IQ is highly heritable

Response to Ian

Ian Jobling is a former White nationalist who wrote for Amren back in 2002-2006, gave up his radical beliefs in 2012, and now advocates against race realism particularly the hereditarian strand of it. He’s come to substack and written an article arguing against the high heritability of IQ:

The primary reason that I became a racist was that the state of behavior genetics research in the 1990s seemed to justify racism. Behavior geneticists believed that human psychological traits were strongly heritable, meaning that they were heavily determined by our genes. Given such strong heritability, it seemed certain that IQ gaps and other psychological differences among the races had a genetic basis and were not likely to change.

In 2012, I still believed in the strong heritability paradigm, so I didn’t criticize it in the Intelligence Report interview. However, more recent research has convinced me that the behavior genetics consensus of the 1990s overestimated the influence of genetic factors on psychological traits.

I generally don’t like bringing up conflicts of interest, since they are a part of life and cannot be controlled, but it’s worth mentioning that he has a massive incentive to attack hereditarianism to improve his image. He might not be lying, but he could be engaged in some Hansonian self-delusion where he is unconsciously adopting thought patterns that lead him to lower levels of belief in hereditarianism. Unfortunately, there is no way this can be fixed — you either have it in you to overcome your delusions, or you don’t.

His case rests on four main arguments: that heritability (proportion of variance in a trait associated with genetic factors) does not imply direct genetic effects, twin studies do not capture certain environmental effects (e.g. cohort associated effects), the Flynn Effect implies a large environmental component to IQ, and molecular genetic studies suggest that the heritability of IQ is lower than previously thought.

I agree that heritability does not imply direct genetic effects, which is why bringing up the high heritability of BMI is misleading, as part of the reason why this trait is heritable is because psychological traits and food preferences are innate, and those then cause BMI to be heritable.

In the case of IQ, IQ may be heritable partly because pre-existing innate intelligence difference may cause people to gain knowledge and skills that magnify the previous differences.

To say that this refutes the idea that heredity matters is nonsense. To analyze the effect of genes on IQ independent of knowledge/skills acquisition, an artificial environment where children cannot be compelled to learn would have to be created. Though I’m not exactly sure if it’s even possible to do this, as children cannot really be stopped from learning, even if they wanted to. And even if the heritability assessed om this environment was lower, these results would not generalize to the real world anyway.

There are environmental effects that are not captured by twin studies, as twins are born on the same year, and typically researchers correct for Flynn Effects or compare twins of similar ages and birth years. Traditionally, observed IQ scores have risen by about 2-4 points per decade. Skepticism of the magnitude of the gain aside, I find it unlikely that intelligence did not increase due to modernity at all. Heights increased, athletes got faster, and people started living longer — to think this would not generalize to intelligence as well, especially when observed scores have increased, would be the alternative hypothesis.

That said, there evidence that the observed Flynn Effects overestimate the true gain in intelligence. Notably, the gains lower substantially after correcting for the violations in measurement invariance that occur from comparing different cohorts:

A meta-analysis of studies that try to test for whether the gains are on g using the method of correlated vectors find that, between subtests, g gains correlated negatively with g-loadings. That is to say, that tests that are better measurements of latent intelligence have smaller Flynn Effects:

Black/White differences in mean IQ have been clearly shown to strongly correlate with g loadings, so large group differences on subtests of high cognitive complexity and small group differences on subtests of low cognitive complexity. IQ scores have been increasing over the last half century, a phenomenon known as the Flynn effect. Flynn effect gains are predominantly driven by environmental factors. Might these factors also be responsible for group differences in intelligence? The empirical studies on whether the pattern of Flynn effect gains is the same as the pattern of group differences yield conflicting findings. A psychometric meta-analysis on all studies with seven or more subtests reporting correlations between g loadings and standardized score gains was carried out, based on 5 papers, yielding 11 data points (total N = 16,663). It yielded a true correlation of − .38, and none of the variance between the studies could be attributed to moderators. It appears that the Flynn effect and group differences have different causes. Suggestions for future research are discussed.

And even if the Flynn Effect was as large as its advocates claim it is, it would not invalidate a high heritability of intelligence, as large gains in height due to modernization have been observed (of about 1.4 SD or so), despite the fact that the high heritability of height is uncontroversial and not disputed.

Molecular genetic studies do generally find lower estimated heritabilities of IQ scores, but the estimated heritability depends substantially on the method used. If heritabilities are estimated based on just the SNP heritability, then the heritability of IQ is 23%.

However, if the effects of rare variants are accounted for using a pedigree-based design, the heritability rises to 54%.

For g, common SNPs (h^2 g) explained 23% (SE = 2%) of the phenotypic variation. Pedigree-associated genetic variants (h^2 kin) added an additional 31% (SE=3%) to the genetic contributions to g, yielding a total contribution of genetic effects on g of 54% (SE=3%). The net contribution of measured environmental factors to phenotypic variance in g was 35%. This was due to two sources of variance, shared sibling environment (es^2 ) and shared couple environment (ec^2 ), that accounted for 9 % (SE=1%), and 22% (SE=2%), respectively.

I read this paper in its entirety, and I couldn’t help but notice that they modeled spouse-specific effects as “environmental”. Later, in the discussion section, the authors implicitly admit that this is not a correct assumption:

The couple component is somewhat complex to interpret. For intelligence and education, there is evidence of assortative mating, which will increase both the genetic and environmental similarity between couples. The couple component may mostly reflect this spousal similarity, along with the effects of more recent environmental influences. Beyond that, intelligence is not perfectly stable across the life course and studies of twins in earlier childhood frequently find a sizeable shared environment component. The importance of shared environment is usually said to decline from childhood to adulthood, as individuals pick their environmental individual niches (i.e., active gene-environment correlation), but this is based only on environment shared with siblings. However, it may also be that the current environment remains important and that the spouse is a better aggregated indicator of the current environment than the sibling with whom one usually no longer shares a home in adulthood. We find no couple component for personality, which is consistent with much weaker assortative mating on personality, especially neuroticism and extraversion.

If the similarity between spouses in intelligence is assumed to be caused by the same proportion of genetic and environmental factors as the ones that cause differences between individuals, then the estimated heritability is 69%, which is roughly in line with the results of twin studies and monozygotic twins reared apart.

54 +22*h2/100 = h2
54 = h2*78/100
h2 = 69

Because of the existence of so many possible effects (additive genes, dominant genes, epistasis, GxE, cultural transmission, shared environment, prenatal environment, twin specific effects, selective adoption, wilson effect, assortative mating), I generally view twin and family models as rough estimates of heritability rather than accurate ones, and am open to the possibility that certain phenotypes (especially socially-associated ones, like income or behaviour) are more caused by environmental factors than is commonly thought. Regardless. the observed family correlations in intelligence are consistent with an estimate of 70%, if the relevant confounds in the relationship (e.g. assortative mating, shared environment) are accounted for.

from here