Journal of the Chinese Institute of Industrial Engineers, Vol. 19, No. 1, pp. 1-15 (2002) 1 AN ELECTRICAL-CIRCUIT MODEL FOR PREDICTING MENTAL WORKLOAD IN COMPUTER-BASED TASKS Urnesh H. Patel and Gavriel Salvendy * School of Industrial Engineering Purdue University 1287, Grissom Hall, West Lafayette, Indiana 47907-1287 Leslie A. Geddes Hillenbrand Biomedical Engineering Center Purdue University Thomas Kuczek Department of Statistics Purdue University ABSTRACT With the increased use of computer-based technology, the work of the human has become more cognitive. As a result, the objective measurement of mental workload has become vital in the design of jobs and the development of adaptive interfaces. This research focused on developing and validating an Ohm’s law analogy for mental workload based on individuality and the human’s ability to process information. In this model, current was defined as the rate of information transmission (in bit/s); voltage was the mental workload per unit of processed information; resistance was a function of task type and individual factors. To test the hypothesis, the analog voltage-current relationship was evaluated in 24 senior nursing students operating a computer-based system. All subjects performed three different, rule-based, self-paced tasks varying in complexity (low, moderate, and high). Regression analysis across all three tasks showed that a linear relationship existed between the analogs of voltage and current ( 97 . 5 ) 69 , 1 ( - = t , 0001 . 0 = p ), indicating that a negative differential resistance was present. Cognitive ability, fatigue, stress, anxiety, computer skill, and content skill were measured as possible components of resistance. Using stepwise regression analysis, cognitive ability was identified as a significant component of resistance ( 2 . 14 ) 22 , 1 ( = F , 0011 . 0 = p ). A model relating voltage to current and cognitive ability produced an 66 . 0 2 = R for the moderate and high complexity task sets. Similar results were also obtained from a more pragmatic model in which the voltage analog was the mental workload per decision and the current analog was the decision rate ( 63 . 0 2 = R ). This electrical-curcuit model could be refined and applied in the workplace for on-line, mental-workload monitoring and in the control of adaptive computer interfaces. Keywords : adaptive interfaces, information processing, mental workload, mental workload modeling * Corresponding author: salvendy@ecn.purdue.edu 1. INTRODUCTION With the increased use of computer technology and the increased automation of physical tasks, the work of the human has become more cognitive. The human operatorperforms less manual control tasks and more supervisory control tasks[17]. In an advanced manufacturing system, the success of the plant operation depends on the performance of the supervisory controller [2]. The supervisory controller, whose tasks include scheduling of jobs, handling machine failures, and monitoring, is responsible for the control of the manufacturing facility. When performing these tasks, it is important that the supervisory controller does not become mentally overloaded or under loaded both of which would lead to a degradation in performance. In hospitals, computer-based monitoring systems are used extensively by nurses. In an intensive care unit, a nurse has information coming from several different monitoring units to assimilate. Increasing the number