I’ve been hesitating a bit to get into this topic, not so much because I’m afraid of sharing my views, but because I don’t want to use this blog too much for discussions that are not directly relevant to open borders. However, given that the topic of “race” seems to have come up in Nathan’s recent post, and in BK’s comments, and since I’ve already commented sharing some of my views, I think it’s best that I go on the record with my views. Note that these are just my personal views. I won’t say that they are completely irrelevant to my case for open borders, but what I would say is that the case for open borders is, to my mind, sufficiently robust under changes to my views on this matter. The goal of this post is merely as a reference I can point to so that I can write in the future about these issues without having to provide long justifications and caveats.

To keep this post short, I will not discuss either the moral issues or the relevance to open borders.

I already signed on to the overall IQ consensus (ignoring the racial issues) at the beginning of this post, so I won’t repeat that here. But you might want to take a look at these two consensus statements, this journalistic survey, and this PDF summary by Linda Gottfredson. So here goes.

My understanding of the genetics-focused human biodiversity/race realist position

The genetics-focused human biodiversity/race realist position can be summed up thus. There are, roughly speaking, three sub-races of the human race: Whites, Blacks, and East Asians. Although most population geneticists don’t use the term “race” in their discussions, this tri-racial scheme is reasonably consistent with genetic distance classifications such as the Cavalli-Sforza tree, which identifies 42 subpopulations of the human species.

The race realist position goes on to argue that the mean “genetic IQs” of the three groups are 100, 85, and 105 respectively (some people use different numbers; I’m just using approximate numbers here). Genetic IQ is an ill-defined term, but roughly, you could think of it as meaning what the average IQ would be if they lived in a well-to-do country such as the US with a middle-class lifestyle. The standard deviations within each group is 15 or lower. Thus, blacks are one standard deviation below whites in average IQ.

Caveat 1: Lynn estimates African IQ as about 67, but says that the “genetic IQ” of Africans is probably 80. But I’m rounding that up to 85 since Lynn is relatively extreme. In the comments, BK has pointed out to the work of Jelte Wicherts which casts some doubt on Lynn’s pessimistic estimates, and suggests that current IQ in South Africa is about 80, which suggests that the “genetic IQ” may be around 85 (see also here).

Caveat 2: Some people, such as Lynn, consider genetic attribution of IQ differences even within sub-races and between ethnically similar populations in different regions of the same country. For instance, Lynn applies genetics to considering differences between Northern and Southern Italy. In a critical piece, Ron Unz writes of Lynn:

Although Lynn attributed this large deficit in Southern Italian IQ to substantial North African or Near Eastern genetic admixture, poverty and cultural deprivation seem more likely explanations.

Still, I think Lynn is a bit of an outlier in the fineness with which he makes genetic distinctions and uses admixture to explain differences between nearby regions.

Best data point: The mean IQ of US blacks, who are mostly of African origin, but have some Caucasian admixture, is estimated to be about 83-90, and the US white mean is estimated at about 100, which fits with the above estimates (more here). While there have been claims of a narrowing of the US white-black IQ gap, the evidence is mixed, though a Flynn effect (a secular trend of rising IQ) seems to have been operational for both races over time. If you believe that US blacks’ current IQ largely measures their “genetic IQ” (whichever interpretation you use of the term) then this vindicates some aspects of race realism.

My understanding of the broader human biodiversity position

Some proponents of human biodiversity say that there is no need to be stuck up on genes. The key point to note, from the public policy perspective, is that these differences are unlikely to “Vanish Real Soon” as Steve Sailer puts it in this article:

For purposes of sensible public policy, arguing over whether genetics plays a role in racial differences in achievement is a red herring. What’s crucial to understand is that racial differences—for whatever reasons—are unlikely to vanish Real Soon Now, as all right-thinking people are supposed to assume.

The point I think Sailer is trying to make is that even if culture or other factors, rather than genes, are responsible for these differences, there is no magic wand to fix these factors in the short run, and most of the “obvious” interventions have been tried and shown not to work (more on that later).

My thoughts on the “genetic IQ” concept

My main beef is that I find the concept of “genetic IQ” ill defined and poorly conceived. Yes, there is a heritable component to variation in IQ, and yes, it’s possible that differences in gene frequencies between races is responsible for differences in observed IQs. In fact, I’d be very surprised if none of the genes that affect IQ were found to vary in frequency between human subpopulations. I think a few genes have already been found that affect IQ and differ in frequency between different human subpopulations (I remember seeing a link from a comment somewhere, but I’m not able to dig it up right now, but I’d be happy to add in a link).

But, talk of genetic IQ presupposes a kind of “holding environment constant” or some kind of maximal, optimal environment in which everybody can fulfill their genetic potential. Most IQ “hard hereditarians” hold this sort of view. I think this is plausible, but I have doubts. It’s possible that talk of “genetic IQ” is like talk of “genetic income” — certainly, genes affect income and part of the variation in income is heritable, but the concept of “genetic income” doesn’t make sense. On the other hand, it may be possible that IQ is more like “number of hands” in that there is a genetic predisposition to a certain value, barring specific environmental catastrophes. Of course, these are both extremes. I would be more inclined to believe that IQ is like income, and that the concept of “genetic IQ” as a maximum potential that individuals can achieve is not yet substantiated by the evidence. Long-term secular trends in IQ (the Flynn effect for one) and in income seem to support my point. Incidentally, the fact that interventions to permanently improve IQ have not been clearly established does not, in my view, cut against my point because I think the same is largely true of income. I discuss interventions a couple paras later.

Still, one can define “US middle-class genetic IQ”: what would a person’s IQ be if he/she were raised in a middle-class US household today? If you subscribe to parental irrelevantism, then knowledge of a person’s genome would give you a distribution whose mean would be the “genetic IQ” and whose variance would reflect non-shared (unique) environment factors. This kind of definition is plausible and useful in the context of discussions of the short-run effects of immigration to the United States and child-bearing in the United States, as long as we don’t confuse it with the concept of a maximum potential.

Heritability or at-birthality of IQ at an individual level

Heritability of IQ is high (0.5-0.8) in the US, Sweden, and wherever else it has been studied (including a few studies in India, I believe), but this is heritability within a country or within a certain type of environment, not cross-national heritability. To be clear, there are two types of heritability numbers: narrow-sense heritability (a lower number, which is relevant when using degrees of genetic relatedness between non-identical twins to compute correlations) and broad-sense heritability (a higher number, relevant for identical twins). The narrow-sense heritability of IQ has been lower-bounded at 0.4 or so and the estimate ranges between 0.45 and 0.7 (see here). The broad-sense heritability estimate ranges are between 0.5 and 0.8, with the consensus probably tending to the higher end of the spectrum. The best criticism of high-end estimates is that identical twin estimates may be suffering from the problem of monochorionic twins (see here). If this concern holds up, then perhaps what we call “broad-sense heritability” is better called “at-birthality” because it measures the squared correlation between a kid’s situation at birth and adult IQ rather than the squared correlation between the genome and adult IQ. The estimate of relative importance of post-birth environmental changes would not be affected, as far as I can make out.

The source of the non-heritable (or post-birthal) component of IQ variation within countries is unclear, but it seems not to be due to parenting or anything else that can be explicitly pinned down, but mostly due to “random” variation, what Bryan Caplan might call “free will” and others might call chance. Here’s how Caplan puts it:

I doubt that scientists will ever account for my sons’ differences, because I think their primary source is free will. Despite genes, despite family, despite everything, human beings always have choices–and when we can make different choices, we often do. Some choices are moment-to-moment: To keep working or give up, lie or tell the truth, abandon or defend your views on immigration policy. Other choices are cumulative: You can’t change your weight, education, or income by snapping your fingers, but in the long run they depend on diet, study, and effort–all of which you’re free to choose.

If you want to be more sophisticated, you might want to talk of path dependency. The most interesting theory regarding the non-genetic contributors to individual variation (not just for IQ but for all personalty traits) that I have come across has been offered by Judith Rich Harris in her book No Two Alike.

Some factors that are clearly implicated over the historical long run include nutrition and disease. Malnutrition clearly affects brain development. But it’s generally believed that these factors don’t have a positive effect beyond a point, i.e., they suffer from diminishing returns . The role of cognitive stimuli may also be quite similar, but there is a possibility that variations in some types of cognitive stimuli may be driving individual variation in IQs even in the developed world.

Interventions to improve IQ

First, adult IQ is reasonably stable, probably less stable than adult height but more than adult weight. Thus, the main intervention point for IQ is in childhood, up to the age of 16 or so.

Tons of research suggests that conscious deliberate attempts to environmentally manipulate IQ don’t work. Parental socio-economic status and other measures of the environment seem to have very little effect on adult IQ, per twin/adoption studies. They do affect measured childhood IQ in the short run, but by the time children become adults, the effects of parenting fade out. (Note: This doesn’t mean that children with high childhood IQ and lower adult IQ lost IQ in absolute terms, they just los in measured IQ because IQ measurements are age-adjusted).

On the other hand, the Flynn effect (secular increases in IQ over the past century) suggests that broad environmental changes can affect IQ. The best reconciliation of these results I have seen is a paper by Dickens and Flynn here. But a lot of their conclusions are speculative. Also of interest is Dennis Garlick’s theory of intelligence and the brain in his book. The main reason they offer for why interventions don’t work is that, however intensive they are, people return to the real world sooner or later, and it’s the broad environment of the real world that ultimately shapes their incentives and stimuli. Fade-out is the key here. People who seek intellectual stimulus can often find it in the broader environment even if the particular sub-environment they are in is lacking in that stimulus. For this reason, large environment, in so far as it offers a “menu of options” is more important than the specific parental environment. (Note: This may not be true of measures other than IQ; for instance, measures of academic and scholastic achievement are probably more directly correlated to the quality of your school, but again, adult academic knowledge is probably less correlated with the quality of the school you attended in childhood).

It’s also important not to assume that the Flynn effect will continue to be operational. It may. Or it may not.

What also seems clear is that the Flynn effect doesn’t seem to be causing significant convergence between groups. It’s a “rising tide that lifts all boats” — the future may be different, or it may not.