This paper presents a probabilistic approach to modeling human performance. Instead of focusing on mean performance, the effects of taskload on the distributions of performance variables are examined. From such data, probabilities of given levels of performance can be derived and methods of measurement that expand the analyses beyond those of the mean developed. Results from two experiments, one abstract, the other realistic, are presented in terms of timely performance on required tasks. As taskload increased, the participants were less likely to act on the experimental tasks at an earliest opportunity than under low taskload, resulting in increase of “too late” errors. Measurement of taskload and performance in temporal terms also allowed for bracketing and making inferences about mental workload, which is not directly measurable.
Electrophysiological and performance variations following driving events involving an increase in mental workload
