EEG correlates of Emotion and Mental Fatigue

Abstract

A change in the brain network necessary to produce a particular neural response is called a neural correlate, and the connectivity amongst various regions of the brain is referred to as the functional connectivity. The two neural correlates, the activation regions in the brain and the functional connectivity, of emotion and mental fatigue have been investigated in many studies using electroencephalography, but the results are inconsistent. In addition, the similarity between functional connectivity in various emotions and mental fatigue has not yet been investigated. This thesis investigates the two neural correlates for four emotions and mental fatigue and also finds the similarity between the two neural correlates of emotion and those of mental fatigue by using the asymmetry index, by proposing emotion-weighted critical networks based on the graph network analysis, and by proposing a subject-independent classifier based on the convolutional neural network. The asymmetry index is used to identify the activation regions in the brain. The emotion-weighted critical networks are generic models for representing the functional connectivity in emotional states. The classifier is used to classify mental fatigue and is pre-trained on power spectral density and graphical strength as the input features. The simulations are performed on public datasets for emotion (SEED IV) and mental fatigue (sustained-driving task), respectively. The results from the simulations are three-fold. Specifically, the frontal region exhibits different activation trends for all emotions and mental fatigue, the functional connectivity for mental fatigue is similar to that for sad emotion, and mental fatigue is correlated with the sad emotional state.