Brain Functional Similarity Between Chinese And Western Personality Traits

Objective:Personality is a relatively stable behavior pattern formed by the combination of genetic factors and cultural environment.The traditional pencil-paper test by the etic NEO test is the main measuring to evaluate individual behavior.However,the phenomenon of"trait shift"of the Big Five personality traits occurs in non-Western culture backgrounds.In addition,the results of traditional paper-pencil test are affected by various factors,such as individual’s subjective initiative,test motivation and current state.How to avoid the subjective influence of penpaper test is a major challenge in personality research.Previous studies have confirmed that whatever how genes and cultural environment shape individual behavior character,the personality can stably reflected in brain function and structure.In order to explore the specific relationship between personality traits and brain features,we comprehensively considered the similarity of personality traits in Chinese and Western cultural backgrounds,and used Latent Profiles analysis（LPA）.And the mediating effect analysis was used to explore the similar characteristics of personality traits and brain local functional structure in different cultural backgrounds.On this basis,the Structural equation model（SEM）and the classification and prediction model based on machine learning methods were established to deeply explore the relationship between the Big Five personality traits and the global functional network,which laid a theoretical foundation for the neurobiological measurement model of personality.Method:1)Latent Profiles analysis of Western personality trait corresponds to Chinese five pattern personality traits.Adult healthy subjects were collected from September 2017 to December 2019 in Taiyuan City and its surrounding areas.General demographic data,NEO-FFI and Five Pattern Personality Inventory（FPPI）were collected.LPA was used to profile the scores of the big five trait dimensions in the subjects’NEO-FFI,and the obtained profile classification was analyzed by ANOVA,and Logsitic regression was conducted with the five pattern trait dimensions of the subjects’FPPI to obtain the relationship between the five-factor trait and the five pattern trait.2)The regional homogeneity and VBM of Western personality traits and five pattern personality traitsAdult healthy subjects from September 2017 to December 2019 in Taiyuan City and its surrounding areas（the same subjects as the first part of the study）were collected.General demographic data,Eysenck personality test（EPQ）,FPPI and Resting state Functional Magnetic Resonance imaging were collected.FMRI)and Functional Magnetic Resonance imaging（MRI）.Regional Homogeneity（Re Ho）was calculated after preprocessing the imaging data.Grey Matter volume（GMV）was calculated by voxel-based Morphometry（VBM）.Correlation analysis,stepwise linear regression model and mediating effect model were used to analyze the relationship between Eysenck personality traits and five pattern traits on local brain function and structure.3)Research on the anti-correlation of neuroticism and extraversion in default networkA total of 236 subjects who participated in the study from December 2019 to May2020 were included.General demographic data,NEO-FFI,f MRI and MRI data were collected.Independent Component Analysis（ICA）was performed to separate the Default mode network（DMN）after preprocessing the imaging data.The dorsal default network（Ventral DMN,v DMN）and dorsal default network（d DMN）were further distinguished based on previous reports.The Network coherence（NC）in each central node region of the two subnetworks was calculated,and the NC value was used as the observation variable to establish the SEM of the Big Five personality traits and DMN and its subnetworks.4)Classification and prediction based on the Big Five personality traits related functional networkA total of 236 subjects who participated in the study from December 2019 to May2020 and 121 subjects who participated in the study from September 2020 to June 2021were included.General demographic data,neo-FFI,f MRI and MRI data were collected.After preprocessing the imaging data,python3.6 platform was used to construct the whole brain functional network.Based on the first batch of sample data,the leave-one-out cross validation（LOO-CV）method was used to calculate the stable canonical correlation functional network of personality traits.Linear Support Vector Machine（LSVM）and Relevance Vector Machine（RVR）in Machine learning classification algorithm were used to classify and predict each personality trait.After 1000 permutation tests,the stable classification and prediction models were obtained,and cross-sample cross-validation was carried out in the second batch of sample data.Results:1)Latent Profiles analysis of Western personality trait corresponds to Chinese five pattern personality traits.Finally,219 subjects were included.Taking gender as a grouping variable,it was found that male subjects had significantly higher scores of conscientiousness than female subjects（t=2.16,P=0.032）,and female subjects had significantly higher scores of neuroticism than male subjects（t=-3.15,p=0.002）.In addition,the scores of sun（Tya）,Yin-Yin-peace（Yy）and Shaoyin（Syi）of male subjects were significantly higher than those of female subjects（p<0.032）,while the scores of Taiyin（Tyi）were significantly lower than those of female subjects（p<0.032）.The results of LPA analysis showed that the Big Five personality traits could be further divided into four potential personality categories（C1,C2,C3,C4）.The fit is good.Entropy is the highest among the four categories and is greater than 0.80,indicating a good model fit.The results of Logistic regression analysis showed that Sya（X²=25.39,p<0.001）,Syi（X²=8.67,p=0.034）and Ty（X²=127.267,p<0.001）were stable predictors of the four latent categories of Western personality.Among them,the characteristics of the lunar system are C1 and C2（β=0.42,P<0.01）,C1 and C3（β=-0.53,p<0.01）,C2 and C3（β=-0.10,p<0.01）,C2 and C4（β=-0.49,p<0.01）,respectively.Stable predictors of potential categories for C3 and C4（β=-0.46,p<0.01）.The characteristics of Sya were stable predictors of potential categories of C1 and C4（β=0.44,p<0.01）,C2 and C3（β=0.53,P<0.01）,and C3 and C4（β=0.29,p<0.05）.Syi trait was a stable predictor of C1and C4（β=-0.28,p<0.05）.Based on the results of LPA,the four latent categories of Western personality were taken as the grouping variables,and the distribution of five-state traits in these four groups was analyzed by one-way ANOVA.Analysis of variance showed that Tya（F=8.873,p<0.001）,Sya（F=20.652,p<0.001）,Yy（F=17.925,p<0.001）,Syi（F=5.786,p=0.001）and Tyi（F=94.963,p=0.001）were significantly different in the four groups.p<0.001)were significantly different.Post-hoc test results showed that there were significant differences in the scores of the five states in the four potential categories（p<0.05）.2)The regional homogeneity and VBM of Western personality traits and five pattern personality traitsA total of 170 subjects were enrolled.The correlation analysis between FPPI and EPQ showed that there was a significant positive correlation between Tya score and E score（r=0.371,p<0.01）.Tyi score was negatively correlated with E score（r=-0.408,p<0.01）,and positively correlated with N score（r=0.688,p<0.01）and P score（r=0.343,p<0.01）.There was a significant positive correlation between Sya score and E score（r=0.503,p<0.01）.Syi score was negatively correlated with E（r=-0.197,p<0.05）and P（r=-0.256,p<0.01）.The Yy score was positively correlated with E（r=0.196,P<0.05）,and negatively correlated with N（r=-0.329,p<0.01）and P（r=-0.348,p<0.01）.Functional correlation analysis showed that Tya was positively correlated with the Re Ho value of the right Superior temporal gyrus（STG）（r=0.391,p<0.0001,Cluster size=30 voxels）.Tyi positively correlated with the Re Ho value of the right medial prefrontal cortex（m PFC）（r=0.292,p<0.0001,Cluster size=22 voxels）.Structural correlation analysis showed no significant correlation between five-state traits and brain GMV.Stepwise regression analysis showed that N（β=0.145,P=0.042）and E（β=0.421,p<0.001）could predict the Re Ho value of right superior temporal gyrus.When N（β=0.341,p<0.001）could predict Re Ho in the right medial prefrontal lobe.The results of the mediating effect model showed that in the first model,E had a significant partial mediating effect between Tya and the Re Ho value of the right superior temporal gyrus（CI=[0.0015,0.0065]）,and in the second model,N had a significant complete mediating effect between Tyi and Reho value in the right medial prefrontal cortex（CI=[0.0009,0.0066]）.3)Research on the anti-correlation of neuroticism and extraversion in default networkA total of 223 subjects were included in the study.The correlation analysis of DMN components and personality traits based on group level showed that neuroticism score was significantly positively correlated with right m PFC（r=0.323,FDR-p<0.001）.Open to experience score was positively correlated with left MTG（r=0.352,FDR-p<0.001）.Multiple linear regression showed that right-sided m PFC and left-sided MTG could stably predict individual neuroticism scores（adjust-R²=0.136 p<0.0001）.However,left m PFC and left MTG could stably predict the score of openness to experience（adjust-R²=0.096,p<0.0001）.The standardized regression coefficients of the regressors in the above model reached the significance level（p<0.05）.The results of the first SME showed that both neuroticism and openness to experience had significant effects on DMN.Neuroticism was negatively correlated with DMN（β=-.452,p=.030）,and openness to experience was positively correlated with DMN（β=.760,p=.014）.The second SEM results showed that neuroticism and openness to experience had opposite patterns on d DMN and v DMN.Neuroticism had a positive predictive effect on v DMN（β=.329,p=.205）and a negative predictive effect on d DMN（β=-.603,p=.012）.On the contrary,openness to experience positively predicted d DMN（β=.309,p=.030）and negatively predicted v DMN（β=-.483,p=.009）.4)Classification and prediction based on the Big Five personality traits related functional networkUltimately,the first dataset included 225 subjects.The second dataset included 118subjects.The leave-one-out method was used to select the relevant features with the occurrence probability greater than 95%into the personality trait network.Finally,five personality networks were obtained.In the feature network of amenity,29 stable connection features are finally obtained,accounting for 0.083%of the total features.In the conscientiousness feature network,31 stable connection features were finally obtained,accounting for 0.1%of the total features.In extraversion network,59 stable connection features were finally obtained,accounting for 0.19%of the total features.In the neurotic network,17 stable connection features were finally obtained,accounting for 0.05%of the total features.In the open experience network,43 stable features were obtained,accounting for 0.14%of the total features.In the test set,we evaluate the classification quality of SVM models.The results showed that among the five personality networks,the model quality of the agreeableness network was the highest,with an accuracy of 0.740,a sensitivity of 0.808,and a specificity of 0.806.The model accuracy was 0.711,sensitivity was 0.72,and specificity was 0.692.After permutation test,the accuracy and AUC of the agreeableness feature network model and the open to experience feature network model were significantly higher than those of the random distribution(p _ACC<0.001,p _AUC<0.001).After cross-sample cross validation,the classification efficiency of the agreeableness feature network is poor.The significance level of classification accuracy and AUC value was greater than 0.05 after permutation test.Although the classification accuracy of open-of-experience feature network decreased,it still maintained good classification efficiency,with an accuracy of 63.3%,a sensitivity of0.53 and a specificity of 46.77.After permutation test,the accuracy and AUC values of the Openness were significantly higher than those of the random distribution(P_ACC<0.001,p _AUC<0.001).RVM was used to train the score prediction model of personality traits according to the characteristic network of each personality trait.The results showed that there was a significant positive correlation between the predicted value of the open-experience feature network prediction model and the real value（r=0.448,p=3.3×10-5）,and there was also a significant positive correlation between the predicted value of the agreeableness feature network prediction model and the real value（r=0.354,p=0.0013）.In addition,the true value of the prediction range of the neurotic feature network model（r=0.244,p=0.030）.After permutation test,the correlation coefficients of the characteristic networks of openness to experience and agreeableness were significantly higher than those of random distribution（P_r<0.001）.The cross-sample cross-validation results showed that the correlation coefficient between the predicted value and the real value of the agreeableness feature network was low and did not reach the significance level（r=-0.095,p=0.305）.The predicted value of EEMN was positively correlated with the real value（r=0.271,p=0.002）.After permutation test,the correlation coefficient between the predicted value and the real value of the EEMN was significantly higher than that of the random distribution（P_r<0.001）.Conclusion:1.The latent profile of the Big Five personality traits can be corresponding to the five-state traits,showing that each five-state trait contains different types and levels of five-factor traits.Therefore,the Big Five personality traits have richer connotations in theory.2.The solar trait in the five-state trait and extraversion and neuroticism in Eysenck personality trait have similar local functional activity characteristics in the brain,and the effect of extraversion on the solar trait and the local consistency in the superior temporal gyrus region is related.Similarly,lunar traits and neuroticism have similar characteristics of local functional activity in the brain,and neuroticism can affect the role of lunar traits and local consistency in the medial prefrontal region.3.Neuroticism and openness to experience affect the functional connectivity of default network in reverse mode.Moreover,neuroticism affected both subnetworks of the default network in the same way as openness to experience.Therefore,the functional separation of the default network may correspond to the behavioral characteristics of neuroticism and openness to experience,respectively.4.The trait related functional network corresponding to agreeableness can distinguish people with different agreeableness levels well.The trait related functional network corresponding to openness to experience can not only realize the classification of people with different trait levels,but also predict the level of individual openness to experience stably.