The main goal of the present study was to examine the feasibility of e
mploying event-related brain potentials to measure dynamic changes in
mental workload. Subjects performed two tasks, monitoring and mental a
rithmetic, both separately and together. Following an analysis of the
performance, subjective workload ratings, and average ERP data in the
single- and dual-task conditions, two different conditions from each o
f the tasks were selected for further analysis. A bootstrapping approa
ch was employed to determine the amount of ERP data required to discri
minate between these conditions. The results of these analyses indicat
ed that (a) 90% correct discrimination could be achieved with from 1 t
o 11 s of ERP data, (b) the best ERP measures varied across tasks and
subjects, and (c) the inclusion of temporal and spatial aspects of the
ERP data improved the ability to discriminate among workload levels.
The data is discussed with respect to real-time assessment of mental w
orkload.