Validity of Mental Workload Measures in a Driving Simulation Environment

Automated in-vehicle systems and related human-machine interfaces can contribute to alleviating the workload of drivers. However, each new functionality can also introduce a new source of workload, due to the need to attend to new tasks and thus requires careful testing before being implemented in vehicles. Driving simulators have become a viable alternative to on-the-road tests, since they allow optimal experimental control and high safety. However, for each driving simulator to be a useful research tool, for each specific task an adequate correspondence must be established between the behavior in the simulator and the behavior on the road, namely, the simulator absolute and relative validity. In this study we investigated the validity of a driving-simulator-based experimental environment for research on mental workload measures by comparing behavioral and subjective measures of workload of the same large group of participants in a simulated and on-road driving task on the same route. Consistent with previous studies, mixed support was found for both types of validity, although results suggest that allowing more and/or longer familiarization sessions with the simulator may be needed to increase its validity. Simulator sickness also emerged as a critical issue for the generalizability of the results.