Raw: –0.98
Adjusted for…
Height: –0.26
Total volume: +0.01
Right thalamus
Raw: –1.03
Adjusted for…
Height: –0.27
Total volume: –0.02
Source: Adapted from Ritchie, Cox, Shen et al. (2018): Tables 1 and S1. Figures in bold indicate that p < 10–4.
*Adjustments for gray and white matter for total brain volume were not performed because of collinearity.
The effect sizes in the “Raw” column represent the expected magnitudes of difference in brain volumes in a randomly chosen man and a randomly chosen woman. The effect sizes in the “Height” column represent the expected differences in brain volumes of a man and woman of the same height. The effect sizes in the “Adjusted for total volume” column represent the expected differences in regional brain volumes for a man and a woman with the same total brain volume.
The raw effect sizes for the subcortical volumes ranged from –0.31 to –1.08, with a median of –0.660. Adjusted for height, the effect sizes ranged from –0.05 to –0.35, with a median of –0.24. The “Adjusted for total volume” column shows men retaining larger volumes in all but two regions, but with effect sizes that are no larger than –0.25 and a median of –0.12.
Which of these three ways of looking at differences in brain volume should we use? This question only raises more questions. It makes sense that we should adjust for body size for some traits. A plausible reason that elephants have brains twice the weight of human brains is that it takes a lot more neurons to control large muscular structures and nervous systems than to control small ones. But the logic of adjusting for body size is not obvious when it comes to the intellect and emotions. It’s all happening in the brain. Why should it take more neurons to solve a quadratic equation in a person who is 5 foot 8 than in a person who is 5 foot 4?
According to neuroscientists whom I have asked, there is oddly little in the technical literature that systematically explores when it is appropriate to adjust for differences in body size. But there is a clear reason to adjust for total brain size for certain purposes: It is the appropriate method for finding interesting sex differences in the relative sizes of different regions. The question that must still be at the back of the investigator’s mind is the extent to which adjusting for total brain size produces another kind of artifact, removing variance that does in fact contribute to sex differences in overt traits. That brings us to the fraught question of brain size and how smart people are.
Brain Size and g
The progress of hominids from chimpanzees to anatomically modern humans has been marked by increases in skull volume.6 Paleontologists, physical anthropologists, evolutionary biologists, and neuroscientists have broadly agreed that greater skull volume means greater brain volume, and greater brain volume across species is associated with greater cognitive capacity.7
It is also established that brain volume in humans is correlated with IQ scores, and hence with g. That knowledge has been hard-won in the face of controversy. The early correlations were based on indirect measures of brain size—for example, measuring head circumference—that left much room for doubt.8 Then MRI technology made it possible to determine in vivo volume of the brain with precision—and not just total brain volume, but the volumes of the dozens of subcortical regions.
The size of the correlation of overall brain volume with IQ has varied from sample to sample. Two meta-analyses of all such studies concluded respectively that the correlation is +.33 and +.24.9 Subsequently, a reanalysis of the literature argued that the scientifically most rigorous studies show an average correlation of +.39.10 Furthermore, this relationship holds within sexes. On average, men with larger brains have higher IQ than men with smaller brains, and women with larger brains have higher IQ than women with smaller brains.11
Is the relationship causal? After all, bigger brains mean more neurons. But what counts for cognitive functioning in mammals is the number of neurons in the cerebral cortex and the subcortical regions.12 On this score, humans stand apart from all other mammals. The human cerebral cortex contains about 16 billion neurons. The next largest, found in gorillas and orangutans, is just 9 billion.13
Several studies have found evidence for this causal link in humans. Cerebellar brain volume has been found to explain variance in g in older adults, even after controlling for frontal lobe volume (which tends to atrophy with age).14 In 2018, neuroscientists reinforced the evidence for a causal link through a study with a large sample (n = 2,904) of brain development in youths and young adults (first author was Kirk Reardon).15 The cortical surface area expands threefold between infancy and adulthood. A priori, one might expect that human brains of different sizes scale uniformly—a brain with larger total volume also has a linearly larger hippocampus, amygdala, and so forth. But that’s not what the Reardon study found. “Rather,” as David Van Essen summarized it, “larger brains show greater expansion in regions associated with higher cognition and less expansion in regions associated with sensory, motor, and limbic (emotion-and affect-related) functions.”16 The Reardon study also found that overall cortical surface area correlated significantly with IQ after factoring out age and sex.17 Thus there is good reason to expect a causal relationship not only between brain volume and IQ, but between brain volume in specific regions and IQ. Other things being equal, more neurons are a good thing for cognitive functioning.[18]
But other things are not equal.
Sexual Dimorphism in Brain Volumes Does Not Necessarily Mean Dimorphism in Cognitive Function
Chapter 5 describes sex differences in connectivity that make female brains more efficient. There’s much more work being done in this area. Sex differences in receptor density could be at work independently of regional brain size or overall brain size. Furthermore, volume isn’t the only relevant measure of size. Two other measures that were considered in the Ritchie study were the convoluted cortical surface areas and cortical thickness. Those two features have been found to be independent of each other, both globally and regionally.19 The male-female differences in surface area for all 68 subcortical regions were even larger than those found for overall volume, with effect sizes ranging from –0.43 to –1.20 and a mean of –0.83.20 This was not true of cortical thickness, however. Consistent with an earlier study of sex differences in cortical thickness,21 females had significantly thicker cortex across most of the brain (47 of the 68 areas). Males had significantly thicker cortex in just 1 of the 68. The differences in the remaining 20 areas did not reach statistical significance. The effect sizes in the 47 ranged from +0.07 to +0.45, with a mean of +0.22 (a positive d indicates greater cortical thickness among females). These differences remained significant after adjusting for total brain size.
Other brain parameters can and do vary by sex; among them, cerebral blood flow, glucose metabolism in the limbic system, dopamine transporter availability, the percentage of gray matter tissue in some parts of the brain, and the percentage of white matter tissue in other parts. The links between these parameters and behavior were summarized by Ruben and Raquel Gur in a review article. “For example,” the Gurs write, “differences in gray and white matter volumes have been related to performance on verbal and spatial tasks, sex differences in hippocampal volume and in dopamine availability have been linked to memory performance, and sex differences in limbic activity and orbitofrontal volume have been associated with differences in emotion regulation.”22 What all this shows, the Gurs conclude, is a set of differences that cannot be ranked from good to bad, but that tend to be complementary.
Neuroendocrinologist Geert de Vries has argued that sex differences in brain structure may work to prevent phenotypic differences. “Intuition tells us that sex differences in brain structure beget sex differences in brain function,” he wrote in 2005. “There is nothing wrong with that. If, for example, a brain area has three times more cells that produce a specific neurotransmitter in
one sex vs. the other, and if these cells send, accordingly, three times denser projections to target neurons in another area, stimulation of these cells will probably cause sex-specific responses in the target neurons.”23 But researchers have drawn their hypothesis too narrowly, de Vries argues. Sex differences in brain structure “may indeed generate differences in overt functions and behavior, but they may just as well do the exact opposite, that is, they may prevent sex differences in overt functions and behavior by compensating for sex differences in physiology”—hence the subtitle to his article, “Compensation, Compensation, Compensation.”24 In 2015, de Vries and Nancy Forger elaborated on such compensatory mechanisms. Every organ in the body is sexually differentiated to some degree, the authors argue, and they present a variety of examples whereby sexual differentiation in organs and tissues throughout the body eventually affect neural function or morphology.25
These are just some of the many reasons for caution. The reality of sex differences in brain volumes is firmly established. Collateral evidence indicates that these myriad differences must have implications at many levels of brain function. But our understanding of the specifics, and what those differences mean for phenotypic traits, is still rudimentary.
Generally Greater Male Variance
If you followed the furor about James Damore’s internal memo at Google that got him fired in 2017, or if you’re old enough to have followed the furor over Larry Summers’s comments about male-female differences in attraction to STEM back in 2005, you’ve encountered the phrase “greater male variance hypothesis.” The reason the hypothesis has gotten so much attention is its implication for explaining male dominance at the highest levels of achievement in the arts and sciences throughout recorded history.26 If general cognitive ability g is normally distributed, then even if males and females have the same mean g, greater variation in males will mean that men are overrepresented at the tails of the normal distribution.
A lot of “ifs” lie between the existence of greater male variance and the explanation for male dominance at the highest levels of achievement. That’s why I am not prepared to defend a statement that begins, “Greater male variance in measures of abilities explains…” That’s a leap too far. But this less ambitious statement is no longer controversial: Greater male variance in a wide variety of traits is a fundamental biological characteristic of humans and of dimorphic species more generally. It doesn’t happen with every trait for which data are available, but with a substantial majority of them. The greater male variance hypothesis isn’t a hypothesis anymore.27 It is now known to be generally true.
The Evolutionary Context for Greater Male Variance
I will break my own rule for this book and introduce a little evolutionary biology into the conversation, because the reasons for greater male variance go so extremely deep into the evolutionary dynamics that have been operating for hundreds of millions of years among all species that reproduce sexually.
The simple observation that the males of many species show more visible variation goes back to the 1700s and was remarked upon by Darwin in The Descent of Man.28 It has an elemental evolutionary driver: the necessity of having progeny if one’s genes are to be passed on. In 1948, A. J. Bateman, a botanist, presented the first hard evidence for what became known as Bateman’s principle: In most species, variability in reproductive success is greater in males than in females. He used that staple of genetic research, the fruit fly, for his evidence.29 Through a series of experiments, Bateman established three important sex differences in reproductive success:
Males’ reproductive success varied much more widely than females’—only 4 percent of females failed to produce offspring, compared to 21 percent of males.
Being able to attract the opposite sex was far more important for males than for females. Female reproductive failure wasn’t because of a failure to attract males. The 4 percent of females who failed to reproduce were vigorously courted; they just weren’t interested. Conversely, the 21 percent of males who failed to reproduce gave every appearance of trying hard to copulate. They just couldn’t get accepted.
Engaging in lots of sex is extremely helpful for male reproductive success, but not for female reproductive success. For males, the number of offspring increased almost linearly with the number of copulations. For females, reproductive success increased only marginally after the first copulation.30
In 1972, Robert Trivers drew on Bateman’s research and collateral findings to make a seminal contribution: “What governs the operation of sexual selection is the relative parental investment of the sexes in their offspring,” with parental investment defined as “any investment by the parent in an individual offspring that increases the offspring’s chances of surviving (and hence reproductive success) at the cost of the parent’s ability to invest in other offspring.”31 The optimal strategy for a sex that made little parental investment, Trivers argued, is to mate with as many partners as possible. The optimal strategy for a sex that made high parental investment is to be choosy about the choice of mate. Evolutionary psychologist David Geary put it this way:
The sex that provides more than his or her share of parental investment is an important reproductive resource for members of the opposite sex. The result is competition among members of the lower-investing sex (typically males) over the parental investment of members of the higher-investing sex (typically females). Competition for parental investment creates demand for the higher-investing sex that in turn allows them to be choosey when it comes to mates.32
In more than 95 percent of mammalian species, males make extremely small parental investments and females make huge ones.33 This imbalance can produce greater male variability via different routes that have been the subject of an extensive literature.34
Reviewing that literature here would take us far afield. Think of it this way: To pass on your genes, you must mate. If you are a male and females are choosy, you need to have traits that attract females in the first place and other traits that enable you to fight off, figuratively or literally, other members of your own sex. If you are a female and the males of your species will copulate with anything that moves, you do not face any of those sources of evolutionary pressure.35 On the contrary, deviations from the normal range may exact fitness costs.
If we are talking about modern human males and females, the dynamic I just described may now be weak. Contemporary human males typically make far higher parental investment than most mammals. Humans have been an example of what Steve Stewart-Williams and Andrew Thomas call the “mutual mate choice” model, in contrast to the “males-compete/females-choose” model.36 But the role that evolution is argued to have played in generating greater male variability is not something that started with Homo sapiens. It has been going on since sexual dimorphism began. In all cases of greater choosiness in one sex, wider variability of traits improves your odds of passing on your genes if you are a member of the less choosy sex. In the overwhelming majority of cases, the less choosy sex has been male. When the process involves unimaginable trillions of reproductive events over millions of years, the fractional fitness advantages of greater variability add up.
A ROLE FOR THE SEX CHROMOSOMES IN PRODUCING GREATER MALE VARIABILITY
The fact that females have two X chromosomes while the male has only one suggests a straightforward explanation for traits that are influenced by genes in the X chromosome. The explanation is based on elementary arithmetic: Women can average, males can’t. Here’s the more precise statement by evolutionary biologists Klaus Reinhold and Leif Engqvist: “Binomial sampling of the large X chromosomes leads to the intuitive prediction that males should show larger variation. In females, the traits that are influenced by X-chromosomal genes will be under the average influence of the two parental copies, whereas in males, the effect of the single X-chromosome will not be averaged. As a result, male mammals are expected to show larger variability than females in all traits that are, at least to some extent, influenced by X-chromosomal al
leles,” given certain conditions described in the note.[37] The overall effect of the sex chromosomes on greater male variability is limited to traits affected by the X chromosome, which means that greater male variability in most traits must be driven by other forces.
So much for the evolutionary explanation of greater male variability. You’re free to ignore it. The topic here is not an explanation of greater male variability, but the empirical evidence for it. Nowhere is that biological truth more unequivocal than in greater male variance in the brain.
Males Have Greater Variance Than Females at Both the Whole Brain and the Regional Levels
The table below shows the variance ratios for the same three versions—raw, adjusted for height, and adjusted for total brain volume—that I reported for effect sizes in regional volumes. Variance ratio (VR) is computed as the variance of one population. In the case of the following table and all other references to VRs in this discussion, male variance is divided by female variance. Therefore a VR greater than 1.0 signifies greater male variance.
VARIANCE RATIOS FOR RAW AND ADJUSTED DIFFERENCES IN BRAIN VOLUMES
Total brain
Raw: 1.22
Adjusted for…
Height
Total volume
Gray matter
Raw: 1.23
Adjusted for…
Height: *
Total volume: *
White matter
Raw: 1.22
Adjusted for…
Height: *
Total volume: *
Left hippocampus
Raw: 1.16
Adjusted for…
Height: 1.15
Total volume: 1.16
Right hippocampus
Raw: 1.30
Human Diversity Page 43