Monday, November 17, 2014

Genes and IQ: some clarifications

I hadn’t anticipated that my article for Prospect on the “language of genes” would spark a discussion so focused on the question of the inheritability of IQ (which was just a small part of what the piece was about). Judging from the reverberations on Twitter, it looks as though some further clarifications might be useful.

The point of my article is not to contest whether cognitive abilities are heritable. Clearly they are. The point is that this fact does not necessarily imply it is meaningful to therefore talk about “genes for” that those abilities. Geneticists might argue that this is a straw man – that they recognize very well that the “genes for” trope is often unhelpful. This is true up to a point, but my argument is that this recognition came later than it needed to, and that even now the narratives and rhetoric used even in academic research on genetics and genomics fails to distance itself sufficiently from that legacy.

As for the issue of genes and IQ, I’ve said already that the evidence is contended, and it seems clear that there is a fair degree of polarization on the matter that is not helping the debate.

First, I should state clearly that there is very good reason to believe that, whatever cognitive abilities IQ tests measure, these abilities are inheritable to a significant degree. Of course one can argue about what exactly it is that IQ testing measures, and the limitations of those tests are widely (and rightly) advertised. But whatever the rights and wrong of using IQ as a measure of “intelligence”, or of using academic achievement as an indication of a person’s intellectual capacity, I believe that such “formal” measures tell us something about an individual’s cognitive abilities and that there is reason to believe that these are to some degree genetic.

There is nothing very surprising about that. The question is what the implications are – if any – for educational policy. One can debate the precise values, but there seems to be rather good reason to think that (1) academic achievement is strongly linked to socioeconomic status (SES), and (2) the relevance of genes to academic achievement becomes less evident as socioeconomic status declines. In cases of significant deprivation, it appears that the impact of genes on IQ might be close to negligible. A likely explanation for this is that, the fewer learning opportunities you have – the more obstructive your environment is to learning – the less you are going to benefit from any advantages your innate cognitive skills confer, since they are less likely to have a chance to manifest themselves. In an excellent learning environment, in contrast, innate differences are able to manifest themselves much more clearly. Again, there seems nothing terribly surprising about this. But the implications for education seem fairly plain: only when opportunities to learn are equal will one see clearly the extent of inherent genetic variation. This is of course very different from saying that when everyone has the same opportunities, everyone will do more or less equally well.

What this means is that, if a child with low SES is doing poorly at school, it would be wrong to say “well, obviously it is because of the disadvantages he faces through being poor.” It is possible that he has inherited limited cognitive resources that would hold him back even if he came from a wealthy background. That sounds harsh, and one might wish it wasn’t so – but it needs to be acknowledged, and it seems (though it’s hard to tell) that one of Dominic Cummings’ aims was to do so. But there are two crucial points to make here:
(1) If a child is doing poorly academically, it seems more likely to be due to inherent cognitive limitations if he has high SES than low SES, because in the former case genetic propensities are less masked by other factors.
(2) While the aforementioned statement might be true for individuals, it seems extremely unlikely as a generalized statement: one would expect inherited variations in cognition for low-SES children to be no different to those from high-SES children, so that any difference in academic achievement seen on average between these two groups would result from social and not genetic factors.

But wait – is that necessarily true? Might it not be that there is a correlation between low SES and genetically determined lower cognitive ability? To put it bluntly, might the poor be inherently dim – indeed, might not the causation actually work that way, i.e. they are poor because they are dim, not that they underachieve because they are poor?

Cummings seems to be raising this possibility. I say this warily because he seems to feel that he is constantly being misunderstood, but I can’t see how otherwise to interpret this passage in his paper (p.74):

“Raising school performance of poorer children is an inherently worthwhile thing to try to do but it would not necessarily lower parent-offspring correlations (nor change heritability estimates). When people look at the gaps between rich and poor children that already exist at a young age (3-5), they almost universally assume that these differences are because of environmental reasons (‘privileges of wealth’) and ignore genetics.”

If genetics is the reason for these gaps, he seems to be saying, then pouring money and resources into education for low-SES groups won’t help all that much, or at least won’t eliminate the gap. He certainly doesn’t say that we should not spend in that direction. Indeed, he suggests that one could argue we should put in more resources to the education of children from poorer backgrounds, precisely because they are more needed – but that we’d need to be realistic about the expected outcomes.

OK, so there are two ways to look at this. One is to say that surely hypothesizing that the correlation between low attainment and low SES is due to genetics shouldn’t be ruled out simply because it is “politically incorrect.” I’d agree with that. But it frustrates me endlessly that there are many scientists who would be content to stop there, i.e. with the notion that there are no questions in science that should not be asked. Science doesn’t work that way. Scientists don’t sit around dreaming up any hypothesis and then testing them. “I wonder if there might really be a race of aliens like the Clangers?” “I wonder if African nations really do have a genetic predisposition to bad government?” (Hello Nicholas Wade.) Hypotheses need to be motivated by observation, but they should also acknowledge what seems plausible on the basis of what we already know – yes, there are problems of governance in parts of Africa, but it is very hard to see how a genetic basis for that might arise, so why would you want to start from that possibility?

This was why I was so dismayed by the response of some scientists to the outcry about James Watson’s suggestions of a link between race and intelligence. If he had been proposing to set up a research programme to explore whether there were correlations between intelligence and some kind of genetic correlates of what we socially recognize as race (if such exist), then one could understand (just about) why criticisms of his comments and cancellations of his talks might seem like a kind of “politically correct” censorship, as some of Watson’s supporters asserted. But not only were Watson’s remarks apparently unmotivated by any well established scientific observations or theories, they were not advanced as a scientific hypothesis at all – they were supported by idle, bigoted anecdote.

Much the same applies here: given the apparent lack (to my knowledge) of any suggestion in the literature that the reasons that social class predicts educational achievement are genetic, it seems a strange (one might even argue dangerous) hypothesis to present a priori. Unless one has an underlying agenda, I’m puzzled by why one would suggest such a thing.

In any event, my real point was that, give the current state of play about links between genetics and measures of “intelligence” (for example, see this recent review), it isn’t clear why the matter should be terribly relevant to education policy at all. Cummings makes a big deal of Robert Plomin’s work – which, I hasten to point out, is totally respectable and worthwhile. Plomin’s hope is that, if one could identify genes linked to cognitive and academic abilities, it might be possible to personalize education to the individual’s inherent intellectual endowments. This is a reasonable and indeed praiseworthy goal. But what we have discovered so far on the matter seems to indicate that such ambitions are not only far from being realised but potentially misplaced in the first place – because there may not be a small set of genes that will predict those endowments. And that might be because we are thinking the wrong way – too linearly – about how many if not most genes actually operate. Personally, I find a little chilling the idea that we might try to predict children’s attainments by reading their genome, and gear their education accordingly – not least because we know (and Plomin would of course acknowledge this) that genes operate in conjunction with their environment, and so whatever genetic hand you have been dealt, its outcomes are contingent on experience. Given the current lack of a strong link between specific gene variants and IQ (and not for want of looking), right now it seems unlikely that we would learn anything more from genomics than smart teachers already discern from noting the strengths and limitations of each child. What’s more, evidence from neuroscience that early experiences affect the wiring of the brain make it seem far more profitable to place emphasis on a child’s learning environment (including that within the family), and not to worry about what your genes “say” about your future abilities.

It’s of course important to recognize that this situation might change in the future, but there seems little reason right now to anticipate that it will. This is why I feel that raising the matter of genetics in educational policy right now seems like a red herring – and worse, that it encourages this misleading and potentially damaging notion of “genes for”. Genetics should certainly not be a taboo subject in discussions about education, but I’ve yet to see a convincing argument for why it has anything to tell us about how to help children learn.


Stuart Ritchie said...

Thank you for the clarifications.

Evidence supporting the argument you make about genes becoming less important for IQ in lower SES brackets does appear in some studies, but there have also been several failed replications of the effect (in the UK:; the Netherlands; and the US: just as three examples among many others).

Even in the papers that find a gene-by-environment interaction, it's very rare for them to show genetic effects are 'negligible' in lower SES groups. The finding you refer to (the Turkheimer paper from 2003) is from a small sample, and has subsequently been found to be an outlier in the size of its effect.

It's a shame that you didn't properly research this area before stating a tentative, unclear, and possibly nation-dependent finding from a single, small study as absolute fact. Perhaps this sort of sloppiness is one reason people familiar with the science get 'touchy' when they read your articles.

Philip Ball said...

The 2012 Turkheimer review that I cite discusses the issues you raise about replication (in a more balanced way than you do). In sny event, so there is some ongoing debate about one sentence in my original article - and one incidental to the main point. I'm not sure what you are trying to prove, or why. And which other "articles" of mine are you referring to?

Stuart Ritchie said...

So what you're saying is, you already knew about the issues of replication, but didn't mention them to your readers? That's not very impressive.

I'm not sure how I could have been more balanced: I said there were findings on both sides and that the issue isn't resolved. That's the truth, I'm afraid. I wish you had conveyed it.

I'm not out to prove anything: I agree with you that the case hasn't been fully made for the implications of genetics for education (and have argued this in print). I'm just a stickler for getting the research right, and you've made the argument a few times in your blog articles here that the gene-environment interaction is definitely the case, without giving any balance.

Philip Ball said...

No, you’re right, we’re being too definitive about this particular issue – me for implying that the case for SES-dependence of the heritability of IQ has been established beyond question, Cummings for suggesting that it can be ignored. This means that my article would certainly be improved by my changing the statement “the association with genetics is much weaker in children from poor backgrounds” to read “some studies have suggested (although others were unable to corroborate it) that the association with genetics is much weaker in children from poor backgrounds”. I will make that change, and it’s good to have cleared it up.

Whether your rhetoric such as “this sort of sloppiness is one reason people familiar with the science get 'touchy' when they read your articles” and “[it’s] one of those articles proving that a small amount of genetics knowledge is dangerous” was all because you just wanted to be a “stickler” about getting this little point (peripheral to my main argument) sorted out is something I’m very happy to leave others to decide. However, if that was really all you wanted, you’ll find that a different attitude works wonders. Just saying.