EEG-based Estimation of Cognitive Workload Across Multiple Tasks.

Area of Science:

• Neuroscience
• Cognitive Science
• Machine Learning

Background:

• Accurate estimation of human cognitive state is crucial for safety and performance in high-risk environments.
• Existing methods for cognitive state assessment can be intrusive or lack real-time applicability.

Purpose of the Study:

• To develop and evaluate a method for predicting cognitive workload levels using electroencephalography (EEG) signals.
• To assess the accuracy of machine learning algorithms in differentiating between high and low cognitive load states.
• To determine if EEG features can predict cognitive load irrespective of the specific cognitive task.

Main Methods:

• Participants completed three distinct cognitive tasks under both high and low workload conditions.
• Electroencephalography (EEG) signals were recorded during task performance.
• Machine learning algorithms, including Support Vector Machine (SVM), Random Forest (RF), and K-Nearest Neighbors (KNN), were employed for analysis.
• The accuracy of the models in predicting cognitive workload levels was evaluated.

Main Results:

• An average accuracy of 82.75% was achieved by the proposed SVM model in identifying cognitive workload levels.
• The study demonstrated the efficacy of EEG features in predicting cognitive load.
• The predictive accuracy of EEG features was found to be independent of the specific cognitive activity.

Conclusions:

• EEG signals contain reliable indicators of cognitive load.
• Machine learning models, particularly SVM, can accurately predict cognitive workload from EEG data.
• The findings support the development of augmented cognition systems for real-time cognitive state monitoring.