Approaches to Modeling the Effects of Fatigue on Cognitive Performance

Behavior representations in simulation environments have become increasingly sophisticated as computational power and understanding of human cognitive processes have increased. Despite improvements in the breadth and depth of models of human behavior, comparatively little research has explored how human cognition is moderated by external and internal factors. In this review paper, we focus on fatigue, a pervasive cognitive moderator that has dramatic and sometimes tragic effects on human cognitive functioning. We discuss prior and current research investigating models of how fatigue impacts human cognition and behavior, sometimes leading to dramatic degradations. The paper ends with a description of our on-going efforts to identify mechanisms for fatigue within the Adaptive Control of Thought – Rational (ACT-R) cognitive architecture. Our approach involves integrating diverse research areas, especially biomathematical modeling of the effects of fatigue on neurobehavioral functioning, and computational cognitive modeling of human cognition and behavior. 1. Motivation and Background Increasingly sophisticated human behavior representations are rapidly becoming a standard component of modern modeling and simulation capabilities. These entities can serve in a variety of roles, including adversaries, teammates, and neutral forces. They engage in complex tasks and have become critical to serving the variety of goals associated with immersive environments, simulations, synthetic tasks, and games, including training, education, analysis, and entertainment. As the tools, techniques, and implementations of human behavior representations have begun to mature, increasing emphasis has been placed on ensuring that they accurately reproduce the cognitive capacities, limitations, and processes of the human participants they replace. One particular area of increased focus is the impact of cognitive moderators (e.g., Gluck, Gunzelmann, Gratch, Hudlicka, & Ritter, 2006). Cognitive moderators are factors that influence human cognition, but are not necessarily part of the goal-directed thought that is the focus of information processing. In this paper we focus on mathematical and computational implementations of fatigue as a cognitive moderator. Psychological research has only begun to uncover the details of how cognitive moderators affect thought. In the case of fatigue, however, there is an extensive empirical literature documenting the negative consequences of sleep restriction, sleep deprivation, and the intrinsic circadian clock (e.g., Durmer & Dinges, 2005; Dinges, Baynard, & Rogers, 2005). In contrast, models of these processes, and the concomitant changes in cognitive functioning associated with these processes, are relatively few. Our goal here is to review significant prior and on-going efforts and accomplishments in the development of mathematical and computational theories of how fatigue impacts cognitive performance. The approaches are varied, but they are linked by a common overarching goal: The desire to make principled, quantitative predictions about how increased levels of fatigue will impact performance in applied settings. The applications of this research are extensive. A surprising number of industrial and commercial disasters have been attributed at least partially to fatigue (e.g., Caldwell, 2003; Mitler et al., 1988), and so have a sizable proportion of other tragedies like automobile accidents (e.g., Horne & Reyner, 1999). One application of this research is to make predictions regarding an individual’s readiness to engage in a particular, cognitively demanding activity. In fact, some research on fatigue has led to