Using Theoretical Models of Fatigue to Design Optimal Fatigue Countermeasures

Jesse Lee Eisert

Advisor: Carryl Baldwin, PhD, CHSSWeb Design Preview

Committee Members: Tyler Shaw, William Helton

David J. King Hall, #1019
May 02, 2018, 02:00 PM to 04:00 PM

Abstract:

Driver fatigue is a prevalent problem that annually results in over 80,000 crashes and 1,000 fatalities annually.  Driver fatigue can be broken down into sleep related and task induced fatigue. This dissertation will focus solely on task induced fatigue, which can further be divided into active and passive. Active fatigue results from continuous and prolonged periods of task related perceptual-motor adjustments, while passive fatigue is the result of prolonged system monitoring with either rare or no overt perceptual-motor response. One key difference between active and passive fatigue is that the adjustments/actions need to be overt. As it relates to driving this is important because simple lane maintenance does not require any overt action. Active and passive fatigue result in different states, with active fatigue being associated with distress and overload whereas passive fatigue is associated with reduced alertness, reduced task engagement, and underload. With the two types of task-induced fatigue being distinct from one another, it is important to consider what the underlying theoretical explanation for fatigue is. Resource theories would attribute the feeling of fatigue to a depletion of resources; effort mobilization theory attributes fatigue to an inability to mobilize effort effectively when tasks appear undemanding; and motivational control theory (MCT) attributes fatigue to a conflict between active and alternate goals as well as activation in effort regulation.

In addition to each theory providing a unique explanation for how one becomes fatigued, they also in turn provide unique predictions for the best countermeasures. Resource theory would predict that taking a break would be the best, while effort mobilization predicts adding an extra task would be best, and MCT would predict that taking a break, switching to a different task, or adding an extra task would all decrease fatigue. The current investigation employed a passive (Experiment 1) and active (Experiment 2) fatigue induction followed by one of three different types of countermeasures:  1) complete break, 2) switch to a different task that used the same working memory resources or 3) addition of a task that used the same working memory resources.  Performance following each countermeasure type was compared to an active control where participants continued the initial fatigue induction task with no change.  The goal was to determine which theoretical account was supported by the resulting performance pattern and to determine if one theory could explain both types of task induced fatigue.  In order to be able to make comparisons between both experiments participants also completed a vigilance task during the drive. This allowed for a comparison of sensitivity across both experiments.

Results of Experiment 1, the passive fatigue experiment, indicated that all three countermeasures were able to restore performance better than the control group, giving support for MCT. Results of Experiment 2, the active fatigue experiment, indicated that the break condition resulted in a significant restoration of performance when compared to both the control group and to the additional task countermeasure. When comparing across the two experiments participants who were actively fatigued saw a greater restoration in performance on the vigilance task than those who were passively fatigued. Additionally participants who were actively fatigued avoided the hazardous event significantly more frequently than those who were passively fatigued. Results from this dissertation suggest that when designing a fatigue intervention system it is extremely important to consider whether participants are fatigued actively or passively. Furthermore, this dissertation shows how dangerous automation failures can be as 35% of participants in the passive fatigue condition failed to respond until they already would have crashed, relative to zero participants in the active fatigue condition.