Research On Brain Network Patterns Of Decision Making Based On EEG

Understanding the decision-making information processing mechanism in the human brain is a hot topic in the field of high-level cognitive neuroscience research.Decision-making has played a crucial role for application in economic,sociology and artificial intelligence.However,the neural mechanism that accout for decision-making is still left unveiled.Decision-making,a high-level cognitive function,involves multiple brain regions to process the related information.Importantly,the decision processing can be effectively performed within milliseconds in our brain.Electroencephalogram(EEG)has been widely applied in the cognitive neuroscience related researches,due to its easy acquisition and high temporal resolution.In this work,we adopted EEG based brain network analysis methods to reveal the mechanism of decision-making.We conducted the systematic research to explore the decision dynamic network information,the diverse network patterns of decision responses,the neural mechanism underpinning decision differences among different age groups or different emotional traits,and the models of intervention and prediction.The contribution of the current dissertation is as follows.1)Based on the method of large-scale time-varying network analysis,we explored the different decision periods and the diverse network structures of the responses of acceptance and rejection.While subjects accepted unfair offers,the brain displayed a more bottom-up flow derived from the visual cortex to the frontal cortices,but when they rejected unfair ones,it recruited a more top-down mechanism with a much stronger information flow from the frontal to the parietal and occipital cortex.Further,we used the transcranial magnetic stimulation(TMS)to modulate the frontal area,which could suggest the decision network model maybe a potential intervention model to alter decision-making behaviours.2)We studied the difference in decision-making between adolescents and adults under unfair situations.Based on the acceptance rates,the event-related potential(ERP)and the reconfigured pattern from resting to task state during the ultimatum game(UG),we found that in adolescents,higher acceptace and smaller medial frontal negativity(MFN),stronger activation in the occipital area as compared to adults.Results of the study deepened our understanding of the developmental changes in decision-making at different ages,especially from adolescence to adulthood,and the individual established a clear fairness consideration gradually.3)Based on the previous evidence that different emotional traits may affect individual cognitive performance,we explored the mechanism underpinning decision differences between emotional stability and instability adolescents in the unfair situations.Our study uncovered the lower acceptance rates of unfair offers for the emotional instability group than the stability group,as well as the higher MFN of acceptance than rejection for the emotional instability group.With respect to the brain functional network,more frontal-parietal activation to acceptance and frontal-occipital reactivity to rejection were revealed for the emotional instability and stability groups,respectively.Results of our study suggested in the unfair social context,the emotional instability adolescents prefered to reject the unfairness while the emotional stability adolescents payed attention to their interests and make rational decisions.4)From the perspective of resting-state EEG network,we investigated the relationship between resting-state network connectivity and decision-making responses,and adopted the EEG network properties to predict individuals' acceptance rates during the UG.This study showed a relationship between the resting-state frontal-occipital connectivity and the acceptance rate in the alpha band.And increased acceptance rates were accompanied by a larger global and local efficiency and clustering coefficient as well as a shorter characteristic path length.The high-acceptance group demonstrated stronger frontal-occipital connectivity compared to the low-acceptance group.Based on resting-state EEG network properties,we utilized a multiple linear regression model to predict the acceptance rates.The correlation coefficient between the actual performance and predicted performance was 0.58,and the root mean square error was 10.24%.5)Based on the information of an individual's single-trial EEG data,we proposed an EEG-based method of discriminative spatial network pattern(DSNP)to predict individual decision responses.To verify the performance of the proposed method,we recruited two independent subject groups,and recorded the EEGs using two types of EEG systems.The proposed DSNP could efficiently extract the decision feature and a linear discriminate analysis was used to predict the subject's response trial-by-trial.The results showed the DSNP features extracted from EEG network achieved better performance when distinguishing acceptance and rejection,compared with network properties and the MFN amplitude.The performances of the trial-by-trial predictors achieved an accuracy of 0.88 for the first dataset,and 0.90 for the second dataset.The trial-by-trial prediction was based on the inherent and implicit spatial information in brain networks derived from single-trial EEG,which helped us to reveal the implicit decision information and to create an intelligent decision-making system.In conclusion,this dissertation was aimed at the decision-making mechanism,linked with the methods of brain network analysis to explore the decision information process of subjects' brain networks.In these works,the multi-modal fusion of decision behaviours and brain network was realized.The processing mechanism of decision responses,and decision differences among different age groups or groups of different emotional traits were explained.And the models of intervention and prediction based on decision-making behaviours were established.