Part 1: Observations that motivate research on sex differences in regional brain anatomy

The idea that humans may show sex differences in brain organization is prompted by a wealth of prior theoretical and empirical information from evolutionary, neuroscientific, and medical research. Evolution provides potential distal causes for why a phenotypic sex difference may exist (e.g., sex differences in body size as a result of competition over mates), whereas genetic and environmental factors provide potential mechanistic causes that address how a sex difference emerges during development (e.g., sex differences in body size being achieved through the sex-biased action of gonadal steroids) [18, 19]. These distinct causal frameworks are complementary (yet sometimes interwoven: [20]), such that attaining an understanding of a mechanistic cause does not preclude the existence of evolutionary explanation [21]. Beyond consideration of causal factors, the high a priori likelihood of human showing sex differences in brain organization is also supported by the existence of very many well-documented sex differences in prevalence, presentation, and prognosis of diverse medical disorders involving the brain. We expand upon each of these motivations for studying sex differences in human brain organization below, before considering why we might expect some of these differences to manifest as sex differences in regional brain anatomy that are resolvable by in vivo sMRI.

Potential distal causes of sex differences in human brain organization

150 years ago, Charles Darwin proposed the concept of sexual selection to explain the numerous sex differences (both physiological and behavioral) that he observed throughout the animal kingdom [22]. He outlined two mechanisms, including: (i) mate choice—individuals of sex A who possess a certain version of a trait are more attractive to (and likely to be chosen as a mate by) members of sex B; and (ii) mate competition—individuals of sex A who possess a certain version of a trait can outcompete other members of sex A for mating opportunities with members of sex B. In either case, the individuals exhibiting these trait varieties produce more offspring, leading them to become more prevalent over generations in sex A but not in sex B (to whom these varieties are not beneficial). These mechanisms have not only produced sex differences in visible, physical traits across the animal kingdom—such as the elaborate tails of male peacocks and the large antlers of male deer—but they have also impacted brain evolution. For example, mate choice is responsible for the elegant songs (and larger brain song nuclei) among the males of many songbird species [23], while competition over locating mates has led to greater spatial navigation skills (and larger hippocampi) among the males of certain vole species [24].

Like all other animals, the human species emerged as a result of evolutionary processes that shaped our bodies and behaviors. Although we are distinct from nonhuman animals in many ways (e.g., theory of mind, language, complex tool use) [25], we are not immune to our own evolutionary history, one that certainly included some aspects of sexual selection. For example, when we look across primate species, mate competition tends to be stronger in species with larger sex differences in body size [26]. Accordingly, an evolutionary history involving competition over mates (among males) is among the most viable evolutionary explanations for sex differences in human height and weight [26]. While hypothetical evolutionary causes are hard to empirically assess, sexual selection has provided a powerful explanatory framework for additional sex-biased traits in humans, including facial hair [27], fat distribution [28], aggression [29], and spatial ability [30]. For example: i) mate competition among males may contribute to male-biased rates of aggression across human cultures [29]; ii) sex differences in foraging throughout early human evolution (and after transitions to agriculture in some populations [31]) may partially explain sex-biased spatial abilities (e.g., labor division: females gather from spatially stable but seasonally variable food sites; males hunt across long distances spanning various routes; modern performance patterns: females outperform on object location memory and navigation by landmark tasks; males outperform on mental rotation tasks, which are associated with throwing accuracy and navigation by orientation) [32, 33]; and (iii) effective mate choice may require behavioral inhibition and contribute to female outperformance on inhibitory tasks [34]. If sexual selection did, in fact, shape adult human behavior over evolutionary time, then this will have necessarily been achieved through sex differences in neurodevelopment and adult brain organization. As discussed below, the leading proximal sources of sex-biased brain development are sex differences in chromosomal complement and gonadal type.

Potential mechanistic causes of sex differences in human brain organization

The hypothesis that humans may show sex differences in brain organization is not only raised by evolutionary theory, but also by genetic and environmental differences between males and females. Individual differences in sex chromosome complement and the concentration of gonadal sex steroids are foundational to the biological definition of sex in humans and nonhuman mammals, and there is substantial experimental evidence from animal studies that these factors directly impact mammalian brain organization [35]. Accordingly, it is reasonable to expect that human brain development is also subject to direct gonadal and sex chromosome effects. As detailed further below, several lines of observational data support the existence of such effects in humans.

In placental (eutherian) mammals, presence of the SRY gene on the Y chromosome of XY males typically leads to differentiation of the embryonic gonadal ridge into testes, whereas the absence of this gene in XX females typically allows differentiation into ovaries [36, 37]. This gonadal divergence tends to lead males and females to exhibit distinct concentration profiles of various sex steroids (e.g., estrogens, androgens) throughout development, although these hormones are not limited to one sex and their concentration distributions overlap [38]. Sex hormones were first shown to shape sex-biased mammalian (rodent) brain organization in the late 1950’s [39], and these findings have been bolstered and extended by the application of a growing suite of experimental methods in murine research [35, 40, 41]. Some of these gonadal influences actually exert experimentally verifiable influences on murine brain anatomy as measured by sMRI [42, 43]. Evidence for similar links between sex steroids and neuroanatomy in humans is largely derived from observational neuroimaging studies that harness non-experimental variations in human sex steroid signaling as a function of: (i) inter-individual variation in circulating sex steroids during development, reflecting variation between individuals in the same developmental period and also across developmental periods (e.g., pre/post puberty, pre/post menopause) [44, 45]; (ii) the menstrual cycle [46]; iii) medical disorders impacting the hypothalamo–pituitary–gonadal axis [47]; or iv) gender-affirming hormone treatment [48].

Although gonadal steroids have been historically viewed as the primary mechanistic drivers of sex differences in mammalian brain organization, there is growing evidence that sex chromosome complement can also have direct effects on mammalian brain organization [49, 50]. For instance, the products of sex chromosome genes are present at different levels in male and female cells [50]: genes on the Y chromosome are expressed only in males and genes that escape X chromosome inactivation (XCI) are often expressed at higher levels females [51, 52]. In addition, XCI in females may partially monopolize cellular epigenetic machinery, altering the expression of other genes [53]. Thus, irrespective of gonadal differences, XY and XX individuals have constitutional differences in the dosage of genes that are known to exert genome-wide regulatory effects that are not mediated by the gonads [54, 55]. This is bolstered by reports that sex-biased expression of autosomal genes in the mouse brain occurs prior to gonadal differentiation [56]. Moreover, there is extensive evidence in both rodents and humans that sex chromosome dosage can influence brain organization [50]. The four core genotype (FCG) model in mice provides experimental evidence that XX and XY groups show several reproducible neuroanatomical differences in sMRI [42, 43], as well differences in vasopressin fiber density within the lateral septum [57]. Evidence for sex chromosome dosage effects on the human brain comes from observational research in groups with differing X and/or Y chromosome doses due to sex chromosome aneuploidy. Specifically, increasing X- and/or Y- chromosome dosage induces highly reproducible changes in diverse aspects of regional brain anatomy, including parieto-occipital cortical surface area (increased), lateral temporal cortical thickness (decreased), and volume of the cerebellum and globus pallidus (both decreased) [58,59,60,61].

Alongside the aforementioned effects of gonadal and sex chromosome dosage on the brain, there are also extensively well-documented physiological sex differences in multiple peripheral organs and tissues that may affect the brain. These include adipose [62], hepatic [63], peripheral immune [64], and renal systems [65]. Such somatic sex differences are notable on two grounds: i) they are far less debated or controversial than the existence of sex differences in the brain; and ii) they represent another potential source of sex differences in the brain since they guide concentrations of various peptide and steroid hormones (e.g., leptin, adiponectin, inflammatory cytokines, neurotransmitter precursors, and estrogens) [66].

In addition to the many potential evolutionary, genetic, and endocrine-related causes of sex differences in brain organization, males and females are (at the group-level) often exposed to systematically different environments across the lifespan. For example, the experience of pregnancy has been linked to changes in brain organization [67, 68], and XY males do not experience pregnancy (although fatherhood also impacts the brain: [69]). There is also extensive evidence that individuals who are outwardly perceived as male or female experience pervasive differences in diverse domains of life [70], spanning the educational [71], professional [72, 73], fiscal [74], and medical [75] arenas, in addition to risk of exposure to different dangerous situations [76]. These considerable, consistent, and enduring gendered socio-environmental factors could conceivably influence brain organization in a manner that would manifest as group differences between males and females. It can, however, be extremely difficult (or impossible) to experimentally verify the action of such experiential sex differences on human brain organization, and it is likely there may well be synergy between experientially and biologically driven sex differences. In particular, gendered experiential influences on the human brain may be intimately coupled with sex-biased genetic and endocrine influences [77] since gendered experiences are correlated with an individual’s perceived gender. For example, observed sex differences in aggression and spatial ability may reflect, or be amplified by, gender-biases in the acceptance of and encouragement towards certain types of behaviors and toys throughout development [30, 78].

Taken together, the lines of research reviewed above identify strong evolutionary priors for the existence of sex differences in human brain organization, and also specify numerous endocrine and genetic factors that differ between typically developing human males and females and are known to influence mammalian neurodevelopment. When considering these potential sources of sex differences, however, it is also important to consider how they inform our understanding of the many phenotypic equivalences between males and females. At one level, the gonadal and chromosomal differences between typical males and females mean that any features of the mammalian brain that lack a sex difference are necessarily being attained from different starting points in each sex. For example, how do two instances of the same neuronal subtype, one in a male and one in a female, arrive at the same structural and functional profile despite having categorically different sex chromosome gene dosages? Similar questions can be asked about organ-level phenotypic equivalence: how do these two populations, with differences in brain size (which reflect differences in body size), arrive at equivalent intelligence levels [79]? One possible path to phenotypic equivalence between the sexes may actually be mechanistic counterbalancing of two opposing sex differences (i.e., compensatory mechanisms that serve to prevent sex differences in brain function and behavior) [80]. For example, experiments in the FCG murine model have identified neuroanatomical regions where sex chromosome and gonadal complement exert opposite effects on volume. These “ying-yang” effects appear to exist in brain areas that exhibit sex-biased volumes (e.g., volume of the medial amygdala is larger in mice with testes than mice with ovaries, but smaller in XY vs. XX mice), as well as regions without any apparent sex differences in mean volume (e.g., the pallidum, where volumetric differences between XX and XY mice only emerge after removal of the gonads, indicating that gonadal effects are “neutralizing” chromosomal effects) [43]. This latter scenario is a specific example of the more generalizable idea that a given phenotype can be similar between males and females but supported by different underlying mechanisms. Additional examples can be found at various biological levels, including interactions between the epigenome and transcriptome (e.g., XCI in females, which results in similar levels of expression for X chromosome genes (that do no escape XCI) in males and females) and between neuroendocrine signaling and behavior (e.g., male and female prairie voles exhibit similar levels of parental care, and while this behavior is produced by pregnancy-related hormonal changes in females, it is driven by vasopressin signaling in males) [80]. Thus, the absence of a sex difference at one level of biological organization cannot be assumed to imply the absence of a sex difference at other levels. Moreover, the preponderance of phenotypic similarities between males and females poses fascinating biological questions in light of the aforementioned sex differences in chromosomal dosage and gonadal status.

Epidemiological and clinical evidence for potential sex differences in human brain organization

Some of the strongest indirect evidence for biologically driven sex differences in human brain organization is provided by epidemiological and clinical studies that demonstrate large sex differences in the prevalence, presentation, and progression of brain-based disorders [81, 82]. These differences cannot in and of themselves be assumed to solely reflect biologically programmed sex differences in brain organization given the profound gendered influences on health [75], but several observations suggest that biologically grounded factors are likely to be an important contributor. First, multiple neurodevelopmental disorders with onset in early childhood are more prevalent in males than females [81, 83], including autism spectrum disorders (ASD) [84], attention-deficit hyperactivity disorder (ADHD) [85], and early-onset obsessive compulsive disorder (OCD) [86]. In contrast, there is a sharp rise in female-biased risk for depression [87], anxiety [88] and eating disorders [89] in adolescence, and of certain neurodegenerative (e.g., Alzheimer's disease and other dementias) [90] and psychiatric disorders (e.g., late-onset schizophrenia) [91] later in life. The concentrated emergence of sex-biased disorders in windows of particularly dynamic human brain development (i.e., early childhood, adolescent, and post-menopausal years) suggests that sex differences in neurobiology contribute to the emergence of these conditions. Sex biases in sociocultural norms and the applicability of diagnostic criteria are likely to contribute to some of the observed sex biases in disease rates. For example, reduced help-seeking behavior in men may explain their lower treatment rates for depression [92], and current diagnostic guidelines lead clinicians to under-detect ASD in females [93]. However, such biases cannot explain multiple diagnostically distinct disorders all showing the same-sex bias during the same developmental window (e.g., male-biased early-onset ASD, ADHD, and OCD). Second, several sex-biased psychiatric disorders display biological features that interact with known sex differences in neurobiology. For example, the profile of altered gene expression from postmortem brain studies in ASD is partly correlated with normative sex differences in brain gene expression [94]. Similarly, the adolescent-emergent female bias in anxiety is closely linked to measures of pubertal progression that track activation of the hypothalamo–pituitary–ovarian axis in females [88]. Third, medical disorders that modify sex-linked biological factors (e.g., sex chromosome dosage) are associated with increased risk for sex-biased brain disorders [95]. Fourth, males and females exhibit differences in the presentation and progression of multiple neurological conditions. For example: i) females with ASD exhibit more camouflaging behaviors (i.e., compensating for and masking autistic characteristics) and a higher rate of co-occurring internalizing disorders that may mask autistic symptoms, and these factors contribute to the under-diagnosis of ASD in females [93]; ii) mania is more prevalent among males with bipolar disorder, while depression and comorbid panic disorders, eating disorders, and borderline personality disorder are more prevalent among females with bipolar disorder [91]; and iii) males with schizophrenia tend to exhibit higher rates of substance abuse and lower rates of recovery and remission [91]. These differences suggest that sex-specific biological and experiential factors may interact with disease-related factors to impact condition progression. Finally, sex differences in treatment responsiveness continue to emerge. The mechanisms underlying these differences include sex effects on drug metabolism (e.g., in Parkinson’s and schizophrenia treatments) [96, 97] and specific sex–genotype interactions that affect treatment response (e.g., APOE genotype x sex effects on Alzheimer’s treatment outcomes) [98]. Taken together, these diverse sex differences across multiple brain-based disorders provide compelling indirect evidence for sex differences in human brain organization.

The case for probing potential sex differences in human brain organization using in vivo structural neuroimaging

The sections above outline a myriad of evolutionary, genetic, and environmental factors that may produce sex differences in human brain organization, potential evidence for which includes documented sex biases in the prevalence and presentation of multiple brain-based diseases. Taken together, these considerations strongly suggest that humans are likely to show sex differences in brain organization. However, this Commentary is specifically focused on the contentious issue of whether humans show sex differences in one particular aspect of brain organization, namely regional brain anatomy as measured by in vivo sMRI [1, 3]. What are the grounds for using sMRI to test for sex differences in human brain organization? This question is fundamental since the human brain offers a vast phenotypic landscape for the manifestation of sex differences, one that spans multiple spatial (from molecules to visible folds) and temporal (from milliseconds to lifespans) scales. We argue that there are both methodological and empirical reasons to test for hypothesized sex differences in human brain organization using sMRI.

From a methodological perspective, in vivo neuroimaging currently offers the only available means of measuring the living human brain in large numbers of individuals. Moreover, in vivo imaging data provide a homogeneous, brain-wide screen for potential sex differences in regional brain anatomy: either to recover effects that have been hypothesized a priori, or to discover novel effects [99, 100]. These properties stand in contrast to postmortem approaches for studying the human brain which, while they enable molecular and microscale measurements, typically require examination of pre-selected regions in small numbers of individuals [