Race, IQ, and Brain Size

In this article I’ll argue that the research on race, brain size, and IQ, constitutes a significant line of evidence supporting a genetic model of racial IQ differences. But before turning directly to that topic we need to differentiate between research on phrenology and research on brain size. Phrenology was largely concerned with predicting people’s psychological traits using the bumps and curves of their skull. Empirically this has been demonstrated to be invalid (Jones et al., 2018).

By contrast, meta-analyses consistently show that brain size is positively correlated with IQ.

CitationKNR
Rushton and Ankey (2009)281,3890.38
Rushton and Ankey (2009)5963,4050.2
McDaniel (2005)371,5300.33
Gignac et al. (2003)128580.43
Pietschnig et al. (2015)888,0360.24

Moreover, a correlation of 0.19 between brain volume and fluid intelligence was found in a pre-registered analysis of 13,608 Brits (Nave et al, 2018). The fact that this study was pre-registered implies that publication bias cannot explain the finding.

Research also shows that this correlation persists within families meaning it is not the result of brain size being a proxy for socio-economic status or any other variable shared by members of the same family.

CitationCorrelationN
Lee et al. (2019)0.18381
Gignac et al. (2003)0.1564
Schoenemann et al. (2000)-0.0772
Jensen (1994)0.13143
Jensen et al. (1994)0.121,936
Jensen et al. (1994)-0.041,377
Lee et al. (2019)0.191,349
N-Weighted Mean0.105,322

The psychologist and noted opponent of hereditarianism Richard Nisbett has denied this. He wrote on page 219 of his popular 2009 book “Intelligence and How to Get It“”: “The correlation found within the white population probably does not indicate that greater brain size causes higher IQ. Within a given family, the sibling with the larger brain has no higher IQ on average than the sibling with the smaller brain”. To substantiate this claim Nisbett cited Schoenemann’s study of 72 subjects, a sample size too small to be worth much, and simply ignored the rest of the research literature.

The fact that brain size also predicts intelligence in non-human species should make us even more confident that this correlation reflects something causal.

CitationAnimalCorrelation
Hamilton (1935)Rats0.35
Anderson (1993)Rats0.48
Deaner et al. (2007)Non-human Primates0.65
Hopkins et al. (2019)Chimpanzees0.23

Still further evidence comes from the fact that research consistently finds brain size to be genetically, rather than just phenotypically, correlated with intelligence.

CitationFinding
Lee et al. (2019)This study of over 10,000 Brits found a genetic correlation of 0.29 between IQ and brain volume. 
Posthuma et al. (2002)The genetic correlations between IQ and brain volume accounts for the entirety of the phenotypic correlation between IQ and brain volume (n=80)
Pennington et al. (2000)This study found a genetic correlation of .48 between brain volume and IQ. This genetic correlation accounted for 80% of the phenotypic correlation between brain volume and IQ (n=96)
Posthuma et al. (2003)Across three measures of brain volume and four measures of cognitive ability, the average genetic correlation between brain volume and ability was 0.20 while the average environmental correlation was -0.09 (n=688).
Van Leeuwen et al. (2009)Across four measures of cognitive ability, the average correlation between brain volume and cognitive ability was .29 while the average environmental correlation was 0.05 (n=224)
Betjemann et al. (2010)Brain volume genetically correlates at 0.28 (ns) with Verbal IQ and 0.71 with Performance IQ.
Elliot et al. (2018)This meta-analysis found that people with higher education based polygenic scores also have larger brains (r=0.06, k=4, n=7,965).
Hagenaars et al. (2016)Education based polygenic scores genetically correlated at 0.44 with intercranial volume in a sample of 111,114 brits. Specific cognitive abilities exhibited weaker genetic correlations.
Jansen et al. (2020)This study estimated the genetic correlation between brain volume and intelligence to be 0.24 (n=269,867)

There also may be a correlation between an IQ subtest’s g loading and its correlation with IQ. I say “may” because we don’t have much evidence on this question and the conclusion that said limited evidence will lead us to depends on whether we include in our analysis a study which is a statistical outliers from the rest.

Woodley et al. (2016)

Regardless, the evidence suggests that there is a pretty robust relationship between IQ and brain size. To counter this claim, various arguments have been offered.

For instance, a common argument goes like this: men have bigger brains than women, but men are not smarter than women. Therefore, larger brains must not cause people to be smarter.

Of course, brain size only explains a fraction of the total variance in IQ so two groups can be of equal intelligence with unequal brain size so long as the smaller brained group is sufficiently advantaged by some other factor, For instance, some autopsy data has suggested that women’s neurons are more tightly packed than men’s are. Thus, men and women may have the same neuron count even though they have different brain sizes (Lipton 2002 page 193). Or, there may be some other difference that we, or at least I, don’t know about. Either way, this is not a very compelling argument.

It’s also worth noting that the assumption behind that argument, that men and women are equally intelligent, may not be true. Depending on which meta-analysis or large study you look at, you can find that men have higher IQs than women or that the sexes have equal IQ (e.g. Jackson and Rushton 2006Flynn and Case 2011, and Irwing and Lynn 2005). If there is an IQ advantage for men, then this argument against brain size and IQ becomes even weaker.

Another argument sometimes made is that people with megalencephaly, a neurological disease, have huge brains but do not have high IQs. Therefore, large brains must not cause high IQ. This argument has been made by some well known academics (Pietschnig et al 2015), and I think it’s about as compelling as saying that bigger hearts don’t pump more blood because people with enlarged hearts pump less blood than average. Both lines of reasoning are flawed for the same reason: people with rare conditions like these have abnormally functioning biology’s and so should not be used as a model for how human biology normally functions. In the case of people with abnormally enlarged brains, any tension between this and brain size research can be eliminated by noting that they we have no reason to think that the brain areas which are involved in intelligence are especially large and normally functioning in such people.

With the link between brain size and intelligence established, let’s turn to race. In 1994 Harvey et al preformed the first study comparing the brains of different racial groups using MRI technology to measure brain size. They confirmed previous findings: Blacks have smaller brains than Whites. The same finding was reproduced by Jones et al (1994), though the difference wasn’t statistically significant. Chee et al (2010) found that Whites have larger brains than East Asians.  Finally, Tang et al (2010) found that Whites have longer brains than East Asians while East Asians have wider and  taller brains than Whites. Unfortunately, all of these studies have very limited sample sizes and, with the exception of Tang et al, failed to control for differences in the sexual composition of racial samples. Thus, if, for instance, the Black samples had a higher female to male ratio than the White samples then, because men have larger brains than women, Black’s brain size will be depressed relative to Whites.

But these findings correspond well with what older studies measuring skull volume found. Beals (1984) aggregated data on roughly 20,000 subjects from past studies which showed East Asians having the largest skulls followed by Whites followed by Blacks.

Brains q

Beals (1984)

A famous critique of this line of work was launched by Gould (1981) who argued that researchers involved in this work, most famously the 19th century anthropologist Samuel Morton, unconsciously allowed their racist views to cause them to back material more tightly into White skulls than into Black skulls and thus inflate the racial difference. Gould also accused Morton of excluding data from his tables that increased the racial disparity in brain size in favor of Whites. Both of these accusations have been shown to be false. Morton’s skulls were re-examined by modern researchers who reproduced Morton’s results, and the data that Gould accused Morton of omitting was actually included by Morton in the same book that Gould cited, just on different pages (Lewis et al. 2011).

Of course, a third way to measure brain size is to rip a brain out of a skull during an autopsy and measure its volume. Autopsy work done in the late 2000’s found that Whites averaged the largest brains, followed by, East Asians, followed by Blacks (Rushton and Ankney, 2009).  This same method has been used to confirm the Back/White/Asian brain size disparity for well over a century:

1

Tobias (1970) offered a highly influential critique of this literature which in turn was cited and popularized by Gould. In it, Tobias argued that comparisons of the brain size of racial groups based on autopsies were invalid because they failed to control for a wide variety of variables that could impact brain size. Said variables included, but were not limited to, age of death, nutritional intake early in life, occupational status, cause of death, time of death, temperature the brain was kept in after death, and the exact place the brain was cut from the spinal cord.

Some of these controls would obviously be misleading because of the correlation between things Tobias wanted held constant and intelligence (occupational status for instance) and so holding them constant would involve holding a part of the racial IQ gap constant. Some of the proposed controls also had nothing to do with measuring brain size but rather had a plausible connection with impacting brain size (e.g. early life nutrition). Such controls would not be appropriate when asking whether, as opposed to why, racial groups differ in mean brain size.

That being said, some of these criticisms are more plausible. However, there is no reason to think that any of these problems would bias the results in favor of one race or another. And random error introduced can be overcome via aggregation. Thus, Rushton (1995) combined all the data from the studies Tobias critiqued and found the now familiar pattern of Whites having the largest brains, followed by Asians, followed by Blacks. Tobias can be said to have pointed out that racial bias in measurement is logically possible, but he did nothing to justify the belief that this was actually taking place to a significant degree.

A fourth way to measure brain size is to estimate it based on external head size. The major advantage to this method is that researchers can cheaply measure the brain sizes of large and representative samples of living people. One of the largest of these studies was Rushton (1992) which had a sample of 6,325 and found that Whites had the largest brains, followed by East Asians, followed by Blacks.

Rushton (1997) analyzed data on 17,000 white Americans, 19,000 black Americans, and 100 Asian Americans taken from The Collaborative Perinatal Project, a longitudinal study which followed subjects from birth to seven years of age. This data set also showed whites having the largest brains, east Asians having the largest relative to body size, and blacks having the smallest brains in both absolute and relative terms.

Rushton combined data from these sources to produce the following graph showing consistent racial differences in relative brain size from birth to adulthood.

These differences are present among new borns and even unborn fetuses. Such has been demonstrated numerous times. (Several of these paper were initially compiled in Kirkegaard (2018)).

CitationFinding
Rushton (1997)At birth, infants exhibited an Asian>White>Black pattern in brain size (n-35,859).
Thomas et al. (2000)White and Hispanic neonates had head circumferences 0.4cm larger than black neonates (n=27,229)
Mujugira et al. (2013)Mothers of infants with large heads were more likely to be white than were mothers of infants with averaged sized heads (82% vs 74%, n=21,500)
Germaine et al. (2015)White fetuses are shown to have larger heads than black, hispanic, and asian, fetuses at various stages of gestational development (n=2,334)
Schultz (1922) Fetuses exhibited a White>Black pattern in brain case size (n=623).
Ho et al. (1981)Due to having less time in gestation, black new borns were foudn to have lower brain weights than white new borns (n=782).

The fact that racial differences in brain size are realized at such early stages of development constitutes the first line of evidence favoring the view that such brain size differences are partly caused by genes rather than just environments.

Several studies have shown that mulattoes have an average brain size in between that of Blacks and Whites (Pearl, 1934Bean, 1906). This finding has been established on multiple occasions and is what a hereditarian hypothesis would predict since mulattoes are half White and half Black genetically speaking.

Furthermore, many traits which tend to co-evolve with larger brains also differ racially in a way that mirrors the body size adjusted brain size pattern. Rushton and Rushton (2003) looked at 41 anatomical features which 3 textbooks on human evolution identified as tending to co-evolve in the hominid line with larger brains.

Brain 3
Brain 4

For instance, larger pelvic size tends to co-evolve with brain size so that mothers can give birth to larger brained infants. The pattern of the distribution of such traits by race fit what we would expect given evolved racial differences in brain size at a rate far greater than what random chance would likely produce.

Finally, there is some evidence that the races evolved different brain sizes in response to climate. Specifically, various studies have found that a population’s brain size correlates with climate related variables. For instance, Pearce and Dunbar (2011)’s data set produces a correlation of .74 between a population’s brain size and its latitude. Similarly, Ash and Gallup (2007) found a correlation of .48 between the size of 109 fossilized human skulls and the latitude at which they were found. Further still, Bailey and Geary (2009) analyzed 175 skulls ranging in age from 10,000 years old to 1.9 million years old and found a correlation of -.41 between brain size and winter temperature and -.61 between size and latitude (larger brains were found in areas more distant from the equator).

Lynn (2015) used Smith and Beals data set of 20,000 skulls from 122 populations to estimate that roughly 30% of the African-European IQ gap can be statistically accounted for by brain size differences. By contrast, brain size differences would actually predict an Asian-European IQ gap 35% larger than the one that actually exists. Thus, brain size is probably one of many factors which account for racial intelligence differences.

The best known remaining counter to the claim that racial groups differ in brain size due to genes comes from Boas (1912). Boas, who was a founder a modern cultural anthropology, looked at the American children of European immigrants and found that their mean brain size was very different from that of their parents and more similar to the brain size of non-immigrant American children. This study has gone on to be widely cited and pointed to as the first crack in the armor of genetic determinism.

However, in his analysis Boas neglected to account for age differences between groups and this distorted his conclusions. Sparks and Jantz (2002) re-analyzed Boas’s data after accounting for the age difference between parents and children and found that the brain size of immigrants and there children did not differ in any statistically significant way and the brain size of immigrant children did significantly differ from those of Americans. In fact, based on the correlations between parent and children brain size, the heritability coefficient derived from Baos’s own data is greater than 50%. This is consistent with the conclusions of modern research on the heritability of brain size (DeStefano et al., 2010).

Thus, when considering the evidence, objections, and counters to those objections, together we are justified in thinking that data on brain size constitutes a significant line of evidence favoring a genetic model of racial difference sin IQ.

3 thoughts on “Race, IQ, and Brain Size

  1. I was reading about early controversies over brain size and intelligence and came across an 1836 paper by Frederick Tiedemann which I believe claimed to disprove differences in brain size between races.

    I haven’t done the work required to understand statistical methods and significance, but it seemed that his paper actually showed the opposite to what he claimed. The data was thin because little was available and I imagine the samples were highly biased (not deliberately). On inspection, it seemed to show meaningful differences – the author simply stated the opposite!

    I read the paper and it seemed to me that the author was motivated by his noble slavery abolitionist politics. I don’t think inferior intelligence justifies slavery, but I imagine this argument was being used at the time and so Tiedemann felt the need to refute intelligence differences in order to put his anti-slavery on a solid footing.

    His final paragraph: “The slave trade was the proximate and remote reason of the innumerable evils which retarded the civilization of the African tribes. Great Britain has achieved a noble and splendid act of national justice in abolishing the slave trade. The chain which bound Africa to the dust, and prevented the success of every effort that was made to raise her, is broken. Hayti and the colony of Sierra Leone can attest that free Negroes are capable of being governed by mild laws, and require neither whips nor chains to enforce submission to civil authority.”

    Plus ça change?

    Like

    • No, because the thesis isn’t “total brain size is the sole determinant of intelligence”. Then a correlation of less than 1 would be very damning. For any biological trait there are going to be multiple factors which impact it, and the correlation between any single factor and that trait would be quite small, and yet nonetheless real.

      Additionally, a difference that explains a relatively small portion of intelligence variation within groups may nonetheless explain a larger portion of intergroup differences, if that factor differs appreciably between groups and at least some of the contributors to variation do not, because of random variation in other factors that cancel out when looking at group averages.

      Like

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s