Detecting functional relations in a computerized discovery environment

Abstract

While many instructional simulation programs concentrate on the difficulties involved in teaching about a specific domain or treat problems of presenting information in specific ways, the computer-based simulation program described in the article, REFRACT, focuses instead on a specific style of inductive knowledge acquisition: learning by hypothesis testing. REFRACT's design is based on a problem-solving model of induction. It puts students in a situation where they can learn about a domain that resembles geometrical optics by means of scientific discovery learning: by stating predictions, running experiments, and analyzing results. The student's interaction with the program is organized in a way that provides both structure to the student's task and a window on her hypothesis generation and testing behavior. The system adapts in some sense to students' knowledge by permitting them to state predictions in several ways, ranging from purely qualitative/graphical to quantitative and precise. Observations on students working the REFRACT are reported, focusing on differences with respect to discovery skills and differences in formulating hypotheses either qualitatively or quantitatively.