EEG-based emotion monitoring in mental task performance

Emotional engagement during mental tasks performance when the difficulty level of mental tasks increases is studied using Electroencephalogram (EEG) recorded by Emotiv Epoch device. A real-time EEG-based emotion recognition algorithm using Valence-Arousal-Dominance emotion model is applied. An experiment with 5 levels of workload is proposed and carried out with 7 subjects. The mental tasks are given to the participants to solve addition problems of increasing complexity. For each task, 3 min is given to complete additions with 2 min rest between the tasks sessions. Performance score for each participant is recorded as well. Emotions are recognized from the EEG during each task using the subject dependent emotion recognition algorithm. The algorithm consists from feature extraction and classification parts. The classifier is trained for each participant using short calibration session with emotion induction before the experiment. The combined features obtained from the training session are used to train a SVM classifier. The following hypotheses are confirmed. Participants have mostly negative, high arousal, low dominance emotions during the mental tasks performance under the stress since high arousal can speed up the mental process and enhance information retrieval, and as the limited time is given to perform the tasks, the feeling of losing control is dominated in the participants that corresponds to low dominance on the scale of Valence-Arousal-Dominance emotion model. At the same time it is found that when the difficulty level of mental tasks increases under the stress, people tend to be more negative and more aroused