Saturday, 10 May 2014

The Economist finds the gene for intelligence

It it difficult to work out what is happening at The Economist. They used to ignore the concept of intelligence, or certainly did not include it in their discussions of economic progress. They never referred to national intelligence levels as a factor in innovation, and did not bother to review those books which linked intelligence and scholastic achievement to a nation’s gross national product.  They were the citadel of Blank Slate-ism.

Now things are changing, and today 10 May under the chapter heading “Genes and intelligence”  their headline reads “The 3% Solution”.

What they are touting is research which suggests that one gene accounts for 3% of the total variance in intelligence, which is a considerable amount for one little gene. You might imagine I would pounce on this finding as part of “I told you IQ was genetic”. To the contrary, this Damascene conversion of an economics magazine, while welcome, does not blind me to the improbable claims being made.

KL-VS did not curb decline, but it did boost cognitive faculties regardless of a person’s age by the equivalent of about six IQ points. If this result, just published in Cell Reports, is confirmed, KL-VS will be the most important genetic agent of non-pathological variation in intelligence yet discovered.

Gene hunters doing their genome wide searches are happy if they find a collection of genes which each account for a fraction of 1% of the variance. The more genes that get compared to a variable like intelligence the higher the chance of spurious associations. False positives require large sample sizes and 100,000 people is now standard. The Dubal and Mucke study has a sample size of 718. Yes, their techniques are different from GWAS but it is a little early to declare a result.

Perplexed, I decided to get into the Psychological Comments network of loyal readers, and quickly got a response from a notable researcher in the field:

I hope the finding is true — and the mouse story is interesting — but we have reasons to be sceptical because the effect size is so large it should have popped out of GWA studies.

The hunt (for many, many genes of small effect) continues.


  1. Pretty unlikely: 1) An allele with such a big effect would have surfaced in other GWAS. 2) It's absent in East Asians and present only in Africans and Europeans, and even then only at a modest (around 20%) frequency.since this mutation is shared by Africans and Europeans it's likely to be pretty old, we would expect such a "gene" to have spread much faster and reach higher frequency among contemporary populations. 3) Sample size is small. 4)

  2. Agree so far. What about point 4?

  3. This comment has been removed by the author.

  4. This comment has been removed by the author.

  5. 4) The association is non-linear across allele status, that is having 0 beneficial alleles is as bas as having 2. Nonlinear associations require greater statistical power to ascertain. It could just be a fluke that only heterozygous state predicts higher IQ, whilst homozygosity for either the ancestral or derived allele (that is, having 0 or 2 copies) predicts lower IQ.

  6. Thanks. Let us see what they come up with when they have a proper sample size which we would estimate at....what sort of number? Same as GWAS?

  7. Yes, a modern GWAS, so I'd say 100K+

  8. @Merculinus:

    2) is a mistaken argument (and James).

    Known height enhancing alleles have shown different selection pressure signatures in different environments (more selection in Northern than in Southern Europe), though some of them are quite old.

    Likewise, educational attainment alleles do not show equal selection pressure in all environments, as evidenced by their different frequencies.

    Any given allele might simultaneously be 1. IQ boosting 2. universally selected against (disadvantageous). 3. selected against in some measure, at some allele frequencies, in some environments, etc.

    Be careful not to fall into the pitfall of arguing that IQ boosting alleles can be assumed to be advantageous, that is an idiosyncrasy. In our current environment (the only one for which we have good data), it is markedly selected against, for example.

    That said, I agree that the Economist obvious is mistaken. It's a shame that they base their conversion upon such a silly study - let's hope they aren't burned too badly.

    1. Redzengenoist: You are apparently forgetting that this allele is supposed to increase not only IQ, but also life expectancy. So why the heck would it be eliminated from East Asian populations, which are NOT known for having short lifespans and low in IQ? The selective pressures among East Asians for high IQ and high life expectancy have probably been higher than among Europeans and Africans, yet they lack this "smart longevity allele". And even among Africans and Europeans, this "beneficial" allele is present at modest (0.2) frequency, whereas if the heterozigous condition were really the most advantageous as the authors claim, the most favorable frequency would be 0.5. You say that the current environment selects against intelligence. 1) Evidence for this? Not much. 2) Even if this were true, it's equally true that human cognitive capacities have increased among Europeans since they left Africa, as shown by 1)technological progress among Cro-Magnon and 2) increase in most of the other alleles affecting IQ. So we're left with a weird picture where the other alleles that increase IQ by only 1 point went up in frequency, but this allele which supposedly has super IQ and health boosting powers was eliminated. This does not make sense at all.

    2. @Merc:

      a) whether the East Asian population is exceptionally high on longevity or intelligence does not speak to whether an allele rare in their population boosts longevity or IQ. There are many good arguments against KL-VS, but this one is misplaced.

      Solomon Islanders have a SNP mutation which gives 10% of them blonde hair. This mutation is rarely present among Norwegians, who are more frequently blonde. The lack of this SNP among Norwegians is a poor foundation on which to assert that this SNP does not boost blondeness, even if one also asserts that there's more selection pressure for blonde hair color in Norway.

      b) "You say that the current environment selects against intelligence. 1) Evidence for this? Not much."

      Come now.

      c) "So we're left with a weird picture where the other alleles that increase IQ by only 1 point went up in frequency, but this allele which supposedly has super IQ and health boosting powers was eliminated. This does not make sense at all."

      There are good arguments against KL-VS, the best of which is that similar studies with larger sample sizes have not replicated it. But the assertion that "it could not be" because an allele suggested to boost IQ and longevity could not be maladaptive brings discredit. Sickle cell anemia boosts Malaria resistance, but it's rare in most environments, because that boost has a hidden cost somewhere else.

    3. You missed the general meaning of my argument. I raised 4 arguments, and each of these makes the false positive thesis more probable. None of them is sufficient per se but each of them reduces the likelihood that this gene is a real hit. Since we've got no hard scientific evidence, we've got to rely on circumstantial evidence. The real test will be future GWAS, so far we're just discussing about remote possibilities. Your arrogance is only matched by your lack of attention to what I wrote

  9. Thanks. I will defer to you both! As to The Economist, I doubt their overall approach has changed at the editorial level. I simply hope they include intelligence as a potentially causal variable in economic outcomes.