Investigating the nature of expertise: analogical thinking, expert systems, and ThinkBack

Abstract

Abstract: Experts solve problems. One significant aspect of problem solving is the approach chosen to solve a problem. An often employed approach, but little studied and understood, is the use of analogical thinking. Experts often see current problems in the light of previous problems, and the application of an analogy (i.e. solving a new problem by using an analogous previous problem‐solution) can lead to improved problem solving in some instances, and poorer problem solving in others. Recent research in cognitive psychology has led to greater understanding of the analogy process, and work in artificial intelligence has helped to automate what we know about analogies—yet still much is unknown about how analogies develop, how they can be automated, and the role of analogies in expertise.This article describes a continuing research effort to learn more about analogical problem solving, particularly from a viewpoint of automating an analogical ‘inference engine”. Building upon previous work in this field of study, results of our own empirical work concerning the nature of analogies—especially as used by experts—are portrayed in a series of conceptual level descriptions. We have studied experts in several domains, including systems analysis, computer programming, and in management. Next, our efforts to create automated analogy support systems is indicated. Finally, experiments being conducted to tie together expert systems and our analogy support system are discussed. The analogical inference engine being developed is aptly called ‘ThinkBack”.