A Study Of Working Memory Based On EEG Brain Networks

Working memory is the ability of individuals to temporarily store and manipulate information during the performance of cognitive tasks,and its deficiencies can also contribute to a number of psychiatric disorders(e.g.schizophrenia)and neurodegenerative diseases(e.g.Alzheimer’s disease).Therefore,the formation and neural mechanisms of working memory has far-reaching implications for improving memory capacity and treating cognitive neurological disorders such as Alzheimer’s disease.In recent years,with the rapid development in the field of Brain-Computer Interface(BCI),the study of working memory based on EEG signals has become more and more mature.In this paper,the community structure studies and node importance analysis are conducted in the persective of brain networks under different working memory states.On this basis,a transcranial electrical stimulation paradigm that enables adaptive modulation of different working memory tasks and subjects is proposed,which effectively enhances the working memory capacity of the subjects.First,the study designed a new set of experimental paradigms to induce word working memory by memorizing phrases.To maintain control variables and prevent the effect of different subjects’ linguistic bases on the results of the experiment,a foreign language that none of the subjects had been exposed was deliberately chosen for this experiment.After pre-processing the EEG signals,their spectrum was analyzed.It was found that the spectrum power of the prefrontal lobe in the theta band and the left prefrontal lobe in the beta,gamma band increased with the difficulty of memory task.To further investigate the operational mechanisms of working memory,we constructed functional brain networks in the characteristic frequency bands based on the Phase-Locked Values(PLV)between channels and proposed an adaptive immunogenetic community structure detection algorithm that satisfies natural selection.This study found that the stability of the brain network community structure increases with the complexity of the working memory task,and the core nodes play an important role in inter-community communication.To this end,we propose a K-order propagation number algorithm to calculate the importance of nodes in brain networks.The algorithm is based on the infectious disease model,setting each node as an infectious source,and measuring node importance by comparing the number of nodes infected by the infectious source within K steps,the K varies from small to large which reflects the local and global network properties of complex networks.Achieving the purpose of multi-scale study of complex networks and compensating for the shortcomings of traditional structural entropy.The results of this algorithm show the importance features of nodes in brain networks are closely related to working memory states,and the highest classification accuracy can reach 92% by using the node importance sequences as feature vectors in a Support Vector Machine(SVM)for state classification.Finally,we designed a new experimental paradigm for transcranial electrical stimulation with reference to the close association between nodal importance features and working memory.Through controlled variable experiments,it was found that the newly established stimulation paradigm has more significant enhancement on working memory compared to the traditional stimulation paradigm.