What kinds of genetic differences should we expect between populations?

Epistemological Status: A little out of my depth.

Kevin Mitchell’s blog came to my attention because of his skeptical posts on epigenetic inheritance in humans. Even more interesting is his recent post on race and IQ, and the likely lack of genetic differences between races.

I think this is an extremely important argument. The strongest arguments for innate racial IQ differences generally run like so:

  1. We observe large differences in IQ between racial groups in the human population.
  2. No known environmental factor could explain such large differences.
  3. Genetic differences between races are large enough to explain the differences.
  4. Therefore, we should assume that most racial IQ differences are due to genetics.

In my posts on the Flynn Effect, I showed that Flynn and Sowell have disproved premise (2) – it’s true that no family-level environmental factors can explain such large differences, but that environmental factors at the level of culture and subculture have demonstrably explained large differences in the past.

But that doesn’t offer racial liberalism much help against an argument for a weaker conclusion:

  1. We observe large differences in IQ between racial groups in the human population.
  2. Most group differences in traits are believed to be partially genetic and partially environmental.
  3. There is no reason to think IQ works differently than most other traits.
  4. Therefore, our default assumption should be that a significant portion of racial IQ differences are genetic.

Before Mitchell’s post, the strongest counterargument I could make was “there are bad consequences for falsely assuming that racial IQ differences are genetic, so public intellectuals should avoid speculating on what our default assumptions should be.” That’s a political argument, though, not a scientific one, and telling public intellectuals to avoid a scientific controversy that is interesting to the public is a tough sell, to say the least.

Mitchell’s argument takes aim at premise (3). One can imagine the genetics of intelligence working one of two ways:

  1. In one model, different human populations have faced a wide variety of evolutionary pressures, some that select for greater intelligence, others that trade intelligence off against other useful traits. The variations in intelligence we see among humans are the result of various, competing evolutionary strategies seeking the ideal balance between intelligence and other traits. Genes for intelligence, in this model, are kind of similar to the single nucleotide polymorphism for sickle-cell – the advantages of malaria resistance trade off against the disadvantages of anemia (don’t be thrown by the word “single” – that doesn’t mean just one gene affects intelligence; rather, it means that a single gene variant is common enough that its effects can be measured.)
  2. In another model, cognitive ability is so widely useful that basically all human populations have faced similar enough pressures that the most intelligent humans in each population are near the theoretical maximum – at least for animals built roughly the same way we are. The variation in intelligence we see among humans within each population are the result of mutational load – basically, a huge number of very tiny “birth defects” that we all have to some degree or another.

Mitchell argues that the latter is true, and as a result, we’re unlikely to see significant, genetic differences in intelligence between populations, even though we do see significant, genetic differences in intelligence between individuals.  I’m a little out of my depth when it comes to evaluating this argument, but I think Mitchell is probably right, for several reasons:

  1. Measured IQ is highly correlated with almost every other trait considered desirable, even ones – like physical attractiveness – that don’t seem like they “should” be correlated with cognitive ability. This suggests that random harmful mutations – “general genetic fitness” – may be a bigger underlying factor than evolutionary selection.
  2. Direct measurement seems to bear out Mitchell’s conclusions. Large-scale studies have found specific genetic variants that explain about 11% of population differences in educational achievement, and the authors of these studies believe this is a correct estimate of the total. Most traits are around 50% heritable, which leaves 39% of variation due to “missing heredity” – likely, due to rare genetic mutations with negative effects. If genetic differences can explain at most 11% of racial differences in IQ, that amounts to less than two points, and is probably socially insignificant.

On the other hand, I can think of several arguments – that I consider weaker – against Mitchell’s conclusion:

  1. Brain size is somewhat correlated with IQ, and seems like it could be a candidate for an evolutionary trade-off, in that large brains make childbirth more risky.  That suggests a possibility that SNPs for “big heads”, essentially, might be correlated to intelligence.  I can think of many reasons this seems unlikely in practice, but the simplest to explain is that we don’t seem to see subpopulations of notably intelligent, large-headed people.  Which brings us to…
  2. …the one possible example we have of a population where – according to some theories – genes for high intelligence trade off against genes for health conditions: Ashkenazi Jews (who, as far as I know, don’t have abnormally large heads.) This is a highly speculative theory, and even if it’s true, I don’t think we have any other examples of similar ethnic effects.  Nothing about Mitchell’s theory suggests that you could never have a subpopulation selected for high intelligence; just that would be very rare.  Racial liberalism can probably survive one exception in the direction of minority advantage.

Interestingly, Mitchell’s theory seems like it should have important implications not just for the study of race and IQ, but also for the possibilities of human genetic engineering.  Under the first model – which we might call the “SNP Model” – it seems like we could potentially genetically engineer people with superhuman intelligence.  We’d need to find lots of SNPs correlated with high IQ and give someone all the right variants.  On the other hand, evolution made certain trade-offs for a reason, so we might expect that people genetically engineered to have superhuman intelligence would have high rates of birth defects and other problematic health conditions.  Ideally, we would need to understand how each and every variant worked, so that we could select only the ones where the trade-offs between intelligence and other traits were worth it.

But under the other model – the “Rare Mutations Model” – it’s paradoxically much harder to engineer superhuman intelligence but much easier to engineer peak human intelligence – and peak human everything-else as well.  Under the Rare Mutations Model, you don’t even need to know what most genes do; you can just go through and replace every rare variation with a common one, and thereby make everyone as cool as Captain America, with little of birth defects or other health problems (or rather, the risks would be due to the side effects of the process, not due to intentional genetic modifications themselves.)

There is a potential problem with genetic engineering under this approach, in that rare variants are occasionally useful.  One can imagine a superplague, a la Stephen King’s The Stand or the Krippin virus in I Am Legend, that wipes out 99% of the human population but leaves the genetically immune untouched; engineering away all rare gene variants increases the likelihood that such a disease would wipe out 100% of the population instead.  I’m not sure whether this is a realistic scenario outside of science fiction.

One last note: Someone looking to support scientific racism in the face of Mitchell’s theory might suggest that mutational load itself varies by race – that is, some races are more “genetically fit” than others.  I don’t know much about the topic, but it seems that if anything, populations whose ancestry is farther from sub-Saharan Africa have higher mutational load, which is the opposite of what a racist theory would predict.


Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s