Wow! So much to blog about, so little time.  There’s a fascinating new scientific paper out by Michael A. Woodley, Shameem Younuskunja, Bipin Balan and Davide Piffer, claiming major natural selection for IQ since agriculture and it couldn’t have come at a better time since I was just arguing about this with commenter MeLo.  Here’s the abstract:

abstract

Although studies of identical twins reared apart show IQ is hugely genetic, at least within specific Western countries, roughly 10,000 genetic variants are involved, so identifying the specific genes with any kind of statistical certainty is extremely tough given how small an effect each one has.

To make sure a gene really does affect intelligence by a fraction of an IQ point with any kind of statistical certainty, you probably need a sample of as many as one MILLION people, and they all need to take a very reliable and g loaded IQ test like the Wechsler, preferably at middle age when heritability is perhaps maximized.  So unless some pro-science gazillionaire makes a huge donation to the field, allowing such a large-scale quality research to be funded, scientific progress is going to be painfully slow.

Because it’s so hard to get large samples who have both taken quality IQ tests and given DNA samples, education level is used as a very crude proxy for IQ and a cognitive polygenic score (POLYcog) used to predict education in 300,000 people, was later found to explain 3.6% of the variance in IQ in a different sample.  This recent paper on post-agriculture cognitive selection pressures largely relied on such POLYcog data.

One interesting part of the paper was a graph showing the increase in POLYcog scores in 66 ancient genomes as a function of time.

genome

I have no idea how these POLYcog scores are being scaled, but judging from the fact that the lowest score (for its time) looks like it’s about 0.5 units below the regression line, and the highest score looks like it’s about 0.72 units above the regression line:  A gap of 1.22 units.  In a normal distribution, the gap between the highest and lowest in a sample of 66 is 4.33 standard deviations (SD), so by dividing 1.22 by 4.33, I estimate the SD of these POLYcog scores is 0.28.  Meanwhile, it looks like the POLYcog scores were increasing at a rate of 0.13 units per 1000 years which would thus be 0.46 standard deviations per 1000 years.

Now it’s a bit of a misnomer to call these POLYcog scores.  I believe they were largely created using a study to predict education (not cognitive ability specifically) and so they probably explain much more of the variance in that (9%).  Taking the square root of 9%, yields correlation of 0.3 between polygenetic scores and education.

Now a 0.3 correlation implies that for every 1 SD increase in POLYcog score, education should increase 0.3 SD which means that the 0.46 SD per thousand year increase in POLYcog scores predict a 0.46(0.3) = 0.14 SD increase in educational achievement per thousand years (if those ancient genomes had been cloned back to life in modern America).  In one large sample of American whites, the standard deviation for years of education was 2.34 years, so multiplying this by 0.14 suggests that for every thousand years of post-agriculture evolution, the expected level of education (if ancient humans were cloned in modern society) increased by 0.33 years per 1000 years.

Assuming this trend was linear from say the start of agriculture in Europe (about 8500 years ago) to the start of the industrial revolution about 200 years ago, we’d expect a European from 8,500 years ago to complete 2.7 fewer years of formal education than a cloned human from 200 years ago.  Whether a cloned human from 200 years ago would complete more or less education than the average European today is hard to say since some argue there’s been a post industrial-revolution dysgenic effect, but the evidence is contradictory.

As to why genetic changes associated with IQ and education have occurred in the Holocene, the paper suggests that after millions of years of living as hunter/gatherers, the novel challenge of adapting to agriculture and all the associated social, political, economic, technological, and military changes, selected for more intelligence.

I also think the huge population boom caused by agriculture created new mutations for selection to select from (see The 10,000 year explosion by Cochran and Harpending), but the study couldn’t confirm that since all the ancient genomes had variants that people today also have.  One fascinating factoid the paper mentions is that the speed of adaptive evolution during the Holocene (end of the ice age and the start of agriculture) increased 100 fold from the preceding Pleistocene.

Near the end of the paper, they discuss the paradox of brain size declining in the Holocene while genes associated with GMA (General Mental ability increased):

gma

However a simpler explanation is that declining brain size in the Holocene had nothing to do with genes and was caused by malnutrition, especially since brain size has arguably returned to its original size.  Richard Lynn made the case back in 1990:

biglynn

biglynn2