Associations between handedness and brain functional connectivity patterns in children

The study investigates whether handedness in children is associated with asymmetries in brain functional connectivity, particularly in hand motor cortical regions, and whether such associations have structural counterparts. Handedness is a common lateralized trait with complex genetic, environmental, and developmental influences and a prevalence of about 9–10% for left-handedness. Prior work in adults and smaller pediatric samples has reported mixed and often inconclusive findings regarding structural asymmetries (e.g., cortical thickness, surface area) and white matter microstructure, as well as task-related functional asymmetries. This cross-sectional study of 9–10-year-old children from the ABCD cohort aims to characterize handedness-related functional connectivity patterns and their lateralization, test reproducibility in discovery and replication subsamples, and assess whether structural morphometrics or diffusion metrics differ by handedness. The authors hypothesized that left-handedness would be associated with higher global functional connectivity density (gFCD) in the left-hand motor area (right hemisphere) and lower gFCD in the right-hand motor area (left hemisphere), with corresponding laterality differences across cortical and cerebellar networks.

Background literature shows that hand preference emerges early in development and relates to broader brain asymmetries. Studies of cortical morphometry and white matter asymmetry in adults have yielded inconsistent results regarding handedness effects, with some reporting regional thickness or surface area asymmetries, while large-scale atlas-based analyses often found minimal associations. A UK Biobank study reported subtle but significant handedness-related asymmetries in surface area and thickness in specific regions. Diffusion studies also show inconsistency, with some adult studies noting lower FA in left-handers and pediatric studies showing no differences. Task fMRI studies indicate handedness-dependent activation/deactivation patterns in primary motor cortex and associated regions during motor tasks, and limited rs-fMRI work suggested weaker interhemispheric connectivity in left-handers. A prior large pediatric study reported both increases and decreases in resting-state laterality with age, and some regions’ lateralization associated with right-handedness. Overall, the literature suggests potential functional lateralization differences by handedness, but structural findings are inconsistent, and robust, reproducible brain-wide functional connectivity associations in children remain underexplored.

Design: Cross-sectional analysis of ABCD Study baseline data (age 9–10 years). From ABCD release 2.0, the authors selected 1,800 children: 600 left-handed (L), 600 right-handed (R), and 600 mixed-handed (M). Groups were matched on sex, age, race, scanner manufacturer, family income, head motion, and total brain volume. Participants with excessive head motion (>50% time points with FD > 0.5 mm) were excluded.

Reproducibility: The matched sample was split into Discovery (N=909; 303 per group) and Replication (N=891; 297 per group) subgroups using ABCC’s matched group designation, balancing key sociodemographic factors.

Handedness: Classified via ABCD Youth Edinburgh Handedness Inventory Short Form (EHIS). A continuous handedness score was computed as the average frequency of dominant hand use for writing, throwing, spoon and toothbrush use (always right=100, usually right=50, both=0, usually left=-50, always left=-100).

MRI acquisition: Multi-site 3T scanners (Siemens Prisma, GE 750, Philips) with harmonized protocols. Structural: T1w and T2w (1 mm isotropic). Diffusion MRI: multiband EPI, b=0/500/1000/2000/3000 s/mm^2, 96 directions, 1.7 mm isotropic. Resting-state fMRI: T2*-weighted multiband EPI, TE/TR=30/800 ms, 2.4 mm isotropic, 60 slices, multiband factor=6, ~20 minutes. Functional data and derived maps were handled in CIFTI space (91,282 grayordinates).

Preprocessing and QA: ABCD-BIDS pipelines (similar to HCP) with ANTs nonlinear registration, motion/global signal/WM/CSF regression, respiratory-motion separation, standard QA procedures, and motion censoring. Structural morphometrics (cortical thickness, curvature, sulcal depth) and cortical myelin (T1w/T2w) were derived and mapped to surfaces. Diffusion metrics (FA, MD, axial/longitudinal diffusivity ID, radial/transverse diffusivity tD) were tabulated. Due to scanner variability for diffusion, DTI analyses were limited to Siemens scanners (N=1177; 392 L, 392 R, 393 M).

Functional connectivity density (gFCD): For each grayordinate xo, Pearson correlations (band-pass 0.01–0.10 Hz) with all other grayordinates were computed; gFCD was defined as log of the number of positive edges with R>0.6, repeated for all 91,282 grayordinates. Ipsilateral (intra-hemispheric) and contralateral (inter-hemispheric) gFCD components were computed by restricting edges within or across hemispheres.

Seeds and rsFC: gFCD-guided seeds in primary sensorimotor cortex for right-hand area (MI; left hemisphere, 10 mm radius around center vertex# 7956) and left-hand area (Mr; right hemisphere, 10 mm radius around center vertex# 7985). For each seed, average time series were correlated brain-wide; Fisher z-transformed rsFC maps were generated.

Laterality metrics: A handedness index IH = (gFCD(MI) - gFCD(Mr)) / (gFCD(MI) + gFCD(Mr)) quantified relative gFCD between hand-motor areas. A rsFC laterality index A = (rsFC(MI) - rsFC(Mr)) / |rsFC(MI) + rsFC(Mr)| captured differential connectivity strength with MI vs Mr while normalizing for global connectivity strength. Brain asymmetry maps computed as RH-LH for gFCD, myelin, curvature, thickness, sulcal depth.

ROIs and specialization: ROIs were defined around clusters showing peak between-group differences in A (10 mm radius). The HCP multimodal 360-area cortical parcellation (plus subcortical/cerebellar parcels) contextualized findings. A functional specialization index (0–1) was derived from the parcellation’s RGB mapping to auditory/sensorimotor/visual domains; higher values indicate unimodal specialization.

Homology validation: Task fMRI activation effect-size maps (Cohen’s d) from HCP S1200 (N=997) for left- vs right-hand movements verified that MI/Mr seeds corresponded to hand-motor activations using stringent thresholds (d>1.3).

Statistical analysis: Site/scanner effects removed; FD, age, sex, race regressed out before vertex-wise tests within groups. Two-sample, two-sided t-tests mapped group differences (L vs R) in gFCD and rsFC; multiple comparisons controlled by FDR at PFDR<0.05 across 91,282 grayordinates. ROI-level analyses used ANCOVA (covariates: age, sex, race, site, brain volume, scanner manufacturer, FD). Diffusion ROI analyses were Bonferroni-corrected across 42 major tracts. Reproducibility assessed by repeating analyses in Discovery and Replication subsamples.

• Cohort characteristics (ABCD 2.0; n=11,875): right-handers 79.4%, left-handers 7.14%, mixed-handers 13.49%. Left-handedness more prevalent in boys (7.85%) than girls (6.37%; χ²=8.5, P=0.0036); mixed-handedness did not differ by sex. European ancestry proportion slightly higher in right-handers (0.75) than non-right-handers (0.73; t=2.75, df=3472, P=0.006). In the matched analytic sample (600 L, 600 R, 600 M), groups were balanced on demographics, head motion, and total brain volume.
• Global functional connectivity density (gFCD): Left-handers showed higher gFCD in the left-hand motor area (Mr; right hemisphere) and lower gFCD in the right-hand motor area (MI; left hemisphere) relative to right-handers (vertex-wise PFDR<0.05). Group differences were robust in Discovery and Replication subsamples and in both sexes. No significant group differences using negative-edge gFCD (R<-0.6), indicating effects were driven by positive connections.
• Handedness index (IH): IH = (gFCD(MI) - gFCD(Mr)) / (gFCD(MI) + gFCD(Mr)) differentiated groups with larger effect sizes than raw gFCD contrasts: left vs right-handers Cohen’s d=0.75; left vs mixed d=0.53; whereas gFCD contrast had d≈0.35. IH correlated with EHIS handedness scores in both Discovery (r=0.37, p=2.2e-16) and Replication (r=0.26, p=2.6e-15) cohorts.
• Seed homology: HCP adult task fMRI maps (d>1.3) for hand movement overlapped with MI/Mr ROIs, validating motor homology.
• Intra- vs inter-hemispheric gFCD: Intra-hemispheric gFCD differed by handedness in MI and Mr (PFDR<0.05), with IH separating L, M, R groups; inter-hemispheric gFCD showed no significant L vs R differences, indicating homotopic connectivity did not drive group effects and that differences predominated within hemispheres.
• Laterality of connectivity (A): Both L and R groups showed similar overall A patterns (e.g., positive A in left somatomotor/premotor/opercular/retro-insular cortices and negative A in their right-hemisphere homologues; cerebellar A reversed). However, vertex-wise L vs R comparisons revealed significant between-group A differences across unimodal sensorimotor regions and heteromodal association cortices and cerebellum (PFDR<0.05), reproducible in Discovery and Replication.
• ROI-level effects: Opposite laterality between R and L was observed bilaterally in FST (visual association), IPS1 (dorsal stream), BA40 (inferior parietal), and in left hemisphere area 6d (premotor) and cerebellum (lobes V and VIII), and right hemisphere BA4 (primary motor). Mixed-handers showed intermediate values. Notable Δ-decreases (lower lateralization in L vs R) occurred in 6d, BA1 (primary somatosensory), 24dd (mid-cingulate motor), and cerebellar ROIs; Δ-increases occurred in FST, IPS1, and BA4/BA40. Approximately 85% of significant grayordinates with L vs R laterality differences were in regions with functional specialization index <0.5 (heteromodal association cortex).
• gFCD asymmetry (RH-LH): Right-handers exhibited asymmetric gFCD favoring LH in somatosensory/motor and medial prefrontal regions and favoring RH in insula/premotor/mid-cingulate, with no hemispheric difference in hand-motor area. Left-handers showed pronounced rightward asymmetry in the hand-motor area (higher gFCD in RH), encompassing the entire Mr ROI (PFDR<0.05). Group differences in asymmetry were significant over hand-motor cortex.
• Structural metrics: No significant handedness effects on brain morphometrics (sulcal depth, cortical thickness, curvature), cortical myelin, or diffusion metrics (FA, MD, ID, tD), including asymmetry analyses. Structural asymmetry patterns were highly consistent across L, R, and M groups (Dice coefficients: curvature >0.86; myelin ~0.65; sulcal depth 0.93; thickness 0.78), and DTI asymmetries were also consistent, without group differences.
• Demographic/statistical controls and reproducibility: All major functional findings replicated in independent Discovery and Replication subsamples, with stringent multiple-comparisons control (PFDR<0.05; Bonferroni for DTI ROIs).

The study demonstrates that in 9–10-year-old children, handedness is strongly associated with lateralized functional connectivity of hand-motor cortices and their broader networks, even in the absence of detectable structural asymmetry differences. Left-handers show enhanced connectivity of the left-hand motor area (right hemisphere) and reduced connectivity of the right-hand motor area (left hemisphere), producing a robust, brain-based handedness index that correlates with behavioral handedness. These effects are driven primarily by intra-hemispheric connectivity, aligning with the notion that motor control is predominantly organized within the hemisphere specialized for the dominant hand. The laterality differences extend beyond primary sensorimotor regions into heteromodal association cortex and cerebellum, indicating that handedness is embedded within broader functional networks affecting attention, visuomotor integration, and cingulate motor control. The strong reproducibility across matched discovery and replication cohorts underscores the robustness of the associations in pediatric populations. The lack of structural differences suggests that functional connectivity measures are more sensitive to behavioral lateralization at this developmental stage and may precede or exist independently of gross morphometric or microstructural asymmetries.

This work provides robust evidence that handedness in children is reflected in the laterality of resting-state functional connectivity within hand-motor cortex and distributed cortical and cerebellar networks. The authors introduce a simple neurobiological handedness index based on relative gFCD between right- and left-hand motor areas, which distinguishes handedness groups and correlates with behavioral handedness scores. Despite pronounced functional lateralization, no corresponding structural or diffusion asymmetry differences were detected, suggesting functional measures are more proximal to hand motor behavior in late childhood. Future research should leverage the longitudinal ABCD data to examine developmental trajectories, age-by-handedness interactions, potential emergence of structural asymmetries with maturation, and the generalization of these functional markers to adolescence and adulthood.

• Cross-sectional design and narrow age range (9–10 years) precluded assessment of developmental trajectories and age-by-handedness interactions in lateralization.
• Structural connectivity analyses limited to Siemens scanners to control scanner-related variability reduced diffusion MRI sample size and may limit generalizability.
• Despite large overall sample, effect sizes for some structural measures are known to be small; the study may be underpowered to detect subtle structural asymmetry differences.
• Handedness in some children may not be fully established (lower left-handedness and higher mixed-handedness prevalence vs adult cohorts), potentially diluting group contrasts.
• Multisite acquisition variability, although mitigated by harmonized protocols and statistical controls, may still contribute residual confounding.