This is a bit of a distraction, but not really. Some time ago Freddie deBoer referenced me in a blog post titled, “Questions for Pure Environmentalists“ as follows:
(You also have Eric Turkheimer, a behavioral geneticist who coined the Three Laws of Behavioral Genetics and yet who has in recent years spent much of his energy sniping at every attempt to quantify the very phenomena he posited. I don’t understand the scientific utility of this effort, although I certainly do understand it from a personal and professional standpoint.)
This quote annoys me in multiple ways, starting with the shitty accusation that I take the points of view that I do to protect my personal and professional reputation. You know, someone who really knows better, but soft-pedals hereditarianism as a way of protecting my lib creds. I don’t think many people who actually know me would say that.
Second, I am not remotely a “Pure Environmentalist” and never have been. Here is a quote from our recent “Spit For Science” paper, a copy of which appears to be here. (I didn’t put it there! A preprint is here.)
Lest readers get the wrong idea, it is necessary to interject that we are not “environmentalists” in the usual sense, any more than the S4S authors are old-fashioned hereditarians. The problem with individual-level psychiatric genetics exposed by S4S is not that variation in alcohol use turns out to be environmental rather than genetic. We have argued that the nearly universal “genes vs. environment” rubric that has been imposed on nature-nurture questions since Francis Galton is a distortion of what is at stake (Turkheimer, 2016). The important question about complex human phenotypes like alcohol use has little to do with partitioning of variance into genetic and environmental components. The real question is whether the causes of those differences can be reduced to molecular explanatory units, whether genetic or environmental. It is this question that S4S has answered in the negative, at least regarding genetics. We do not think that a program of research aimed at the specific family dynamics underlying problematic alcohol use would be any more successful (Turkheimer, 2004).
I am not an environmentalist; I am an anti-essentialist. I don’t think people with alcohol problems have an essential genetic nature (they don’t have an essential environmental nature either), and I don’t think that kids who struggle in school have an essential unintelligent nature.
There is a school of heterodox thinkers— Haidt, deBoer, Steven Pinker— who misread my three laws paper in a hereditarian direction, as a declaration that everything is ultimately genetic. Then they are disappointed when it turns out I don’t really think that. But the title of that paper was “Three Laws of Behavior Genetics and What They Mean.” What they mean is that the heritability of behavior is misleading— not imaginary or spurious, but misleading— in particular because it doesn’t mean that essential behavioral traits are being passed down from parents to children.
In particular, heritability does not “quantify” the characteristic that deBoer wants it to quantify, the extent to which a trait is “genetic.” What deBoer perceives as me taking back his hereditarian interpretation of the first law is really me denying the status of heritability coefficients as descriptive statistics of “genetic.” More recently, it has been me predicting that GWAS was not going to succeed in exposing the genetic basis of complex behaviors, and even more recently it has been me pointing out that I was right.
The funny thing is, I agree with the fundamental premise of The Cult of Smart, which I take to be this: It is wrong for both progressives and No Child Left Behind conservatives to blame the educational difficulties faced by some kids exclusively on the schools. You take schools with kids from the worst backgrounds, in the worst neighborhoods, with the worst economic problems, observe that they do badly in school, blame it on the schools themselves, and then (in NCLB) punish the schools for the failure.
For progressives, this a way of avoiding blaming the kids for their own failure, which I understand. But still, you have to be a pretty committed blank slatist (environmentally, genetically, whatever) to think every kid is equally ready to succeed on Day 1 of first grade, if only the schools did their job. deBoer wants to insist that genetic differences are part of the reason some kids are less ready than others. Is this right? It is a question about genetic essentialism. It’s not a simple argument.
Here is what heritability shows us. Genetic variance is associated with variance in behavioral traits (educational attainment or EA for now). More genetically similar people are more similar for ALL behavioral traits. That is the first law of BG. But that assertion is only about variances. On a causal level, all we know is that if people were more genetically similar— ultimately if we were all clones— the variance of EA would go down. That is what I call a non-directional hypothesis heritability tells us nothing about how people with certain genes will do better or worse in school than people with other genes. A finding that people with certain genes reliably do better in school would show that they had an essential inner characteristic that makes them smarter, but that is what we never know.
GWAS was supposed to solve this problem, but it didn’t. In the real world, what does deBoer want? Presumably he wants more than just that we think about genetics when we ponder all the reasons that people have variable outcomes. What could we actually do with genetic knowledge? Suppose we decided to start genotyping little kids and computing their EA polygenic score. To do that right, we would absolutely have to use a within-family score, so we would not just be estimating their educability on the basis of their family. That means we would have a predictor that accounts for maybe 3% of the variance. Would that be good enough to leave deBoer sanguine about declaring, the kids in School X get low scores on standardized tests, but of course that is in part because they are genetically unfit? And if not, why bother?
This brings me back to Spit for Science. (I’ll be shameless: Please read that paper!). S4S is about alcohol use, not childhood education, but the same principle applies. S4S happened because some very competent people took deBoer’s ideas seriously. We know that alcohol use is heritable in the twin sense, we therefore believe that alcohol use has a “genetic basis,” so let’s get serious and start genotyping some students. 12K students later, they accomplished exactly nothing. That is not me offering a harsh reinterpretation of their findings, it is literally what they reported. Whatever the virtues of alcohol genetics may turn out to be in the long run, right now it has no value in understanding why some college kids drink more than others. That is exactly what would happen if we tried to do the same thing to understand why some kids do better in the third grade than others. That is, as deBoer dismissively put it, “the scientific utility of my effort.”