Modeling Individual Differences at a Process Control Task

Abstract

Individual differences at process control tasks are difficult to analyze because similar system performance may result from different control strategies. This paper introduces a methodology for describing rule and skill based components of performance directly, using machine learning techniques. Similarity among strategies is shown to have a greater effect on performance than training interventions. The behavior of two subjects who are nearly identical in performance but vary greatly in strategy is examined to illustrate the potential of this approach for assessing individual differences at process control tasks.