Document Type


Publication Date





Physiologically-based cognitive workload assessment can discriminate changing levels of operator functional state in complex task environments. In this paradigm, electroencephalography (EEG) is a commonly used physiological measure. Spectral power in clinical frequency bands is used to derive features to train an artificial neural network (ANN) classifier to recognize changes in cognitive workload. Recent research has suggested that power in high frequency bands may be influenced by electromyographic artifact. In a previous study, nineteen channels of EEG were recorded (from 10 participants) during a complex uninhabited air vehicle (UAV) control simulation in which task difficulty was manipulated to induce changes in cognitive demand. In offline analysis, an ANN classifier was trained using feature sets which included and excluded features from high frequency bands. Excluding the high frequency bands reduced classification accuracy, suggesting that, while potentially of electromyographic origin, these features are still important features in physiology-based cognitive workload assessment.