A methodology for modeling and predicting mental workload in single- and multi-task environments

Bin Xie, Purdue University


Mental workload has long been recognized as an important factor in human performance in complex systems. It is documented that either overload or underload may degrade human performance. Therefore, system designers need, at an early design phase, some explicit models to predict the mental workload imposed on individuals so that alternatives of system designs can be evaluated. This research aims to develop a practical framework for predicting mental workload in both single- and multi-task environments with particular consideration of individual factors. In order to describe mental workload more precisely and completely, a framework for mental workload definitions, containing instantaneous workload, average workload, accumulated workload, peak workload and overall workload is proposed. In order to model individual factors, two new variables--effective workload and ineffective workload--are introduced to conceptually model task-generated and individual-generated workloads. Under the conceptual model, the operational models for predicting human mental workload for human-computer interaction tasks are developed.^ The results of two experimental studies, utilizing 60 subjects, aimed at validating the model, indicated the followings: (1) Average workload, accumulated workload, and instantaneous workload are different from overall workload. Hence, combined use of them describes the workload more precisely. (2) The predictive mental workload model explains 42 percent of the variances associated with NASA-TLX subjective mental workload ratings and explains 78 percent of the variances associated with performance time. Hence, the conceptual and operational model for predicting mental workload have been validated. (3) The relationships between the effective/ineffective workload and the four independent variables have not been fully validated. The results showed that both task-related factors and individual-related factors can affect mental workload significantly. (4) Mental workload is significantly affected by time pressure. The workload in self-paced multi-task environment is 29% lower than the workload in system-paced multi-task environment. The workload in a self-paced, multi-task environment is 19% lower than the workload in a system-paced, single task environment.^ The models developed and validated for predicting mental workload can be used by designers to allocate tasks and to increase task performance and job satisfaction. ^




Major Professor: Gavriel Salvendy, Purdue University.

Subject Area

Engineering, Industrial|Psychology, Industrial

Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server