The new EA3 study is out today. Or I should say it is unembargoed today, the basic findings have been floating around the genomics community in one form or another for quite a while. I should add that it is first authored by GHA participant James Lee, so I take most of the credit for it. A couple of weeks ago, a paper by Belsky et al came out, applying the EA3 to several big datasets, with a focus on socioeconomic mobility, that is to say relations of EA3 out outcomes in offspring, conditional on their parents score on the outcome, or computed within sibling pairs. There was quite a bit of reaction to that paper when it came out, and I want to start there. Various parts of the hereditarian right (for lack of a better term, I’m not trying to insult anyone) were dancing in the endzone. Robert Plomin, promoting his upcoming book, announced that “Nature defeats nurture by a landslide.”
Steven Hsu (following Plomin, who always refers to polygenic scores as a game-changer) tweeted:
Game Over! Genomic Prediction of Social Mobility. Out-of-sample validation; polygenic score predicts life outcome (socioeconomic index). Cohorts from NZ / US, born different decades. Higher SES families have higher polygenic scores on average. PNAS paper.https://t.co/yGHO7E8WJw pic.twitter.com/qrdAwu8eUg
— steve hsu (@hsu_steve) July 10, 2018
Charles Murray didn’t go with the game metaphor, but suggested that Belsky et al. vindicatedRichard Herrnstein’s syllogism about IQ and social inequality.
Richard Herrnstein took lots of grief for his syllogism, first published in 1971. One of the kindliest men I have ever known, he never took any joy in the prospect of a cognitively stratified society. But he was right. His often-vicious critics were wrong. That’s something. https://t.co/FuwKE6bV6c
— Charles Murray (@charlesmurray) July 10, 2018
What game is being changed and ended here? For Plomin and Hsu it is the old nature-nurture game. I don’t want to get into that pointless argument (Though for insiders, I would say that Plomin is sort of correct in the sense that A has won in a landslide over C; it has fought E to a draw)
I am not here to criticize EA3 or the work it has enabled. It is a great study, using amazing new technology that permits all kinds of new analyses that weren’t possible in the twin and family era of behavior genetics. The authors of the study are cautious and thoughtful in their interpretation of the results, far more so than the people above. But what EA3 does not do is change the basic outcome of the nature-nurture equation from where we already knew it to be. (With one possible exception: as I have said elsewhere, modern DNA-based genomics should finally put an end to the old argument that the only reason behavioral differences are heritable is flaws in the twin design like the equal environment assumption.) . But modern genomics has not changed the basic nature-nurture equation at all. The world is no more genetic after EA3 than it was before.
There are two ways to look at the progress made in EA3. As a very successful application of new genetic technology, it is extraordinary. Scientists interested in the complex interaction of genes and environments in the development of human differences can now do all sorts of things that were not possible before. We can, for example, ask about how genes children don’t inherit from their parents affect them anyway, via the environments the parents create. And although there is still a very long way to go, we can begin to ask questions about how individual genes contribute to biological pathways that eventually lead to brains and learning. I have my doubts about how successful these efforts will be, but that is no reason not to try.
On the other hand, it must be remembered that the basic theoretical insight that genes have a role to play in human differences like intelligence and education has been well-established for decades. Identical twins separated at birth are correlated for their educational attainment and IQ, and identical twins raised together are substantially more similar than fraternal twins who are less alike genetically. These classical studies demonstrated that genetic differences are related to educational differences, and the findings reported here– that there are units of DNA that have tiny correlations with EA, and aggregates of DNA that have larger correlations– are exactly what would have been predicted on the basis of the twin studies. If anything, the genetic relations reported here are smaller than what had been known before from twins and other family members.
The authors of the EA3 study do an excellent, responsible job of emphasizing the enormous complexity of the genetic relations they have uncovered. A substantial portion of the genetic effect appears to operate via environmental pathways, so the prediction among siblings in the same family is substantially smaller. Once again, although the new technology is once again remarkable, in a general way we have known this for a long time. Measures of the family environment are correlated with genes, and genes have their effect via the family environment. The genomic effects that are observed between and within families are starting to diverge. The authors also show that the effects of DNA depend to a great extent on the context in which the DNA is collected– these are not genes that have the same effect everywhere, like the gene for Huntington’s Disease. Instead they influence outcomes in subtle, contextually sensitive, and hard to trace ways, with effects that can only be detected in enormous samples that average across the multiple contexts. This dependence of genes on environmental and cultural context is most important when the genetic effects are evaluated across racial groups. The genomic composites, developed primarily in Europeans, do a poor job of predicting EA in other racial groups.
So the study should be celebrated for all the new things it allows scientists to do. But it should always be emphasized– as the authors do– that the study does not break new ground on the general question of whether complex human outcomes are related to genetic differences. We knew that already. We knew that some genetic relationship existed, and we also knew that the relationship is extremely complicated, too complex to permit simple deterministic explanations. So this study does not show, pace Plomin, that children should be assigned to schools based on their genomic scores (would you assign children based on their parents’ IQs?); it does not show that quality education is pointless; it does not show that The Bell Curve was right all along; it does not show that differences in outcomes among racial and ethnic groups are genetically determined. In fact none of these things are any more likely today than they were yesterday.