The Diagnostic Utility of Fuzzy System Modeling for Application in Training Systems

Abstract

While there are many different computational modeling techniques capable of predicting human decision-making outcomes, training applications require modeling techniques that are also diagnostic of human decision-making processes. Multiple linear regression, a commonly used modeling technique in Psychology, makes overly restrictive processing assumptions such as that of additivity. A relatively new modeling approach, fuzzy system modeling, bears some striking similarities to current theories of categorization and cognition. In this research, we compare the diagnostic utility of multiple linear regression to fuzzy system models. Specifically, decision-making data are modeled using either linear regression or fuzzy system models, and trainee models are compared to an expert model built with the same technique. Discrepancies between the trainee and expert models are noted and qualitative feedback is generated. The diagnostic utility of each technique is evaluated by measuring changes in performance after model-based feedback is provided to the trainees.