An evaluation of the left-brain vs. right-brain hypothesis with resting state functional connectivity magnetic resonance imaging

1,011 individuals aged 7-29 (mean 18.3, SD 5.6; 587 male, 424 female) pooled from 36 sites across the 1000 Functional Connectomes Project and the ADHD-200 typically-developing samples; functional lateralisation measured across 7,266 grey-matter ROIs at 5 mm resolution, giving ~14.1 million intrahemispheric connections

Children, adolescents and young adults, ages 7-29 - overlapping the school-age band but with a mean age of 18.3, so weighted toward late adolescence and young adulthood

Cross-sectional resting-state fMRI on open-access data, recorded as 'quasi-experiment' because the controlled vocabulary has no observational-imaging category. Tests whether strongly lateralised connections COVARY WITHIN individuals, which is what a 'left-brained' or 'right-brained' personal type would require. Grey-matter density differences between homotopic coordinates were regressed out so that structural asymmetry could not drive the result (the authors note a residual structural-functional relationship survives). Motion scrubbing was applied and no motion measure correlated with lateralisation after FDR correction. Internal reproducibility is good - the largest single site (Beijing) matched the rest of the sample at r = 0.85, and dropping the five sites with under 10 subjects changed nothing (r = 0.999). All multiple-comparison correction is FDR q < 0.05. The design's limit is that it tests a neuroimaging phenotype, not behaviour - it shows there is no whole-brain lateralisation dimension for a cognitive style to sit on, but it never measures personality, cognitive style or learning outcomes, so it refutes the mechanism rather than measuring the trait.

Nine left- and eleven right-lateralised hubs were identified, forming two separable networks - so lateralisation of specific FUNCTIONS is real, with left hubs covering language (Broca, Wernicke) and default-mode regions and right hubs covering attention regions. But the covariance test kills the whole-brain phenotype: of 1,620 cross-network comparisons (left-hub connections against right-hub connections) only 16 (1.0%) correlated positively and 20 (1.2%) negatively, and most of even those involved a right hub within 10 mm of a left hub's mirror position. Within networks the correlations are much commoner - 144 of 630 left-hub comparisons (22.9%) and 329 of 990 right-hub comparisons (33.2%) - but almost all of those shared a node (141/144 and 314/329). Lateralisation is therefore a LOCAL property of individual nodes and subnetworks, not a global one; the data are not consistent with a whole-brain phenotype of greater left- or right-brained network strength across individuals. Lateralisation increased slightly but significantly with age (left r = 0.08, p = .009; right r = 0.09, p = .004), and no sex differences were found either on mean laterality or across the 195 hub-connection comparisons.

Not applicable to the claim at issue; the finding is that the proposed individual-difference dimension does not exist, so there is nothing for heredity to explain.

Partly confirms and partly contradicts the two prior whole-brain rs-fcMRI studies. The spatial distribution of lateralised connectivity is "broadly consistent" with Liu et al. (2009) and Tomasi & Volkow (2012) - left for language and default-mode regions, right for attention regions. But BOTH of those studies reported males more strongly lateralised than females, and this paper finds no sex effect at all, so that null is a failed replication of the prior literature rather than a fresh confirmation of it. The central local-not-global finding is a novel test that the prior studies did not run; we have not checked whether it has since been independently replicated, so no claim is made either way.