Psychological resilience mediates the protective role of default-mode network functional connectivity against COVID-19 vicarious traumatization

The COVID-19 pandemic increased psychological stress through rapid spread, lockdowns, and uncertainty. Media overexposure to others’ suffering can produce vicarious traumatization (VT), which entails stress, burnout, and symptoms overlapping with anxiety, depression, and PTSD. Neurobiological mechanisms of COVID-related VT are largely unknown. The study’s primary research question was whether pre-pandemic resting-state brain functional connectivity predicts VT during the pandemic. A second question was whether psychological resilience mediates the relationship between brain connectivity and VT. The authors hypothesized that lower functional connectivity density (FCD) and resting-state functional connectivity (RSFC) in default mode network (DMN) regions would predict higher VT, and that psychological resilience would mediate these brain–VT associations.

VT is influenced by situational and personal factors including socioeconomic status, trauma history, coping, affect tolerance, and particularly psychological resilience. Resilience is protective against PTSD and vicarious trauma, and is negatively associated with COVID-related distress indicators. Neuroimaging research implicates large-scale networks, especially the DMN (including limbic and prefrontal regions), in stress and stress-related psychopathology. Altered DMN activity/connectivity relates to perceived stress, PTSD, mood and anxiety disorders, while greater DMN integrity and volume in regions such as mPFC and hippocampus have been associated with resilience. These findings suggest that DMN connectivity may underlie individual differences in VT, potentially via resilience as a mediator.

Design: Prospective cohort with pre-pandemic MRI/behavioral assessment (T1: Oct 2019–Jan 2020) and pandemic behavioral assessment (T2: Feb–Apr 2020). Participants: 151 healthy, right-handed Chinese university students (19–27 years) with no psychiatric/neurological history. MRI and questionnaires at T1; 115 provided T2 data; after motion exclusion (mean FD > 0.25 mm), final N=100 (42 male/58 female). Behavioral measures: VT assessed at T2 using the 38-item Vicarious Traumatization Questionnaire (VTQ; total 38–190; higher=worse VT; Cronbach’s α=0.95). Psychological resilience measured at T1 and T2 with 10-item CD-RISC (10–50; higher=greater resilience; α=0.83 at T1, 0.88 at T2). Mean of T1 and T2 CD-RISC used as resilience index (no significant T1–T2 difference; r=0.65, p<0.001). Covariates included family socioeconomic status (SSS; α=0.76) and Self-Rating Life Events Checklist (SRLEC; number and impact; α=0.91) at T1. MRI acquisition: 3T Siemens Trio with 12-channel head coil. Resting-state fMRI: EPI, TR=2000 ms, TE=30 ms, 240 volumes, 30 slices, voxel size 3.75×3.75×5 mm3, slice thickness 5 mm, FOV 24×24 cm2, matrix 64×64, flip angle 90°, eyes closed. Structural T1: TR=1900 ms, TE=2.26 ms, FA=9°, 176 slices, 1 mm isotropic. Preprocessing: DPABI—discard first 10 volumes, slice timing, realignment, co-registration, DARTEL normalization, resampling to 3 mm isotropic, 6 mm FWHM smoothing, linear trend removal, band-pass 0.01–0.08 Hz. Nuisance regression: white matter, CSF, motion parameters. Motion: mean FD computed; exclude FD>0.25 mm; motion scrubbing at FD>0.50 mm (mean 16 frames scrubbed; ~7%). FCD computation: Whole-brain voxelwise Pearson correlations; threshold r>0.6 to define suprathreshold connections; binary FCD per voxel equals count of suprathreshold correlations. Grand mean scaling applied to improve normality; normality verified across networks. Statistical analyses: Whole-brain FCD–VTQ correlation controlling age, sex, FD; sex-by-VT interaction tested. GRF correction: voxel p<0.001, cluster p<0.05. RSFC analysis: Seed ROI from FCD–behavior cluster (right ITG); seed time series correlated with all voxels; Fisher z-transform; VTQ–RSFC correlation with same covariates and GRF thresholds. Prediction: 4-fold balanced cross-validated linear regression using FCD or mean seed-based RSFC to predict VTQ; age, sex, FD regressed out; prediction performance assessed as correlation between predicted and observed VTQ with 5000-permutation significance. Network mapping: Overlap of identified clusters with 7-network atlas (Yeo et al., 2011) to quantify relative distribution. Mediation: PROCESS macro—X: ITG FCD or ITG-DMN mean RSFC; M: mean CD-RISC (T1+T2); Y: VTQ; covariates age, sex, FD; 5000-bootstrap 95% CI for indirect effects. Specificity tests added SSS and SRLEC as additional covariates.

• Descriptives: VT at T2 negatively correlated with resilience (mean CD-RISC) r=-0.32, p<0.001; remained after adjusting for age, sex, FD (r=-0.31, p=0.002). VT unrelated to age, sex, FD.
• FCD: Whole-brain analysis identified a single cluster—right inferior temporal gyrus (ITG; BA 20/21)—where lower FCD associated with higher VT (r=-0.35, p<0.001). Cluster size 1809 mm3; peak MNI [63, -6, -30], peak Z=-3.88. No significant sex-by-VT interaction. Cross-validated prediction: correlation between predicted and observed VTQ using ITG FCD = 0.33, p<0.001. Network mapping: ITG cluster overlapped mainly with DMN (RD 52.5%) and AFN (RD 27.5%).
• RSFC (ITG seed): VT negatively associated with RSFC between right ITG and left medial prefrontal cortex (mPFC; BA 32/10), left orbitofrontal cortex (OFC; BA 11/10), right superior frontal gyrus (SFG; BA 6/8), right inferior parietal lobule (IPL; BA 39/40), and bilateral precuneus (BA 7/31). Overall mean ITG–network RSFC correlated with VT: r=-0.42, p<0.001. Cross-validated prediction: predicted–observed VTQ = 0.39, p<0.001. Network mapping of connected regions: majority DMN (RD 33.6%), smaller overlaps with CEN (7.4%) and AFN (1.3%).
• Mediation by resilience: Resilience positively associated with ITG FCD (r=0.24, p=0.02) and ITG–DMN RSFC (r=0.30, p=0.003) after covariate adjustment. Mediation models (controlling age, sex, FD): For ITG FCD → VT, total effect c=-0.39 (p<0.001), direct effect c'=-0.33 (p<0.001), indirect effect via resilience = -0.06, 95% CI [-0.12, -0.01]. For ITG–DMN RSFC → VT, c=-0.44 (p<0.001), c'=-0.38 (p<0.001), indirect effect = -0.06, 95% CI [-0.13, -0.01]. Thus, partial mediation.
• Specificity/robustness: Findings held after additionally controlling for SES (SSS) and life events (SRLEC): VT remained linked to resilience (r=-0.26, p=0.01), ITG FCD (r=-0.37, p<0.001), and ITG–DMN RSFC (r=-0.45, p<0.001); mediation indirect effects remained significant (both -0.04, 95% CIs excluding 0).

Pre-pandemic lower functional connectivity hubness (FCD) in right ITG and weaker ITG coupling with DMN regions (mPFC, OFC, SFG, IPL, precuneus) prospectively predicted higher COVID-19 VT. The ITG is a DMN node with broad cortical and subcortical connectivity supporting higher-order cognition (visual recognition, semantic memory, language), self-referential and social-emotional processes (empathy, mentalizing). Reduced ITG–DMN connectivity may reflect diminished internally oriented thought and empathic engagement, contributing to greater susceptibility to vicarious stress. Overlap with AFN and CEN suggests contributions of affective processing and cognitive control; decreased ITG–CEN coupling may indicate impaired top-down regulation. Psychological resilience was inversely related to VT and positively related to ITG FCD and ITG–DMN RSFC, and mediated their relations to VT, supporting a model where stronger DMN connectivity fosters resilience that protects against VT. These results integrate with prior literature linking DMN structure and function to stress-related psychopathology and resilience, extending the protective role of DMN connectivity to media-driven vicarious trauma during a global stressor.

This study prospectively demonstrates that default mode network connectivity—particularly right ITG hubness and ITG–DMN coupling—protects against COVID-19 vicarious traumatization via psychological resilience. ITG/DMN may be targets for prevention and intervention in stress- and trauma-related risk, including noninvasive brain stimulation and psychotherapy approaches aimed at enhancing resilience.

• MRI was only acquired before the pandemic, precluding analysis of longitudinal changes in brain function and VT.
• VT and resilience were measured via self-report instruments; objective/multimethod assessments are needed.
• The sample consisted of college students, limiting generalizability to broader populations.
• Only ITG emerged in FCD analyses; other DMN regions implicated in trauma (e.g., hippocampus, amygdala, mPFC) were not identified—other modalities/analyses may reveal additional regions.
• Regions were defined using a group-level atlas, which may dilute individual-specific brain–behavior associations.
• The exploratory nature limits immediate clinical applicability; replication and clinical translation studies are required.