Modeling of human emotion with effective frequency band during a test of sustained mental task

Abstract

Modeling of human emotion with the effective frequency band of EEG signal plays a significant role in brain signal analysis and physiological research area. This paper describes an approach to model human emotion with the variations of different effective frequency bands of EEG signal during a test when subjected to sustained mental task. For this purpose, the EEG signals are collected when different types of mental task were performed. During a test the subjects were asked to read the text or mathematical problems to solve. The effective frequency band was detected from the variation of functional activities of brain when different mental tasks were performed. The emotional states were identified from the variations of effective band of EEG signal. Then the states were modeled with the effective bands using the subband coefficients of wavelet analysis. When the subjects were reading the text, the most effective band is alpha because of the higher estimation accuracy which indicates the relax state. For problem reading and solving the beta band indicates the higher estimation accuracy which indicate stress. So to model emotions during a test, alpha band is more effective for relax state and beta band is more effective for the state of stress.