Turkheimer's Projects: Genetics and Human Agency | Behavior Genetic Revanchism

Haven’t blogged in a long time.  As of a couple of months from now I will be under some contractual obligation to blog, so I guess I had better get started.

First of all, the location of the blog has moved to the Genetics and Human Agency website, geneticshumanagency.org.  Genetics and Human Agency is a project funded by the Templeton Foundation to support research and scholarship in the philosophy of behavior genetics.  We will be making final decisions on applications to the project this week.  One component of the project is to build a blog network on the topic, and funded participants will be required to participate.  The site will also be open to other interested parties, so get in touch with me (ent3c@virginia.edu) if you think you might be interested.

One process note.  I find that one of the (many) things that holds me back from writing here more often is that I have an ongoing urge to self-cite, which is kind of embarrassing and makes me feel old.  Nevertheless, I am officially giving myself permission to do so.  It’s an old person’s feeling, but I confess to having the impression that a lot of what is getting written about behavior genetics these days involves conveniently forgetting some complications I have introduced into the discussion over the last twenty years.  I think the main reason for this, which I’ll try to explore over the next few weeks, is the slow replacement of the twin studies at which my criticisms (really, reinterpretations) were originally directed are now all getting re-conducted as molecular studies of one kind or another.  Doing so gives the investigators license to return to innocent surprise at the First Law of Behavior Genetics (Wow, this has turned out to be genetic!) and then to trot out simplistic accounts for what the universal heritability of human behavior implies for individual self-concepts and the methodology of psychological science.

My example for today was just published in the online webzine Quartz, with the unfortunate title, IN YOUR GENES: One of the fastest growing fields in science still makes a lot of people very uncomfortable. The article is motivated by a recent paper published in Molecular Psychiatry by Robert Plomin’s team, linked here.  This is a complicated article using a lot of new methods that I will comment on later.  For now, suffice it to say that they used a molecular method called a polygenic risk score to show that we can predict educational attainment from people’s genes (this is not new) and that we can do so at with somewhat higher accuracy than has been reported before, getting close to 10% of the variance.

First of all, the overall take of the article– that the heritability of educational attainment is this newly discovered thing, making people (and presumably people on the left, more in a second) uncomfortable, is ridiculous.  We have known that educational attainment was heritable for at least 50 years on the basis of twin and family studies.  If there is anything to be surprised about based on new polygenic risk score studies, it is that heritabilities are much LOWER than had been estimated by twin studies.  There are lots of reasons why this would be so, which I will get into later.  I don’t mean to say that this study shouldn’t have been done– it’s very interesting to see what happens when we examine an old phenomenon with a new method.  But the general idea, that genetic differences are correlated with differences in educational attainment, is not new, so it can’t really be a fresh source of irritation to liberal sensitivities.

That brings me to the important part, thinking about what it means that something like educational attainment turns out to be partially heritable.  Goldhill states that,

Though these scientific findings could be alarming for anyone, the seemingly deterministic perspective seems to contradict the left-wing emphasis on the role of privilege in any person’s success.

and then,

But though the scientific findings in molecular genetics don’t directly contradict this, many who care deeply about the effects of privilege seem to flinch at the research. I’ve spoken to people with no knowledge of the scientific literature but a strong political perspective who insist that such studies about the role of genes simply cannot be true.

Well I don’t know who Goldhill has spoken to, but really?  I think any of our (liberal or conservative) grandparents, if asked whether people are born with any characteristics that make them more or less likely to get educated, would have answered yes. Does Goldhill really suppose that liberal beliefs about equality of educational opportunity depend on denying genetic differences?  We liberals are not that stupid.

There are even more important reasons to be skeptical of Goldhill’s conclusions.  She (and unfortunately, the behavior geneticists she speaks to) make the classic mistake of confounding heritability with (low) malleability of outcomes.  She quotes Nancy Segal as follows:

But the belief that anyone can be anything they choose—that you could take a child and train them towards literally any goal with success is, says Segal, “such an old-fashioned notion, with absolutely no backing whatsoever.”

So let’s see.  Phenylketonuria is a recessive genetic disorder that causes mental disability and death if affected people consume foods containing phenylalanine.  So does having this gene make it impossible to train children to live and prosper?  No, because it can be prevented via diet.  Height is nearly perfectly heritable in the modern world.  Does this mean that it would have been hopeless for a pre-war Dutch person, three inches shorter than the average American, to have children as tall as future Americans?  No, and in fact Dutch people are now substantially taller.

If this happens to be the first time you have read my writing on this subject, it is important to understand that I am not disputing the heritability of anything, I am disputing what that heritability means, on two related accounts:

  1. There is nothing unusual about it.  It is extremely easy to get caught up in the reaction, Holy Mackerel, educational attainment is heritable!  But EVERYTHING is heritable.  So it isn’t like we have opened up some new avenue in understanding EA, we have just shown that it is subject to the human condition like everything else.
  2. There is a hidden theoretical assumption under all this which is incorrect.  A simplistic model of the heritability of EA says that people are born with a genetic propensity for education, which then plays out environmentally in their lives.  Polygenic risk scores make this interpretation all the more likely, because it seems as though they ARE the genetic propensity, but this whole assumption is wrong.  The heritability of EA shows that people with some genotypes are more likely to wind up more educated than others; it does not show that some people have greater CAPACITY for education.  The fact that in 1920 height was heritable and greater in Americans than Dutch did not show that Americans had greater capacity for height, it meant that under then current conditions Americans were more likely to wind up tall than Dutch people.  That is not some pointy-headed philosophical argument, it is the key to understanding human behavior genetics.

I don’t want to oversimplify- it is certainly possible to think of genetic or environmental (eg brain damage) examples that would represent diminished capacity, rather than simply a different probabilistic outcome. For now, I just mean that a polygenic risk score that correlates .25 with EA does not represent capacity.  I’ll save further explanation for a later post.

Scroll to Top