EEG-based Estimation of Cognitive Workload Across Multiple Tasks

Key to the efficacy of working in high-risk environments is the reliable estimation of the human's cognitive state for improving safety and to maintain high performance longer. In this study, we developed an experimental protocol in which participants completed three cognitive tasks under two different levels (High, Low) of workload. We then evaluated the effect of the different cognitive activities on EEG signals and its accuracy in predicting respective cognitive load. The analysis was conducted using well-known machine learning algorithms such as SVM, RF, and KNN. An average accuracy of 82.75% was obtained through the proposed SVM model to identify the participant's cognitive workload level. The results obtained through this study indicated the efficacy of the EEG features in predicting the level of cognitive load irrespective of the activity. The proposed set of EEG features represents the cognitive indicators that form the basis for developments of augmented cognition systems in our future works.