Cro-magnons were the first truly human Europeans and have long been romanticized as the first truly modern humans, in part because of a Euro-centric bias, but also because they’re the people most associated with the Upper Paleolithic revolution and the replacement of the Neanderthals who had held our ancestors captive in Africa for hundreds of thousands of years.

They had wonderful lives, wandering through the forests of Europe as the seasons changed from summer to snowy winters, sleeping under the stars, exploring a vast uncharted terrain with a child-like awe of the unknown and rich mythology to explain it, vastly out-competing their only rivals the Neanderthals for food and shelter, and hunting so much mega fauna that they reached their full genetic potential for height and brain size, unlike the stunted malnourished people who would follow them once agriculture was invented. These were some of the happiest people to have ever lived and they did so in a true Garden of Eden, before tempted by the metaphoric apple into the “sin” that was agriculture. A “sin” we would pay dearly for.

But how intelligent were they? This is a tricky question because based on brain size, they appear a bit more intelligent than Europeans today, but based on polygenic education scores, they appear substantially less intelligent.

Brain size suggests Cro-Magnons had an IQ of 104

Over the 20th century, brain size, IQ and height in the developed World has increased thanks to advances in nutrition and reduction in disease allowing us to recover the genetic potential not realized since the Upper Paleolithic, before agriculture had stunted us (Lynn, 1990). Despite brain size in 21st century America reaching an impressive 1418 cm³, (Pumpkin Person, 2018) it’s still less than it was 40,000 years ago, when brain size peaked (mostly in Europe) at about 1487 cm³ (after correcting for preservation bias). So it seems the remaining 69 cm3, shortfall (0.76 SD) is a genetic loss.

But why did genetic potential for brain size decrease 0.76 SD? Perhaps in part because genetic height in Europe decreased 0.5 SD (Piffer & Kirkegaard, 2024) and polygenic height scores correlate 0.64 with height, and height correlates 0.4 with cranial capacity (Rushton, 1992), then this fact alone might explain (0.5 SD)(0.64)(0.4) = 0.13 SD of the brain size decrease, leaving 0.63 SD (57 cm³) not explained.

Perhaps the remaining shrinkage can be explained by a genetic drop in muscle mass that paralleled the drop in genetic height, but until ancient DNA studies explore SNPs associated with lean body mass, this remains speculation.

Given the 0.4 correlation between IQ and brain size (Jensen 1998)(Gignac & Bates, 2017), this would predict an IQ loss 0.4(0.63 SD) = 0.25 SD or 4 IQ points since the Upper Paleolithic, suggesting Cro-Magnon would have scored an impressive IQ 104 if raised in today’s America!

Genetic education predicts Cro-Magnon had an IQ of 89

Although brain size may have decreased in Europe by as much as 0.63 SD (after adjusting for body size), polygenic education scores (EA3 PGS) seem to have increased from what looks like -1.63 in Upper Paleolithic Europe type areas, to what looks like about -0.2 in Medieval England to about +0.2 in contemporary England (Piffer & Kirkegaard, 2024) for a total increase of about 1.83 SD since the Upper Paleolithic. Given a correlation of 0.4 between IQ and polygenic education scores (Pumpkin Person, 2024), this suggests a gain of 0.4(1.83 SD) = 0.73 SD since the Upper Paleolithic (11 IQ points), suggesting Cro-Magnon would have scored IQ 89 if raised in today’s America.

Indeed when I recently asked Peter Frost what their IQs would be if reared in today’s America, he guessed “high 80s”:

Combining brain size with genetic education predicts Cro-Magon had an IQ of 93

If brain size predicts Cro-Magnons have an IQ of 104 (if raised in today’s America) and genetic education level predicts an IQ of 89, and there’s no a priori reason to think one prediction is more valid than the other, then the only reasonable solution is to predict their IQs based on both variables.

The multiple regression equation predicting IQ from brain size and genetic education is:

Phenotypic IQ gap = 0.37(phenotypic brain size gap) + 0.37(genetic education gap) (Pumpkin Person, 2024).

When this formula was tested on the 15 point black-white IQ gap, it accounted for an incredible 13 of the 15 points! (Pumpkin Person, 2024).

Now applying it on the gap between today’s U.S. whites and Cro-Magnon man:

Phenotypic IQ gap = 0.37(+0.63 SD) + 0.37(-1.83 SD)

Phenotypic IQ gap = +0.23 SD – 0.68 SD

Phenotypic IQ gap = -0.45 SD

0.45 SD = 7 IQ points and is equivalent in SD units to a one inch increase in within sex height over 40,000 years. Kind of trivial. On a scale where today’s U.S. whites scores 100, Cro-Magnon would have scored IQ 93, consistent with the sophistication of their drawings; when I objectively scored these on the draw-a-man IQ test, they clocked in at IQ 90, though the sample size was tiny (n = 2).

Although some evidence suggests adaptive evolution increased ten thousand percent in the last 40,000 years (Hawks et al, 2007). Europe only gaining 7 IQ points in 40,000 years suggests evolution more or less ended in the stone age, consistent with the punctuated equilibrium model. This is consistent with the emergence of brain genes peaking 55,000 years ago (Libedinsky et al, 2023).

Richard Klein states:

What happened 40,000 or 50,000 years ago was the last major change in the genotype. At least the last major biological change. Evolution continues, but the evolution that’s involved in making us capable of wielding this vast variety of cultures–that probably stopped around 40,000 or 50,000 years ago and there’s been no essential change since.

Stephen Jay Gould agreed , famously stating:

There has been no biological change in humans in 40,000 or 50,000 years. Everything we call culture and civilization we’ve built with the same body and brain.

One of the great mysteries of human evolution is why after four million years of brains getting bigger in the hominin line leading to humans, did they suddenly start to shrink sometime after modern humans colonized the World and replaced all rival homo species. While Lynn (1990) attributes this to malnutrition and disease caused by agriculture, others have argued it was a genetic change.

However Peter Frost argues that it might (at least in part) be a statistical artifact:

Perhaps smaller skulls are more likely to decompose faster. The skulls we unearth would therefore be a biased sample, and this bias toward preservation of larger skulls would gradually increase for skulls that have been in the ground longer.

The problem of “preservation bias” has already been noted with respect to female and infant remains

This makes perfect sense but how big is the effect and is there anyway to adjust for it? Frost points us to a paper by Walker et al. (1988) that describes a California cemetery for the mission’s Indian neophytes who died circa 1831. Although the mission records showed that 53% of the buried people were female, the skeletal record showed only 50% were. The inference to be drawn from this is that because females are smaller, they decompose faster.

Now imagine if we tried to estimate the average cranial capacity of the mission Indians from the skeletal record. Considering men have brains about 12% bigger than women do, such a decrease in the percentage of women alone would increase the mean cranial capacity of the entire sample by 0.34% in just the 157 years after death that the data was published. Assuming the process is linear, that leads to a 87% increase in 40,000 years. So instead of Cro-Maagnons having brains of 1498 cm³(80 cm3 bigger than the average American today), they’d have brains that averaged 1498 cm³/1.86 = 805 cm³

A more reasonable hypothesis is that instead of preservation bias inflating brain size estimates by 0.34% per 157 years, it inflates it by 0.34% per the natural logarithm of 157 years, so 5.06 log years. And if Cro-Magnons lived around 40,000 years ago, or 10.6 log years ago, this would make the 1498 cm³ brain size estimate for Cro Magnon 0.71% inflated, and thus the true value 1498 cm³/1.0071 = 1487 cm³. One should note that weighting the samples by sex will not entirely solve this problem because sex is simply being used here as a proxy for big vs small brains and for most skulls, the sex is probably not known, they just assume the bigger skulls are male.

The U.S. black white IQ gap has been about 1 standard deviation (σ) since WWI (Jensen 1998, Pumpkin Person 2016). How much of this can be explained by biology?

Well we know the brain size gap is about 1.08 σ (at least as of 1980) (Pumpkin Person 2024)

The gap in polygenic education scores between U.S. whites and Nigerians is 1.6 σ (Piffer, June 4, 2020), however because U.S. blacks are about a quarter white genetically (Bryc et al. 2015), we can arguably reduce this by a quarter to 1.2 σ.

In other words, the U.S. black-white gap in brain size, polygenic education level, and IQ are all about 1 σ.

Predicting the black-white IQ gap from brain size & genetic education

Polygenic education scores now explain about 8.7% of the variation in Peabody verbal scores (Okbay et al 2022) suggesting a correlation of 0.29 (the square root of 8.7%), but since vocabulary tests “only” correlate 0.74 with general intelligence ( McCrea & Robinson, 2011) we can arguably divide this correlation by 0.74 and estimate that polygenic education scores (at least the best ones), much like brain size, correlate about 0.4 with IQ.

So knowing both phenotypic brain size and genetic education both correlate about 0.4 with phenotypic IQ, how well do these variables predict the U.S. black-white IQ gap?

Well assuming these two variables are independent, aside from their shared correlation with IQ, we’d expect them to correlate about 0.16 with each other (the product of their shared 0.4 correlation with IQ) and thus each would independently correlate 0.37 with IQ.

In other words:

Phenotypic IQ gap = 0.37(phenotypic brain size gap) + 0.37(genetic education gap)

Phenotypic IQ gap = 0.37(1.08 σ ) + 0.37(1.2 σ)

Phenotypic IQ gap = 0.4σ + 0.44σ

Phenotypic IQ gap = 0.84 σ

In other words, about 84%, or 13 of the 15 point U.S. black-white IQ gap can be explained by known biology. Of course that doesn’t prove it’s 84% genetic as the phenotypic brain size gap could be at least partly environmental in origin.

Jensen predicted 80% of the variation in adult IQ was explained by genes and the remaining 20% was explained by biological environment. Jensen’s default hypothesis thus predicts √ 80% = 89% of the 15 point gap (13 points) will ultimately be explained by DNA, and the remaining 2 points will probably be explained by racial differences in prenatal environment.

The single best study of U.S. race and brain size ever done was by Ho et al (1980). They found that at age 25, the average sex-combined brain weight of white Americans (at autopsy) was 1455 g and the average brain weight of black Americans was 1333 g. I suspect the brain weights of both races have since increased thanks to the nutrition that causes the Flynn effect (Lynn 1990)(Figure taken from Jensen (1988), though even in 1980, the white male brain weight at 25 was absolutely colossal:

Unfortunately Ho et al only reported standard deviations for the entire sample and not specific ones for different age groups. However a study of South African black brain weights found that at age 21 to 30, the SD is 122 g for men and 103 g for women for a within sex average of 113 g (Govender et al, 2018)

If we assume a similar SD within U.S. races, then the sex combined age 25 black-white brain weight gap gives a within sex brain weight gap of 1.08 SD which is identical to the 1 SD black-white IQ gap observed since WWI.

Of course that’s not to say that the black-white IQ gap is entirely caused by the black-white brain size gap. The correlation between brain size and IQ is “only” about 0.4 (Jensen 1998)(Gignac & Bates, 2017) so only about 40% of the gap might be caused by brain size and that’s assuming the correlation is entirely causal.

However individual level correlations can more than double at the group level because individual sources of error cancel out. For example, Rushton (2009) found brain size correlated 0.91 with IQ across 10 population groups.