Predicting operator mental workload using a time-based algorithm

Abstract

Existing workload algorithms based on the Multiple Resources Theory (MRT) provide an effective approach for diagnosing operator overload caused by interference among concurrent tasks, however, their ability to handle overload in single task conditions is limited. We argue a time-based algorithm, developed on the Information Processing (IP) model of workload (Hendy, Liao, & Milgram, 1997), provides a viable solution to address this limitation. In this study, we proposed a new algorithmic implementation of the IP model in the context of task network modeling. The new algorithm was implemented in a JAVA program and tested on an existing model of a Bakan vigilance task. The results obtained from the new algorithm demonstrated the feasibility of this solution. By integrating resource-based and time-based algorithms, analysts will be able to diagnose more accurately system performance breakdowns caused by operator overload.

Mental Workload Modeling in an Integrated Cognitive Architecture

“…This feature makes ACTR-QN especially suitable for early-stage evaluation without any full-scale system, quick evaluation of a large amount of design alternatives, and evaluation in high-risk domains. In this regard, taskbased (e.g., Cassenti, Kelley, & Carlson, 2010) and tasktime-based (e.g., Wang et al, 2010) methods usually need expert judgments to estimate workload scales and time constraints for unit cognitive tasks. In contrast, ACTR-QN's prediction of mental workload can complement or substitute expert subjective judgments, as it is based on a synthesis of a large amount of related findings and governed by the psychological and neuroscience theories underlying the cognitive architecture.…”

“…Human-in-the-loop empirical testing methods include performance measures such as secondary task methods, physiological measures such as blink rate, and subjective measures such as NASA-Task Load Index (TLX, Hart, 2006). The other group is computational modeling methods, which include task-analysis-based methods (e.g., Mitchell & Samms, 2009;Wang, Cain, & Lu, 2010) and cognitive-architecture-based methods (e.g., Gray, Schoelles, & Sims, 2005;Wu & Liu, 2007).…”