Individual Differences In Loneliness And Its Neural And Genetic Basis

Loneliness is generally defined as an aversive experience that resulting from a perceived discrepancy between expected and actual social relationships,either quantitatively or qualitatively.Chronic loneliness can lead to many physical and mental health problems.With an increasingly aging population in contemporary societies,loneliness has gradually become a public health problem.Investigating the mechanisms by which loneliness occurs and is maintained can lead to a better understanding of how loneliness impairs health,so that effective interventions can be developed.To date,the most complete theoretical model of loneliness is the evolutionary theory of loneliness.According to this theory,loneliness increases hypervigilance to social threats.The hypervigilance to social threats may result in a vicious circle that is detrimental to long-term physical and mental health.Although research evidence supports the evolutionary theory of loneliness well,existing research still has many limitations.The evolutionary theory of loneliness focuses on explaining the physiological mechanisms by which loneliness impairs health,but not on the psychological factors,cognitive neurological and genetic bases of individual differences in loneliness.To address the above issues,this thesis uses 6 studies to investigate individual differences in loneliness from behavior,brain function,and genetics levels.In the behavioral research section,we used network analysis to examine psychological factors that influence individual differences in loneliness and interactions between loneliness and daily emotions with cross-sectional and longitudinal research designs,respectively.Study 1 collected 32 psychological measures from 523 college students on socio-emotional,personality,self-esteem,and emotional intelligence.The relationships between these variables were analyzed using Spearman’s correlation and Gaussian graph network model methods,respectively.Compared to the direct correlation analysis,the network model eliminated most of the indirect connections.Nodes with positive connections to loneliness included hostility,feelings of rejection,neuroticism,depression,feelings of danger and loss of control,and social avoidance;nodes with negative connections included two dimensions of self-esteem(appear and social),emotional social support,and friendship quality.Other variables were not directly connected to loneliness,but mainly indirect connected through the neuroticism node.In study 2,we measured state loneliness and momentary positive and negative affect in 211 college students over a period of 13 days,using an experience sampling method.The network analysis revealed that state loneliness has a strong autoregressive effect and is positively correlated to sadness,fear,alertness,and hostility in the temporal and contemporaneous networks.In the temporal network,state loneliness formed mutually reinforced feedback loops with the feelings of alertness and hostility.The between-subject network suggested that the feeling of fear mediated the connections between loneliness and alertness,hostility,and anxiety.The centrality analysis showed that loneliness has a high strength centrality in the temporal and contemporaneous networks.In the brain function research section,we used two task-based f MRI experiments and machine learning methods to study the prediction of loneliness by the brain functional connectivity pattern in social cognitive processing.The prediction analysis process included signal extraction of brain ROI time series,feature matrix construction,Elastic Net algorithm-based predictive model fitting,and cross-validation.Two ROI templates were used,one for the 36 ROI brain regions involved in social cognitive and emotional processing(Social ROI)and another for the cortical template containing 500 ROIs(Schaefer ROI).In the predictive analysis,we constructed the functional connections within the Social ROI and their connections with the Schaefer ROI as functional connectivity features.In addition,based on the above connectivity matrix,we calculated the positive and negative connection strengths of each node as node strength features,respectively.Then,the loneliness score was predicted and analyzed under different task states using the functional connectivity features and node strength features.The 10-fold cross-validation method was used in the model evaluation to examine the generalization ability of the prediction model,and the explained variance(EV)was used as an evaluation metric of the prediction performance.Study 3 used a Theory of Mind(To M)paradigm and 157 subjects participating in the task.The paradigm used videos as material to record the functional brain activity of the subjects while watching videos of social interaction(Mental)and random movements(Random).We analyzed and compared the predictive performance of functional connectivity and node strength on loneliness in different task conditions.The results revealed that,with functional connections as features,the Task > Rest condition had the best performance(EV = 0.1112,p = 0.006),and the correlation coefficient between the predictived values and the observed loneliness score was r = 0.408.For all connections with predictive ability,connections between Social ROIs and the default network had the largest proportion,followed by connections between Social ROIs and the sensorimotor network.With node strength as features,the Random > Rest condition had the best performance(EV = 0.101,p = 0.004),and the correlation coefficient between the predictived values and the observed loneliness score was r = 0.332.Similar to the results of functional connectivity,nodes with predictive weights were more distributed within the default network and the sensory-motor network.Study 4 used a social exclusion task with 107 participants.The experiment was a virtual 3-player online ball-toss game(Cyber ball).The experiment was divided into two stages,with the first stage being the inclusion condition and the second stage being the exclusion condition.The data processing procedure of the experiment was identical to the To M paradigm,and we analyzed and compared the predictive ability of functional connectivity and node strength on loneliness in different task conditions.In the prediction analysis with functional connectivity,none of the predictions reached the level of statistical significance.In the prediction analysis of node strength,the best predictive performance was found in the Inclusion condition(EV = 0.1249,p = 0.006),and the correlation coefficient between the predictived values and the observed loneliness score was r = 0.332.The nodes with predictive ability were mainly distributed in the default mode,sensorimotor,and control networks.In the genetic research section,we used an imaging genetics research paradigm to explore the influence of genetic factors on individual differences in loneliness.Study 5 used a candidate gene approach to analyze the role of the BDNF gene Met/Val polymorphism in modulating the relationship between loneliness and brain white matter microstructure.The study collected DTI data from 162 individuals and analyzed voxel levels of FA and RD indices of brain white matter structure using the TBSS method.The results showed that the Met/Val polymorphism modulates the correlation between loneliness and white matter structure.The relationship between loneliness and FA values(or RD values)was different in the Val/Met group and Val/Val group,and the most significant interaction effect was located in the superior longitudinal tract(SLF).In addition,a global effect between FA(RD)and loneliness was found,with whole-brain mean FA values negatively correlated with loneliness within the Val/Met group(r =-0.415,p < 0.001).Study 6 performed an inter-subject representational similarity analysis between the similarity of the polygenic score of loneliness and the similarity of resting-state functional connectivity,and explored the relationship between the spatial distribution of this similarity and brain gene expression.A total of 550 subjects participated in the study.After constructing polygenic scores for loneliness,we estimated the inter-subject similarity matrix on the genetic predisposition to loneliness,and the inter-subject brain functional connectivity similarity matrix by resting-state functional connectivity patterns.Inter-subject representational similarity analysis were then obtained for 500 brain regions by analyzing the Spearman correlation of the two similarity matrices in each brain region.After obtaining the distribution of representational similarity across the whole brain,we analyzed the gene expression basis of this spatial distribution using the Allen brain atlas data.The algorithm used for gene expression analysis was partial least squares regression(PLSR).After obtaining the association weights for gene expression and brain representational similarity,the list of genes was ranked according to their relative contribution.To explain the biological meaning of gene expression patterns,we performed an enrichment analysis based on the Gene Ontology(GO)database for the ranked genes list.The results showed that the brain regions with the strongest positive correlation in the inter-subject representational analysis were mainly concentrated in the default network,including the left m PFC,bilateral PCC,and temporal pole.PLSR analysis showed that,the PLS1 component,with the strongest correlation effect,explained 0.175 of the variance in the inter-subject representational similarity in the brain,and 0.186 of the variation in gene expression.GO enrichment analysis indicated that the genes most strongly associated with the PLS1 component may be involved in biological processes such as cellular communication,immune system processing,leukocyte activation,inter-synaptic chemical signaling.By integrating the results of these studies,this thesis provided new evidence to support the evolutionary theory of loneliness at different levels.First,at the behavioral level,based on experience sampling data,we found mutually reinforced feedback loops between state loneliness and feelings of "hostility" and "alertness," suggesting that state loneliness is associated with emotional hypervigilance.This result extends the hypothesis of hypervigilance in loneliness to daily emotions.Then,using f MRI techniques,we found that brain functional network patterns during social cognitive and emotional processing significantly predicted individual differences in loneliness,and that functional connections between the default network and social brain regions contributed most to the prediction.This result provides additional details on the neural basis of individual differences in loneliness.Finally,by characterizing similarities in polygenic scores and patterns of functional connectivity between individuals,we found that genetic factors associated with loneliness may modulate the functional activity patterns of the default network,and further analysis suggests that genetic and brain function similarity patterns are associated with immune-related gene expression.This result provides a possible neural basis for the association between loneliness and inflammatory gene expression.