Study Of Cognitive Mechanism Based On EEG Cortical Network Analysis

EEG is a comprehensive reflection of brain electrical activity on the scalp surface,which contains abundant physiological and psychological information.In cognitive science,EEG analysis method has become an important research tool.Generally speaking,time and space are two basic aspects of brain analysis.In time,EEG has high resolution and can achieve millisecond speed.However,it is not satisfactory in spatial resolution.In spite of this,the spatial resolution of EEG signals can be improved properly by source location.Therefore,in this paper,we used the methods of EEG source localization and effective connectivity,with pattern recognition methods as auxiliary means,to constructed cortical networks of the task states(visual and auditory)and resting state of scalp EEG signal,and further explored the cognitive mechanism.The main contents of the study include the following three aspects:First,based on the task of facial emotion discrimination,the cognitive mechanism of N170,a negative component whose peak appeared at about 150~200ms after face stimulation,was explored.Firstly the scalp N170 time-varying network was constructed and the effects of EEG reference technologies(average reference and reference electrode standardization technology)on N170 were studied.These results showed that compared to the average reference,the N170 network was more stable and its time characteristic was closer to the source level under the reference electrode standardization technology.Then we made comparative analysis among the cortical time-varying networks of N170 induced by the positive,neutral and negative emotional faces,respectively,and found the difference in time among the sources of N170 induced by different emotional faces,revealing the influence of facial emotion on N170.Second,based on the auditory cocktail effect experiment,we explored the cognitive differences between the tracking error and the tracking correct states.Firstly,a EEG feature extraction method based on rhythmic entropy was developed,which made the two states of the subjects can be well discriminated.Then we reconstructed the brain regions according to classification rate and these brain regions were as network nodes,thus establishing cortical time-varying causal connectives and analyzing their difference.It found that the medial frontal cortex,the left auxiliary motor area and the right auxiliary motor area are the brain regions with the largest difference in two cognitive status,revealing the left dominance of the brain in the correct state and the right dominance of the brain in the wrong state.Third,based on the resting state EEG data,the cognitive effects of music assisted therapy on students with left behind experience were explored.Firstly,power spectrum analysis of resting state EEG data in different frequency bands was carried out,finding that the EEG power spectrum has a significant differences among different stages of music assisted therapy under theta band and a tendency of spectral energy from frontal area to occipital region was obtained by support vector machine classification.Then,based on the obtained frequency band,granger causality analysis in frequency domain was performed and found significant changes of both the network core nodes and the effect connectivity values caused by the music assisted therapy,thus revealing the network plasticity change of the college students with left behind experience before and after music assisted therapy.