foundation for all of physics and it is a required course for most science majors there is a clear need to improve the instruction of the subject. To address this problem I developed a intelligent tutoring system, called the Conceptual Helper, which coaches students during homework problem solving. The tutor uses a unique cognitive based approach to teaching physics, which presents innovations in three areas. (1) The teaching strategy, which focuses on teaching those links among the concepts of the domain that are essential for conceptual understanding yet are seldom learned by the students. (2) The manner in which the knowledge is taught, which is based on a combination of effective human tutoring techniques (e.g., hinting), effective pedagogical methods (e.g., a microscopic view of matter), and less cognitively demanding approaches (e.g., anthropomorphism). (3) The way in which misconceptions are handled which uses the underlying scientific correct line of reasoning to describe to the student the phenomenon that is the basis for the misconception. From a technological point of view the Conceptual Helper was implemented as a model-tracing tutor which intervenes when students make errors and after completion of each problem, at which time the tutor scaffolds the students on post-problem reflection. The remediation is guided by probabilistic assessment of mastery and the interventions are adapted to the errors. The thesis also presents the results of the evaluation of the system which revealed that the gain scores of the experimental group were statistically significantly higher than those of the control group, suggesting that the Conceptual Helper was indeed capable of effectively teaching the conceptual aspects of physics as well as helped students abandon common misconceptions. Furthermore, the evaluation showed that the students' performance on a standardized test was comparable to those of other more complex approaches.