The Development and Evaluation of a Methodology for Assessing Mental Models of Complex Decision Tasks

Abstract

Models of decision making behavior in complex systems are hampered by an incomplete understanding of the way in which operators internally represent information about decision tasks. This paper describes the development and evaluation of a methodology for assessing mental models by humans to represent complex decision tasks. The methodology is based on the analysis of proximity measures using multidimensional scaling (MDS). In this research, six subjects were trained to classify a visually presented stimulus into one of four categories. The MDS methodology was used to track the evolution of the subject's internal model and identify the dimensional information used to classify the stimuli. The relation between the dimensional information revealed by the analysis and subject performance is discussed.