VanLehn, K., & Jones, R. M. (1993). Better learners use analogical problem solving sparingly. In P. E. Utgoff (Ed.), Proceedings of the Tenth International Conference on Machine Learning (pp. 338-345). San Mateo, CA: Morgan Kaufmann.

When solving homework exercises, human students often notice that the problem they are about to solve is similar to an example. They then deliberate over whether to refer to the example or to solve the problem without looking at the example. We present protocol analyses showing that effective human learners prefer not to use analogical problem solving for achieving the base-level goals of the problem, although they do use it occasionally for achieving meta-level goals, such as checking solutions or resolving certain kinds of impasses. On the other hand, ineffective learners use analogical problem solving in place of ordinary problem solving, and this prevents them from discovering gaps in their domain theory. An analysis of the task domain (college physics) reveals a testable heuristic for when to use analogy and when to avoid it. The heuristic may be of use in guiding multi-strategy learners.

For a PDF full article version, click here (128KB).