Tracing the development and lifespan change of population-level structural asymmetry in the cerebral cortex

  1. James M Roe  Is a corresponding author
  2. Didac Vidal-Pineiro
  3. Inge K Amlien
  4. Mengyu Pan
  5. Markus H Sneve
  6. Michel Thiebaut de Schotten
  7. Patrick Friedrich
  8. Zhiqiang Sha
  9. Clyde Francks
  10. Espen M Eilertsen
  11. Yunpeng Wang
  12. Kristine B Walhovd
  13. Anders M Fjell
  14. René Westerhausen
  1. Center for Lifespan Changes in Brain and Cognition (LCBC), Department of Psychology, University of Oslo, Norway
  2. Groupe d'Imagerie Neurofonctionnelle, Institut des Maladies Neurodégénératives-UMR 5293, CNRS, CEA, University of Bordeaux, France
  3. Brian Connectivity and Behaviour Laboratory, Sorbonne University, France
  4. Institute of Neuroscience and Medicine, Research Centre Jülich, Germany
  5. Language and Genetics Department, Max Planck Institute for Psycholinguistics, Netherlands
  6. Donders Institute for Brain, Cognition and Behaviour, Radboud University, Netherlands
  7. Department of Human Genetics, Radboud University Medical Center, Netherlands
  8. PROMENTA Research Center, Department of Psychology, University of Oslo, Norway
  9. Department of Radiology and Nuclear Medicine, Oslo University Hospital, Norway
  10. Section for Cognitive and Clinical Neuroscience, Department of Psychology, University of Oslo, Norway

Abstract

Cortical asymmetry is a ubiquitous feature of brain organization that is subtly altered in some neurodevelopmental disorders, yet we lack knowledge of how its development proceeds across life in health. Achieving consensus on the precise cortical asymmetries in humans is necessary to uncover the developmental timing of asymmetry and the extent to which it arises through genetic and later influences in childhood. Here, we delineate population-level asymmetry in cortical thickness and surface area vertex-wise in seven datasets and chart asymmetry trajectories longitudinally across life (4–89 years; observations = 3937; 70% longitudinal). We find replicable asymmetry interrelationships, heritability maps, and test asymmetry associations in large–scale data. Cortical asymmetry was robust across datasets. Whereas areal asymmetry is predominantly stable across life, thickness asymmetry grows in childhood and peaks in early adulthood. Areal asymmetry is low-moderately heritable (max h2SNP ~19%) and correlates phenotypically and genetically in specific regions, indicating coordinated development of asymmetries partly through genes. In contrast, thickness asymmetry is globally interrelated across the cortex in a pattern suggesting highly left-lateralized individuals tend towards left-lateralization also in population-level right-asymmetric regions (and vice versa), and exhibits low or absent heritability. We find less areal asymmetry in the most consistently lateralized region in humans associates with subtly lower cognitive ability, and confirm small handedness and sex effects. Results suggest areal asymmetry is developmentally stable and arises early in life through genetic but mainly subject-specific stochastic effects, whereas childhood developmental growth shapes thickness asymmetry and may lead to directional variability of global thickness lateralization in the population.

Editor's evaluation

Roe et al. provide a large-sample analysis of hemispheric lateralisation in brain structure, synthesising local cortical thickness and surface area data from 7 different datasets. The study provides a rich descriptive catalogue of phenomena related to hemispheric anatomical asymmetries. These results are convincing and will prove an important point of reference to neuroscientists who might want to compare their own future results to the ones from this large and varied data set.

https://doi.org/10.7554/eLife.84685.sa0

Introduction

The brain’s hemispheres exhibit high contralateral symmetry (van Kesteren and Kievit, 2021; Stark et al., 2008), homotopic regions are under similar genetic influence (Schmitt et al., 2018; Chen et al., 2013; Eyler et al., 2014) and show highly correlated developmental change (Schmitt et al., 2018; Raznahan et al., 2011). Despite this, structural asymmetry is also a ubiquitous aspect of brain organization (Kong et al., 2018; Chiarello et al., 2016). Cortical thickness and surface area are known to exhibit distinct asymmetry patterns (Kong et al., 2018; Meyer et al., 2014), albeit reported inconsistently (Kong et al., 2018; Chiarello et al., 2016; Roe et al., 2021; Li et al., 2015; Luders et al., 2006; Lyttelton et al., 2009; Sha et al., 2021a; Zhou et al., 2013; Maingault et al., 2016; Shaw et al., 2009; Lou et al., 2020; Hamilton et al., 2007; Zhou et al., 2018; Koelkebeck et al., 2014; Plessen et al., 2014). Yet achieving consensus on cortical asymmetries in humans is a prerequisite to uncover the genetic-developmental and lifespan influences that shape and alter them. Although an extensive literature in search of structural asymmetry deviations in various conditions and disorders is in several cases being challenged by newer data (Kong et al., 2022), at least some aspects of cortical asymmetry are confirmed to be subtly reduced in neurodevelopmental disorders such as autism (Postema et al., 2019; Sha et al., 2022), but also through later life influences such as aging, and Alzheimer’s disease (Roe et al., 2021; Thompson et al., 2007). Hence, altered cortical asymmetry at various lifespan stages may be associated with reduced brain health. However, it is currently unknown how cortical asymmetry development proceeds across life in health, because no previous study has charted cortical asymmetry trajectories from childhood to old age using longitudinal data.

Compounding the lack of longitudinal investigation, previous large-scale studies do not delineate the precise brain regions exhibiting robust cortical asymmetry, relying on brain atlases with predefined anatomical boundaries that may not conform well to the underlying asymmetry of cortex (Kong et al., 2018; Sha et al., 2021b). Taking an atlas-free approach to delineate asymmetries that reliably reproduce across international samples as starting point (i.e. population-level asymmetries) would better enable mapping of the developmental principles underlying structural cortical asymmetries, as well as the genetic and individual-specific factors associated with cortical lateralization. Furthermore, such an approach would help resolve the many reported inconsistencies for cortical asymmetry maps – for example reports of both right- (Maingault et al., 2016; Lou et al., 2020; Hamilton et al., 2007; Williams et al., 2022) and left- (Roe et al., 2021; Li et al., 2015; Luders et al., 2006; Plessen et al., 2014) thickness lateralization in medial and lateral prefrontal cortex (PFC; Chiarello et al., 2016; Zhou et al., 2013; Maingault et al., 2016; Shaw et al., 2009; Koelkebeck et al., 2014 and Roe et al., 2021; Luders et al., 2006; Sha et al., 2021a; Plessen et al., 2014; Williams et al., 2023), and of right- (Maingault et al., 2016; Li et al., 2014) and left areal lateralization of superior temporal sulcus (STS) (Kong et al., 2018; Bain et al., 2019; Remer et al., 2017) – while serving as a high-fidelity phenotype for future brain asymmetry studies to complement existing low-resolution atlases (Kong et al., 2018).

Determining the developmental and lifespan trajectories of cortical asymmetry may shed light on how cortical asymmetries are shaped through childhood or set from early life, and provide evidence of the timing of expected brain change in normal development. Although important in and of itself, this would also provide a useful normative reference, as subtly altered cortical asymmetry – in terms of both area and thickness – has been linked at the group-level to neurodevelopmental disorders along the autism spectrum, suggesting altered lateralized neurodevelopment may be a relevant outcome in at least some cases of developmental perturbation (Postema et al., 2019; Sha et al., 2022). For areal asymmetry, surprisingly few studies have charted developmental (Li et al., 2014; Remer et al., 2017) or aging-related effects (Kong et al., 2018; Williams et al., 2022), although indirect evidence in neonates suggests adult-like patterns of areal asymmetry are evident at birth (Li et al., 2014; Williams et al., 2023) and may exhibit little change from birth to 2 years (Li et al., 2014) despite rapid and concurrent developmental cortical expansion (Li et al., 2013). For thickness asymmetry, longitudinal increases in asymmetry have been shown during the first two years of life (Li et al., 2015), with suggestions of rapid asymmetry growth from birth to 1 year (Li et al., 2015), and potentially continued growth until adolescence (Nie et al., 2013). However, previous lifespan studies mapped thickness asymmetry linearly across cross-sectional developmental and adult age-ranges (Zhou et al., 2013; Plessen et al., 2014), mostly concluding thickness asymmetry is minimal in infancy and maximal age ~60. In contrast, recent work established thickness asymmetry shows a non-linear decline from 20 to 90 years that is reproducible across longitudinal aging cohorts (Roe et al., 2021). Thus, although offering viable developmental insights (Zhou et al., 2013; Plessen et al., 2014), previous lifespan studies of thickness asymmetry do not accurately capture the aging process, and likely conflate non-linear developmental and aging trajectories with linear models. A longitudinal exploration of the lifespan trajectories of thickness asymmetry accounting for dynamic change is needed to further knowledge of normal human brain development.

Correlations between cortical asymmetries may provide a window on asymmetries formed under common genetic or developmental influences. Contemporary research suggests brain asymmetries are complex, multifactorial and independent (i.e. uncorrelated) traits (Rentería, 2012; Francks, 2015; Neubauer et al., 2020), contrasting earlier theories emphasizing a single (Annett, 1998; Annett, 1964) or predominating factor controlling various cerebral lateralizations (McManus and Bryden, 1991; Geschwind and Galaburda, 1985). Yet while there has been much research on whether asymmetries of various morphometric measures (Chiarello et al., 2016; Maingault et al., 2016; Koelkebeck et al., 2014) or imaging modalities relate to one another (Bain et al., 2019), few have focused on interregional relationships between asymmetries derived from the same metric. Where reported, evidence suggests cortical asymmetries are mostly independent (Sha et al., 2021b; Guadalupe et al., 2015; Chiarello et al., 2013) – in line with a multifactorial view (Bain et al., 2019; Rentería, 2012; Francks, 2015; Liu et al., 2009). Currently, it is unknown whether or which cortical asymmetries are reliably correlated within individuals, though this may signify coordinated development of left-right brain asymmetries through genetic or lifespan influences – a genetic or later developmental account depending on trait heritability and whether phenotypic correlations are underpinned by genetic correlations.

Finally, altered development of cerebral lateralization has been widely hypothesized to relate to average poorer cognitive outcomes (Plessen et al., 2014; Crow et al., 1998; Hirnstein et al., 2010). Specifically in the context of cortical asymmetry, however, although one previous study reported larger thickness asymmetry may relate to better verbal and visuospatial cognition (Plessen et al., 2014), phenotypic asymmetry-cognition associations have been rarely reported (Plessen et al., 2014; Moodie et al., 2020; Yeo et al., 2016), conflicting (Moodie et al., 2020; Yeo et al., 2016), not comparable (Plessen et al., 2014; Moodie et al., 2020), and to date remain untested in large-scale data. Still, recent work points to small but significant overlap between genes underlying multivariate brain asymmetries and those influencing educational attainment and specific developmental disorders impacting cognition (Sha et al., 2021b), indicating either pleiotropy between non-related traits or capturing shared genetic susceptibility to altered brain lateralization and cognitive outcomes. However, most large-scale studies examining factors widely assumed important for asymmetry have used brain atlases with limited spatial precision (Kong et al., 2018; Kong et al., 2022; Sha et al., 2021b). Accordingly, such studies did not detect associations with handedness (Kong et al., 2018; Wiberg et al., 2019) that were not found until a recent study applied higher resolution (i.e. vertex-wise) mapping in big data (Sha et al., 2021a). Therefore, as a final step, we reasoned that combining an optimal delineation of population-level cortical asymmetries with big data would optimize detection and quantification of the effects of factors purportedly related to asymmetry, namely cognitive ability, handedness, and sex.

Here, we first aimed to delineate population-level cortical areal and thickness asymmetries using vertex-wise analyses and their overlap in seven international datasets. With a view to gaining insight into cortical asymmetry development, we then aimed to trace a series of lifespan and genetic analyses. Specifically, we chart the developmental and lifespan trajectories of cortical asymmetry for the first time longitudinally across the lifespan. Next, we examine phenotypic interregional asymmetry correlations, under the assumption correlations indicate coordinated development of left-right asymmetries through genes or lifespan influences. To shed light on the extent to which differences in asymmetry are genetic, we test heritability of asymmetry using genome-wide single nucleotide polymorphism (SNP) and extended twin data, and examine whether or not phenotypic correlations are underpinned by genetic correlations suggestive of coordinated development through genes. Finally, we screen our set of robust, population-level asymmetries for association with general cognitive ability and factors purportedly related to asymmetry in UK Biobank (UKB) (Miller et al., 2016). Based on findings of aging-related dedifferentiation in thickness asymmetry (Roe et al., 2021), we hypothesized trajectories of cortical thickness would show developmental growth in thickness asymmetry (i.e. differentiation), but remained agnostic regarding lifespan areal asymmetry development.

Results

Population-level asymmetry of the cerebral cortex

First, to delineate cortical regions exhibiting population-level areal and thickness asymmetry, we assessed asymmetry vertex-wise in 7 independent adult samples and quantified overlapping effects (Methods). Areal asymmetries were highly consistent across all 7 datasets (Figure 1A): the spatial correlation between surface AI maps ranged from r=0.88 to 0.97 (Figure 1C). Across all seven datasets (Figure 1D), overlapping effects for strong leftward areal asymmetry were observed in a large cluster in supramarginal gyrus (SMG) that spanned postcentral gyrus, planum temporale and primary auditory regions (and conformed well to their anatomical boundaries; see Figure 1—figure supplement 1A for significance), anterior insula and temporal cortex, rostral anterior cingulate, medial superior frontal cortex, and precuneus, the latter spanning parahippocampal and entorhinal cortex. Overlapping effects for strong rightward areal asymmetry were evident in cingulate, inferior parietal cortex, STS, medial occipital cortex, medial PFC (mPFC) and rostral middle frontal cortex (Figure 1D). This global pattern agrees with previous reports (Kong et al., 2018; Lyttelton et al., 2009; Sha et al., 2021a).

Figure 1 with 5 supplements see all
Population-level asymmetry of the cerebral cortex.

(A) Mean areal and (B) thickness asymmetry in each dataset. Warm and cold colours depict leftward and rightward asymmetry, respectively. (C) Spatial overlap (Pearson’s r) of the unthresholded maps between datasets for areal (lower matrix) and thickness asymmetry (upper). (D) Overlapping effects across datasets were used to delineate clusters exhibiting population-level areal (lower threshold = 5%) and (E) thickness asymmetry (lower threshold = 1%) based on a minimum 6-dataset overlap (black outlined clusters). (F, H) Raw distribution of the individual-level asymmetry index (AI) in adults extracted from clusters exhibiting areal and thickness asymmetry, respectively. Mean AI’s are in black, Raw distributions are shown for the LCBC (18–55 years) dataset with mixed effects data (cluster-wise outliers defined in lifespan analysis removed on a region-wise basis; Methods; Supplementary file 1E-F). X-axis denotes the AI of the average thickness and area of a vertex within the cluster. (G, I) Proportion of individuals with the expected directionality of asymmetry within each cluster exhibiting areal and thickness asymmetry, respectively, shown for the three largest adult datasets. The X-axes in G and I are ordered according to the clusters shown in F and H, respectively. Lat = lateral; Med = medial; Post = posterior; Ant = anterior; Sup = superior; Inf = inferior.

For thickness, an anterior-posterior pattern of left-right asymmetry was evident in most datasets (Figure 1B), consistent with more recent reports (Kong et al., 2018; Roe et al., 2021; Sha et al., 2021a; Plessen et al., 2014; Williams et al., 2022). Though spatial correlations between AI maps from independent datasets were high, they were notably more variable (r=0.33 - 0.93; Figure 1C); HCP showed lower correlation with all datasets (r=0.33 - 0.46) whereas all other datasets correlated highly (min r=0.78). Consistent leftward thickness asymmetry effects were evident in cingulate, postcentral gyrus, and superior frontal cortex (Figure 1E), whereas consistent effects for rightward thickness asymmetry were evident in a large cluster in and around STS (Figure 1E; Figure 1—figure supplement 1B), insula, lingual gyrus, parahippocampal and entorhinal cortex. Of note, both areal and thickness asymmetry extended beyond these described overlapping effects (Figure 1—figure supplements 1 and 2 & Figure 1—figure supplement 5).

Based on effect size criteria (Figure 1D–E; Methods), we derived a set of robust clusters exhibiting population-level areal (14 clusters) and thickness asymmetry (20 clusters) for further analyses (see Supplementary file 1E-F for anatomical descriptions). The proportion of individuals lateralized in the population direction in each cluster was highly similar across datasets, on average ranging between 61% and 94% for area, and 57–90% for thickness (Figure 1F–I). We then formally compared our approach to asymmetry estimates derived from a gyral-based atlas often used to assess asymmetry (Kong et al., 2018; Koelkebeck et al., 2014; Sha et al., 2021b), finding fairly poor correspondence with the vertex-wise structure of cortical asymmetry for atlas-based regions, particularly for thickness asymmetry (Figure 1—figure supplement 3).

Lifespan trajectories of population-level cortical asymmetries

Having delineated regions exhibiting population-level areal and thickness asymmetry, we aimed to characterize the developmental and lifespan trajectories of cortical asymmetry from early childhood to old age, using a lifespan sample incorporating dense longitudinal data (Methods). For this, we used the mixed-effects LCBC lifespan sample covering the full age-range (4–89 years). To account for non-linear lifespan change, we used Generalized Additive Mixed Models (GAMMs) to model the smooth left- (LH) and right hemisphere (RH) age-trajectories in our robust clusters (Methods).

In all clusters, the homotopic areal trajectories revealed areal asymmetry was strongly established already by age ~4, and the lifespan trajectories of both leftward (Figure 2A) and rightward (Figure 2B) asymmetries were largely parallel. Specifically, a large left-asymmetric region in and around SMG/perisylvian (#1; Figure 2A) showed strong asymmetry by age ~4 that was largely maintained throughout life through steady aging-associated decline of both hemispheres, whereas leftward asymmetry of temporal cortex (#2,6) and anterior insular (#4) was maintained through developmental expansion and aging-associated decline of both hemispheres. Others (retrosplenial #5; mPFC #3,7) showed growth from pre-established asymmetry and more variable lifespan trajectories. On the other side, rightward asymmetries showed largely preserved asymmetry through aging-associated decline of both hemispheres (Figure 2B; medial occipital #1; lateral parietal #2; STS #5; orbitofrontal #7), through bilateral developmental expansion and aging-associated decline (mPFC #6), or steadily expanding bilateral surface area until mid-life (cingulate; #3). There was also little indication of relative hemispheric differences during cortical developmental expansion from 4 to 30 years (Figure 3—figure supplement 3A) or aging from 30 to 89 years (Figure 3—figure supplement 5). Though lifespan areal asymmetry trajectories did show significant change at some point throughout life in most clusters (Supplementary file 1E), factor-smooth GAMM interaction analyses confirmed that areal asymmetry was significantly different from zero across the entire lifespan in all clusters (Figure 3—figure supplements 12), and the average trajectories across all leftward and rightward clusters were clearly parallel (although still both exhibited a significant difference; bordered plots in Figure 2A–B; Supplementary file 1E).

Lifespan trajectories of population-level areal asymmetry.

Homotopic lifespan trajectories of surface area in clusters exhibiting population-level (A) leftward (yellow plots; yellow clusters) and (B) rightward (pink plots; blue clusters) areal asymmetry (mm2). Larger plots on the left show the mean age trajectory across all clusters exhibiting leftward (top) and rightward (bottom) asymmetry. Note that the unit of measurement is the average surface area of a vertex within the cluster. Dark colours correspond to LH trajectories. All age trajectories were fitted using GAMMs. Data is residualized for sex, scanner and random subject intercepts. Clusters are numbered for reference. As outliers were removed on a region-wise basis (Methods), the number of observations underlying the plots range from 7862 to 7874 (see Supplementary file 1E).

In contrast, though homotopic trajectories of thickness clusters were more variable, they were non-parallel, and mostly characterized by developmental increase in thickness asymmetry from age 4–30, through seemingly unequal rates of continuous thinning between hemispheres (Figure 3; Figure 3—figure supplements 13). Importantly, in 10/20 clusters the data indicated developmental increase in thickness asymmetry corresponded to a significant relative hemispheric difference in the rate of developmental thinning. Mostly, these conformed to a pattern whereby the thicker homotopic hemisphere thinned comparatively slower (Figure 4): leftward thickness asymmetry developed through comparatively slower thinning of the LH that was significant in 6 clusters (superior, lateral and medial PFC #2 #8 #10, precentral #4, inferior temporal #6, calcarine #11; Figure 3A; Figure 4), whereas rightward asymmetry developed through significantly slower RH thinning (STS #1, planum temporale #7, anterior insula #9; Figure 3A; Figure 4), or significantly faster RH thickening (#5 entorhinal). Only one other cluster exhibited a relative hemispheric difference seemingly driven by faster thinning of the thicker hemisphere (#8 posterior cingulate; Figure 4). In these clusters, asymmetry development was generally evident until a peak in early adulthood (median age at peak = 24.3; see Figure 4) for both leftward and rightward clusters, around a point of inflection to less developmental thinning (see also Figure 3—figure supplement 4). The average trajectories across all leftward and rightward clusters also indicated developmental asymmetry increase (bordered plots; Figure 3). Despite the developmental growth, factor-smooth GAMMs nevertheless confirmed the developmental foundation for thickness asymmetry was already established by age ~4 (95% of clusters exhibited small but significant asymmetry at age ~4; Figure 3—figure supplement 2B), and again asymmetry trajectories showed significant change at some point throughout life (Supplementary file 1F). Across clusters delineated here we observed little evidence aging-related change from 30 to 89 years corresponded to a relative hemispheric difference in the rate of aging-related thinning, except in regions overlapping with our previous report (Roe et al., 2021; e.g. mPFC #10; Figure 3—figure supplement 5B; Figure 3—figure supplement 4). Thus, across population-level thickness asymmetries, the data indicated either developmental growth in asymmetry, or conserved relative asymmetry through development and aging despite absolute asymmetry change. Results were robust to varying the number of knots used to estimate trajectories (Figure 3—figure supplement 1).

Figure 3 with 5 supplements see all
Lifespan trajectories of population-level thickness asymmetry.

Homotopic lifespan trajectories of cortical thickness in clusters exhibiting population-level (A) leftward (yellow plots; yellow clusters) and (B) rightward (pink plots; blue clusters) thickness asymmetry (mm). Larger plots on the left show the mean age trajectory across all clusters exhibiting leftward (top) and rightward (bottom) asymmetry. Dark colours correspond to LH trajectories. All age trajectories were fitted using GAMMs. Data is residualized for sex, scanner, and random subject intercepts. Clusters are numbered for reference. As outliers were removed on a region-wise basis (Methods), the number of observations underlying the plots range from 7856 to 7874 (see Supplementary file 1F).

Relative developmental trajectories of homotopic cortical thickness in clusters exhibiting population-level thickness asymmetry.

To highlight development, the x-axis covers the age-range 4–30 years, although relative change was calculated from full lifespan models (Methods). Darker trajectories indicate LH trajectories. Shaded areas indicate 95% CI. A relative difference in developmental thinning rates between hemispheres is suggested if the CI’s of relative LH and RH trajectories diverge to be non-overlapping (denoted with *). These indications should be interpreted in combination with the full lifespan GAMM trajectories shown in Figure 3. Where relative hemispheric differences are indicated, age at the point of maximum thickness asymmetry across life is denoted in grey (Methods). Since one cluster (R7) was estimated to exhibit maximum thickness asymmetry at age 89.4 (see Figure 3), age at maximum asymmetry for the developmental peak is also given in parentheses (Figure 3). As outliers were removed on a region-wise basis (Methods), the number of observations underlying the plots range from 7856 to 7874 (see Supplementary file 1F).

Interregional asymmetry correlations

We then investigated which cortical asymmetries correlate within individuals (AI’s corrected for age, sex, scanner; Methods). For areal asymmetry, a common covariance structure between asymmetries was detectable across datasets: Mantel tests revealed the correlation matrices derived independently in LCBC, UKB and HCP data all correlated almost perfectly (r ≥ 0.97, all p<9.9e-5; Figure 5A; Figure 5—figure supplement 1A). The highest correlations (or ‘hotspots’) all reflected positive correlations between regions that are on average left-asymmetric and regions that are on average right-asymmetric (i.e. higher leftward asymmetry in one region related to higher rightward asymmetry in another; Figure 5A black outline); leftward asymmetry in SMG/perisylvian (#1 L) was related to higher rightward asymmetry in inferior parietal cortex (#2 R; r=0.48 [LCBC]), leftward anterior cingulate asymmetry (ACC; #3 L) was related to higher rightward asymmetry in mPFC (#6 R, r=0.47), and leftward asymmetry in a superior frontal cluster (#7 L) was related to rightward asymmetry in the cingulate (#3 R, r=0.68). None of the relationships could be explained by brain size, as additionally removing the effect of intracranial volume (ICV) from cluster AI’s had a negligible effect on their interrelations (max correlation change=0.009). Post-hoc tests confirmed that opposite-direction areal asymmetries were more correlated if closer in cortex (Methods); geodesic distance was lower between cluster-pairs that were more correlated (rho=–0.35, p=0.01 [LCBC]; –0.38, p=0.007 [UKB; Figure 5C]; –0.34, p=0.02 [HCP]), though this was driven by the aforementioned ‘hotspots’. By contrast, same-direction areal asymmetries were not more correlated if closer in cortex (leftward [all p>0.5]; rightward [all p>0.5]). This suggests specific areal asymmetries that are closer in cortex and opposite in direction may show coordinated development.

Figure 5 with 4 supplements see all
Interregional asymmetry correlations.

Interregional correlations between (A) areal asymmetries and (B) thickness asymmetries for each replication dataset (AI’s residualized for age, sex, scanner). Individual AI’s in rightward clusters are inversed, such that positive correlations reflect positive asymmetry-asymmetry relationships, regardless of direction of mean asymmetry in the cluster (i.e. higher asymmetry in the population-direction). Yellow and blue brain clusters/colours denote leftward and rightward asymmetries, respectively (clusters numbered for reference). A consistent covariance structure was evident both for areal (r ≥ 0.97) and thickness asymmetry (r ≥ 0.46; results above matrices). Black box in A highlights relationships between opposite-direction asymmetries (i.e. leftward vs rightward regions). (C) For areal asymmetry, asymmetry in opposite-direction cluster-pairs that were closer in cortex was more positively correlated (datapoints show cluster-pairs; geodesic distance in mm). (D) A single component explained 21.9% variance across thickness asymmetries in UKB (inset plot). Accordingly, we found a correlation of r=–0.61 (p<2.2e–16) in UKB between mean asymmetry across leftward clusters (Y-axis) vs. mean asymmetry across rightward clusters (X-axis; AI’s in rightward clusters inversed). Lines of symmetry (0) are in dotted grey (see also Figure 5—figure supplements 13).

For thickness asymmetry, the correlation matrix exhibited a clear pattern in UKB that was less visible but still apparent in LCBC and HCP (Figure 5B; Figure 5—figure supplement 1B). Mantel tests confirmed that the covariance structure replicated between all dataset-pairs (LCBC-UKB r=0.49, p=0.007; LCBC-HCP r=0.62, p<2.9e–4; UKB-HCP r=0.46, p=0.009). The observed pattern suggested higher leftward asymmetry in regions that are on average left-asymmetric was associated with less rightward asymmetry in regions that are on average right-asymmetric. However, given that the AI measure is bidirectional, closer inspection of the correlations revealed that higher leftward asymmetry in regions that are left-asymmetric actually corresponded to more leftward asymmetry in right-asymmetric regions, and vice versa (and on average; see Figure 5—figure supplement 2). In other words, individuals may tend towards either leftward lateralization or rightward lateralization (or symmetry) on average, irrespective of the region-specific direction of mean thickness asymmetry in the cluster. Similarly, asymmetry in left-asymmetric regions was mostly positively correlated, and asymmetry in right-asymmetric regions was mostly positively correlated. Again, additionally removing ICV-associated variance had negligible effect (max correlation change=0.01). Post-hoc principal components analysis (PCA) in UKB revealed PC1 explained 21.9% of the variance in thickness asymmetry and suggested a single component may be evident for thickness asymmetry (Figure 5D). Accordingly, we found a correlation of r=–0.61 between mean asymmetry across all leftward vs. mean asymmetry across all rightward clusters in UKB p<2.2e-16 [means weighted by cluster size]; see Figure 5D; r=–0.56, p<2.2e-16 [unweighted means]; r=0.66, p<2.2e-16 [PC1 across all leftward vs. PC1 across all rightward]. Although less strong, all relationships were significant in LCBC (r=–0.12; p=7.4e-11 [weighted]; r=–0.05; p=0.005 [unweighted]; r=0.19; p<2.2e-16 [PC1 vs. PC1]) and significant or trend-level in HCP (r=–0.11; p=1.6e-4 [weighted]; r=–0.04; p=0.15 [unweighted]; r=0.12, p=3.3e-5 [PC1 vs. PC1]; see Figure 5—figure supplement 3). These results suggest thickness asymmetry may be globally interrelated across the cortex and show high directional variability in the adult population.

Heritability

Although heritability of the global measures for each hemisphere was high (area h2SNP ~67%, h2twin ~92%; thickness h2SNP ~36%, h2twin ~81%), heritability of global asymmetry measures was substantially lower (area h2SNP=7% [95% CI=3–10%], h2twin=16% [5–27%]; thickness h2SNP=1% [0–5%], h2twin=16% [5–27%]). Except the SNP-based estimate for global thickness asymmetry (P=.22), all estimates for global measures were post-corrected significant (p<0.008; see Supplementary file 1G). Of our robust asymmetry clusters, four areal asymmetries showed significant twin-based heritability in HCP (post-correction; leftward SMG/perisylvian and anterior insula, and rightward cingulate and STS), all of which were also significant in the SNP-based analyses (h2twin range = 19–27%; h2SNP range = 8–19%). Additionally, SNP-based analyses revealed 9/14 (64%) areal asymmetry clusters exhibited significant heritability (post-correction; Supplementary file 1H). Importantly, highest SNP-based heritability was observed for leftward areal asymmetry in the anterior insula cluster (h2SNP = 18.6%, p<1.78e–15; see Supplementary file 1H), which was substantially higher than the next highest estimates in SMG/perisylvian (h2SNP = 10.7%, p=3.01e–9), retrosplenial cortex, gyrus rectus and the cingulate (all h2SNP = 8–10%). For thickness, two robust asymmetries survived correction in the twin-based analyses (rightward STS and lingual gyrus). Of these, only STS was also significant in the SNP-based analysis but did not survive correction. Moreover, only 3/20 (15%) thickness asymmetries exhibited significant SNP-based heritability (h2SNP = 3–7%; Supplementary file 1I). Cluster-wise heritability estimates were significantly lower for thickness asymmetry than for area using a SNP-based approach (β=–1.1, p=0.0008) but not a twin-based approach (β=–0.03, p=0.33).

We then estimated asymmetry heritability cortex-wide (Schaefer et al., 2017; Figure 6). For areal asymmetry, 69 parcels survived FDR-correction in HCP, 84% (58) of which were also FDR-corrected significant in the SNP-based analysis. Moreover, a total of 267 (53%) parcels exhibited significant FDR-corrected SNP-based heritability for areal asymmetry in UKB data (significant p[FDR]<0.05 parcels in each sample are depicted with black outlines in Figure 6A). Beyond significance, a consistent heritability pattern for areal asymmetry was clearly evident using SNP- and twin-based data from independent samples, with higher heritability notably in anterior insula, SMG, Sylvian fissure, STS, calcarine sulcus, cingulate, medial and orbitofrontal cortex, and fusiform. This overlap was substantiated by a spatial correlation of r=0.46 between maps; p<2.2e–16. Moreover, maximum SNP-heritability (cyan parcel in Figure 6) was observed in anterior insula (parcel h2SNP = 16.4%; p<1.78e–15), confirming this region constitutes the most heritable cortical asymmetry in humans (and not improving on the cluster-based estimate). For thickness asymmetry, 15 parcels survived FDR-correction in the twin-based analysis, 7 of which were also post-corrected significant in the SNP-based analysis, and these were typically low estimates (Figure 6B). Moreover, significant FDR-corrected SNP-heritability was observed in only 11% (57) of parcels, including around superior temporal gyrus, planum temporale, the posterior insula/Sylvian fissure, anterior insula, and in orbitofrontal cortex (max h2SNP = 16.6%), along the cingulate and in medial visual cortex. Beyond significance, we observed little obvious visual overlap in twin- and SNP-based heritability patterns (spatial correlation was significant but low; r=0.24; p=0.01), and higher estimates pertained to regions that were limited in extent but generally showed no clear global pattern common to both datasets, with the possible exception of calcarine, cingulate and orbitofrontal cortex. Furthermore, cortex-wide heritability estimates were significantly lower for thickness asymmetry than for area using both a SNP-based (β=–0.71, p<2.2e–16) and twin-based approach (β=–0.33, p=1.08e–7).

Figure 6 with 1 supplement see all
Asymmetry heritability.

Heritability of areal (A) and thickness asymmetry (B) estimated cortex-wide using SNP-based (UKB data; top rows) and twin-based methods (HCP data; bottom rows). Unthresholded effect maps are shown. Parcels in black outline show significance at p[FDR]<.05. Cyan parcels depict the point of maximum observed SNP-heritability (area h2=16.4%; thickness h2=16.6%). (C) Significant SNP-based genetic correlations (FDR-corrected) between areal (lower matrix) and thickness asymmetries (upper matrix). For area, SNP-based genetic correlations explained several phenotypic correlations (Figure 5A). For thickness, one pair survived FDR-correction (shown). See Figure 6—figure supplement 1 for comparison with genetic correlation estimates from the twin-based HCP sample. Individual AI’s in rightward clusters are inversed, such that positive genetic correlations reflect asymmetry-asymmetry genetic relationships, regardless of direction of mean asymmetry in the cluster (i.e. higher asymmetry in the population-direction). Yellow and blue brain clusters/colours denote leftward and rightward asymmetries, respectively (clusters numbered for reference).

For areal asymmetry, large SNP-based genetic correlations explained several phenotypic correlations evident in Figure 5A (Figure 6C–D; Methods). For example, high SNP-based genetic correlations were found between leftward asymmetry in SMG/perisylvian and higher rightward asymmetry in lateral parietal cortex (LPC; rGSNP = 0.83; p[FDR]=6.58e–5), between leftward superior frontal cortex asymmetry and rightward asymmetry along the cingulate (rGSNP = 0.82; p[FDR]=0.01), and between leftward anterior temporal/parahippocampal asymmetry and rightward asymmetry in LPC (rGSNP = 0.68; p[FDR]=0.01). SNP-based genetic correlations between anterior insula and two rightward superior frontal clusters were also observed (rGSNP = 0.86; p[FDR]=1.21e–6; rGSNP = 0.42; p[FDR]=6.64e–4) in the absence of phenotypic correlations (Figure 5), and several same-direction asymmetries showed moderate genetic correlation. In contrast, only one post-corrected significant SNP-based genetic correlation was found for thickness asymmetry (see Figure 6C; rGSNP = 0.68; p[FDR]=0.03). We tested replication in the twin-based HCP sample across all cluster-pairs tested in the SNP-based analysis (Methods; note not all of these exhibited significant twin-based heritability; Supplementary file 1H-I). Two twin-based genetic correlations survived FDR-correction, both for areal asymmetry (Figure 6—figure supplement 1). One was the high genetic correlation between leftward SMG/perisylvian and rightward LPC areal asymmetry observed in the SNP-based analysis (rGtwin = 1.0 [95% CI=0.69–1.0], p[FDR] = 0.03). The second was between leftward anterior insula and leftward SMG/perisylvian areal asymmetry and was not post-corrected significant in the SNP-based analysis (rGtwin = 0.59 [0.24–1.0], p[FDR]=0.03). Of note, while several areal asymmetry cluster-pairs exhibiting post-corrected significance in the SNP-based analysis were estimated to show high twin-based genetic correlation, these did not survive correction (see Figure 6—figure supplement 1). For thickness asymmetry, no twin-based genetic correlation survived correction, and there was little overlap in estimates produced betweeen methods.

Associations with cognition, handedness, sex, and ICV

Finally, we found several significant associations between asymmetry and factors purportedly related to it in UKB data (Figure 7; Supplementary file 1J-K). Notably, all effect sizes were small. For general cognitive ability, we found one association, wherein higher areal asymmetry in the largest leftward cluster (SMG/perisylvian) was significantly associated with better cognition (β=0.03 [CI = 0.02 –0.04], p=7.4e–7, Figure 7C). Although small, we note this association was far from only just surviving correction at our predefined alpha level (⍺=0.01; 136 tests; Methods). This was checked in the substantially reduced non-imputed subset of data with no missing cognitive variables and retained the lowest p-value (N=4696; β=0.04 [CI = 0.01 -0.07]; p=6.9e–3). We also post-hoc tested areal asymmetry associations with the 11 separate cognitive tests (Figure 7—figure supplement 1). For handedness, reduced leftward areal asymmetry in anterior insula and thickness asymmetry along postcentral gyrus was evident in left-handers, in line with our recent vertex-wise mapping in UKB (Sha et al., 2021a). For sex effects, which were also small, males typically exhibited slightly stronger areal asymmetry in large clusters (e.g. leftward SMG/perisylvian and temporal pole; rightward inferior parietal and superior frontal) but reduced leftward and rightward asymmetry in mPFC. For thickness, males exhibited more rightward asymmetry in STS and posterior insula, more leftward thickness asymmetry in superior frontal cortex, but reduced rightward thickness asymmetry in entorhinal cortex and anterior insula, and reduced leftward asymmetry in caudal superior frontal cortex. As ICV effects were typically most nominal, these are shown in Figure 7—figure supplement 2. None of the reported associations changed appreciably when controlling for additional brain-size related covariates (Williams et al., 2022; Supplementary file 1J-K).

Figure 7 with 2 supplements see all
Asymmetry associations with general cognitive ability (first principal component [PC1]), handedness, sex, and intracranial volume (ICV) in UKB, in clusters exhibiting population-level areal (upper) and thickness asymmetry (lower).

(A, D) Significance of associations (negative logarithm; corrected [p<7.4e–5] and uncorrected threshold [p=0.01] shown by dotted and non-dotted line, respectively). X-axis displays the test for each cluster-association. As maximum sample size was used to test each association, effects of general cognitive ability were tested in separate models with fewer observations (N=35,198; separated association plots) than handedness, sex and ICV (N=37,569). (C) Visualization of the found association between leftward areal asymmetry in the large supramarginal cluster with general cognitive ability. The line of null association is shown for comparison (dotted) (B, E) Right plots denote effect sizes, 95% confidence intervals (error bars) and cortical location of associations surpassing Bonferroni-corrected significance. Individual AIs in rightward clusters were inversed. Right handers and females are coded 0, such that a negative effect for general cognitive ability / handedness / sex / ICV / reflects less asymmetry in higher cognition / left handers / males / larger brains. Associations with ICV are shown in Figure 7—figure supplement 2. Yellow and blue clusters denote leftward and rightward asymmetries, respectively.

Discussion

Combining the strengths of a vertex-wise delineation of population-level cortical asymmetry in 7 international datasets and dense longitudinal data, we offer the first description of the longitudinal developmental and lifespan trajectories of cortical asymmetry, advancing knowledge on normal human brain development. We show areal asymmetry is predominantly stable across life, whereas we trace developmental growth in many thickness asymmetries signifying differentiation of thickness asymmetry from early childhood to the mid-20s. We further demonstrate the replicable interregional relationships between asymmetries within individuals, provide the most detailed heritability maps for cortical asymmetry to date, and uncover novel and confirm previously-reported associations with factors purportedly related to asymmetry – all with small effects. All maps are available in Supplementary file 2.

Population-level asymmetry

Our vertex-wise analysis of cortical asymmetries that reproduce across adult cohorts replicates and completes a recent low-resolution meta-analysis (Kong et al., 2018), and can serve as a high-fidelity phenotype for future brain asymmetry studies. The marked consistency across samples suggests consensus may now be reached regarding cortical asymmetry phenotypes in humans, as our results agree with most reported results for areal asymmetry (Kong et al., 2018; Chiarello et al., 2016; Lyttelton et al., 2009; Maingault et al., 2016; Li et al., 2014), as well as several, typically more recent reports for thickness asymmetry (Kong et al., 2018; Roe et al., 2021; Li et al., 2015; Plessen et al., 2014; but see Figure 1—figure supplement 4 and Limitations for an outstanding issue regarding the biological origin of thickness asymmetry). Indeed, for thickness asymmetry – for which findings have been particularly mixed (Kong et al., 2018; Chiarello et al., 2016; Roe et al., 2021; Li et al., 2015; Luders et al., 2006; Zhou et al., 2013; Maingault et al., 2016; Shaw et al., 2009; Lou et al., 2020; Hamilton et al., 2007; Zhou et al., 2018; Koelkebeck et al., 2014; Plessen et al., 2014) – the left-right patterning here is compatible with low-resolution meta-analyses (Kong et al., 2018), asymmetries evident in the first months of life (Li et al., 2015; Williams et al., 2023), a large-scale mapping in mid-old age (Sha et al., 2021a), reports using alternative analysis streams Li et al., 2015; Plessen et al., 2014, and possibly the overall pattern of brain torque (Kong et al., 2018; Hugdahl, 2011) – a gross hemispheric twist leading to frontal and occipital bending at the poles (Toga and Thompson, 2003). The high overlap in effects between seven datasets from four countries here suggests these results likely apply universally. This evident consensus suggests genetic-developmental programs regulate mean brain lateralization with respect to both area and apparent thickness in humans. However, the genetic findings presented herein suggest these may have reached population fixation, as heritability of even our optimally delineated asymmetry measures was generally low. This indicates either subject-specific stochastic mechanisms in early neurodevelopment or later developmental influences primarily determine cortical asymmetry. Tracing their lifespan development, we show the trajectories of areal asymmetry primarily suggest this form of asymmetry is developmentally stable at least from age ~4, maintained throughout life, and formed early on – possibly in utero (Sha et al., 2021b; Li et al., 2014; Williams et al., 2023) (while we cannot extrapolate to ages before our sample begins, we note this agrees with findings in neonates Li et al., 2014; Williams et al., 2023). One interpretation of lifespan stability combined with low heritability may be stochastic early-life developmental influences determine individual differences in areal asymmetry more than later developmental change, but work linking prenatal and childhood trajectories is needed to affirm this. Still, we also found relatively stronger heritability for areal asymmetry (notably, anterior insula exhibited ~19% SNP-heritability). This also illustrates region-dependent genetic effects upon areal asymmetry, and high genetic correlations suggest specific areal asymmetries are formed under common genetic influence. In contrast, childhood development of thickness asymmetry until a peak around age ~24 (Roe et al., 2021), higher directional variability in adult samples, and lower heritability all converge to suggest thickness asymmetry may be more shaped through subject-specific effects in later childhood, possibly through interaction with the environment. This interpretation applied to asymmetry also agrees with work suggesting cortical area in general may trace more to early-life factors (Walhovd et al., 2016; Grasby et al., 2020; Rakic, 1995) whereas thickness may be more impacted by lifespan influences (Grasby et al., 2020; Fjell et al., 2019).

Lifespan trajectories

Our longitudinal description of cortical asymmetry lifespan trajectories gleaned novel insight into normal brain maturation. For areal asymmetry, adult-like patterns of lateralization were strongly established before age ~4, indicating areal asymmetry traces back further and does not primarily emerge through later cortical expansion (Wierenga et al., 2014). Rather, the lifespan trajectories predominantly show stability from childhood to old age, as asymmetry was maintained through periods of developmental expansion and aging-related change that were region-specific and bilateral. This may align with evidence indicating areal asymmetry may be primarily determined in utero (Li et al., 2014; Williams et al., 2023), including evidence suggesting little change in areal asymmetry from birth to 2 years (Li et al., 2014; Li et al., 2013; Wierenga et al., 2014), and little difference between maps derived from neonates and adults (Li et al., 2014; Williams et al., 2023). It may also fit with the principle that the primary microstructural basis of cortical area (Rakic, 1995) – the number of and spacing between cortical minicolumns – is determined in prenatal life (Rakic, 1995; Fjell et al., 2019), and agree with work suggesting asymmetry at this microstructural level may underly hemispheric differences in surface area (Chance et al., 2006). The developmental trajectories agree with studies indicating areal asymmetry is established and strongly directional early in life (Li et al., 2014; Remer et al., 2017). That change in surface area later in development follows embryonic gene expression gradients may also agree with a prenatal account for areal asymmetry (Fjell et al., 2019). Our results may therefore constrain the extent to which areal asymmetry can be viewed as a plastic feature of brain organization, and may even suggest areal asymmetry may sometimes be a marker for innate hemispheric specializations shared by most humans. Although future research is needed to assess structure-function relationships, the degree of precision with which leftward areal asymmetry follows the contours of auditory regions in the Sylvian fissure (Figure 1—figure supplement 1) that show left functional lateralization may be one example (Chance et al., 2006; Tzourio-Mazoyer et al., 2018; Ocklenburg et al., 2018); we found ~94% of individuals exhibited leftward areal asymmetry here – the most consistently lateralized cortical region in humans (Figure 1F).

In contrast, although weak thickness asymmetry was evident by age ~4, we observed childhood developmental growth in many thickness asymmetries. Developmental trajectories showed non-linear asymmetry growth by virtue of accelerated thinning of the non-dominant hemisphere (10/20 clusters showed this relative hemispheric difference in the rate of developmental thinning; Figure 4). This led to maximally established asymmetry around ~24 years of age. These trajectories clearly suggest differentiation of the cortex occurs with respect to thickness asymmetry in development, possibly (though not necessarily) suggesting it may be more amenable to experience-dependent change. Indeed, as cortical thinning in childhood is thought to partly reflect learning-dependent processes such as intracortical myelination (Natu et al., 2019), pruning of initially overproduced synapses (Petanjek et al., 2011; Faust et al., 2021) and neuropil reduction, thickness asymmetry growth may suggest hemispheric differences in the developmental optimization of cortical networks at least partly shaped by childhood experience. This raises the possibility thickness asymmetry may be a marker of ontogenetic hemispheric specialization within neurocognitive networks, possibly in line with animal models suggesting lateralized training can alter the balance in homotopic thickness (Díaz et al., 1994). However, this is speculative, and intervention approaches are needed to experimentally test plasticity of thickness asymmetry in humans. Our lifespan results are difficult to reconcile with earlier lifespan reports finding different mean thickness asymmetry patterns (Zhou et al., 2013; Shaw et al., 2009; Zhou et al., 2018) and modelling asymmetry linearly across cross-sectional childhood and adult samples (Zhou et al., 2013; Zhou et al., 2018; Plessen et al., 2014). However, our findings agree with work finding a similar left-right thickness asymmetry pattern shows rapid longitudinal increase in the first years of life (Li et al., 2015), particularly in mPFC (Li et al., 2015). As we observed rapid asymmetry differentiation spanninng childhood and adolescence in most prefrontal regions and others (Figure 3; Figure 4; Figure 3—figure supplements 3 and 4), we extend these earlier findings in neonates (Li et al., 2015). Likely, longitudinal data and nonlinear modelling was critical to capture early developmental growth in thickness asymmetry within a lifespan perspective, because large variation in hemispheric thickness estimates at any age may hinder detection of subtle change effects even in large cross-sectional samples (Fjell et al., 2020), rendering follow-up data likely critical in the context of cortical asymmetry change. As prefrontal thickness asymmetry seems particularly vulnerable in some neurodevelopmental disorders (Postema et al., 2019), aging, and Alzheimer’s disease (Roe et al., 2021), these trajectories provide a useful normative reference regarding the timing of expected brain change in development. With regards to aging, most clusters delineated here did not exhibit the relative hemispheric difference in rates of cortical thinning we have previously shown is a feature of aging in heteromodal cortex (Roe et al., 2021), except in clusters overlapping with our previous analysis (Figure 3—figure supplement 5). This fits with our previous work showing aging-related loss in thickness asymmetry is specific to heteromodal regions vulnerable in aging. In these, we also found strong evidence of early developmental growth in thickness asymmetry (Figure 3—figure supplement 4). Developmental differentiation and aging-related dedifferentiation of thickness asymmetry underscores its proposed role in supporting optimal brain organization and function, though it seems not all thickness asymmetries that grow in childhood development decline in aging.

Interregional asymmetry correlations

For areal asymmetry, we uncovered a covariance structure that almost perfectly replicated across datasets. In general, this fit with a multifaceted view (Bain et al., 2019; Rentería, 2012; Francks, 2015), in which most asymmetries were either not or only weakly correlated – but reliably so – contrasting views emphasizing a single biological (Annett, 1998; Annett, 1964) or overall anatomical factor (Crow, 2010) controlling cerebral lateralization. However, we also identified several regions wherein areal asymmetry reliably correlated within individuals, showing the variance in cortical asymmetries is not always dissociable, as often thought (Bain et al., 2019; Rentería, 2012; Francks, 2015). The strongest relationships all pertained to asymmetries that were proximal in cortex but opposite in direction. Several of these were underpinned by high asymmetry-asymmetry SNP-based genetic correlations, illustrating some lateralizations in surface area exhibit coordinated genetic development.

For thickness asymmetry, we also uncovered a common covariance structure – particularly clear in UKB – that nevertheless replicated with moderate precision across datasets (Figure 5C). Furthermore, a single component explained 21.9% variance in thickness asymmetry in UKB, and a high correlation across 38,171 individuals further suggested thickness asymmetry may be globally interrelated across the cortex (Figure 5D). These data for thickness indicate individuals may tend towards either leftward asymmetry, rightward asymmetry, or symmetry, both globally across the cortex and irrespective of the region-specific average direction of asymmetry (Figure 5—figure supplements 23). Though it is unclear why the relationships were weaker in the other datasets, we nevertheless found similarly significant relationships in each (Figure 5—figure supplement 3). This result may be in broad agreement with the notion that some lateralized genetic-developmental programs may trigger lateralization in either direction or lose their directional bias through environmental interaction (Francks, 2015). As thickness asymmetry seems established at but minimal from birth (Li et al., 2015), genetic effects may determine the average region-specific hemispheric bias in the population, but later developmental change may subsequently confer major increases upon its directional variance (Francks, 2015). Overall, our data suggests developmental change in thickness asymmetry may lead to directional variability of its lateralization across individuals. Thus, far from being independent phenotypes (Bain et al., 2019; Rentería, 2012), thickness asymmetries may be globally interrelated across cortex and their direction coordinated through childhood.

Genetic influences

For areal asymmetry, we found replicable patterns of low-moderate heritability across datasets and methods. We also found areal asymmetry in anterior insula is, to our knowledge, the most heritable asymmetry yet reported with genomic methods (Sha et al., 2021a; Sha et al., 2021b; Carrion-Castillo et al., 2020; Cuellar-Partida et al., 2021; Elliott et al., 2018), with common SNPs explaining ~19% variance. This is notably higher than in our recent report (<5%) (Sha et al., 2021a), illustrating a benefit of our approach. As we reported recently (Sha et al., 2021a), we confirm asymmetry here associates with handedness. Furthermore, highest SNP and twin-based heritability for areal asymmetry was found in regions constituting the earliest emerging cortical asymmetries in utero (Dubois et al., 2010; Habas et al., 2012; Kasprian et al., 2011; Hill et al., 2010): anterior insula, STS, PT, medial occipital cortex, and parahippocampal gyrus (Figure 6A). However, heritability was not restricted to these regions, as most areal asymmetries exhibited significant – albeit often lower – SNP-based heritability, as did most parcels when estimated cortex-wide. SNP-based heritability was also evident in regions not found in the present analyses to show strong areal asymmetry, such as Broca’s area (but see Figure 1—figure supplement 5). The effects agree with and elaborate on two previous genetic explorations (Kong et al., 2018; Sha et al., 2021b) and reports of heritable areal asymmetry in handedness-associated clusters (Sha et al., 2021a). However, while heritability of either hemisphere was high, areal asymmetry heritability was still only moderate at best, suggesting genetic but primarily subject-specific stochastic effects underly its formation. By contrast, thickness asymmetry was generally not heritable, or showed low and localized heritability effects with no clear global pattern, as well as more divergent results using twin and genomic methods, possibly due to low-power for twin-models.

Together, lifespan stability (Li et al., 2014), higher heritability, and phenotypic and genetic correlations suggest higher genetic influence upon individual differences in areal asymmetry. By contrast, childhood developmental growth, directional variability and low heritability suggest thickness asymmetry may be more shaped through individual lifespan exposures (Fjell et al., 2019). Whether region-specific thickness asymmetry change relates to the maturation of lateralized brain functions is thus an interesting question for future research (Bishop and Bates, 2019; Somers et al., 2015). Regardless, these results support a differentiation between early-life (i.e. before age ~4) and later developmental factors in shaping areal and thickness asymmetry, respectively.

Individual differences

Other factors commonly espoused to be important for asymmetry were associated with only small average effects in adults. For example, we found one region – SMG/perisylvian – wherein higher leftward areal asymmetry related to subtly higher cognitive ability. Since interhemispheric anatomy here is likely related to brain torque (Toga and Thompson, 2003; LeMay, 1976), this may agree with work suggesting torque relates to cognitive outcomes (Kong, 2019; Zhao et al., 2021). Interestingly, that ~94% of humans exhibit leftward asymmetry in this region (Figure 1G) suggests tightly regulated genetic-developmental programs control its lateralized direction in humans (see Figure 6). This result may therefore suggest disruptions in areal lateralization early in life are associated with cognitive deficits detectable in later life as small effects in big data (Deary et al., 2004). While speculative, this may also agree with evidence that differences in general cognitive ability that show high lifespan stability (Deary et al., 2004) relate primarily to areal phenotypes formed early in life (Walhovd et al., 2016; Rakic, 1995; Fjell et al., 2019).

Consistent with our recent analysis in UKB (Sha et al., 2021a), we confirmed leftward areal asymmetry of anterior insula, and leftward somatosensory thickness asymmetry is subtly reduced in left-handers. Sha et al. (Sha et al., 2021a) reported shared genetic influences upon handedness and asymmetry in anterior insula and other more focal regions. Anterior insula lies within a left-lateralized functional language network (Labache et al., 2019), and its structural asymmetry may relate to language lateralization (Chiarello et al., 2013; Biduła and Króliczak, 2015; Keller et al., 2011) in which left-handers show increased atypicality (Wiberg et al., 2019; Mazoyer et al., 2014; Westerhausen et al., 2006; Carey and Johnstone, 2014). Since asymmetry here emerges early in utero (Dubois et al., 2010) and is by far the most heritable (Figure 6), we agree with others (Chiarello et al., 2013) that this ontogenetically foundational region of cortex may be fruitful for understanding genetic-developmental mechanisms influencing laterality (Afif et al., 2007; Kalani et al., 2009). Less leftward somatosensory thickness asymmetry in left-handers also echoes our recent report (Sha et al., 2021a) and fits a scenario whereby thickness asymmetries may be partly shaped through use-dependent plasticity and detectable through group-level hemispheric specializations of function. Still, the small effects show cortical asymmetry cannot predict individual handedness. Associations with other factors typically assumed important were similarly small, and mostly compatible with the ENIGMA report (Kong et al., 2018) and elsewhere (Williams et al., 2022; Guadalupe et al., 2015). Concerning sex effects – which were small, differing in direction, and more predictive than ICV (Guadalupe et al., 2015) – inconsistencies between ours and ENIGMA include findings of increased (here) and decreased (Kong et al., 2018) LPC areal asymmetry in males, and increased (Kong et al., 2018) and decreased (here) entorhinal thickness asymmetry in males, and our approach detected other regions slightly more asymmetric in males (e.g. STS). Possibly, differences in sample median age (UKB = ~64; Kong et al., 2018 = 26) and potential sex-differences in brain decline (McCarrey et al., 2016) may underlie some inconsistencies.

Limitations

Some limitations should be noted. First, our delineation of population-level asymmetry used a single analysis software. We used FreeSurfer’s default ‘recon-all’ to delineate the cortex, which has been extensively validated against postmortem data (Cardinale et al., 2014) and is the software underlying most large-scale studies with brain measures. It is currently unclear to what extent differences in pipelines account for previous mixed results. Although we highlight there are clear commonalities between our results and studies using alternative pipelines, suggesting they generalize across analysis streams (Li et al., 2015; Luders et al., 2006; Lyttelton et al., 2009; Plessen et al., 2014), we found one instance where the MRI pipeline leads to different results for thickness asymmetry (HCP pipeline; Figure 1—figure supplement 4). It is not known what underlies this difference, though it is unrelated to the cross-hemispheric registration methods employed here, as our results reproduce using standard methods and thus are likely evident in most FreeSurfer-derived datasets (Figure 1—figure supplement 5). One possibility could be that thickness asymmetry may not reflect cortical thickness differences per se, but rather reflect biologically meaningful hemispheric differences in intracortical myelination that are consistently picked up on via FreeSurfer’s delineation procedure. Future research is required to resolve this. However, that the thickness asymmetry pattern we observe shows a clear developmental trajectory suggests it is a true biological effect (Figure 3; Figure 4; Figure 3—figure supplements 14), as do the replicable aging-related changes therein we have shown previously (Roe et al., 2021). Other sources of inconsistently reported results for cortical asymmetries likely include varying age-distributions (Roe et al., 2021), and multiple asymmetries within atlas-based parcels (e.g. we observed discrepant results to ENIGMA for insula thickness asymmetry Kong et al., 2018). Still, this does not explain other discrepancies, such as rightward areal STS asymmetry here but not elsewhere (Kong et al., 2018; Bain et al., 2019; Remer et al., 2017; Figure 1—figure supplement 2). Thickness asymmetries seem also more variable compared with area (Kong et al., 2018), though it should be noted thickness in general may be slightly less reliable (Hedges et al., 2022), the asymmetry effect is smaller, and thus possibly contains more error. Relatedly, while areal and thickness asymmetry patterns and magnitudes using cross-hemispheric methods agree with standard analysis (Figure 1—figure supplement 5), the magnitude of some asymmetries near the subcortical boundary may be exaggerated via this approach (Sha et al., 2021a). And while we did not find (strong) areal asymmetry in inferior frontal regions (Kong et al., 2018), both the unthresholded significance maps and standard parcellation analyses were compatible with this (Figure 1—figure supplement 2 and Figure 1—figure supplement 5), highlighting a limitation of our thresholding approach. Second, while GAMMs are considered optimal for modelling lifespan data and are robust to nonuniform age distributions (Sørensen et al., 2021), relative underrepresentation in mid-adulthood may drive trajectory inflection points around this age (Roe et al., 2021), urging caution around interpreting these as reflecting real change. Relatedly, as individual-level mean estimates likely contain more error when extracted from smaller clusters, the trajectories from smaller clusters may be more prone to deflection by age density differences. And while spatially averaging across asymmetries helps smooth out some of this noise to better reveal developmental principles underlying cortical asymmetries (Figures 23), it also trades off with accuracy, since trajectories from distinct regions somewhat differ. Third, although differing heritability methods enabled a replication test, methodological differences should be considered when interpreting heritability estimates, and may partly explain the common observation of higher estimates from twins, which we also observed. For example, twin methods implicitly incorporate additive genetic effects and gene-gene interactions amongst other terms, whereas SNP-based methods incorporate only additive effects. Still, twin methods may be prone to overestimating heritability due to unmet assumptions (Dalmaijer, 2020), whereas SNP-based methods may not capture all phenotype-relevant genetic variance and have their own assumptions (Hibar et al., 2017). Adding to this complexity, the samples used for twin- and SNP-based estimation consist of young adults and older adults, respectively. Hence, low SNP-based estimates for thickness asymmetry in UKB may be partly due to reduced thickness asymmetry in older adults (Roe et al., 2021; Williams et al., 2022). However, as cortex-wide twin-based estimates were also significantly lower for thickness asymmetry in the young HCP sample, age is likely not the main driver behind this difference. Again, we also cannot rule out that reliability differences between measures may partially explain some magnitude differences in heritability. Further, genetic correlations can be high even where heritability of either trait is low, and are typically higher than phenotypic correlations – known phenomena also observed here (Cheverud, 1988; Sodini et al., 2018). It may therefore be prudent to not overinterpret their magnitudes, though the veracity of the reported SNP-based genetic correlations seems well-supported, as one expects true genetic correlations between developmentally-related traits (here, sampled from nearby in the same organ), and to track the phenotypic correlations (see also Figure 6—figure supplement 1). Fourth, we imposed a cluster size limit for overlapping asymmetry effects, and thus more focal asymmetries may also be informative for the factors tested here (Sha et al., 2021a). Fifth, only dichotomous handedness self-reports are available with UKB, and future studies might benefit from more nuanced handedness assessments. Relatedly, because UKB cognitive data is not exhaustive (Fawns-Ritchie and Deary, 2020), we extracted the common variance across tests to index general cognitive ability. This approach does not permit testing associations with well-operationalized or specific cognitive domains, and it remains to be seen whether cortical asymmetry may be more informative for lateralized cognition (Ocklenburg et al., 2014).

Overall, we track the development of population-level cerebral cortical asymmetries longitudinally across life and perform analyses to trace developmental principles underlying their formation. Developmental and lifespan trajectories, interregional correlations and heritability analyses converge upon a differentiation between early-life and later-developmental factors underlying the formation of areal and thickness asymmetries, respectively. By revealing hitherto unknown principles of developmental stability and change underlying diverse aspects of cortical asymmetry, we here advance knowledge of normal human brain development.

Methods

Samples

We used anatomical T1 -weighted (T1w) scans from 7 independent international MRI datasets originating from 4 countries (see Supplementary file 1A for an overview of samples used for each analysis). Note that with the exception of vertex-wise analyses in UKB (see below), all analyses made use of all available observations from each sample meeting the stated age-range criteria for each analysis. Studies conducted at the Center for Lifespan Changes in Brain and Cognition (LCBC Vidal-Pineiro et al., 2020) were approved by the Regional Ethical Committee of South-East Norway (2017/653) and complied with all relevant ethical regulations. Ethical approval for the other datasets was granted by the relevant authorities (Supplementary file 1M).

Reproducibility across samples: population-level asymmetry

To delineate average adult patterns of whole-cortical areal and thickness asymmetry, we restricted the age-range of all samples used in the vertex-wise analyses to 18–55. Dataset 1: Here, the LCBC sample comprised 1572 mixed cross-sectional and longitudinal scans (N longitudinal = 812; timepoint range = 1–6) from 923 participants (mean age = 30.6 ± 9.6) collected across 2 scanners. Additionally, 125 individuals were double-scanned at the same timepoint on both scanners. Dataset 2: The Cambridge Centre for Ageing and Neuroscience (Cam-CAN) (Shafto et al., 2014) sample comprised cross-sectional scans of 321 individuals (mean age = 38.7 ± 9.7) (Taylor et al., 2017). Dataset 3: The Dallas Lifespan Brain Study (DLBS) (Kennedy et al., 2012) sample comprised cross-sectional scans of 160 individuals (mean age = 37.5 ± 10.7). Dataset 4: The Southwest University Adult Lifespan Dataset (SALD) (Wei et al., 2018) sample comprised cross-sectional scans of 301 individuals (mean age = 33.7 ± 11.5). Dataset 5: The IXI sample comprised cross-sectional scans of 313 healthy individuals collected across 3 scanners (mean age = 36.8 ± 9.6; http://brain-development.org/ixi-dataset). Dataset 6: Here, the Human Connectome Project (HCP) 1200 (Van Essen et al., 2013) sample comprised 1111 scans (mean age = 28.8 ± 3.7). Dataset 7: Here, the UKB sample consisted of 1000 randomly sampled cross-sectional scans (mean age = 52.1 ± 1.9), restricted to be comparable in size to the other datasets in this analysis.

Lifespan trajectories

Here, we used the full age-range of the longitudinal lifespan LCBC sample (4.1–89.4 years), 3937 cross-sectional and longitudinal scans (N longitudinal = 2762) from 1886 individuals (females = 1139; mean age = 36.8) collected across 4 scanners (271 double-scans) (Fjell et al., 2020; Vidal-Pineiro et al., 2019).

Interregional asymmetry correlations

Here, we used the three largest datasets: LCBC (N=1263; N obs = 2817), UKB (N=38,171), and HCP (N=1109; two outliers removed; see below), excluding the childhood age-range from the LCBC sample not covered in any other dataset.

Heritability and individual differences

For twin heritability, we used HCP 1200 extended twin data (1037 scans from twins and non-twin siblings; age-range=22–37; mean age = 28.9 ± 3.7). The various kinships are described in Supplementary file 1B. All included twin pairs were same-sex. For SNP-heritability, we used the UKB imaging sample with genome-wide data surpassing quality control (N=31,433). For individual differences analyses, we used the UKB imaging sample with the maximum number of available observations for each variable-of-interest (see below).

MRI preprocessing

T1w anatomical images (see Supplementary file 1C for MRI acquisition parameters) were processed with FreeSurfer (v6.0.0) (Fischl and Dale, 2000) and vertex-wise areal and thickness morphometry estimates were obtained for each MRI observation. As the LCBC sample also contained longitudinal observations, initial cross-sectional reconstructions in LCBC were subsequently ran through FreeSurfer’s longitudinal pipeline. As HCP data was acquired at a higher voxel resolution (0.7 mm isotropic), the T1w scans were processed with the --hires flag to recon-all (Zaretskaya et al., 2018). Areal and thickness maps of the LH and RH of each participant in each dataset were resampled from the native cortical geometry to a symmetrical surface template (‘LH_sym’) (Maingault et al., 2016; Marie et al., 2016) based on cross-hemispheric registration (Greve et al., 2013). This procedure achieves vertex-wise alignment of the data from each participant and homotopic hemisphere in a common analysis space, enabling a whole-cortical and data-driven analysis of cortical asymmetry. Areal values were resampled with an additional Jacobian correction to ensure preservation of the areal quantities (Winkler et al., 2012). We then applied an 8 mm FWHM Gaussian kernel to surface-smooth the LH and RH data.

Data analysis

All analyses were performed in FreeSurfer (v6.0) and R (v4.1.1).

Population-level asymmetry

We assessed areal and thickness asymmetry vertex-wise using FreeSurfer’s Linear Mixed Effects (LME) tool (Bernal-Rusiel et al., 2013). Asymmetry was delineated via the main effect of Hemisphere (controlling for Age, Age × Hemisphere, Sex, Scanner [where applicable], with a random subject term). For each sample and metric, we computed mean Asymmetry Index maps (AI; defined as (LH-RH) / ((LH + RH)/2)). Spatial overlap of AI maps across datasets was quantified by correlating the one-dimensional surface data between every dataset pair (Pearson’s r). Next, to delineate regions exhibiting robust areal and thickness asymmetry across datasets, we thresholded and binarized the AI maps by a given absolute effect size (areal = 5%; thickness = 1%; achieving p[FDR]<0.001 in most datasets with FreeSurfer’s two-stage FDR-procedure Bernal-Rusiel et al., 2013), and summed the binary maps. After removing the smallest clusters (<200 mm2), a set of robust clusters was defined as those exhibiting overlapping effects in six out of seven samples. We then extracted area and thickness data in symmetrical space for each cluster, subject, and hemisphere, spatially averaging across vertices.

Lifespan trajectories

Factor-smooth GAMMs (‘gamm4’, v0.2–6 Wood and Scheipl, 2017) were used to fit a smooth Age trajectory per Hemisphere, and assess the smooth Age × Hemisphere interaction in our clusters. GAMMs incorporate both cross-sectional and longitudinal data to capture nonlinearity of the mean level trajectories across persons, resulting in population estimates that are intermediate between cross-sectional and longitudinal trajectories (Sørensen et al., 2021). The linear predictor matrix of the GAMM was used to obtain asymmetry trajectories and their confidence intervals, computed as the difference between zero-centered (i.e. demeaned) hemispheric age-trajectories. We included Hemisphere as an additional fixed effect, sex and scanner as covariates-of-no-interest, and a random subject intercept. We did not consider sex differences in lifespan asymmetry change because our elected method was not well-suited to testing three-way interactions between nonlinear smooth terms. A low number of basis dimensions for each smoothing spline was chosen to guard against overfitting (knots = 6; see Figure 3—figure supplement 1). LCBC outliers falling >6 SD from the trajectory of either hemisphere were detected and removed on a region-wise basis (Supplementary file 1E-F). To calculate relative change, we refitted lifespan GAMMs adding an ICV covariate, then scaled the LH and RH fitted lifespan trajectories by the prediction at the minimum age (i.e. ~4 years). Age at peak thickness asymmetry was estimated where the CI’s of absolute hemispheric trajectories were maximally non-overlapping.

Interregional asymmetry correlations

We assessed covariance between asymmetries, separately for areal and thickness asymmetry. All individual AI’s in clusters with rightward mean asymmetry were first inversed, such that positive correlations would reflect asymmetry-asymmetry relationships regardless of the direction of mean asymmetry in the cluster (i.e. higher asymmetry in the population-direction). Then, we regressed out age, sex and scanner (where applicable) from each AI, using linear mixed models after collating the data from each sample (adding random intercepts for LCBC subjects), to ensure the correction was unaffected by differences in sample age-distribution. Separately for each dataset, we then obtained the cluster-cluster correlation matrix. At this point, two strong outliers in HCP data were detected and discarded for this and all subsequent analyses (Figure 5—figure supplement 4). Replication was assessed using the Mantel test (‘ade4’ R package v1.7–18 Dray and Dufour, 2007) between each dataset-pair (LCBC, UKB, HCP) across 10,000 permutations. We then post-hoc tested whether covariance between areal asymmetries was related to proximity in cortex, obtaining the average geodesic distance between all clusters along the ipsilateral surface (‘SurfDist’ Python package v0.15.5 Margulies et al., 2016), and correlating pair-wise distance with pair-wise correlation coefficient (Fisher’s transformed coefficients; Spearman’s correlation). To post-hoc assess whether observed covariance patterns for thickness asymmetry reflected a global effect, we ran a PCA across z-transformed AI’s for all thickness clusters (pre-corrected for the same covariates). Based on the results, we computed the mean AIs across all leftward clusters, and across all rightward clusters, and tested the partial correlation between mean leftward thickness asymmetry in left-asymmetric clusters and mean rightward thickness asymmetry in right-asymmetric clusters, in each of the three cohorts.

Heritability

We assessed heritability of areal and thickness asymmetry using both SNP- and twin-based methods, both for our set of robust clusters and cortex-wide across 500 parcels (Schaefer et al., 2017). For cluster analyses, significance was considered at Bonferroni-corrected p<0.05 applied separately across each metric. Cortex-wide significance was considered at p(FDR)<0.05 (500 tests per map). For SNP-based analyses in UKB data, the final genetic sample consisted of 31,433 UKB participants (application #32048) with imaging and quality checked genetic data. We removed subjects that were outliers based on heterozygosity [field 22027] and missingness (>0.05), mismatched genetic and reported sex [22001], sex chromosome aneuploidies [22019], and those not in the ‘white British ancestry’ subset [22006] (Bycroft et al., 2018). At variant level, after removing SNPs with minor allele frequency <0.01, data from 784,256 autosomal SNPs were used to compute a genetic relationship matrix using GCTA (v1.93.2) (Yang et al., 2010). For each phenotype, we first regressed out age and sex and computed z-scores. Genome-based restricted maximum likelihood (GREML) methods as implemented in GCTA were then used to compute SNP-heritability for each AI measure, applying a kinship coefficient cut-off of .025 (excluding one individual from each pair), and controlling for genetic population structure (first 10 principal components). Bivariate GREML analysis was used to test genetic correlations between asymmetry clusters (Yang et al., 2010). These estimate the proportion of variance two asymmetries share due to genetic influences through pleiotropic action of genes (van Rheenen et al., 2019). We tested genetic relationships only for cluster-pairs where both clusters exhibited significant SNP-heritability (p<0.05; pre-corrected; 78 tests for area, 55 for thickness). Significance was assessed at p(FDR)<0.05. Replication of heritability results was assessed using twin-based analyses in HCP data, applying AE models in ‘OpenMx’ (v2.19.8) (Neale et al., 2016). These use observed cross-twin and cross-sibling covariance to decompose the proportion of observed phenotypic variance into additive genetic effects [A], and unique environmental effects or error [E]. Data were reformatted such that rows represented family-wise observations. As is standard, we set A to be 1 for MZ twins assumed to share 100% of their segregating genes, and 0.5 for DZ twins and siblings that share 50% on average. For each phenotype we first regressed out age and sex and computed z-scores. Statistical significance was assessed by comparing model fit to a submodel with the A parameter set to 0. To test replication of genetic correlations in HCP data, bivariate twin models were employed in OpenMx. We tested twin-based genetic correlations across the same set of cluster-pairs tested in the SNP-based analyses (Supplementary file 1H–I; note that while this approach permitted a replication attempt, some clusters did not exhibit significant twin-based heritability). Model significance was assessed by comparing model fit to submodels with the genetic covariance parameter set to 0, and significance was assessed at p(FDR)<0.05.

Associations with cognition, sex, handedness, and ICV

Finally, we assessed relationships between asymmetry in our robust clusters and general cognitive ability, handedness, sex, and ICV. For general cognition, we used the first principal component across the following 11 core UK Biobank cognitive variables (Fawns-Ritchie and Deary, 2020): Mean reaction time (log transformed) [field 20023], Numeric memory [4282], Fluid reasoning [20016], Matrix completion [6373], Tower rearranging [21004], Symbol digit substitution [23324], Paired associate learning [20197], Prospective memory [20018] (recoded as 1 or 0, depending on whether the instruction was remembered on the first attempt or not), Pairs matching (log) [399], Trail making A (log) [6348], Trail making B (log) [6350]. Prior to the PCA, for participants with cognitive data, data was imputed for missing cognitive variables via the ‘imputePCA’ function (number of estimated components tentatively optimized using general cross validation; ‘missMDA’ R package v1.18 Josse and Husson, 2016). PC1 (explaining 39.2%; Supplementary file 1L) was inversed to correlate negatively with age (r=–0.39), ensuring higher values reflected higher cognition. As fewer participants had cognitive data relative to the other variables, for each cluster we ran one set of linear models to assess the marginal effect of cognition (PC1 as predictor; age, sex, ICV controlled; N=35,198), and one set of linear models to assess the marginal effects of Handedness, Sex, and ICV in a model including all three predictors (age controlled, N=37,569 with available handedness data). For the cognitive analysis, effects identified in the imputed dataset were checked against the confidence intervals for the effect in the subset of the data with no missing cognitive variables (N=4696). Participants who self-reported as mixed handed were not included (Sha et al., 2021a). Individual AI’s in rightward clusters were first inversed, such that higher values reflect higher asymmetry in the population-direction. Significance was considered at Bonferroni-corrected α=p < 7.4e–5 (.01/136 [34 clusters ×4]).

Data sharing/availability

All summary-level maps are available in Supplementary file 2. All code underlying the main analyses is available at https://github.com/jamesmroe/PopAsym (copy archived at Roe, 2023) and on the Open Science Framework (OSF; https://osf.io/dv9um/). Derived source data underlying figures is available on the OSF. All datasets used in this work are openly available, with the exception of LCBC where participants, which include many children, have not consented to share their data publicly online. Other datasets used in this work are available without restrictions and are not subject to application approval (DLBS; https://fcon_1000.projects.nitrc.org/indi/retro/dlbs.html; CC BY-NC; SALD; http://fcon_1000.projects.nitrc.org/indi/retro/sald.html; CC BY-NC; IXI; https://brain-development.org/ixi-dataset; CC BY-SA 3.0). Accordingly, we have made the individual-level data for these samples available and our code can be used to reproduce vertex-wise analyses in these samples. Individual-level data for the remaining samples (LCBC; Cam-CAN, HCP; UKB) may be available upon reasonable request, given appropriate ethical, data protection, and data-sharing agreements where applicable. Requests must be submitted and approved via the relevant channel (details are provided in Supplementary file 1M).

Data availability

All summary-level maps are available in Supplementary file 2. All code underlying the main analyses is available at https://github.com/jamesmroe/PopAsym (copy archieved at Roe, 2023) and on the Open Science Framework (OSF; https://osf.io/dv9um/). Derived source data underlying figures is also available on the OSF. All datasets used in this work are openly available, with the exception of LCBC where participants, which include many children, have not consented to share their data publicly online. Other datasets used in this work are available without restrictions and are not subject to application approval (DLBS; https://fcon_1000.projects.nitrc.org/indi/retro/dlbs.html; CC BY-NC; SALD; http://fcon_1000.projects.nitrc.org/indi/retro/sald.html; CC BY-NC; IXI; https://brain-development.org/ixi-dataset; CC BY-SA 3.0). Accordingly, we have made the individual-level data for these samples available and our code can be used to reproduce vertex-wise analyses in these samples. Individual-level data for the remaining samples (LCBC; Cam-CAN, HCP; UKB) may be available upon reasonable request, given appropriate ethical, data protection, and data-sharing agreements where applicable. Requests must be submitted and approved via the relevant channel (details are provided in Supplementary file 1).

The following data sets were generated
The following previously published data sets were used
    1. Tyler LK
    (2017) Cambridge Centre for Ageing and Neuroscience
    ID camcan-archive. Cambridge Centre for Ageing and Neuroscience (Cam-CAN) dataset inventory.

References

    1. Annett M
    (1998) Handedness and cerebral dominance: the right shift theory
    The Journal of Neuropsychiatry and Clinical Neurosciences 10:459–469.
    https://doi.org/10.1176/jnp.10.4.459
    1. Grasby KL
    2. Jahanshad N
    3. Painter JN
    4. Colodro-Conde L
    5. Bralten J
    6. Hibar DP
    7. Lind PA
    8. Pizzagalli F
    9. Ching CRK
    10. McMahon MAB
    11. Shatokhina N
    12. Zsembik LCP
    13. Thomopoulos SI
    14. Zhu AH
    15. Strike LT
    16. Agartz I
    17. Alhusaini S
    18. Almeida MAA
    19. Alnæs D
    20. Amlien IK
    21. Andersson M
    22. Ard T
    23. Armstrong NJ
    24. Ashley-Koch A
    25. Atkins JR
    26. Bernard M
    27. Brouwer RM
    28. Buimer EEL
    29. Bülow R
    30. Bürger C
    31. Cannon DM
    32. Chakravarty M
    33. Chen Q
    34. Cheung JW
    35. Couvy-Duchesne B
    36. Dale AM
    37. Dalvie S
    38. de Araujo TK
    39. de Zubicaray GI
    40. de Zwarte SMC
    41. den Braber A
    42. Doan NT
    43. Dohm K
    44. Ehrlich S
    45. Engelbrecht HR
    46. Erk S
    47. Fan CC
    48. Fedko IO
    49. Foley SF
    50. Ford JM
    51. Fukunaga M
    52. Garrett ME
    53. Ge T
    54. Giddaluru S
    55. Goldman AL
    56. Green MJ
    57. Groenewold NA
    58. Grotegerd D
    59. Gurholt TP
    60. Gutman BA
    61. Hansell NK
    62. Harris MA
    63. Harrison MB
    64. Haswell CC
    65. Hauser M
    66. Herms S
    67. Heslenfeld DJ
    68. Ho NF
    69. Hoehn D
    70. Hoffmann P
    71. Holleran L
    72. Hoogman M
    73. Hottenga JJ
    74. Ikeda M
    75. Janowitz D
    76. Jansen IE
    77. Jia T
    78. Jockwitz C
    79. Kanai R
    80. Karama S
    81. Kasperaviciute D
    82. Kaufmann T
    83. Kelly S
    84. Kikuchi M
    85. Klein M
    86. Knapp M
    87. Knodt AR
    88. Krämer B
    89. Lam M
    90. Lancaster TM
    91. Lee PH
    92. Lett TA
    93. Lewis LB
    94. Lopes-Cendes I
    95. Luciano M
    96. Macciardi F
    97. Marquand AF
    98. Mathias SR
    99. Melzer TR
    100. Milaneschi Y
    101. Mirza-Schreiber N
    102. Moreira JCV
    103. Mühleisen TW
    104. Müller-Myhsok B
    105. Najt P
    106. Nakahara S
    107. Nho K
    108. Olde Loohuis LM
    109. Orfanos DP
    110. Pearson JF
    111. Pitcher TL
    112. Pütz B
    113. Quidé Y
    114. Ragothaman A
    115. Rashid FM
    116. Reay WR
    117. Redlich R
    118. Reinbold CS
    119. Repple J
    120. Richard G
    121. Riedel BC
    122. Risacher SL
    123. Rocha CS
    124. Mota NR
    125. Salminen L
    126. Saremi A
    127. Saykin AJ
    128. Schlag F
    129. Schmaal L
    130. Schofield PR
    131. Secolin R
    132. Shapland CY
    133. Shen L
    134. Shin J
    135. Shumskaya E
    136. Sønderby IE
    137. Sprooten E
    138. Tansey KE
    139. Teumer A
    140. Thalamuthu A
    141. Tordesillas-Gutiérrez D
    142. Turner JA
    143. Uhlmann A
    144. Vallerga CL
    145. van der Meer D
    146. van Donkelaar MMJ
    147. van Eijk L
    148. van Erp TGM
    149. van Haren NEM
    150. van Rooij D
    151. van Tol MJ
    152. Veldink JH
    153. Verhoef E
    154. Walton E
    155. Wang M
    156. Wang Y
    157. Wardlaw JM
    158. Wen W
    159. Westlye LT
    160. Whelan CD
    161. Witt SH
    162. Wittfeld K
    163. Wolf C
    164. Wolfers T
    165. Wu JQ
    166. Yasuda CL
    167. Zaremba D
    168. Zhang Z
    169. Zwiers MP
    170. Artiges E
    171. Assareh AA
    172. Ayesa-Arriola R
    173. Belger A
    174. Brandt CL
    175. Brown GG
    176. Cichon S
    177. Curran JE
    178. Davies GE
    179. Degenhardt F
    180. Dennis MF
    181. Dietsche B
    182. Djurovic S
    183. Doherty CP
    184. Espiritu R
    185. Garijo D
    186. Gil Y
    187. Gowland PA
    188. Green RC
    189. Häusler AN
    190. Heindel W
    191. Ho BC
    192. Hoffmann WU
    193. Holsboer F
    194. Homuth G
    195. Hosten N
    196. Jack CR
    197. Jang M
    198. Jansen A
    199. Kimbrel NA
    200. Kolskår K
    201. Koops S
    202. Krug A
    203. Lim KO
    204. Luykx JJ
    205. Mathalon DH
    206. Mather KA
    207. Mattay VS
    208. Matthews S
    209. Mayoral Van Son J
    210. McEwen SC
    211. Melle I
    212. Morris DW
    213. Mueller BA
    214. Nauck M
    215. Nordvik JE
    216. Nöthen MM
    217. O’Leary DS
    218. Opel N
    219. Martinot MLP
    220. Pike GB
    221. Preda A
    222. Quinlan EB
    223. Rasser PE
    224. Ratnakar V
    225. Reppermund S
    226. Steen VM
    227. Tooney PA
    228. Torres FR
    229. Veltman DJ
    230. Voyvodic JT
    231. Whelan R
    232. White T
    233. Yamamori H
    234. Adams HHH
    235. Bis JC
    236. Debette S
    237. Decarli C
    238. Fornage M
    239. Gudnason V
    240. Hofer E
    241. Ikram MA
    242. Launer L
    243. Longstreth WT
    244. Lopez OL
    245. Mazoyer B
    246. Mosley TH
    247. Roshchupkin GV
    248. Satizabal CL
    249. Schmidt R
    250. Seshadri S
    251. Yang Q
    252. Alzheimer’s Disease Neuroimaging Initiative
    253. CHARGE Consortium
    254. EPIGEN Consortium
    255. IMAGEN Consortium
    256. SYS Consortium
    257. Parkinson’s Progression Markers Initiative
    258. Alvim MKM
    259. Ames D
    260. Anderson TJ
    261. Andreassen OA
    262. Arias-Vasquez A
    263. Bastin ME
    264. Baune BT
    265. Beckham JC
    266. Blangero J
    267. Boomsma DI
    268. Brodaty H
    269. Brunner HG
    270. Buckner RL
    271. Buitelaar JK
    272. Bustillo JR
    273. Cahn W
    274. Cairns MJ
    275. Calhoun V
    276. Carr VJ
    277. Caseras X
    278. Caspers S
    279. Cavalleri GL
    280. Cendes F
    281. Corvin A
    282. Crespo-Facorro B
    283. Dalrymple-Alford JC
    284. Dannlowski U
    285. de Geus EJC
    286. Deary IJ
    287. Delanty N
    288. Depondt C
    289. Desrivières S
    290. Donohoe G
    291. Espeseth T
    292. Fernández G
    293. Fisher SE
    294. Flor H
    295. Forstner AJ
    296. Francks C
    297. Franke B
    298. Glahn DC
    299. Gollub RL
    300. Grabe HJ
    301. Gruber O
    302. Håberg AK
    303. Hariri AR
    304. Hartman CA
    305. Hashimoto R
    306. Heinz A
    307. Henskens FA
    308. Hillegers MHJ
    309. Hoekstra PJ
    310. Holmes AJ
    311. Hong LE
    312. Hopkins WD
    313. Hulshoff Pol HE
    314. Jernigan TL
    315. Jönsson EG
    316. Kahn RS
    317. Kennedy MA
    318. Kircher TTJ
    319. Kochunov P
    320. Kwok JBJ
    321. Le Hellard S
    322. Loughland CM
    323. Martin NG
    324. Martinot JL
    325. McDonald C
    326. McMahon KL
    327. Meyer-Lindenberg A
    328. Michie PT
    329. Morey RA
    330. Mowry B
    331. Nyberg L
    332. Oosterlaan J
    333. Ophoff RA
    334. Pantelis C
    335. Paus T
    336. Pausova Z
    337. Penninx B
    338. Polderman TJC
    339. Posthuma D
    340. Rietschel M
    341. Roffman JL
    342. Rowland LM
    343. Sachdev PS
    344. Sämann PG
    345. Schall U
    346. Schumann G
    347. Scott RJ
    348. Sim K
    349. Sisodiya SM
    350. Smoller JW
    351. Sommer IE
    352. St Pourcain B
    353. Stein DJ
    354. Toga AW
    355. Trollor JN
    356. Van der Wee NJA
    357. van ’t Ent D
    358. Völzke H
    359. Walter H
    360. Weber B
    361. Weinberger DR
    362. Wright MJ
    363. Zhou J
    364. Stein JL
    365. Thompson PM
    366. Medland SE
    367. Enhancing NeuroImaging Genetics through Meta-Analysis Consortium (ENIGMA)—Genetics working group
    (2020) The genetic architecture of the human cerebral cortex
    Science 367:eaay6690.
    https://doi.org/10.1126/science.aay6690
    1. Hibar DP
    2. Adams HHH
    3. Jahanshad N
    4. Chauhan G
    5. Stein JL
    6. Hofer E
    7. Renteria ME
    8. Bis JC
    9. Arias-Vasquez A
    10. Ikram MK
    11. Desrivières S
    12. Vernooij MW
    13. Abramovic L
    14. Alhusaini S
    15. Amin N
    16. Andersson M
    17. Arfanakis K
    18. Aribisala BS
    19. Armstrong NJ
    20. Athanasiu L
    21. Axelsson T
    22. Beecham AH
    23. Beiser A
    24. Bernard M
    25. Blanton SH
    26. Bohlken MM
    27. Boks MP
    28. Bralten J
    29. Brickman AM
    30. Carmichael O
    31. Chakravarty MM
    32. Chen Q
    33. Ching CRK
    34. Chouraki V
    35. Cuellar-Partida G
    36. Crivello F
    37. Den Braber A
    38. Doan NT
    39. Ehrlich S
    40. Giddaluru S
    41. Goldman AL
    42. Gottesman RF
    43. Grimm O
    44. Griswold ME
    45. Guadalupe T
    46. Gutman BA
    47. Hass J
    48. Haukvik UK
    49. Hoehn D
    50. Holmes AJ
    51. Hoogman M
    52. Janowitz D
    53. Jia T
    54. Jørgensen KN
    55. Karbalai N
    56. Kasperaviciute D
    57. Kim S
    58. Klein M
    59. Kraemer B
    60. Lee PH
    61. Liewald DCM
    62. Lopez LM
    63. Luciano M
    64. Macare C
    65. Marquand AF
    66. Matarin M
    67. Mather KA
    68. Mattheisen M
    69. McKay DR
    70. Milaneschi Y
    71. Muñoz Maniega S
    72. Nho K
    73. Nugent AC
    74. Nyquist P
    75. Loohuis LMO
    76. Oosterlaan J
    77. Papmeyer M
    78. Pirpamer L
    79. Pütz B
    80. Ramasamy A
    81. Richards JS
    82. Risacher SL
    83. Roiz-Santiañez R
    84. Rommelse N
    85. Ropele S
    86. Rose EJ
    87. Royle NA
    88. Rundek T
    89. Sämann PG
    90. Saremi A
    91. Satizabal CL
    92. Schmaal L
    93. Schork AJ
    94. Shen L
    95. Shin J
    96. Shumskaya E
    97. Smith AV
    98. Sprooten E
    99. Strike LT
    100. Teumer A
    101. Tordesillas-Gutierrez D
    102. Toro R
    103. Trabzuni D
    104. Trompet S
    105. Vaidya D
    106. Van der Grond J
    107. Van der Lee SJ
    108. Van der Meer D
    109. Van Donkelaar MMJ
    110. Van Eijk KR
    111. Van Erp TGM
    112. Van Rooij D
    113. Walton E
    114. Westlye LT
    115. Whelan CD
    116. Windham BG
    117. Winkler AM
    118. Wittfeld K
    119. Woldehawariat G
    120. Wolf C
    121. Wolfers T
    122. Yanek LR
    123. Yang J
    124. Zijdenbos A
    125. Zwiers MP
    126. Agartz I
    127. Almasy L
    128. Ames D
    129. Amouyel P
    130. Andreassen OA
    131. Arepalli S
    132. Assareh AA
    133. Barral S
    134. Bastin ME
    135. Becker DM
    136. Becker JT
    137. Bennett DA
    138. Blangero J
    139. van Bokhoven H
    140. Boomsma DI
    141. Brodaty H
    142. Brouwer RM
    143. Brunner HG
    144. Buckner RL
    145. Buitelaar JK
    146. Bulayeva KB
    147. Cahn W
    148. Calhoun VD
    149. Cannon DM
    150. Cavalleri GL
    151. Cheng C-Y
    152. Cichon S
    153. Cookson MR
    154. Corvin A
    155. Crespo-Facorro B
    156. Curran JE
    157. Czisch M
    158. Dale AM
    159. Davies GE
    160. De Craen AJM
    161. De Geus EJC
    162. De Jager PL
    163. De Zubicaray GI
    164. Deary IJ
    165. Debette S
    166. DeCarli C
    167. Delanty N
    168. Depondt C
    169. DeStefano A
    170. Dillman A
    171. Djurovic S
    172. Donohoe G
    173. Drevets WC
    174. Duggirala R
    175. Dyer TD
    176. Enzinger C
    177. Erk S
    178. Espeseth T
    179. Fedko IO
    180. Fernández G
    181. Ferrucci L
    182. Fisher SE
    183. Fleischman DA
    184. Ford I
    185. Fornage M
    186. Foroud TM
    187. Fox PT
    188. Francks C
    189. Fukunaga M
    190. Gibbs JR
    191. Glahn DC
    192. Gollub RL
    193. Göring HHH
    194. Green RC
    195. Gruber O
    196. Gudnason V
    197. Guelfi S
    198. Håberg AK
    199. Hansell NK
    200. Hardy J
    201. Hartman CA
    202. Hashimoto R
    203. Hegenscheid K
    204. Heinz A
    205. Le Hellard S
    206. Hernandez DG
    207. Heslenfeld DJ
    208. Ho B-C
    209. Hoekstra PJ
    210. Hoffmann W
    211. Hofman A
    212. Holsboer F
    213. Homuth G
    214. Hosten N
    215. Hottenga J-J
    216. Huentelman M
    217. Hulshoff Pol HE
    218. Ikeda M
    219. Jack CR Jr
    220. Jenkinson M
    221. Johnson R
    222. Jönsson EG
    223. Jukema JW
    224. Kahn RS
    225. Kanai R
    226. Kloszewska I
    227. Knopman DS
    228. Kochunov P
    229. Kwok JB
    230. Lawrie SM
    231. Lemaître H
    232. Liu X
    233. Longo DL
    234. Lopez OL
    235. Lovestone S
    236. Martinez O
    237. Martinot J-L
    238. Mattay VS
    239. McDonald C
    240. McIntosh AM
    241. McMahon FJ
    242. McMahon KL
    243. Mecocci P
    244. Melle I
    245. Meyer-Lindenberg A
    246. Mohnke S
    247. Montgomery GW
    248. Morris DW
    249. Mosley TH
    250. Mühleisen TW
    251. Müller-Myhsok B
    252. Nalls MA
    253. Nauck M
    254. Nichols TE
    255. Niessen WJ
    256. Nöthen MM
    257. Nyberg L
    258. Ohi K
    259. Olvera RL
    260. Ophoff RA
    261. Pandolfo M
    262. Paus T
    263. Pausova Z
    264. Penninx BWJH
    265. Pike GB
    266. Potkin SG
    267. Psaty BM
    268. Reppermund S
    269. Rietschel M
    270. Roffman JL
    271. Romanczuk-Seiferth N
    272. Rotter JI
    273. Ryten M
    274. Sacco RL
    275. Sachdev PS
    276. Saykin AJ
    277. Schmidt R
    278. Schmidt H
    279. Schofield PR
    280. Sigursson S
    281. Simmons A
    282. Singleton A
    283. Sisodiya SM
    284. Smith C
    285. Smoller JW
    286. Soininen H
    287. Steen VM
    288. Stott DJ
    289. Sussmann JE
    290. Thalamuthu A
    291. Toga AW
    292. Traynor BJ
    293. Troncoso J
    294. Tsolaki M
    295. Tzourio C
    296. Uitterlinden AG
    297. Hernández MCV
    298. Van der Brug M
    299. van der Lugt A
    300. van der Wee NJA
    301. Van Haren NEM
    302. van ’t Ent D
    303. Van Tol M-J
    304. Vardarajan BN
    305. Vellas B
    306. Veltman DJ
    307. Völzke H
    308. Walter H
    309. Wardlaw JM
    310. Wassink TH
    311. Weale ME
    312. Weinberger DR
    313. Weiner MW
    314. Wen W
    315. Westman E
    316. White T
    317. Wong TY
    318. Wright CB
    319. Zielke RH
    320. Zonderman AB
    321. Martin NG
    322. Van Duijn CM
    323. Wright MJ
    324. Longstreth WT
    325. Schumann G
    326. Grabe HJ
    327. Franke B
    328. Launer LJ
    329. Medland SE
    330. Seshadri S
    331. Thompson PM
    332. Ikram MA
    (2017) Novel genetic Loci associated with hippocampal volume
    Nature Communications 8:13624.
    https://doi.org/10.1038/ncomms13624

Decision letter

  1. Jonas Obleser
    Reviewing Editor; University of Lübeck, Germany
  2. Christian Büchel
    Senior Editor; University Medical Center Hamburg-Eppendorf, Germany

Our editorial process produces two outputs: (i) public reviews designed to be posted alongside the preprint for the benefit of readers; (ii) feedback on the manuscript for the authors, including requests for revisions, shown below. We also include an acceptance summary that explains what the editors found interesting or important about the work.

Decision letter after peer review:

[Editors’ note: the authors submitted for reconsideration following the decision after peer review. What follows is the decision letter after the first round of review.]

Thank you for submitting the paper "Population-level asymmetry of the cerebral cortex: reproducibility, lifespan changes, heritability, and individual differences" for consideration by eLife. Your article has been reviewed by 3 peer reviewers, and the evaluation has been overseen by a Reviewing Editor and a Senior Editor. The following individual involved in the review of your submission has agreed to reveal their identity: Dorothy V M Bishop (Reviewer #3).

Comments to the Authors:

We are sorry to say that, after consultation with the reviewers, we have decided that this work in its current form will not be considered further for publication by eLife.

However, note that in the light of the very constructive comment and in tune with our general belief that these data can potentially provide an important reference for future studies, we as editors would be willing to consider a future, thoroughly revised revision of this paper. We would treat it as a new submission (with potentially new reviewers being invited, depending on the availability of reviewers.) – All of this of course depending on the nature and extent of your revisions to the manuscript and your documentation of these. *

Specifically, the reviewers and editors agreed that the paper provides a useful descriptive catalogue of phenomena related to hemispheric anatomical asymmetries and that it does so in a largely sound manner (but see specific methodological concerns below). However, there was no consensus to be reached that this study has a specific impact on our learning about the mechanisms of hemispheric lateralisation. After reviews and an elaborate discussion amongst reviewers and editors, concerns remained over the mechanistic impact of the results presented and the degree of conceptual synthesis provided, so we decided not to invite a revision of this study for eLife.

Reviewer #1 (Recommendations for the authors):

I have several comments and suggestions that I hope will be of use to the authors.

1. In discussing the differences between regional asymmetries in cortical thickness and cortical area, it is prudent to mention that CT is less reliable than SA and this reliability difference may be amplified in the computation of asymmetry.

2. In claiming that "disrupted cortical asymmetry is a confirmed feature of… aging…" the authors cite one study and ignore others that not only lend little support to this conclusion but also warn that regional asymmetry in the aging brain is not well-replicated even across similar samples.

3. Unclear how GAMM captures nonlinearity of change over time with only two measurement occasions.

4. Discovering a capture of 21% of common variance by the first component of the PCA hardly signals the existence of a single factor. CFA would be a more appropriate way of testing single vs multiple factors hypotheses.

5. The authors consistently characterize correlations as "strong" – unclear why. The characterization is somewhat hyperbolic: for example, r=.56 does not seem particularly strong, with about 70% of the variance not shared by the variables. Why not just write "correlated," report the magnitude and the direction and let the readers make a qualitative judgment if they feel like it.

Some stylistic editing will improve this manuscript. For example, the sentence "Areal asymmetry correlates in specific regions…" in the Abstract is quite unwieldy and can be split into two or three.

Reviewer #2 (Recommendations for the authors):

I found some of the graphs hard to read.

For example, in Figure 5 the color scale was hard to read. It goes from one shade of blue to another shade of blue and I could not really see much distinction.

Figure 6, it is not clear what is being plotted on the abscissa. There is no axis label.

The authors use the term "longitudinal" to describe their age-related analyses, but my understanding of that term is that it refers to the same individuals tested over time, which is not the case here; it's a cross-sectional approach.

The word "course" should be "coarse" (line 70).

Reviewer #3 (Recommendations for the authors):

Overall, while I think the attempt at a comprehensive approach by the authors is commendable, the paper, which includes a great deal of supplementary material, is just too dense and unwieldy. My recommendation would be to break it up. I'm not a fan of authors who turn papers into 'minimal publishable units', but here we have the opposite – a paper so dense with information that it is hard to follow.

I don't know if eLife would consider a series of related papers like this but the work would be easier to follow, and potentially more influential if it were split up. There is a nice logical progression, with the initial analyses delineating brain areas that are consistently asymmetric, and subsequent analyses then using these, so a series of papers would work well I think – something like this:

Paper 1: qs 1 and 3 – replicability of population-level asymmetries, and inter-regional correlations.

Paper 2: consideration of causal factors, incorporating longitudinal analysis of lifespan trajectories of asymmetry (q2) and heritability (q4). The longitudinal analysis could potentially incorporate analysis of interregional correlations across age – given that connectivity changes with age, there is a rationale for looking at that (LCBC sample), but even without that, this would make a meaty paper.

Paper 3 (q5) Consideration of significance of individual differences in terms of correlations with sex, handedness, and general cognitive ability (UKB sample). Potentially more could be done with this; it could look at associations with absolute as well as raw asymmetry indices. But I think it would be most valuable if preceded by a balanced review of prior work in this area, culminating in a list of claims of associations that could be tested in this sample. That would make it more evident which claims are NOT supported, which would be very useful for the field.

Note: I am not recommending jettisoning the null results – on the contrary, I would like to see them given more serious treatment, as they could be helpful in moving the field forward.

Since breaking up the paper this way would be a fairly radical move, I'll assume that the authors and/or editors will decide against this, and so my remaining comments will be written as if we are dealing with a paper that does incorporate all the current material.

I would suggest rewriting the Abstract to give a more balanced account of the findings, including an indication of the null as well as positive results.

I don't know whether eLife insists on the particular structuring of the paper used here, with Methods at the end. I would hope this could be modified, as for a multi-question paper like this, it is particularly frustrating for the reader. The introduction ends with five specific questions clearly laid out, but we then jump into Results, which are hard to understand without seeing the Methods. So the reader goes forward to Methods, where the dataset for each of the five questions is first described, then the data analysis methods for each of the five questions. This involves a great deal of jumping about. I found myself wanting to literally cut and paste the material so I could read it in a sensible sequence.

Interregional correlations

The material on interregional correlations was fascinating but hard to follow. The numbers (Figure 4 Figure supplement 1) are hard to read, but I was very surprised at just how low these correlations were for SA, given that people have tended to interpret asymmetries in terms of a general 'torque' effect.

The CT asymmetries were hard to interpret, given the relatively low Mantel scores – and the way the Biobank data stand out even just on the heatmap. This made me wonder if the more restricted age range of the Biobank sample might affect the covariances. I wasn't sure how the age residualisation worked, and whether it was within-sample or across samples, so hard to know if and how the fact that Biobank has a much higher age minimum than others might have an effect?

One thing that makes the results confusing is the flipping of the hemisphere so that the emphasis is on strength of laterality in the population direction. It would be worth taking some time in the introduction to discuss predictions from the 'multifactorial' account versus a 'single factor' account – relating this to discussions of cerebral torque.

An alternative approach would be to just present the unflipped data, explaining that a positive correlation then indicates agreement in left-wardness (regardless of whether the population bias is left or right), and a negative correlation indicates agreement between the strength of lateralisation in the population direction.

The subsequent discussion of these results was pretty clear – it's just the account in Results that was hard to follow.

Heritability

The account of results in the Abstract again focused on the positive findings and ignored the null results as if they were of no interest.

So the Abstract states: "Areal asymmetry is moderately heritable (max h2SNP ~19%), and phenotypic correlations are reflected by high genetic correlations, whereas heritability of thickness asymmetry is low. "

My reading of the results is that none of the twin-based heritability estimates – either global or regional – was statistically significant. Consistent with previous reports by Eyler et al. and others, the strongest influence on asymmetry indices is the E-term (nonshared environment), which may be interpreted in this context as chance (stochastic) influences.

The SNP-based estimates from UK Biobank give a handful of significant heritability estimates for the SA asymmetry, but none for CT. I agree it is of interest to document these, but the genetic effects that are found are not high (19% is the highest estimate, with the remaining 80% of variance accounted for by the e2 term).

I guess it is a case of one's perspective, but my take on this is not "isn't it interesting that if we have a huge sample we can pick up some small genetic effects on a handful of asymmetry measures", but rather, "isn't it interesting that brains end up with similar patterns of structural asymmetry, even though genes explain so little individual variation ". This does suggest that there are genetic influences that create these general biases to one direction or the other, but which have reached fixation in the population and so show little individual variation.

Associations with cognition

As noted above, I felt this treatment was too brief. There is a rich literature on associations with asymmetry for the variables considered here, much of it focusing on the degree rather than the direction of asymmetry. So I would like to see a separate paper dealing with associations with AIs from this analysis, relating the findings more systematically to prior studies.

The discussion did a phenomenal job in pulling everything together but would benefit from some subheadings.

[Editors’ note: further revisions were suggested prior to acceptance, as described below.]

Thank you for submitting your article "Tracing the development and lifespan change of population-level structural asymmetry in the cerebral cortex" for consideration by eLife. Your article has been reviewed by 1 (new) peer reviewer who had also access to your rebuttal letter. The evaluation has been overseen by Jonas Obleser as the Reviewing editor and Christian Büchel as the Senior Editor. The reviewer has opted to remain anonymous. The Reviewing Editor has drafted this to help you prepare a revised submission.

Essential revisions (for the authors):

1) Thorough editions for clarity. This new reviewer has overall praise for your work, but they also expose the remaining lack of clarity and a relative lack of depth in coverage of the genetic results. Please use the detailed comments below as a guide when revising your study prior to final acceptance.

2) Reproducibility: In addition to revising along the reviewer's comments, …

– Please also fix issues such as "All maps are available at neurovault.org/XXXX."

­

­– Please add version numbers where applicable (eg to R packages).

Reviewer #1 (Recommendations for the authors):

Overall I believe the authors have done a solid job in assessing brain asymmetry and addressing the comments of the reviewers.

Nevertheless, here and there some elements may still be (considered to be) tidied up a bit further, so that the aim of the study is even more clear. Currently, there are still a lot of sub-analyses. For example, 2.3 is interesting but possibly a slight deviation from the general goal.

Moreover, with respect to the analysis part 2.5, did the authors consider running these analyses also in HCP?

Also – with respect to the heritability analyses; have genetic correlations also been performed in the HCP sample?

https://doi.org/10.7554/eLife.84685.sa1

Author response

[Editors’ note: the authors resubmitted a revised version of the paper for consideration. What follows is the authors’ response to the first round of review.]

Reviewer #1 (Recommendations for the authors):

I have several comments and suggestions that I hope will be of use to the authors.

1. In discussing the differences between regional asymmetries in cortical thickness and cortical area, it is prudent to mention that CT is less reliable than SA and this reliability difference may be amplified in the computation of asymmetry.

We agree and have added text to reflect this in the Discussion (see below). Furthermore, we now also acknowledge that this potential reliability difference may be a contributor to some of the differences in results for areal and thickness asymmetry (e.g. heritability), stating this in the limitations:

615 – “Conceivable sources of inconsistently reported results for cortical asymmetries likely include varying age distributions (particularly for thickness) (Roe et al. 2021), and multiple asymmetries within atlas-based parcels (e.g. we observed notably discrepant results to ENIGMA for insula thickness asymmetry (Kong et al. 2018)) … Thickness asymmetries seem also more variable compared with area (Kong et al. 2018), though it should be noted thickness in general may be slightly less reliable (Hedges et al. 2022), the asymmetry effect is smaller, and thus possibly contains more error”

646 – “Again, we also cannot rule out that potential reliability differences between thickness and areal asymmetry may partially explain some magnitude differences in heritability estimates.”

2. In claiming that "disrupted cortical asymmetry is a confirmed feature of… aging…" the authors cite one study and ignore others that not only lend little support to this conclusion but also warn that regional asymmetry in the aging brain is not well-replicated even across similar samples.

We apologize for the lack of clarity. In the context of aging, this statement is specific to cortical thickness asymmetry. While we agree previous cross-sectional studies provided mixed support for this (e.g. (Zhou et al. 2013; Plessen et al. 2014)), our previous study which we cite here showed that both leftward and rightward thickness asymmetry exhibit widespread reduction in aging that is replicable in multiple longitudinal samples (Roe et al. 2021). This previous study was the first to address the question longitudinally, which is likely the key reason why earlier cross-sectional studies lent little support to this conclusion. We agree the few studies assessing cross-sectional age effects on areal asymmetries do not lend support for the claim in aging, and indeed, a key motivation for the current study is that we know little about lifespan changes in areal asymmetry. As this is an introductory sentence, it is perhaps too early to specify that in the context of neurodevelopmental disorders, the evidence suggests alterations in both area and thickness, whereas in the context of aging, current evidence supports changes in thickness only. Therefore, we opted for the following change that highlights only some aspects of cortical asymmetry have been shown to be subtly altered in aging, without going into the specifics at this point – which we do later in paragraphs dealing with these issues in depth. We agree reported patterns of average regional asymmetry in the aging brain in general have been inconsistent, especially for thickness asymmetry (e.g. Zhou et al. 2013; Plessen et al. 2014; Roe et al. 2021). We believe this point only highlights the need for our study and approach.

55 – ”Although an extensive literature in search of structural asymmetry deviations in various conditions and disorders is in several cases being challenged by newer data (Kong et al. 2020), at least some aspects of cortical asymmetry are confirmed to be subtly reduced in neurodevelopmental disorders such as autism (Postema et al. 2019), but also through later life influences such as aging (Roe et al. 2021), and Alzheimer’s disease (Thompson et al. 2007; Roe et al. 2021)”

74 – “Determining the developmental and lifespan trajectories of cortical asymmetry may shed light on how cortical asymmetries are shaped through childhood or set from early life, and provide evidence of the timing of expected brain change in normal development. Although important in and of itself, this would also provide a useful normative reference, as subtly altered cortical asymmetry – in terms of both area and thickness – has been linked at the group-level to neurodevelopmental disorders along the autism spectrum (Postema et al. 2019; Sha et al. 2022), suggesting altered lateralized neurodevelopment may be a neurobiologically relevant outcome in at least some cases of developmental perturbation (Postema et al. 2019; Sha et al. 2022). For areal asymmetry, surprisingly few studies have charted developmental (Li et al. 2014; Remer et al. 2017) or aging-related effects (Kong et al. 2018; Williams et al. 2022), although indirect evidence in neonates suggests adult-like patterns of areal asymmetry are evident at birth and may exhibit little change from birth to 2 years (Li et al. 2014) despite rapid and concurrent developmental cortical expansion (Li et al. 2013). For thickness asymmetry, longitudinal increases in asymmetry have been shown during the first two years of life (Li et al. 2015), with suggestions of rapid asymmetry growth from birth to 1 year (Li et al. 2015), and potentially continued growth until adolescence (Nie et al. 2013). However, previous lifespan studies mapped thickness asymmetry linearly across cross-sectional developmental and adult age-ranges (Zhou et al. 2013; Plessen et al. 2014), mostly concluding thickness asymmetry is minimal in infancy and maximal age ~60. In contrast, recent work established thickness asymmetry shows a non-linear decline from 20 to 90 years that is reproducible across longitudinal aging cohorts (Roe et al. 2021). Thus, although offering viable developmental insights (Zhou et al. 2013; Plessen et al. 2014), previous lifespan studies of thickness asymmetry do not accurately capture the aging process, and likely conflate non-linear developmental and aging trajectories with linear models. A longitudinal exploration of the lifespan trajectories of thickness asymmetry accounting for dynamic change is needed to further knowledge of normal human brain development.”

3. Unclear how GAMM captures nonlinearity of change over time with only two measurement occasions.

We agree this is useful to clarify. GAMMs capture the non-linearity of the mean-level trajectories across persons, incorporating data from both cross-sectional and longitudinal measurements. They do not model non-linearity within persons. Hence, the non-linearity in GAMM is achieved in the same way as in a Generalized Additive Model that does not incorporate repeat measures (essentially a weighted sum of basis functions and their coefficients). Since both single and repeat measures over time are taken into account, the estimated population trajectories will be an intermediate estimate of cross-sectional and longitudinal trajectories. We have added this clarification in the Methods:

743 – “GAMMs incorporate both cross-sectional and longitudinal data to capture non-linearity of the mean level trajectories across persons (Sørensen et al. 2021).”

4. Discovering a capture of 21% of common variance by the first component of the PCA hardly signals the existence of a single factor. CFA would be a more appropriate way of testing single vs multiple factors hypotheses.

We have changed all statements pertaining to a “single global factor” to a “single principle component” or similar, as we agree this better represents our findings.

5. The authors consistently characterize correlations as "strong" – unclear why. The characterization is somewhat hyperbolic: for example, r=.56 does not seem particularly strong, with about 70% of the variance not shared by the variables. Why not just write "correlated," report the magnitude and the direction and let the readers make a qualitative judgment if they feel like it.

We have removed such examples where this refers to a correlation. We now only use the word “strong” when describing asymmetries.

Some stylistic editing will improve this manuscript. For example, the sentence "Areal asymmetry correlates in specific regions…" in the Abstract is quite unwieldy and can be split into two or three.

In line with this suggestion and others, we have made extensive amendments to the abstract. However, since we must be mindful of word count and taking the care to highlight our main findings, we cannot split this sentence into multiple sentences. The abstract now reads:

Abstract:

31 – “Cortical asymmetry is a ubiquitous feature of brain organization that is subtly altered in some neurodevelopmental disorders, yet we lack knowledge of how its development proceeds across life in health. Achieving consensus on the precise cortical asymmetries in humans is necessary to uncover the genetic and later influences that shape them, such as age. Here, we delineate population-level asymmetry in cortical thickness and surface area vertex-wise in 7 datasets and chart asymmetry trajectories longitudinally across life (4-89 years; observations = 3937; 70% longitudinal). We find replicable asymmetry interrelationships, heritability maps, and test asymmetry associations in large-scale data. Cortical asymmetry was robust across datasets. Whereas areal asymmetry is predominantly stable across life, thickness asymmetry grows in childhood and peaks in early adulthood. Areal asymmetry correlates phenotypically and genetically in specific regions, and is low-moderately heritable (max h2SNP ~19%). In contrast, thickness asymmetry is globally interrelated across the cortex in a pattern suggesting highly left-lateralized individuals tend towards left-lateralization also in population-level right-asymmetric regions (and vice versa), and exhibits low or absent heritability. We find less areal asymmetry in the most consistently lateralized region in humans associates with subtly lower cognitive ability, and confirm small handedness and sex effects. Results suggest areal asymmetry is developmentally stable and arises in early life through genetic but mainly subject-specific stochastic effects, whereas childhood developmental growth shapes thickness asymmetry and may lead to directional variability of global thickness lateralization in the population.”

Reviewer #2 (Recommendations for the authors):

I found some of the graphs hard to read.

For example, in Figure 5 the color scale was hard to read. It goes from one shade of blue to another shade of blue and I could not really see much distinction.

We have changed the scale in Figure 5 to a warm colour scale, and changed the gradient to better distinguish between numerical estimates. Since we have also added a new figure 4, please see revised Figure 6.

Figure 6, it is not clear what is being plotted on the abscissa. There is no axis label.

We have added a label that makes clear the x-axis plots the individual association tests per cluster (“Cluster associations”). That is, for areal asymmetry which has 14 clusters, 14 associations are conducted for general cognitive ability, handedness etc. Please see revised Figure 7.

The authors use the term "longitudinal" to describe their age-related analyses, but my understanding of that term is that it refers to the same individuals tested over time, which is not the case here; it's a cross-sectional approach.

As stated above, we apologize this very important point was not clear, but the LCBC sample used for the lifespan trajectory analysis did indeed incorporate dense longitudinal data (70% of scans were longitudinal in nature, with 1-6 timepoints per person). In fact, all of the analyses ran with the LCBC sample incorporated all longitudinal data within the given age-range. Therefore, our use of this term, as well as other longitudinal terms – such as “aging” and “change” – is justified and correct. We have extensively amended the revised manuscript to highlight the longitudinal lifespan aspect to these data, which we propose was key to uncovering the developmental trajectories we show here. Some example amendments/statements:

35 – “… and chart asymmetry trajectories longitudinally across life (4-89 years; observations = 3937; 70% longitudinal)”.

61 – “no previous study has charted cortical asymmetry trajectories from childhood to old age using longitudinal data”.

63 – “Compounding the lack of longitudinal investigation, previous large-scale studies do not …”.

90 – “A longitudinal exploration of the lifespan trajectories of thickness asymmetry accounting for dynamic change is needed to further knowledge of normal human brain development.”

121 – “To gain insight into their development, we then trace a series of lifespan and genetic analyses. Specifically, we chart the developmental and lifespan trajectories of cortical asymmetry for the first time longitudinally across the lifespan.”

179 – “we aimed to characterize the developmental and lifespan trajectories of cortical asymmetry from early childhood to old age, using a lifespan sample incorporating dense longitudinal data (Methods). For this, we used the mixed effects LCBC lifespan sample covering the full age-range (4-89 years)”.

416 – “Combining the strengths of a vertex-wise delineation of population-level cortical asymmetry in 7 international datasets and dense longitudinal data, we offer the first description of the longitudinal developmental and lifespan trajectories of cortical asymmetry, advancing knowledge on normal human brain development.”

The word "course" should be "coarse" (line 70).

Fixed.

Reviewer #3 (Recommendations for the authors):

Overall, while I think the attempt at a comprehensive approach by the authors is commendable, the paper, which includes a great deal of supplementary material, is just too dense and unwieldy. My recommendation would be to break it up. I'm not a fan of authors who turn papers into 'minimal publishable units', but here we have the opposite – a paper so dense with information that it is hard to follow.

1. I don't know if eLife would consider a series of related papers like this but the work would be easier to follow, and potentially more influential if it were split up. There is a nice logical progression, with the initial analyses delineating brain areas that are consistently asymmetric, and subsequent analyses then using these, so a series of papers would work well I think – something like this:

Paper 1: qs 1 and 3 – replicability of population-level asymmetries, and inter-regional correlations.

Paper 2: consideration of causal factors, incorporating longitudinal analysis of lifespan trajectories of asymmetry (q2) and heritability (q4). The longitudinal analysis could potentially incorporate analysis of interregional correlations across age – given that connectivity changes with age, there is a rationale for looking at that (LCBC sample), but even without that, this would make a meaty paper.

Paper 3 (q5) Consideration of significance of individual differences in terms of correlations with sex, handedness, and general cognitive ability (UKB sample). Potentially more could be done with this; it could look at associations with absolute as well as raw asymmetry indices. But I think it would be most valuable if preceded by a balanced review of prior work in this area, culminating in a list of claims of associations that could be tested in this sample. That would make it more evident which claims are NOT supported, which would be very useful for the field.

Note: I am not recommending jettisoning the null results – on the contrary, I would like to see them given more serious treatment, as they could be helpful in moving the field forward.

2. Since breaking up the paper this way would be a fairly radical move, I'll assume that the authors and/or editors will decide against this, and so my remaining comments will be written as if we are dealing with a paper that does incorporate all the current material.

Thank you for your thoughtful comments. We do not feel the work will be as impactful if split up over several papers in the current publishing landscape, and so respectfully maintain our decision not to. However, we have taken on board the Reviewer’s many comments for how to improve the paper as it is.

(Regarding the specific point about association analysis being a paper on its own, please see below):

We agree the literature review regarding asymmetry-cognition associations lacked some necessary detail. However, although a rich literature has attempted to associate diverse aspects of brain asymmetry with cognitive ability, almost none of these studies focuses specifically on cortical asymmetry. Indeed, we are only aware of three other studies testing cognitive links with cortical asymmetry or derivations of it (beyond our previous one which included cognitive change analysis in aging; (Roe et al. 2021)). Two of these focused on fluctuating asymmetry (Yeo et al. 2016; Moodie et al. 2020), operationalizing this as the average across absolute asymmetry deviance scores from the mean per-ROI, and are therefore not directly comparable to the present study. The remaining one found that greater thickness asymmetry in the expected direction may relate to better cognitive outcomes (Plessen et al. 2014). We now explicitly mention this study in the Introduction and Discussion, as well as summarize all previous work attempting to relate cortical asymmetry specifically to variations in cognitive ability:

106 – “Finally, altered development of cerebral lateralization in general has been widely hypothesized to relate to average poorer cognitive outcomes (Crow et al. 1998; Hirnstein et al. 2010; Plessen et al. 2014). Specifically in the context of cortical asymmetry, however, although one previous study reported larger thickness asymmetry may relate to better verbal and visuospatial cognition (Plessen et al. 2014), phenotypic asymmetry-cognition associations have been rarely reported (Plessen et al. 2014; Yeo et al. 2016; Moodie et al. 2020), conflicting (Yeo et al. 2016; Moodie et al. 2020), not directly comparable (Plessen et al. 2014; Moodie et al. 2020), and to date remain untested in large-scale data. Still, recent work points to small but significant overlap between genes underlying multivariate brain asymmetries and those influencing educational attainment and specific developmental disorders impacting cognition (Sha, Schijven, et al. 2021), indicating either pleiotropy between non-related traits or capturing shared genetic susceptibility to altered brain lateralization and cognitive outcomes.”

574 – “Also of note, no other asymmetry showed evidence of cognitive association (Figure 7—figure supplement 1), suggesting previously reported associations in small samples will likely not replicate (Plessen et al. 2014)”

Since we are careful not to generalize our findings to asymmetry measures not studied here, we do not agree we should separate this into another paper. Methodologically, we maintain our decision to reduce the 11 UK Biobank cognitive variables to their shared variation (Lyall et al. 2016), because the brief nature of UKB cognitive tests does not permit an overly thorough investigation into how different aspects of cognition relate to cortical asymmetry. That is, although it seemed like a brief treatment, it is a defensible and possibly close to optimal approach to take with this data. Note that we now consistently refer to our PC1 measure in terms of “general cognitive ability”. Regarding the distinction with absolute asymmetry, the statements we opt for are always specific to cortical asymmetry, and we are not aware of a large literature looking at absolute strength of cortical asymmetries when testing associations with cognition. However, we have added text to acknowledge this remains to be tested:

657 – “Relatedly, because UKB cognitive data is not exhaustive (Fawns-Ritchie and Deary 2020) (e.g. fluid IQ ranges from 1-13), we extracted the common variance across core tests to index general cognitive ability. This approach does not permit testing associations with well-operationalized or specific cognitive domains (for which UKB cognitive data may not be sufficient), and it remains to be seen whether cortical asymmetry may be informative in the context of specific forms of lateralized cognition (Ocklenburg et al. 2014), or using absolute non-directional measures. However, the subtlety of the only effect on cognition we find here may suggest we might expect similarly weak effects, though not necessarily if brain and phenotypic measurement accuracy is jointly optimized (Aki Nikolaidis et al. 2022).”

3. I would suggest rewriting the Abstract to give a more balanced account of the findings, including an indication of the null as well as positive results.

As per a previous comment by the Reviewer, we now highlight that all effects we find are small in the Abstract and Discussion, including heritability findings.

Though we acknowledge we must be careful not to interpret the null, we agree that in data of this size, we can tentatively suggest that the null relationships we find suggest any effect of a given cortical asymmetry on e.g. general cognitive ability is arguably negligible if we are not able to detect it here. We also regret including the last sentence the reviewer mentions, as we completely agree it is very unlikely measures of specific cognitive abilities will be highly informative in terms of association with cortical asymmetry. We have added this argumentation and amended the closing statement accordingly:

658 – “This approach does not permit testing associations with well-operationalized or specific cognitive domains (for which UKB cognitive data may not be sufficient), and it remains to be seen whether cortical asymmetry may be informative in the context of specific forms of lateralized cognition (Ocklenburg et al. 2014), or using absolute nondirectional measures. However, the subtlety of the only effect on cognition we find here may suggest we might expect similarly weak effects, though not necessarily if brain and phenotypic measurement accuracy is jointly optimized (Aki Nikolaidis et al. 2022)”

Though we agree with the reviewer’s points, the Abstract is arguably not the right place to present the argument for null associations, also given word limits, which also influence bias towards positive results. However, we have made changes to the Abstract to highlight the found effects were small, including heritability effects (note that in line with other Reviewer suggestions, we have reduced emphasis on the individual differences analysis and thus no longer highlight these findings to the same degree in the Abstract). In accordance with the Reviewer’s later comments, we also highlight the interpretation of stochastic influences, further highlighting that (heritability) effects were predominantly small.

38 – “Areal asymmetry correlates phenotypically and genetically in specific regions, and is low-moderately heritable (max h2SNP ~19%). In contrast, thickness asymmetry is globally interrelated …, and exhibits low or absent heritability. We find less areal asymmetry in the most consistently lateralized region in humans associates with subtly lower cognitive ability, and confirm small handedness and sex effects. Results suggest areal asymmetry is developmentally stable and arises in early life through genetic but mainly subject-specific stochastic effects, whereas childhood developmental growth shapes thickness asymmetry and may lead to directional variability of global thickness lateralization in the population.”

We also added Discussion around null/small effects in the context of the small cognitive association we find:

572 – “However, as seems typical for brain associations in big data (Marek et al. 2022) and may be expected for any single structural measure explaining a complex phenomenon, the association was notably small. Also of note, no other asymmetry showed evidence of cognitive association (Figure 7—figure supplement 1), suggesting previously reported associations in small samples will likely not replicate (Plessen et al. 2014).”

And amended the Results and Discussion to highlight that the found associations are indeed small:

381 – “Notably, all effect sizes were small … Although small, we note this association was far from only just surviving correction at our predefined α level (⍺ = .01; Methods).”

422 – “… and uncover novel and confirm previously-reported associations with factors purportedly important in the context of asymmetry – all with small effects.”

576 – That the association we find was specific to the most lateralized areal asymmetry in humans may suggest disruptions in early life cerebral lateralization leads to cognitive deficits – detectable in later life as small effects in big data.

590 – “the small effects highlight cortical asymmetry cannot predict individual hand preference.”

591 – “Similarly, asymmetry-relationships with other factors typically assumed important were all small, despite our optimization of asymmetry phenotypes”

4. I don't know whether eLife insists on the particular structuring of the paper used here, with Methods at the end. I would hope this could be modified, as for a multi-question paper like this, it is particularly frustrating for the reader. The introduction ends with five specific questions clearly laid out, but we then jump into Results, which are hard to understand without seeing the Methods. So the reader goes forward to Methods, where the dataset for each of the five questions is first described, then the data analysis methods for each of the five questions. This involves a great deal of jumping about. I found myself wanting to literally cut and paste the material so I could read it in a sensible sequence.

We enquired, but eLife mostly prefer to stick to a methods last format. They did say they sometimes allow a change in structure “where it makes sense to do so”, but the general impression was that it would not be deemed justifiable. Therefore, to avoid futile work, we chose to refine the new submission to hopefully be easier to follow, in the format they predominantly accept. While we sympathize with the Reviewer that it is not always preferable, we hope for understanding in this matter, as the decision does not lie with us.

Interregional correlations

5. The material on interregional correlations was fascinating but hard to follow. The numbers (Figure 4 Figure supplement 1) are hard to read, but I was very surprised at just how low these correlations were for SA, given that people have tended to interpret asymmetries in terms of a general 'torque' effect.

We reduced the number of technical analyses in this section (i.e. geodesic distance) and streamlined the results to be more readable. We have also increased the resolution of the figure (now Figure 5—figure supplement 1). We agree many of the correlations are indeed surprisingly low for areal asymmetry, and highlight multifactorial explanations in the Discussion.

510 – “For areal asymmetry, we uncovered a covariance structure that almost perfectly replicated across datasets. In general, this fit with a multifaceted view (Rentería 2012; Francks 2015; Bain et al. 2019), in which most asymmetries were either not or only weakly correlated – but reliably so – contrasting views emphasizing a single (Annett 1964, 1998) or overall anatomical factor (Crow 2010) controlling cerebral lateralization. However, we also identified several regions wherein areal asymmetry reliably correlated within individuals, showing the variance in cortical asymmetries is not always dissociable, as often thought (Rentería 2012; Francks 2015; Bain et al. 2019). The strongest relationships all pertained to asymmetries that were proximal in cortex but opposite in direction. Several of these were underpinned by high asymmetry-asymmetry genetic correlations, illustrating cerebral lateralizations in surface area that are formed under common genetic influence, and in agreement with likely prenatal origins for areal asymmetry (Rakic 1995; Li et al. 2014).”

6. The CT asymmetries were hard to interpret, given the relatively low Mantel scores – and the way the Biobank data stand out even just on the heatmap. This made me wonder if the more restricted age range of the Biobank sample might affect the covariances. I wasn't sure how the age residualisation worked, and whether it was within-sample or across samples, so hard to know if and how the fact that Biobank has a much higher age minimum than others might have an effect?

The age residualisation was previously performed separately within each sample, so it was plausible differences in age-ranges could have affected the results. Since we agree a better approach is to perform the age/sex/site/residualisation across collated samples, we followed the suggestion. The overall pattern of results did not change, with the UKB data still standing out. However, this approach improved results of the Mantel tests, increasing the whole-matrix correlation from r = .43 to.49 for the LCBC-UKB comparison, and from r = .45 to.62 for the LCBC-HCP comparison. This revised approach supports a stronger replication, with the effects clearly similar across samples, albeit still weaker in LCBC and HCP.

We have also added the following in the Discussion to mention the differences between datasets for thickness asymmetries:

526 – “Though it is unclear why the relationships were weaker in the other datasets tested, we nevertheless found similarly significant relationships in each (Figure 5—figure supplement 3).”

7. One thing that makes the results confusing is the flipping of the hemisphere so that the emphasis is on strength of laterality in the population direction.

Please see our response in point 9 below.

8. It would be worth taking some time in the introduction to discuss predictions from the 'multifactorial' account versus a 'single factor' account – relating this to discussions of cerebral torque.

We have added the following in the introduction to briefly mention these contrasting views:

93 – “Correlations between cortical asymmetries may provide a window on asymmetries formed under common genetic or developmental influences. Contemporary research suggests brain asymmetries are complex, multifactorial and largely independent (i.e. uncorrelated) traits (Rentería 2012; Francks 2015; Neubauer et al. 2020), contrasting earlier theories emphasizing a single (Annett 1964, 1998) or predominating factor (Geschwind and Galaburda 1985; McManus and Bryden 1991) controlling various cerebral lateralizations”

In addition, we also briefly mention the overall cerebral torque factor in the Discussion:

510 – “For areal asymmetry, we uncovered a covariance structure that almost perfectly replicated across datasets. In general, this fit with a multifaceted view (Rentería 2012; Francks 2015; Bain et al. 2019), in which most asymmetries were either not or only weakly correlated – but reliably so – contrasting views emphasizing a single biological (Annett 1964, 1998) or overall anatomical torque factor (Crow 2010) controlling cerebral lateralization.”

9. An alternative approach would be to just present the unflipped data, explaining that a positive correlation then indicates agreement in left-wardness (regardless of whether the population bias is left or right), and a negative correlation indicates agreement between the strength of lateralisation in the population direction.

10. The subsequent discussion of these results was pretty clear – it's just the account in Results that was hard to follow.

Although we of course agree this is a viable alternative approach, we argue it is a less obvious interpretation for the average reader that a positive correlation between cluster AI’s with a negative mean (i.e. rightward) and one with a positive mean (i.e. leftward) indicates agreement in leftward asymmetry. At least, this will likely be less obvious to those outside of the laterality community, such as developmental scholars, and readers who skim the explanation in the figure caption that the “opposite hemisphere” relationships are evidence of an arguably different phenomenon may easily miss this point (see below heat maps). Instead, we chose to highlight the qualitatively different nature of the relationships we see between opposite-direction asymmetries. We argue that the most unambiguous and clear way to present these relationships is that a positive correlation should indicate an asymmetry-asymmetry relationship, regardless of population-direction in the cluster (i.e. higher asymmetry in the population-direction, as the Reviewer points out). As is made clear in the below heat maps, the flipping of course only affects the opposite-direction asymmetry relationships. The negative correlations immediately stand out as a different phenomenon, and indicate that more leftward asymmetry relates to less rightward asymmetry, without having to consider the AI formula. We chose to consistently do this also to aid interpretation of genetic correlations (again to asymmetry-asymmetry genetic relationships), and for the analysis of cognitive ability, handedness, sex etc. For the latter, it greatly facilitates the communication of results to the average reader to not have to consider that e.g. the interpretation of the relationship between handedness and asymmetry changes depending on whether the brain cluster visualized in Figure 7 and Author response image 1 is yellow or blue. For these reasons, we respectfully opt to maintain this decision. Where relevant, we have added the following text or similar:

“AI’s in rightward clusters are inversed, such that positive correlations denote positive asymmetry-asymmetry (genetic) relationships, regardless of direction of mean asymmetry in the cluster (i.e. higher asymmetry in the population-direction)”

Author response image 1
Example of results in UKB using proposed alternative approach.

Heritability

11. The account of results in the Abstract again focused on the positive findings and ignored the null results as if they were of no interest.

So the Abstract states: "Areal asymmetry is moderately heritable (max h2SNP ~19%), and phenotypic correlations are reflected by high genetic correlations, whereas heritability of thickness asymmetry is low. "

My reading of the results is that none of the twin-based heritability estimates – either global or regional – was statistically significant. Consistent with previous reports by Eyler et al. and others, the strongest influence on asymmetry indices is the E-term (nonshared environment), which may be interpreted in this context as chance (stochastic) influences.

The SNP-based estimates from UK Biobank give a handful of significant heritability estimates for the SA asymmetry, but none for CT. I agree it is of interest to document these, but the genetic effects that are found are not high (19% is the highest estimate, with the remaining 80% of variance accounted for by the e2 term).

Following the Reviewers’ suggestion of switching to an AE model, we now have several FDR-corrected results also for the twin models. This makes the replication much clearer, at least for areal asymmetry. We agree that not enough attention was previously paid to the fact the E term soaks up most variance, and the potential theoretical interpretations of this. Please see our full response in the below comment (point 12). As mentioned above, the abstract now highlights that areal asymmetry was also low-moderately heritable, and invokes an interpretation of stochastic influences around these still relatively low estimates. However, we maintain our decision to highlight the 19% estimate, because to our knowledge this represents the most heritable brain or behavioural asymmetry yet reported with genomic methods, and this estimate is as high as many other structural brain imaging phenotypes that are not asymmetry-focused (Elliott et al. 2018). The fact that this same cluster associates with handedness will thus be of great interest to asymmetry researchers.

38 – “… Areal asymmetry correlates phenotypically and genetically in specific regions, and is low-moderately heritable (max h2SNP ~19%). In contrast, thickness asymmetry is globally interrelated …, and exhibits low or absent heritability. … Results suggest areal asymmetry is developmentally stable and arises in early life through genetic but mainly subject-specific stochastic effects, whereas childhood developmental growth shapes thickness asymmetry …”

12 I guess it is a case of one's perspective, but my take on this is not "isn't it interesting that if we have a huge sample we can pick up some small genetic effects on a handful of asymmetry measures", but rather, "isn't it interesting that brains end up with similar patterns of structural asymmetry, even though genes explain so little individual variation ". This does suggest that there are genetic influences that create these general biases to one direction or the other, but which have reached fixation in the population and so show little individual variation.

Thank you. We fully agree this is a likely interpretation, and now also emphasize this in the Discussion (in addition to the Abstract):

435 – “This evident consensus suggests genetic-developmental programs regulate mean brain lateralization with respect to both area and thickness in humans. However, the genetic findings presented herein suggest these may have reached population fixation, as heritability of even our optimally delineated asymmetry measures was generally low. This indicates either subject-specific stochastic mechanisms in early neurodevelopment or later developmental influences primarily determine cortical asymmetry. Tracing their lifespan development, we show the lifespan trajectories of areal asymmetry primarily suggest this form of cerebral asymmetry is developmentally stable at least from age ~4, maintained throughout life, and formed early on – possibly in utero (Li et al. 2014; Sha, Schijven, et al. 2021). One interpretation of lifespan stability combined with low heritability may be stochastic early life developmental influences determine interindividual differences in areal asymmetry more than later developmental change. However, future work linking prenatal and childhood trajectories is needed to affirm this. Still, we also found relatively stronger heritability and reproducible heritability maps for areal asymmetry (notably, anterior insula exhibited ~19% SNP-based heritability). This also illustrates region-dependent interindividual genetic effects upon areal asymmetry, and the fact that phenotypic correlations were underpinned by high genetic correlations suggests specific areal asymmetries are formed under common genetic influence. In stark contrast, our findings of childhood development of thickness asymmetry until a peak around age ~24, higher directional variability in adult samples, and lower heritability, converge to suggest thickness asymmetry may be more shaped through subject-specific effects in later childhood development, possibly as the brain grows in interaction with the environment. This interpretation applied to asymmetry agrees with work suggesting cortical area in general may trace more to early life factors (Walhovd et al. 2016; Grasby et al. 2020) whereas thickness may be more related to and impacted by lifespan influences (Rakic 1995; Walhovd et al. 2016; Grasby et al. 2020)”

549 – “However, as noted above, although heritability of either hemisphere was high, areal asymmetry heritability was still only moderate at best, suggesting both genetics but primarily subject-specific stochastic effects likely underly its formation.”

Associations with cognition

13. As noted above, I felt this treatment was too brief. There is a rich literature on associations with asymmetry for the variables considered here, much of it focusing on the degree rather than the direction of asymmetry. So I would like to see a separate paper dealing with associations with AIs from this analysis, relating the findings more systematically to prior studies.

We agree the literature review regarding asymmetry-cognition associations lacked some necessary detail. However, although a rich literature has attempted to associate diverse aspects of brain asymmetry with cognitive ability, almost none of these studies focuses specifically on cortical asymmetry. Indeed, we are only aware of three other studies testing cognitive links with cortical asymmetry or derivations of it (beyond our previous one which included cognitive change analysis in aging; (Roe et al. 2021)). Two of these focused on fluctuating asymmetry (Yeo et al. 2016; Moodie et al. 2020), operationalizing this as the average across absolute asymmetry deviance scores from the mean per-ROI, and are therefore not directly comparable to the present study. The remaining one found that greater thickness asymmetry in the expected direction may relate to better cognitive outcomes (Plessen et al. 2014). We now explicitly mention this study in the Introduction and Discussion, as well as summarize all previous work attempting to relate cortical asymmetry specifically to variations in cognitive ability:

106 – “Finally, altered development of cerebral lateralization in general has been widely hypothesized to relate to average poorer cognitive outcomes (Crow et al. 1998; Hirnstein et al. 2010; Plessen et al. 2014). Specifically in the context of cortical asymmetry, however, although one previous study reported larger thickness asymmetry may relate to better verbal and visuospatial cognition (Plessen et al. 2014), phenotypic asymmetry-cognition associations have been rarely reported (Plessen et al. 2014; Yeo et al. 2016; Moodie et al. 2020), conflicting (Yeo et al. 2016; Moodie et al. 2020), not directly comparable (Plessen et al. 2014; Moodie et al. 2020), and to date remain untested in large-scale data. Still, recent work points to small but significant overlap between genes underlying multivariate brain asymmetries and those influencing educational attainment and specific developmental disorders impacting cognition (Sha, Schijven, et al. 2021), indicating either pleiotropy between non-related traits or capturing shared genetic susceptibility to altered brain lateralization and cognitive outcomes.”

574 – “Also of note, no other asymmetry showed evidence of cognitive association (Figure 7—figure supplement 1), suggesting previously reported associations in small samples will likely not replicate (Plessen et al. 2014)”

Since we are careful not to generalize our findings to asymmetry measures not studied here, we do not agree we should separate this into another paper. Methodologically, we maintain our decision to reduce the 11 UK Biobank cognitive variables to their shared variation (Lyall et al. 2016), because the brief nature of UKB cognitive tests does not permit an overly thorough investigation into how different aspects of cognition relate to cortical asymmetry. That is, although it seemed like a brief treatment, it is a defensible and possibly close to optimal approach to take with this data. Note that we now consistently refer to our PC1 measure in terms of “general cognitive ability”. Regarding the distinction with absolute asymmetry, the statements we opt for are always specific to cortical asymmetry, and we are not aware of a large literature looking at absolute strength of cortical asymmetries when testing associations with cognition. However, we have added text to acknowledge this remains to be tested:

657 – “Relatedly, because UKB cognitive data is not exhaustive (Fawns-Ritchie and Deary 2020) (e.g. fluid IQ ranges from 1-13), we extracted the common variance across core tests to index general cognitive ability. This approach does not permit testing associations with well-operationalized or specific cognitive domains (for which UKB cognitive data may not be sufficient), and it remains to be seen whether cortical asymmetry may be informative in the context of specific forms of lateralized cognition (Ocklenburg et al. 2014), or using absolute non-directional measures. However, the subtlety of the only effect on cognition we find here may suggest we might expect similarly weak effects, though not necessarily if brain and phenotypic measurement accuracy is jointly optimized (Aki Nikolaidis et al. 2022).”

14. The discussion did a phenomenal job in pulling everything together but would benefit from some subheadings.

Thank you. We have added subheadings.

[Editors’ note: further revisions were suggested prior to acceptance, as described below.]

Essential revisions (for the authors):

1) Thorough editions for clarity. This new reviewer has overall praise for your work, but they also expose the remaining lack of clarity and a relative lack of depth in coverage of the genetic results. Please use the detailed comments below as a guide when revising your study prior to final acceptance.

We thank the editor for their valuable input. We have made many editions for clarity as pointed out by the Reviewer. We have also substantially streamlined and shortened the Discussion.

2) Reproducibility: In addition to revising along the reviewer's comments, …

– Please also fix issues such as "All maps are available at neurovault.org/XXXX."

­

All maps have been made available in Supplementary file 2 and the text links have been updated.

­– Please add version numbers where applicable (eg to R packages).

This issue has been fixed.

https://doi.org/10.7554/eLife.84685.sa2

Article and author information

Author details

  1. James M Roe

    Center for Lifespan Changes in Brain and Cognition (LCBC), Department of Psychology, University of Oslo, Oslo, Norway
    Contribution
    Conceptualization, Data curation, Software, Formal analysis, Validation, Investigation, Visualization, Methodology, Writing - original draft, Project administration, Writing – review and editing
    For correspondence
    j.m.roe@psykologi.uio.no
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-8008-902X
  2. Didac Vidal-Pineiro

    Center for Lifespan Changes in Brain and Cognition (LCBC), Department of Psychology, University of Oslo, Oslo, Norway
    Contribution
    Conceptualization, Supervision, Writing – review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0001-9997-9156
  3. Inge K Amlien

    Center for Lifespan Changes in Brain and Cognition (LCBC), Department of Psychology, University of Oslo, Oslo, Norway
    Contribution
    Data curation
    Competing interests
    No competing interests declared
  4. Mengyu Pan

    Center for Lifespan Changes in Brain and Cognition (LCBC), Department of Psychology, University of Oslo, Oslo, Norway
    Contribution
    Data curation
    Competing interests
    No competing interests declared
  5. Markus H Sneve

    Center for Lifespan Changes in Brain and Cognition (LCBC), Department of Psychology, University of Oslo, Oslo, Norway
    Contribution
    Formal analysis, Writing – review and editing
    Competing interests
    No competing interests declared
  6. Michel Thiebaut de Schotten

    1. Groupe d'Imagerie Neurofonctionnelle, Institut des Maladies Neurodégénératives-UMR 5293, CNRS, CEA, University of Bordeaux, Bordeaux, France
    2. Brian Connectivity and Behaviour Laboratory, Sorbonne University, Paris, France
    Contribution
    Conceptualization, Writing – review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-0329-1814
  7. Patrick Friedrich

    Institute of Neuroscience and Medicine, Research Centre Jülich, Jülich, Germany
    Contribution
    Conceptualization, Writing – review and editing
    Competing interests
    No competing interests declared
  8. Zhiqiang Sha

    Language and Genetics Department, Max Planck Institute for Psycholinguistics, Nijmegen, Netherlands
    Contribution
    Conceptualization, Writing – review and editing
    Competing interests
    No competing interests declared
  9. Clyde Francks

    1. Language and Genetics Department, Max Planck Institute for Psycholinguistics, Nijmegen, Netherlands
    2. Donders Institute for Brain, Cognition and Behaviour, Radboud University, Nijmegen, Netherlands
    3. Department of Human Genetics, Radboud University Medical Center, Nijmegen, Netherlands
    Contribution
    Conceptualization, Writing – review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-9098-890X
  10. Espen M Eilertsen

    PROMENTA Research Center, Department of Psychology, University of Oslo, Oslo, Norway
    Contribution
    Formal analysis, Writing – review and editing
    Competing interests
    No competing interests declared
  11. Yunpeng Wang

    Center for Lifespan Changes in Brain and Cognition (LCBC), Department of Psychology, University of Oslo, Oslo, Norway
    Contribution
    Conceptualization, Formal analysis, Writing – review and editing
    Competing interests
    No competing interests declared
  12. Kristine B Walhovd

    1. Center for Lifespan Changes in Brain and Cognition (LCBC), Department of Psychology, University of Oslo, Oslo, Norway
    2. Department of Radiology and Nuclear Medicine, Oslo University Hospital, Oslo, Norway
    Contribution
    Funding acquisition, Writing – review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-1918-1123
  13. Anders M Fjell

    1. Center for Lifespan Changes in Brain and Cognition (LCBC), Department of Psychology, University of Oslo, Oslo, Norway
    2. Department of Radiology and Nuclear Medicine, Oslo University Hospital, Oslo, Norway
    Contribution
    Funding acquisition, Writing – review and editing
    Competing interests
    No competing interests declared
  14. René Westerhausen

    Section for Cognitive and Clinical Neuroscience, Department of Psychology, University of Oslo, Oslo, Norway
    Contribution
    Conceptualization, Supervision, Writing – review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0001-7107-2712

Funding

European Research Council (283634)

  • Anders M Fjell

European Research Council (725025)

  • Anders M Fjell

European Research Council (313440)

  • Kristine B Walhovd

Norwegian Research Council (249931)

  • Anders M Fjell
  • Kristine B Walhovd

Norwegian Research Council (302854)

  • Yunpeng Wang

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Acknowledgements

Scripts were run on the Colossus processing cluster at the University of Oslo, and on resources provided by UNINETT Sigma2 (nn9769k). LCBC funding: European Research Council under grants 283634, 725025 (to A.M.F.), and 313440 (to K.B.W.); Norwegian Research Council (to A.M.F. and K.B.W.) under grants 249931 (TOPPFORSK) and 302854 (FRIPRO; to Y.W.), The National Association for Public Health’s dementia research program, Norway (to A.M.F). Data used in the preparation of this work were obtained from the MGH-USC Human Connectome Project (https://ida.loni.usc.edu/login.jsp). Data used in this work was also provided by the Cambridge Centre for Ageing and Neuroscience (Cam-CAN). This research has been conducted using the UK Biobank Resource (February 2020 data release). The authors also express thanks to Liyuan Yang for notifying us about the outstanding issue in HCP data.

Ethics

All studies were conducted in accordance with the Declaration of Helsinki. Ethical approval was obtained from the relevant authorities, and all participants provided informed consent. Studies conducted at the Center for Lifespan Changes in Brain and Cognition (LCBC) were approved by the Regional Ethical Committee of South-East Norway (2017/653) and complied with all relevant ethical regulations. Ethical approval for the other datasets was granted by the relevant authorities (Supplementary file 1M).

Senior Editor

  1. Christian Büchel, University Medical Center Hamburg-Eppendorf, Germany

Reviewing Editor

  1. Jonas Obleser, University of Lübeck, Germany

Version history

  1. Preprint posted: November 25, 2021 (view preprint)
  2. Received: November 3, 2022
  3. Accepted: June 16, 2023
  4. Accepted Manuscript published: June 19, 2023 (version 1)
  5. Version of Record published: July 25, 2023 (version 2)

Copyright

© 2023, Roe et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

Metrics

  • 698
    Page views
  • 70
    Downloads
  • 0
    Citations

Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Open citations (links to open the citations from this article in various online reference manager services)

Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)

  1. James M Roe
  2. Didac Vidal-Pineiro
  3. Inge K Amlien
  4. Mengyu Pan
  5. Markus H Sneve
  6. Michel Thiebaut de Schotten
  7. Patrick Friedrich
  8. Zhiqiang Sha
  9. Clyde Francks
  10. Espen M Eilertsen
  11. Yunpeng Wang
  12. Kristine B Walhovd
  13. Anders M Fjell
  14. René Westerhausen
(2023)
Tracing the development and lifespan change of population-level structural asymmetry in the cerebral cortex
eLife 12:e84685.
https://doi.org/10.7554/eLife.84685

Further reading

    1. Neuroscience
    Jing Wang, Hamid Azimi ... Gregor Rainer
    Research Article

    The lateral geniculate nucleus (LGN), a retinotopic relay center where visual inputs from the retina are processed and relayed to the visual cortex, has been proposed as a potential target for artificial vision. At present, it is unknown whether optogenetic LGN stimulation is sufficient to elicit behaviorally relevant percepts, and the properties of LGN neural responses relevant for artificial vision have not been thoroughly characterized. Here, we demonstrate that tree shrews pretrained on a visual detection task can detect optogenetic LGN activation using an AAV2-CamKIIα-ChR2 construct and readily generalize from visual to optogenetic detection. Simultaneous recordings of LGN spiking activity and primary visual cortex (V1) local field potentials (LFP) during optogenetic LGN stimulation show that LGN neurons reliably follow optogenetic stimulation at frequencies up to 60 Hz, and uncovered a striking phase locking between the V1 local field potential (LFP) and the evoked spiking activity in LGN. These phase relationships were maintained over a broad range of LGN stimulation frequencies, up to 80 Hz, with spike field coherence values favoring higher frequencies, indicating the ability to relay temporally precise information to V1 using light activation of the LGN. Finally, V1 LFP responses showed sensitivity values to LGN optogenetic activation that were similar to the animal's behavioral performance. Taken together, our findings confirm the LGN as a potential target for visual prosthetics in a highly visual mammal closely related to primates.

    1. Neuroscience
    Anna C Geuzebroek, Hannah Craddock ... Simon P Kelly
    Research Article Updated

    Decisions about noisy stimuli are widely understood to be made by accumulating evidence up to a decision bound that can be adjusted according to task demands. However, relatively little is known about how such mechanisms operate in continuous monitoring contexts requiring intermittent target detection. Here, we examined neural decision processes underlying detection of 1 s coherence targets within continuous random dot motion, and how they are adjusted across contexts with weak, strong, or randomly mixed weak/strong targets. Our prediction was that decision bounds would be set lower when weak targets are more prevalent. Behavioural hit and false alarm rate patterns were consistent with this, and were well captured by a bound-adjustable leaky accumulator model. However, beta-band EEG signatures of motor preparation contradicted this, instead indicating lower bounds in the strong-target context. We thus tested two alternative models in which decision-bound dynamics were constrained directly by beta measurements, respectively, featuring leaky accumulation with adjustable leak, and non-leaky accumulation of evidence referenced to an adjustable sensory-level criterion. We found that the latter model best explained both behaviour and neural dynamics, highlighting novel means of decision policy regulation and the value of neurally informed modelling.