Highly consistent anatomical asymmetry in a small primate brain: left is always larger in the marmoset monkey

Abstract

Although lateralisation of brain function is relatively common in animals, this is rarely accompanied by overt anatomical asymmetries. Morphological differences between the cerebral hemispheres are well established in Hominoids and other large primate species, but regarded as subtler, or inconsistent, in other primates. We demonstrate that the left hemisphere is reproducibly larger than the right hemisphere in one of the smallest primates, the marmoset. Among 208 adult marmosets the left hemisphere was on average 2.31% larger than the right (range: 0.22%- 4.37%). This asymmetry develops during early postnatal life, persists throughout adulthood, and is primarily due to differences in the volume of cortex associated with social cognition, including in the lateral and inferior temporal lobe areas and ventral premotor areas. These results challenge the notion that the morphological asymmetry between the hemispheres is linked to the evolution of large brains and progressive reduction in the relative connectivity between hemispheres.

Analysis of a large cohort of marmosets across 4 countries revealed that the cortex in the left hemisphere is consistently larger than in the right one, raising questions about theories that associate large brain size to the emergence of asymmetry.

Introduction

Although lateralisation of brain function is common across the animal kingdom¹, this is not usually linked to morphological differences between the two cerebral hemispheres. For example, while many species show lateralisation according to criteria such as the preferred use of one of the limbs², development and production of communication calls^3,4, spatial cognition⁵, and motor expression of emotional cues^6,7, this is rarely accompanied by overt anatomical asymmetries.

Large primates, including humans, apes, and baboons, show consistent morphological differences between hemispheres, including (but not restricted to) cortical areas related to the evolution of language, such as those in the planum temporale^8–11. In baboons, a planum temporale asymmetry is present at birth¹², develops further postnatally into a consistent pattern of thicker and larger left hemisphere cortex¹³, and has been shown to predict laterality in the use of the right hand for communicative gestures¹¹. In macaque monkeys, population-level asymmetries have also been reported, including the superior temporal sulcus and superior temporal gyrus^14,15, which have been shown to consolidate postnatally¹⁴. However, the precise distribution of this trait across large populations of individuals has not been described in detail.

The presence of clear temporal cortex asymmetry in Hominoids, together with the presence of subtler asymmetry in larger simian primates and the rarity of overt asymmetry in other mammalian species, has been linked to the idea that reproducible changes in the gross anatomy of the cortex may be linked to the evolution of larger brains^16,17. In a nutshell, the development of a larger brain would impose constraints related to the efficiency of axonal connections and availability of synaptic space in single neurons; this would lead to the gradual segregation of neural structures into more specialised subnetworks¹⁸, particularly due to the increase in volume of the cortex relative to the cross-section of the corpus callosum, and ultimately morphological asymmetries as the hemispheres become more specialised^16,17. In other words, increasing evolutionary pressure for functional specialisation would only be expressed as consistent anatomical differences between hemispheres in selected instances.

We explored the occurrence of hemispheric asymmetry using high-resolution magnetic resonance imaging (MRI) in marmoset monkeys (Callithrix jacchus), which are among the smallest primates, with a brain volume of only 8 g¹⁹. Marmosets have become increasingly used as an animal model in the neurosciences due to their highly social behaviour²⁰, which relies on a rich repertoire of vocal communication²¹ and a well-developed visual system²², as well as a relatively short reproductive cycle, which facilitates the generation of transgenic lines^23,24. Like many other mammals, marmosets show behavioural evidence for lateralisation of motor and sensory function, which, however, tends to vary across individuals^25,26. According to the hypothesis that links anatomical asymmetry to brain size and considering the relatively subtle differences described in old world monkeys, one would expect that marmoset brains would show little evidence of hemispheric asymmetry. The present results do not support this, instead revealing a previously unreported and remarkably consistent pattern whereby the left hemisphere is larger.

Results

The primary sample included MRI scans acquired from 302 marmosets obtained either for the purpose of ongoing studies in our laboratories or from publicly available databases (Supplementary Data 1). The dataset originated from four geographically isolated colonies, located in Australia (Monash; n = 9), China (ION; n = 60), Japan (RIKEN; n = 206), and the USA (NIH; n = 17). This included both in vivo (Australia, Japan, USA) and ex vivo scans (China). We found that adult marmosets ( > 2 years old, n = 208; Supplementary Table S1) show a remarkably consistent anatomical asymmetry, whereby the left hemisphere is invariably larger than the right hemisphere (Fig. 1a). When quantified according to a commonly used metric, the hemispheric asymmetry index (AI), the left hemisphere is on average 2.31% larger than the right (range 0.22– 4.37%), corresponding to a difference in volume of 96.89 ± 27.18 mm³ (Wilcoxon Signed-Rank Test: W = 0, p = 7.0131e − 36; paired t-test: t = 51.42, p = 7.54e − 120; Cohen’s d = 3.57, 95% CI [3.1, 4,21]). Male and female marmosets did not differ significantly in this respect (Mann–Whitney U test: U = 4203.0000, p = 7.8553e − 02; rank-biserial correlation: r = 0.1549, 95% CI [-0.0036, 0.3256]) (Fig. 1b). Data from the different colonies demonstrate the same trend, despite the individuals therein being genetically isolated for at least 10 generations. While statistical differences in the AI distribution were observed between colonies (Kruskal–Wallis: H = 52.4961, p = 2.3478e − 11; ε² = 0.2067, 95% CI [0.1707, 0.2963]), AI was consistently positive, indicating that leftward hemispheric asymmetry is a universal feature in adult marmosets irrespective of sex, age, or colony of origin (Supplementary Fig. S1).

Fig. 1. Distribution of hemispheric asymmetry in adult marmosets.

a Violin plots of the distribution of the hemispheric Asymmetry Index (AI) in all adult marmosets ( > 2 years old) included in the study (n = 208). AI quantifies the relative volume between brain hemispheres, with positive values indicating leftward asymmetry. Each dot represents an individual, with red diamond outlines indicating group medians. b Distribution of AI values according to sex. Both male (n = 90) and female (n = 118) marmosets exhibit positive AI values, indicating leftward hemispheric asymmetry. c Voxel-based morphometry (VBM) analysis comparing grey matter volume between hemispheres in adult marmosets (n = 208). Statistical maps (yellow-red) indicate cortical regions where grey matter volume is significantly greater in the left hemisphere compared to the right, thresholded at p < 0.05 (FWE-corrected), with a minimum cluster extent of 100 voxels (k = 100). The left panel shows a lateral view of the left hemisphere, while the right panel shows a ventral view. Overlaid anatomical boundaries correspond to the Marmoset Brain Mapping Version 3 (MBMv3 cortical parcellation²⁷. Anatomical orientation: A- anterior, D- dorsal, L- left, P-posterior, R- right, V-ventral. For abbreviations of the names of cortical areas, see ref.³⁷.

The hemispheric asymmetry is not present throughout the brain. Voxel-based morphometry according to registration to a MR-based template²⁷, revealed that the differences are attributed to larger left hemisphere volumes in visual and polysensory areas of the lateral and inferior temporal cortical regions (cytoarchitectural areas TE, TPO, IPa/ PGa), the high-order auditory cortex (belt, parabelt, and the rostral superior temporal auditory association area), perirhinal cortex, and the foveal representations of some extrastriate visual areas (V3, V4, and MST) (Fig. 1c). Further asymmetries were observed in the frontal lobe (ventral premotor area, and precentral opercular cortex), and in the ventral posterior parietal cortex (areas PG, PFG, and PF).

Considering these observations, a natural question is whether the hemispheric asymmetry is present from birth or develops during postnatal life. For this we studied additional individuals (Supplementary Tables S2 and S3), including animals within the first 12 months of postnatal life (n = 64), and peripubescent animals (12-24 months; n = 30), and further segregated the adult cohort according to age (Fig. 2a). We found that there is a slight, but inconsistent trend towards a larger left hemisphere already in the youngest animals. The left hemisphere bias becomes consolidated into the emergence of a marked asymmetry during the first year of postnatal life. This is maintained throughout adulthood, including old age (Fig. 2b).

Fig. 2. Age-related differences in cerebral asymmetry across the lifespan in marmosets.

The asymmetry index (AI) according to developmental stages in marmosets. Postnatal: <1 year old (n = 64). Peripubescent: 1–2 years (n = 30). Adult: >2 years (n = 208). Marmosets are typically considered to be old beyond 7-8 years and rarely survive beyond 12 years, even in controlled colony conditions. a All groups demonstrate leftward hemispheric asymmetry (positive AI values), but there is higher variability in the Postnatal group. b Relationship between age and the cortical asymmetry index (AI) across individual marmosets, using a generalised additive model (GAM) to account for potential nonlinear trends. Each point represents one animal. A red dashed horizontal line indicates equal volume in the 2 hemispheres (AI = 0). The model captures subtle age-related patterns in asymmetry, particularly during early development, while highlighting the overall stability of AI across later stages of life.

Discussion

Perhaps the most surprising aspect of our data is the high reproducibility of the anatomical bias towards the left hemisphere. This finding contrasts with reports on most mammals to date, where functional, behavioural and anatomical asymmetries, when present, tend to be variable across individuals. Although the full extent of leftward hemispheric asymmetry in other primates clearly requires more research, differences in gross anatomical features such as sulcal extent and depth have been reported as subtle or inconsistent in smaller primate species, including all New World monkey species studied so far (e.g., refs. ^28,29). However, many of the older studies were based on manual measurement of parameters such as the length and depth of the Sylvian fissure, which could be reliably estimated with the technology then available. The present analysis, taking advantage of a much larger sample of individuals and being based on computerised analyses that do not depend on sulcal morphology, demonstrates that a leftward bias is clear even in marmosets, one of the smallest simian primates. The present results raise the likelihood that a left hemisphere bias is a symplesiomorphic evolutionary trait, shared in some form by most primates. This hypothesis, which remains to be tested, would be in line with the demonstration of planum temporale asymmetries already in newborn baboons¹², and evidence of heritability of some aspects of hemispheric differences in various species, which suggest a genetic basis (e.g., ref. ¹⁷).

There have been several previous studies suggesting that region-specific lateralisation may be bidirectional. For example, in macaque monkeys, a longitudinal analysis revealed that some asymmetries favour the right hemisphere (cingulate cortex, insula)¹⁴. The present study differs in two important respects: first, the primary finding concerns hemispheric volume asymmetry, not isolated regional effects. At the whole-hemisphere level, marmosets show a highly consistent leftward bias across individuals, colonies, sexes, and ages. Voxel-based morphometry (VBM) was used secondarily to localise which cortical regions contribute most strongly to this hemispheric difference. The VBM analysis does not claim the absence of rightward asymmetries per se but rather identifies regions where leftward volume differences are sufficiently consistent across animals to survive stringent paired statistical testing. Thus, whereas the predominance of leftward effects in VBM reflects the whole-hemisphere bias, it does not necessarily imply that right hemisphere-biased areas do not exist in the marmoset brain.

The present results show that higher-order auditory and polysensory areas are among those showing asymmetry that favours the left hemisphere. These include the homologues of the areas located in the planum temporale and superior temporal gyrus in larger primates, which are among the regions most consistently associated with overt anatomical asymmetry. However, our analyses also reveal asymmetries in other cortical areas, including those related to higher-order visual recognition, olfactory processing, and episodic memory. These are all important requirements for social interaction, which are distinctively developed in marmoset groups³⁰. In the frontal and parietal regions, the only areas showing asymmetry are those linked to the “mirror neuron” system in other primate species³¹. Taken altogether, the anatomical distribution of asymmetry in the marmoset cortex suggests the possibility that the allocation of increased neural resources for social interaction may have been one of the evolutionary drivers of the functional specialisation that led to consistent asymmetry in cortical structure and function in primates. Moreover, our results demonstrate that hemispheric asymmetry in marmosets is inconsistent in newborns, where only a very subtle population-level bias towards a larger left hemisphere is evident. Thus, in addition to possible genetic biases¹⁷, it is likely that postnatal developmental mechanisms, including hormonal influences and epigenetic modification, could account for this, given the strong influence of social context in marmoset behaviour.

We have considered several potential methodological factors that could have influenced the present analyses. Among these are potential scanner and sequence variability across imaging centres. The dataset includes MRI data acquired across four independent colonies using different scanners and acquisition protocols, which indeed introduces heterogeneity. However, the hemispheric asymmetry index (AI) is computed within subject, directly comparing left and right hemisphere volumes from the same scan. This paired formulation makes the AI inherently robust to global scaling differences, scanner-specific intensity profiles, or absolute volume offsets introduced by acquisition parameters. Crucially, despite inter-site variability, all four colonies show a consistent leftward asymmetry, with AI values remaining positive in every cohort (Fig. 2 and Supplementary Fig. S1). This convergence across geographically isolated populations, scanners, field strengths, and acquisition strategies strongly argues against scanning-specific artefacts as a primary driver of the observed effect.

It is also important to exclude artifacts related to preprocessing (e.g., skull stripping, template-based extraction) and segmentation or warping asymmetries associated with the template. We explicitly minimised left–right bias during preprocessing. For example, skull stripping was performed using atlasBREX with the MBMv3 atlas. Hemisphere-specific volumes were derived using predefined left and right masks from the same atlas, which is bilaterally symmetric by construction, and applied identically across subjects. Because brain extraction and hemisphere masking are based on a symmetric reference, any systematic left–right bias introduced by skull stripping would be expected to cancel out at the group level, rather than generate a consistent unilateral effect. In addition, only rigid-body registration (translations and rotations) was used to align individual brains to the MBMv3 template, with no affine scaling, shearing, or nonlinear warping being applied during hemisphere volume estimation. Rigid-body registration does not alter voxel size, shape, or volume, nor does it introduce scaling, shearing, or warping. In other words, hemisphere volumes were computed after rigid reorientation only, meaning that native inter-hemispheric volume relationships are preserved exactly. Finally, it is important to reiterate that the MBMv3 atlas is explicitly symmetric, eliminating any intrinsic atlas bias between hemispheres. Because the same rigid transform is applied uniformly to both hemispheres, registration cannot introduce differential left–right volume changes. Volumes were computed directly from each subject’s brain mask using identical voxel geometry for both hemispheres.

We considered the desirability of a manual ROI-based validation, but given the large sample size and the multi-institutional nature of the dataset, manual segmentation is not feasible. Importantly, the observed asymmetry magnitude ( ~ 2.3%) is well above expected numerical precision or interpolation errors introduced by rigid resampling and voxel discretization, especially given the consistency of the effect across animals, ages, sexes, and colonies. In addition, because the present study is based on a multi-centre retrospective dataset, test–retest MRI data are unfortunately not available, and rescanning individual animals is not feasible. Future studies, including those focused on other species, would benefit from including this as a step.

Finally, our data leave open the question of whether fixation may affect the degree of measured asymmetry. The bulk of the data obtained in adults comes from in vivo scans (Australia, Japan, USA datasets), with only 3 ex vivo adult scans being available (China). Overall, we believe that any effects of fixation, if any, are slight, given that the latter results fall within the distribution measured in other sites (Supplementary Fig. S1), as do the results obtained ex vivo in the peripubescent group (Fig. 2). This can be in part explained by our acquisition protocol: the brains were not physically extracted before imaging, with all MRI scans performed with the brains remaining within the skull. By maintaining the cranial structural support, we preserved the brain’s natural physical architecture and successfully prevented the hemispheric distortions typically caused by gravity or handling during the extraction process.

Whereas the precise distribution of morphological asymmetry in primates and the developmental mechanisms behind this asymmetry remain a subject for future studies, our results clearly reveal that consistent anatomical asymmetries are not an exclusive domain of larger brains, calling into question the exclusive association between these and brain volume and the inability of the corpus callosum to maintain proportional connectivity. Previous findings on the baboon planum temporale have been interpreted as indicating that the leftward-favouring asymmetry reflects a predisposition for lateralisation of function that was already present in the last common ancestor of Old-World monkeys and Hominids^11–13, between 25 and 30 million years ago³². Although the small brain of the marmoset does not have an anatomically distinct planum temporale, we find that the areas in the homologous parts of the brain show a similar asymmetry, which suggests that the origin of this primate brain trait may extend even further in evolutionary time, to at least 40 million years³³.

Among other implications, our findings highlight the need to consider laterality whenever anatomical or physiological data are obtained in various animal models. Many studies do not report whether observations in non-human primates originated in the left or right hemispheres, or treat observations obtained on both sides of the brain as interchangeable. Moving forward, this is an issue that will require additional attention, possibly including revised editorial guidelines for publication in comparative neuroscience.

Methods

Animals and study demographics

A total of 302 common marmosets (Callithrix jacchus) were included in this study, classified into three age groups: Postnatal ( < 1 year), Peripubescent (1–2 years), and Adult ( > 2 years). Supplementary Tables S1, S2, and S4 summarise the demographics by colony, age, and sex. Animals were obtained from four independent colonies: RIKEN Centre for Brain Science (Japan), ION (Institute of Neuroscience, China), NIH (National Institutes of Health, USA), and Monash University (Australia). Animals from each colony were bred and group housed according to institutional animal care protocols, each subject to approval by a local Animal Ethics Experimentation Committee.

MRI acquisition

Australia

All procedures were according to protocols approved by the Monash University Animal Experimentation Ethics Committee. MRI data were acquired in vivo on a 9.4 T Bruker BioSpec scanner equipped with a Bruker 86 mm volume coil for transmission and a Rapid 8-channel array coil for reception. Structural MRI was performed using a two-dimensional T2-weighted Rapid Acquisition with Relaxation Enhancement (RARE) sequence (TR = 3000 ms, TE = 47.4 ms, rare factor = 12, slice thickness = 0.2 mm, FOV = 40 × 40 mm², matrix = 200 × 200, in-plane resolution = 0.2 × 0.2 mm², 100 slices, three signal averages, total scan time = 25 min 30 s). Animals were anaesthetised with alfaxalone (8 mg/kg, i.m.) and diazepam (3 mg/kg, i.m.), and anaesthesia was maintained with 0.5–2.0% isoflurane during scanning. Animals were positioned in the sphinx posture within a water-heated, MRI-compatible cradle. Body temperature and respiratory rate were continuously monitored throughout the imaging session. During scanning, anaesthesia was maintained with 0.5–2.0% isoflurane delivered in medical-grade oxygen via a custom-fitted face mask.

China

All animal studies and procedures were approved by the Animal Care and Use Committee of the Centre for Excellence in Brain Science and Intelligence Technology (CEBSIT), Chinese Academy of Sciences (Shanghai, China). Ex vivo MRI data for marmoset brain development were acquired at the Chinese Academy of Sciences Institute of Neuroscience (CAS-ION) on a 9.4 T Bruker BioSpec 94/30 MRI scanner using a FLASH sequence. Prior to scanning, all brains were preserved in situ within the skull and immersed in a solution of 4% paraformaldehyde (PFA) with 0.3% gadolinium (Gd) for 2-4 weeks, followed by a one-week immersion in phosphate-buffered saline (PBS). The dataset comprises two cohorts. For the first cohort (16 male, 10 female) acquisition parameters for this cohort were individually adjusted to account for differences in brain size and to optimise image contrast for animals of varying ages, resulting in a range of values: TR = 20 ms; TE = 3.6 ms to 4.7 ms; matrix size = 225×290×210 to 285×360×270; FOV = 21.6×27.84×20.16 to 22.8 × 28.8 × 21.6 mm; flip angle = 15° to 25°; voxel size = 80–96 μm isotropic; number of average = 1; scan time = 9:57 min to 18:10 min. The second cohort (23 male, 13 female) was scanned with a fixed parameter: TR = 30 ms; TE = 20 ms; matrix = 278×360×270; FOV = 22.2×28.8×21.6 mm; flip angle = 35°; voxel size = 80 μm isotropic; number of averages = 2; scan time = 37:58 min.

Japan

The data acquisition was approved by the Experimental Animal Committee of RIKEN. Images from the from the Japanese colony were obtained from the BRAIN/MINDS marmoset Brain MRI dataset³⁴. In vivo T2-weighted MRI data were obtained from the BRAIN/MINDS marmoset MRI dataset. Scans were performed using a 9.4 T Bruker BioSpec 94/30 MRI scanner (TR = 4000 ms, TE = 22 ms, RARE factor = 4, voxel size = 270 × 270 × 540 µm, scan time = 7 min 24 s). Animals were anaesthetised with 2% isoflurane in an oxygen-air mixture and monitored throughout.

USA

All experiments were approved by the Animal Care and Use Committee of the National Institute of Neurological Disorders and Stroke. In vivo MRI data were acquired as described previously²⁷. Briefly, animals were anesthetised with an intramuscular injection of 10 mg/kg ketamine and ventilated with 1.5–2% isoflurane. The animals were placed in an MR-compatible cradle, and their vital signs were monitored throughout the MRI scanning. MRI scanning was performed in a Bruker 7 T/300 mm magnet with a 2D RARE sequence: TR = 30 s, TE = 8 s with three effective TE (TE1 = 16 ms, TE2 = 48 ms and TE3 = 80 ms), FOV = 36 × 28 × 24 mm, matrix size = 144 × 112, slice thickness = 0.25 mm, and the scanning duration was about 10 min 30 s.

Image preprocessing and volume calculation

All MRI data were processed using a standardised pipeline implemented in Advanced Normalisation Tools (ANTs) and the FMRIB Software Library (FSL)³⁵. T2-Weighted images were first skull-stripped using the template-based brain extraction tool atlasBREX³⁶, using the Marmoset Brain Mapping V3 (MBMv3) atlas²⁷ as a reference to the areas defined by Paxinos et al. ³⁷. Brain-extracted images were denoised (DenoiseImage, Rician noise model, spatial radius = 3), followed by N4 bias field correction (N4BiasFieldCorrection, shrink factor = 3, B-spline fitting). To avoid introducing artificial hemispheric asymmetries, image registration to the MBMv3 template was restricted to rigid-body alignment only (translations and rotations; antsRegistrationSyNQuick.sh, transform type: rigid). No affine scaling, shearing, or nonlinear deformation was applied. The MBMv3 atlas is bilaterally symmetric by design, and hemisphere-specific volume masks were derived from this symmetric reference. This approach preserves native inter-hemispheric volume relationships while enabling standardised orientation across subjects. Binary brain masks were generated using fslmaths, and hemisphere-specific volumes were computed using fslstats with predefined left and right hemisphere masks from the MBMv3 atlas.

Asymmetry index calculation

Hemispheric asymmetry was quantified using the Asymmetry Index (AI):

$$\mathrm{AI}=\frac{V_L-V_R}{0.5\times \left(V_L+V_R\right)}$$

where $V_L$ and $V_R$ represent the left and right hemisphere volumes, respectively. This metric normalises inter-hemispheric differences by the mean total hemispheric volume. Positive AI values indicate leftward asymmetry (larger left hemisphere), while negative values indicate rightward asymmetry.

Voxel based morphometry

To assess which specific cortical areas contributed to the hemispheric asymmetries in grey matter (GM) volume, a paired voxel-based morphometry (VBM) analysis was employed using SPM12 (Wellcome Trust Centre for Neuroimaging, London, UK) in MATLAB. Preprocessed images were resampled to 1×1×1 mm³ isotropic resolution. Then, images were left–right flipped along the x-axis using fslswapdim to generate mirrored versions of each brain. Flipped images were then registered again to their original unflipped images to ensure anatomical correspondence and segmented into grey matter, white matter and cerebrospinal fluid in SPM12 (old segmentation). Grey matter maps were then normalised to the MBMv3 marmoset template using DARTEL, with Jacobian modulation applied to preserve local volumetric information. Smoothed GM maps (6×6×6 mm FWHM) were compared using a voxel-wise paired t-test between original and flipped images. Results were thresholded at p < 0.05, family-wise error (FWE) corrected, with a cluster extent threshold of 100 voxels. Significant values were displayed on the MBMv3 T2-template and cortical areas identified with the MBM_cortex_vM atlas labels.

Statistics and reproducibility

All statistical analyses were conducted using Python (version 3.10.15) scipy, statsmodels, scikit-posthocs and pyGAM libraries and were two-sided unless otherwise stated. All analyses were performed at the individual-animal level. Nonparametric statistical tests were used because the Shapiro–Wilk tests indicated nonnormality. To assess whether the left hemisphere was significantly larger than the right in adult marmosets, we used the Wilcoxon Signed-Rank Test, a non-parametric paired test. Differences in the asymmetry index (AI) between colonies and age groups were analysed using the Kruskal–Wallis test, followed by Dunn’s post-hoc test with Bonferroni correction to evaluate pairwise group differences. To examine sex differences in AI, we employed the Mann–Whitney U test. Additionally, we investigated the relationship between age and AI using both linear regression and a non-parametricGeneralized Additive Model (GAM). The GAM framework allowed for the detection of potentialnonlinearage-related trends in asymmetry by fitting a smooth spline function to the data. To evaluate the combined effects of age, sex, and colony on inter-hemispheric volume differences, we employed multiple linear regression using thestatsmodels.OLSframework. Effect sizes were calculated to complement statistical significance testing, including epsilon-squared (ε²) for Kruskal–Wallis comparisons and rank-biserial correlation for the Mann–Whitney U test. All results were considered statistically significant at p < 0.05. Visualisations were generated using the matplotlib and seaborn libraries, employing violin plots overlaid with individual data points (swarm/strip plots) and red diamond markers to indicate median values.

All data used in the calculations are available by request to the authors.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Peer review

Peer review information

Communications Biology thanks the anonymous reviewers for their contribution to the peer review of this work. Primary Handling Editor: Benjamin Bessieres. A peer review file is available.

Funding

National Health and Medical Research Council Ideas grants (APP2011733, APP2019011), and Investigator grant (APP1194206). Australian Research Council Discovery Projects (DP210103865, DP210101042, DP250101768). National Natural Science Foundation of China (No. 32427802 and W2512020). Lingang Lab Grant (LGL-6672-08).

Data availability

The numerical data used to generate the graphs in the present paper are available as Supplementary Data 1. They can also be accessed from the Marmoset Brain Connectivity Atlas web site (https://www.marmosetbrain.org/hemispheric-asymmetry). All other data are available from the corresponding author on reasonable request.

Code availability

The present analyses did not require customised code, being based on existing software libraries. All software used and applicable versions are detailed in the Methods section.

Competing interests

The authors declare no competing financial or non-financial interests.

Supplementary information

The online version contains supplementary material available at https://doi.org/10.1038/s42003-026-10576-3.