Cognitive modelling approach to diagnose over-simplification in simulation-based training

Abstract

Simulation-based training has become a standard in operational knowledge training for supervisory control in safety-critical environments. But traditional simulators do not support mental model formation of automated systems though these systems are a dominant part in modern control systems. Recently various”intelligent components” for this support have been suggested. But these approaches neglect the dynamic character of mental models. They focus on building a normative model at the beginning of the training but do not consider how it evolves due to knowledge acquisition processes. In this paper we present a model-based approach to diagnose success-driven learning in simulator training and to predict dangerous over-simplifications. Our research focuses on pilot training for automated cockpits. © Springer-Verlag Berlin Heidelberg 2002.