From Rules to Prototypes

Abstract

Traditional expert systems attempt to represent logically the complex decision rules used by experts. This approach has some defined successes but has been less effective with multi-dimensional problems such as characterize clinical situations. Research suggests that effective expertise is based more on memory than on analysis and logic. The article discusses the nature of memory-based expertise and questions the feasibility of capturing it through expert systems. Training based on computerized case simulations is suggested as an alternative. The paper also discusses the need for knowledge on the specific impact interventions have on distinct subgroups of consumers. This knowledge can be used to guide training and assist consumers in making key decisions.