No strong evidence that social network index is associated with gray matter volume from a data-driven investigation

Abstract

Recent studies in adult humans have reported correlations between individual differences in people's Social Network Index (SNI) and gray matter volume (GMV) across multiple regions of the brain. However, the cortical and subcortical loci identified are inconsistent across studies. These discrepancies might arise because different regions of interest were hypothesized and tested in different studies without controlling for multiple comparisons, and/or from insufficiently large sample sizes to fully protect against statistically unreliable findings. Here we took a data-driven approach in a pre-registered study to comprehensively investigate the relationship between SNI and GMV in every cortical and subcortical region, using three predictive modeling frameworks. We also included psychological predictors such as cognitive and emotional intelligence, personality, and mood. In a sample of healthy adults (n = 92), neither multivariate frameworks (e.g., ridge regression with cross-validation) nor univariate frameworks (e.g., univariate linear regression with cross-validation) showed a significant association between SNI and any GMV or psychological feature after multiple comparison corrections (all R-squared values ≤ .1). These results emphasize the importance of large sample sizes and hypothesis-driven studies to derive statistically reliable conclusions, and suggest that future meta-analyses will be needed to more accurately estimate the true effect sizes in this field.

Keywords: Cross-validation; Gray matter volume; Predictive modeling; Social network index.

Fig. 1.. Illustration of three predictive modeling frameworks.

(A) Framework 1 performed ridge regression with cross-validation using selected features. Features were selected within the cross-validation loop based on univariate correlations. The hyperparameter of ridge regression was tuned using a nested cross-validation loop. (B) Framework 2 performed ridge regression with cross-validation using all features. (C) Framework 3 performed univariate ordinary least-squares linear regression between each feature and SNI within the cross-validation loop.

Fig. 2.. Predicting SNI with all GMV and psychological/demographic features.

(A) Results from analyses of Framework 1. The selection frequency (blue bars) of the top (most frequently selected) eighteen features over the 2000 iterations of the outer cross-validation loop (left) and the mean prediction accuracy (red vertical line, assessed with Pearson’s r) averaged over the 2000 outer cross-validation iterations compared to the null distribution generated with permutation (right). The mean prediction accuracy assessed with prediction R² = 0.060, p = 0.136. (B) Results from analyses of Framework 2. Model coefficients (blue dots) and standard deviations (black bars) of the top eighteen features (left) and the mean prediction accuracy (red vertical line, assessed with Pearson’s r) averaged over the 2000 outer cross-validation iterations compared to the null distribution generated with permutation (right). The mean prediction accuracy assessed with prediction R² = −0.023, p = 0.404.

Fig. 3.. Descriptive effect sizes between SNI and every cortical GMV.

The descriptive effect size of the univariate associations between all cortical regions and SNI are shown to provide background for future studies that could test hypotheses based on these results. Four renderings of the univariate Pearson correlations (uncorrected) between individual cortical regions and SNI are projected on the pial surface for (A) the lateral view of the left hemisphere, (B) the superior view of both hemispheres, (C) the lateral view of the right hemisphere, (D) the medial view of the left hemisphere, (E) the inferior view of both hemispheres, and (F) the medial view of the right hemisphere. These effect sizes provide recommendations for the sample sizes required to test associations between specific cortical regions and SNI, shown in Appendix E.