Cognitive Factors in User/Expert-System Interaction

Abstract

Expert systems are a type of decision aid that has emerged from the field of artificial intelligence. In this paper, it is hypothesized that two cognitive factors are important mediators of the quality of user/expert-system interaction: (1) the degree of consistency between the user's problem-solving and that of the expert system, and (2) the user's mental model of the system's operating processes. It is further hypothesized that when a user possesses an accurate mental model of expert-system processing, user/expert-system performance will remain high even when there is significant inconsistency between the problem-solving processes of the user and the expert system. Three experiments testing the impact of a user's mental model on user/expert interaction are summarized. Results support the preceding hypotheses.