Monday, July 8, 2013

The significance of (looking for) genes for educational achievement

Genes for educational achievement? So claims a paper in the 21 June issue of Science.
A genome-wide association study (GWAS) of educational attainment was conducted in a discovery sample of 101,069 individuals and a replication sample of 25,490. Three independent single-nucleotide polymorphisms (SNPs) are genome-wide significant (rs9320913, rs11584700, rs4851266), and all three replicate. Estimated effects sizes are small (coefficient of determination R2 ≈ 0.02%), approximately 1 month of schooling per allele. A linear polygenic score from all measured SNPs accounts for ≈2% of the variance in both educational attainment and cognitive function.
Dan Graur (of Encode critique fame) sums up the import of this paper very nicely, in 12 languages even, here. Here's the gist of his argument:
The German-born American Architect Ludwig Mies van der Rohe famously proclaimed that “God is in the details.” So, let us look carefully at the findings. All the 2,515,021 autosomal SNPs explain 2% of the variation. The largest estimated SNP effect size was 0.02%. Thus, the best genetic variant in the population explains 0.02% of the variation, and the addition of 2.5 million SNPs only adds 1.98%. By extrapolation, using the entire genome, would optimistically explain 3% of the variation in educational attainment.
I was positively floored with excitement.
His multi-lingual critique?
In the following, I shall attempt to summarize the findings of this study in all the languages of all the authors of this article: niets, nothing, mitte midagi, nix, ei mitään, ingenting (Norwegian), ingenting (Swedish), rien, ekkert, niente, τίποτα, and لا شيء.
Graur being Graur here, but he does have a point. The amount of variation explained is on the low end--even for GWAS!--so it's pretty clear that Science didn't publish this thing for its scientific impact. They published it, surely, for its splash factor. Evolutionary psychologists, behavioral geneticists and so forth will gush they're finally making headway: Now we're beginning to see the real truth about human intelligence! Finger-pointing at generations of IQ-testing naysayers. And think of the billions of dollars they'll demand to follow up these strikingly powerful results.

We have often criticized the spending of public money on genetic studies of traits or diseases for which environmental influences are much more significant -- heart disease, asthma, some psychiatric diseases, stroke, and so on, and educational attainment is surely another such trait.  Even just from a genetic point of view, the contributions of specific genes are generally environment and context dependent, and individually trivially uninformative.  Taken to an extreme, imagine someone with no opportunity to attend school at all, or to go to college -- no matter his or her genetic make-up, and let's even suppose it's completely determinative, this person will not reach his or her genetically determined potential.

Other factors
At the same time that this study appeared, a paper was published in Archives of Disease in Childhood, reporting on a factor with a large effect on social mobility. Analysis of data from the 1950s and 70s in Britain suggests that infants who were breast fed were 24% more likely to have moved up the social ladder (their job in adulthood compared with their father's job when the subjects were age 10 or 11), and 20% less likely to have moved down.

The authors suggest that breast feeding is associated with improvements in neurological development, cognitive improvement and emotional stability. They controlled for selection bias influencing who chooses to breast feed, so presumably it's not confounding variables that explain their results.

Maybe. But whatever it is that increases social mobility, whether it's breast feeding itself, or something associated with it, or even something unmeasured, something does, and it's unlikely to be genes for social mobility (or educational attainment, which is highly correlated with social mobility). For whatever reason, breast fed kids are 24% more likely to move up the social ladder -- that swamps the effect of whatever alleles reported by the educational attainment GWAS. Reinforcing the idea that whatever genetic component there is to winning the social race, it's miniscule compared with social and environmental factors. But will such papers, not heralded and published in Science, have any tempering effect? Unlikely.

Elementary school classroom, USA; Wikimedia
Breast fed or not, it is clear that social factors can have a large effect on educational attainment -- very early childhood education, small class sizes, good teachers (i.e., money), involved families and so on. We know that there are gene variants with major effect on cognitive development -- Fragile X, PKU, Down syndrome, e.g., there are perhaps gene variants associated with exceptional musical ability and there must be as yet unidentified alleles associated with exceptional intellectual ability as well. Just as there are genes associated with extreme short and tall stature, but hundreds if not thousands of genes associated with the rest of the distribution, the same will be true of cognitive ability. Your mother's nutritional level during pregnancy, and what and how much you ate as a child has a lot to do with your final height -- birth cohorts have grown taller over several generations, while genetic contributions haven't changed.

The morality of such studies
There are a number of issues that acceptance of such a study, much less acting on it or hyping it as important, raises. Such studies have societal impact of all sorts. They can affect how funds are spent and how people are viewed in terms of their inherent worth. Many scientists, especially those benefiting from the grant largesse (and this is not too harsh a way to put it) take the standard amoral position and say that these are issues for society to decide, and of course we must do more and bigger studies of this subject. Science is about finding the facts. You can't outlaw what you don't like about the real world, and the facts are what they are. Science's job is to find the facts. Leave us alone and go make your own social judgments in the political arena.

We'd like to suggest that studies of the genetics of intelligence are not just convenient abstract amorality. They are dangerously immoral. Just as GWAS of diseases like cardiovascular disease or type 2 diabetes, that take the focus off lifestyle which clearly has a large effect on risk of disease, emphasis on the genetics of intelligence takes the responsibility from society to ensure that each child meets his or her intellectual potential.

It is hard to imagine that this study isn't going to be just one of many more. While we don't think they will ever find much of significance, others believe otherwise, and one can't dismiss the likelihood that they will, or are eager to, interpret them differently. Indeed, GWAS believers have declared victory with studies of many other traits where we see pretty trivial results. But studies of the genetics of intelligence are almost certainly to be used to evaluate people's inherent intellectual worth -- the most fundamental worth, criteria that have widespread formal and informal use in our society

We aren't the only people to worry about a new era of intolerance. Voting rights are being challenged in the US. How long will it be before we see a new era of using such data for group characterizations whether formally or not--that is, racial judgment based on 'scientific' documentation of different average inherent mental abilities? And similar conclusions are always waiting in the wings to be used between as well as within nations. Have we not learned the disastrous lessons of the eugenic era in the last century? If you think this time it's different, then you haven't read enough of what was written a century ago.

This kind of non-sensical study should not have been funded, and should not be published in a major journal as if it were more than a non-result gussied up to seem profound. We may not be able to censor research formally, but we don't, as a society, have to pay for it.

One might say that this is a heavy duty over-reaction on our part, to a rather bland (if hyped) study that found hardly anything at all. Why get all heated up and panicky about a return to biological rationales for segregation (and worse)? Our response is that, as the Germans learned in the 1930s and later wrote widely about, the time to keep the horse in the barn is before it has inched its way out, one un-noticed step at a time. But lessons have been forgotten, and distant amorality is again offered, often with pious expressions of benign intent (again, as they were a century ago).


Stuart Ritchie said...

This post is written as if there wasn't a whole century's worth of twin and adoption studies showing that intelligence and educational attainment are under substantial genetic influence. Here's just one relatively recent paper confirming this:

The debate as to whether there are genetic influences on education and/or intelligence is over, and was over a long time ago. Now the search is on for the specific genes, and as this Science paper shows, some modest progress is being made. That there are also environmental influences on these traits is another question entirely, and nobody is denying that these should be studied too (in fact, that's what I'm currently doing).

I'm slightly disturbed by the discussion in this post of the author's fear of 'racial judgements' based on intelligence/education GWAS results. The author seems to be arguing that there exist genetic, racial differences in intelligence, a rather presumptuous (and some would say pretty offensive) statement given the lack of evidence for such differences.

Finally, the conflation of genetic studies of education and intelligence - which could in future be used to identify and help children at risk of learning disabilities, which is the explicit aim of many of the researchers in the field - with the Third Reich is outrageous, and is far beneath even a blog like this one.

Ken Weiss said...

We didn't expect a neutral reaction to this post. Finding trivial gene regions that affect intellectual achievement is not the same as studying major genetic effects, and there are tens if not hundreds of such genes already known.

This kind of intrusive prying by geneticists is not new and anyone is free to read the last century's early writings on this subject, writing and theory that later led to the abuses, step by step.

Anyone who does not think such results can be, and are likely to be, extended to group differences also is looking at the world through rose (or is it grant?) colored glasses.

Almost every trait has substantial heritability, and there is as you say plenty of evidence that there there are such effects in intellectual achievement. But they are dangerous to try to enumerate in this way, as the past has shown, and we know that the major effects, by far, for the normal population are sociocultural and economic, not genetic.

The drumbeat to list genes that affect, no matter how trivially, any trait one can imagine, and then to trumpet these findings as if they had importance, is a sign of the way things are being done in our society. Much more money will now be poured into this area, likely, and we happen to think it shouldn't be done. If one wants to study learning disabilities, one doesn't need mega GWAS to do it.

Of course, that's just my view. You're welcome to yours. At least, the propriety of this kind of study should be debated, with all the plausible societal issues raised, not just some possible salubrious uses of the results.

Anne Buchanan said...

Among the questions that our post and your response raise, Stuart, is the important one of the influence of politics on science. You and we have read the same paper -- you see progress, we see nonsense. From our post you surely can glean a lot about our politics. As, I think, we can from your comment. That science, and the interpretation of scientific findings, is not done in a vacuum, but rather by people with values and beliefs, is exactly why we worry about the use and misuse of the 'science' of intelligence.

Unknown said...

Debate is good! I just don't think scientific debate is particularly well advanced by (a) listing the word 'nothing' in several languages, as the puerile Graur does; or (b) comparing one side to the Nazis, as your post does. That's not too controversial a position, is it?

You put 'science' in inverted commas, but you haven't provided a scientific critique of the GWAS paper, or any reason for us to think it is 'nonsense' or in any way unscientific.

(As an aside, what could you learn about my politics from my comment above, exactly? That I am disturbed by your apparent belief in genetically-determined racial intelligence differences? That I devote my energies to studying environmental effects on education and intelligence?)

Ken Weiss said...

If people were sincere about raising educational achievement they could make vastly greater gains by raising the standard of schools of education, rewarding teachers better, investing heavily in areas where education is failing, and the like. We all know the issues.

Technologizing the subject as being largely genetic is a way of satisfying professional interests (of professors) rather than really addressing the main problem. In a sense, it is a way of providing excuses to legislators why they need not vote major funding for needy areas: because scientists tell us that this is genetic, not societal!

I would say that if one simply wants predictive evidence showing which kids to help, all one needs to do is use parent-offspring correlations, which are far more informative.

Unfortunately, that is problematic in ways that show the epistemic hollowness of studies like this one: heritability changes. If its genomic basis, which hasn't changed enough to make much difference during say the past century, the reason that h^2 has changed is that genetic effects are context-dependent. Therefore, GWAS-like predictions must be context-dependent as well, and their retrospectively ascertained location-specific risk estimates are of unreliable and to some extent unknowable accuracy, even in principle.

As to group differences, of course there are group differences--no matter how one defines or samples the group, if one does a large enough study. That doesn't mean they are major, exhaustively huge studies can always find something, and 'significance' is subjective and history shows the tendency of scientists to peer irresistibly into the subject, and there are a number of them around today, including prominent names who have the media's ear, who are very interested in showing group intelligence differences.

Nobody is comparing today's GWASers to Nazis. The point is that this is an incremental kind of area that, history has shown clearly, is highly vulnerable to abuse, by leading scientists based on benign social-good rationales. And it's not necessary to geneticize the subject, for the above reasons, if the true objective is to help student achievement.

Anne Buchanan said...

Ah, if you search this blog for other GWAS posts you will see it's not the method per se that we critique, it's the belief that it's yielding important results. As here. And the continued commitment of large amounts of funding to a method that has clearly not yielded the promised results. Yes, Dan has a singular way of putting things, but he does make some good points.

The impression I got from your previous comment is that you are more sanguine about genetic determinism than I. If I misread your comment, I apologize.

Stuart Ritchie said...


1) Nobody disagrees with this, and vast sums of money are being spent on researching how to improve education. That isn't an argument against looking at genetics, which we know from twin studies are majorly important. The two are not mutually exclusive.

2) If by 'professional interests' you mean 'intellectual interests', then yes! The question of how genes make brains, and make smart brains, is a fascinating one. Given the substantial genetic influences on education, we might need to rethink how we do things, not ignore the genes and continue doing things the way we always have. This paper is particularly interesting in this regard:

3) Agreed. But that's in part due to genetics, so logically you should agree with me that a better understanding of how the genetics of ability/attainment work would make identifying these kids easier (nobody is claiming that such an understanding is nigh, by the way. Just that getting the basic science done now is useful).

4) Well, heritability might change. Are you referring to the GxSES papers? Sometimes it's found that SES moderates heritability, but very often no effect is found. See, e.g.

5) Too vague to respond to - who are these prominent names who want to show group differences?

6) Oh do come on - the final paragraph of the post above is heavy with innuendo, essentially arguing that these studies are the first step down an inevitable slippery slope towards a repeat of Nazi eugenics. I don't think anyone would read it and not come to that conclusion. I think that's not just an offensive argument, but a pretty unconvincing one.

Holly Dunsworth said...

Well put Anne.

Bau Dong said...

I think that individual genes related to IQ may have such small effects as to be rather uninteresting, but casting ethical aspersions on the researchers (as the authors of this piece have done) is pretty silly and pretty distasteful. In the US we have a problem with Republican congressmen trying to block research they find odious--the attitudes expressed in this post are pretty much identical, except for the choice of what to find most distasteful. Too bad we can't have respect and rationality.

Anonymous said...

The paper (1) provides scientific evidence that intelligence is likely to be extremely polygenic, (2) further demonstrates that the genes involved will probably all have small effect sizes, and (3) thus refutes entire decades of "successful" but spurious candidate-gene studies of intelligence. How is that not a worthy set of findings?

Holly Dunsworth said...

This comment thread demonstrates exactly one of the main points of the post: There are human geneticists who are succeeding at their research without being aware of context.

Anne Buchanan said...

Let's return to the important point. Just as we ask with respect to heart disease or type 2 diabetes, or asthma or many other traits with strong environmental components, the bottom line question is this: Why spend so much time and money on the genetics of intelligence, when the environmental influences are large, manipulable, and much more tractable?

And, we're not the only people who think about how easy it would be to go from here to a new era of eugenics. Again, science is not free of context, and in the US at least, it's not farfetched to think it's possible.

Holly Dunsworth said...

should say: "significant, anthropology-based (and evolutionarily-based as well)" context.

Holly Dunsworth said...

This. "Why spend so much time and money on the genetics of intelligence, when the environmental influences are large, manipulable, and much more tractable?"

Ken Weiss said...

We did not, I think, cast aspersions on the authors of this paper, though we did criticize the attention being given to the results. What we said is that this is a dangerous area to go into, and if the objectives are as some comments say then this is not the kind of study one needs. One maybe should ask Republicans to invest in education, which is risky since one might suspect that people they don't like might benefit.

There are many limits to what science can be funded to do, and it is perfectly legitimate to ask whether this kind of study should be funded.

These issues are never settled by academic arguments, for better or for worse; they are sociopolitical issues.

There are plenty of clearly genetic causes of major educational impairment. Maybe the genetic resources should primarily go there, to develop effective remedies. Maybe society could find lots of boondoggle spending that could be stopped and redirected towards educational improvement generally. We certainly need it.

Anne Buchanan said...

We have failed here if it seems that we are casting aspersions on the authors of this piece per se. Rather it's the idea that the genetics of intelligence is something that society should be investing in, when social, environmental influences on educational attainment are so much greater. This is true of so many traits, including chronic diseases, but the belief that genetics will solve these problems doesn't wane. To me, this is what is irresponsible.

Stuart Ritchie said...

The actual question here, if I may rephrase your comment, is:

"Why spend so much time and money on one massively important and interesting question when another massively important and interesting question--which already gets huge amounts of cash and attention from governments and private bodies worldwide--is already being answered by different scientists working in different contexts? Especially when the extent to which the latter question can be answered is dependent upon answers to the former one?"

Doesn't seem such a killer question now.

Joe Pickrell said...

The authors have a "FAQ" about their results here:

In my reading, it seems like they have a reasonable understanding of both their research and the context. I think this post could be summarized as "these researchers and I disagree about whether the genetics of intelligence is an interesting topic".

Stuart Ritchie said...

Is there a contradiction between your stance that the GWAS results are meaningless, and your stance that genetics might be used for nefarious eugenic ends? If one is right then the other can't be. Or have I missed something?

Anonymous said...

The point made by the article is that the variation in educational achievement due to genetic heterogeneity is swamped by other sources of variation.

I disagree with your assertion that twin studies show that genetics is majorly important for the educational achievement. Virtually all human traits will have a genetic component, so will show up positive in twin studies, but that does not mean that we should expect to find large amounts of genetic variation underlying the population level differences in the trait being measured.

I doubt very much that looking at someones genotype will be helpful in identifying children at risk of underachieving. The resources required for such an undertaking would clearly be much better employed in other ways. The Science paper clearly shows that.

On the whole using genetics as a predictor of complex human traits, ignores the huge ignorance that we have about how genes influence human behaviour generally, confuses necessary with sufficient and does genetics a diservice.

JayMan said...

I was very disappointed with this post. I initially thought it would be another discussion of the issue of "missing heritability", that is, the gap between the known very high heritability of IQ and other traits (as Stuart Ritchie points out) and the paltry findings from GWAS analyses. Instead, this posts goes off on a moralistic tirade similar to Kevin Mitchell's post decrying eugenics. Indeed, this is an even less serious enterprise, because Mitchell from the outset only intended to evaluate the ethical implications of eugenics and he certainly didn't make egregiously false claims like those made here.

I think the factual objections are well covered by Stuart Ritchie's comments. I would however like to add that whatever the situation is with genetic links to IQ and group differences in IQ and other psychological traits, there is a reality there whether you like it or not. To suggest that we should not investigate this reality, as John Horgan did, would be to effectively suggest that we should stop conducting science, as Greg Cochran succinctly explains. Obviously, that's a foolish notion.

Holly Dunsworth said...

Criticizing science is not disrespectful. It's how science is done.

Publicly funded science for the sake of science isn't a shared belief of all scientists.

And to require all scientists to believe that it is, is being just as political as the Republicans.

Anne Buchanan said...

Thanks, yes.

Stuart Ritchie said...

And so that huge ignorance should be addressed by... NOT doing more work attempting to understand the genetics of complex traits?

Anonymous said...

Sorry, my last comment was aimed at Stuart Ritchie. This thread has suddenly become very busy.

Anne Buchanan said...

Here's where social context comes in. As we say in the post, while we don't see significant findings here, others (such as you) do.

Ken Weiss said...

I think this exchange series is going nowhere fast. People don't agree about these sorts of things.

What I would say you have 'missed' (but I wouldn't have put it that way) is that there is already huge investment in similar kinds of research, with very widespread promises (often with deniability caveats here and there) about personalized genomic medicine and the like. That means individual predictions.

Even with weak evidence, the belief system that genomes provide widespread and accurate predictive power, and the interests that defend 'Big Data' approaches, persist.

Many subjects are interesting. Many subjects can be studied in a harmless or beneficent way. But some subjects can be abused, or resources invested in more predictably productive ways.

I think that if you read carefully from the eugenics era you would see that it was a slowly creeping stepwise path towards horrendous abuses. Whether Santayana was right in any rigorous sense or not, history's lessons are worth paying attention to, in my personal opinion.

When we have real solutions to the genuine genetic intellectual impairments, and when we have truly done all we can to improve funding and societal equity, the things that make the major differences in educational achievement for the nonclinically impaired population, then what's left may be more genetically cogent and worthy of study.

But I think this is the end of any useful exchanges about this, since clearly we don't agree.

Anonymous said...

No-one is (and I'm certainly not) arguing against doing more basic research! This is about how science is applied to policy - i.e. should we genotype children to find underachievers (a waste of money), or invest in social programmes for which there is very good evidence that they have a positive effect.

Holly Dunsworth said...

Not saying you're naive. But back when I was more naive than I am now, this is how I reacted to Ken and Anne's perspective they shared with me/us back in grad school. Having read a lot about the history of science and anthropology and learned quite a bit more about genetics and about behavior... I completely shed this reaction, this almost instinct to prize Science at all cost.

Stuart Ritchie said...

A false dichotomy, of course. As I said above, identifying kids who need extra help is something that will only be possible waaaaay down the line, and is certainly not the only possible future benefit of this research (and all this is above and beyond just knowing how genes build intelligent brains, a worthy goal in itself).

Nobody is saying 'do genetics and not social stuff' - as I've said several times, I mainly do the social stuff myself (and I've never touched a GWAS - in fact, I wish some gosh-darn GWAS people would get in here and defend themselves!).

Holly Dunsworth said...

Don't tests and quizzes identify those that need extra help? Not trying to be too glib but for identifying those with more subtle needs (than textbook disorders, etc), that's just basic education: If they're not getting it yet, then they need more time and help and training. And that they don't get it is because we don't have infinite resources. It might be my limited imagination, but I can't imagine how GWAS within a normal population will give better strategies than *spend more quality time with children to help them learn better.*

Anonymous said...

I seriously doubt that common genetic variation (as opposed to rare alleles, which GWAS cannot find) in educational attainment will ever be shown to be significant enough to be worth spending ANY money on compared to the other, demonstrably significant, sources of variation.

This article pretty much sums up my view of the issue:

Holly Dunsworth said...

And of course, feed them better and treat them with better healthcare, etc etc etc...

JayMan said...

Believe me, I have given the impact of knowledge of the biological and genetic roots of behavior more than a little bit of thought.

That said, in the end, I don't think there is a such a thing forbidden knowledge (obviously, since I blog about this). As many the commenters in the previous links would note, (and indeed, Cochran himself) the implications of not researching or acknowledging this topic can be quite dire (and are not headed in a good direction at present) – indeed, possibly quite worse than the fears discussed in this post – which were speculative anyway.

As I and others (particularly Razib Khan) told Kevin Mitchell when he wrote his post, the potential benefits of knowledge of the genetic roots of important traits to individuals is quite large, perhaps even largest for those on the lowest end of the socioeconomic scale. But if we are venture into speculation of what would ensue as science gains knowledge of the genetic links to important human traits, then my two cents on that will be that it probably would be a bit messy, because, unfortunately, such is the nature of man. Though I suspect the worst fears echoed here will not come to pass.

Holly Dunsworth said...

Aren't we supposed to cheer when negative results are published? If so, HOORAY FOR SCIENCE for doing science a service.

Kevin Mitchell said...

I strongly agree that the results of the GWAS on education are effectively useless. Not completely, because it's actually a very convincing negative result from a huge sample, with a strong implication: there are no common variants making even a modest contribution to the genetic variation affecting educational achievement.

And I share the authors' frustration with the GWAS approach in general, which I think has indeed been massively hyped far beyond the degree to which it has paid off in terms of actually useful discoveries. (Though I know many disagree and those opinions may yet be proved right).

But I strongly disagree with the idea that studying the genetics of intelligence is somehow immoral, in itself. There is a wealth of evidence that intelligence is a real trait, reflecting real, innate differences in intellectual ability or potential. There is also a wealth of evidence that many people don't come close to reaching that potential to due highly unequal access to education. (And that changes in nutrition, education and other social factors have led to an increase in IQ over past decades). Those are not contradictory points. They are both important, certainly both scientifically interesting and valid areas of study.

I don't think GWAS will get to the genetics of intelligence, because they are looking in the wrong place. ( But this is an interesting and important topic to understand, if only because of what it will tell us about how the brain develops and operates.

We can legitimately investigate that without necessarily leaping to the idea of selecting embryos based on predicted intelligence, which some researchers seem to be advocating (

Anonymous said...

Good point.

Holly Dunsworth said...

(a point made simultaneously but with better wording by Anonymous who posted far far up there)

Ken Weiss said...

I would just say because we had diverse data already that showed these points very clearly, also that intelligence is not even easy to define (or, rather, is not a unitary trait), and therefore we did not need to spend the funds and divert attention from working on the major aspects of the problem that we know so well about.

And we DO know that there are specific genes that, when mutated in particular ways, have major effect on educational abilities or achievement. Funds should go to the science of trying to see what can be done about that.

Finally, it provides grist for the mill of those who want to stress the importance of group or even of individual differences in a policy-related sense.

But the work has been done and it will be used by countless investigators to do even more extensive genomics, in various ways, I predict. It will be a boon for BigPharma and the therapist industry who can 'treat' those with a few months' lag time predicted by their genomes. Etc.

Anne Buchanan said...

Thanks for your comments, Kevin. One thing that needs to be mentioned, I think, is that there are limited resources for science. Society can't invest in research on everything. This is the age of genetics, but it's clear that GWAS of almost anything aren't a great investment. Intelligence (however you define it) is clearly due to polygenes and environmental factors.

Indeed, Ken and I have been writing for years about this with respect to most other traits. Why spend so much money on the genetics of heart disease, or asthma, or schizophrenia etc? Intelligence is no different. Except that it's a loaded subject, as the comments on this post show.

Ken Weiss said...

I can only add that these kinds of results were explicitly predictable, even before the GWAS era was in full-capturing of resources. Some of us did that in writing, as long ago as 1993. So that explains, if perhaps a bit in terms of vanity, the strength of our feelings about the way resources are being used, or squandered in the name of various excuses and vested interests, when more serious problems remain under-studied.

Kevin Mitchell said...

I also think that GWAS are not the most useful way to spend scarce scientific funds. But conflating genetic research generally with GWAS does the wider approach a disservice. GWAS are intellectually bereft, in my opinion, because, among other things, they ignore evolutionary genetics, which suggests that common variants are common because they're not doing much of anything. But geneticists taking other approaches really are finding rare mutations predisposing to some of the conditions you mention, especially schizophrenia. These discoveries are providing real insights into the complex etiology and pathogenesis of such conditions and also reinforce their extreme genetic heterogeneity. This further undermines any approaches - especially GWAS - which treat them as single entities. (See here for more on this:

Anne Buchanan said...

I will not banish genetics entirely when I am queen. We have long said that genetics should legitimately deal with truly genetic disorders. But getting from the age of everything being genetic to only truly genetic traits being genetic is proving to be a long haul. (For those who didn't see our Aeon piece on genetic determinism, which might be relevant to this discussion, or at least help expand on our ideas a bit, it's here.)

David Colquhoun said...

Having only last night been through yet another IQ twitter storm with Stuart Ritchie, I'll keep this short.

It seems to me that IQ is correlated with all sorts of things, but the problem of causality makes the correlations impossible to interpret.

It seems self-evidently silly to suppose that the attributes of a human can be described by a single number (whether IQ or any other number).

The paper failed to discover anything useful about the genetics of IQ. That itself is interesting (but the hype isn't).

The fact that UK politicians were silly enough to listen to the exaggerated claims of psychologists in the 1930s led to selection of children at 11 and that did much social harm. Hype is not a game: it has consequences.

As I think rather often, whatever happened to the ability of scientists to say "I don't know. It's too complicated to understand"?

Ken Weiss said...

There's no magic answer. Many do argue that what we should be seeking is rare variants with major effect. That's like going back in circles, because that was the basis of genetics in the first 2/3 of the 20th century.

Clearly there are rare variants, when they are important they segregate in families, with high penetrance. That's not the general explanation for high heritability of complex traits. And segretation can arise in purely polygenic traits (that's been clear only since Falconer and others pointed it out in the '60s).

Common variants for common disease was the even-then BS rationale for mega genomic funding. It never had a serious theoretical basis, but of course there are instances where that's the case.

The retreat (never so acknowledged, of course) of some was to say that GWAS was a bust and what we need is even more whole-genome sequence data, and we'll find the rare variants and combinations of rare variants that are responsible for your-favorite-trait. But the role of rare combinations of variants is something discussed (and controversial) roughly 80 years ago.

Rare and hence essentially unique genotype combinations will not account in epistemically useful medical application, and probably have little evolutionary relevance, for various reasons.

That almost everything that's not cultural has substantial heritability is not at all new, and indeed was implicitly the basis of Darwin's argument for how evolution worked. But complex genomic etiology is not best studied by enumeration of individually unique genotypes even if (as we said in a series of posts not too long ago) each is highly determinitive.

The truth is that life is a mix of these and that we have no a prior theory for what that mix will be. And the truth of the structure of science is that me-too projects with predictable but incremental results and exaggeration of importance and the like, are the way our system works. That tends also to obfuscate what we know and to alter priorities from accurately reflecting that.

Anne Buchanan said...

Thanks for this. Yes, hype is not a game. There are consequences.

And indeed, sometimes "It's complicated" is the answer. We do understand.

Ken Weiss said...

Here I'd say that we're human, and we're these days bourgeois: we need our careers and status and so on. And, thanks to Thatcherism even in the UK you need a steady stream of grants. The bigger and less-endable they are, the better. And then big investment has to be justified.

Thinking originally is very hard if a problem is a substantial one.

Thinking of consequences of this kind of research doesn't lead to tempering. Thinking of history rarely works in human societies, hence our repeatedly falling victim to Santayana's warning about repeating past mistakes. Past abuses have resulted from otherwise seemingly beneficent research,but we usually only close the door after some horse has gotten out. Some of us are old enough to have direct memories of such things.

Even our IRBs are less than strict watchdogs, and tolerate all sorts of research that is thinly justified but causes all sorts of torment to animals, and is under pressure not to turn down faculty requests for human research either, certainly not if nobody is immediately harmed by it. We can always envison great positive benefits for work we propose.

But self-censoring for societally appropriate reasons is not very effective.
And many justify laissez faire because we can blithely say that our job is to understand Nature however she is, and censorship does, of course, have its negative side.

Mike Miller said...

The authors who planned this study intended to show that all individual genetic effects are vanishingly small and that studies finding large effects are not credible. Maybe the authors were reading Weiss. Read this:

Kevin Mitchell said...

First, for so-called "complex disorders", genetic heterogeneity completely confounded efforts to find really rare mutations, and our inability to define a biologically relevant phenotype (especially in psychiatry) confounds all attempts to define segregation patterns and penetrance. So early failures or apparent lack of Mendelian segregation do not argue against the existence of many rare mutations with large effects.

For quantitative traits, it is not clear whether the phenotype is more affected by a combination of (very many) common variants or by a much smaller number of rare mutations in each individual. Either way, the inheritance patterns would be very complex and would look pretty much the same. We have not had a method to look for rare mutations and assess their impact on phenotype to date but, given: (i) traits like intelligence are highly heritable, and (ii) common variants don't seem to explain much of this, it makes sense to look at the rare ones. That being said, there is another possible explanation for intelligence in particular, which is that it is a general fitness indicator, reflecting a very general mutational load, rather than (or in addition to) the effects of mutations specifically affecting (mostly presumably lowering) intelligence.

Mike Miller said...

Regarding funding -- how much did this study cost? It was piggy-backing on a series of other studies designed to find genes influencing other traits. This study should have been very inexpensive for its large sample size. It shows what it intended to show -- that Weiss is right about vanishingly small effects on complex traits. Given the pro-Weiss design, pro-Weiss finding and low cost, it's hard to see what you all are complaining about. Do you think the authors disagree with you on some point?

Disclaimer: I'm one of the 100+ coauthors.

Have you read their FAQ?:

R. Jones said...

Do not be afraid. "That which can be destroyed by the truth should be." -P.C. Hodgell

The brain is extremely complex and something like 80% of our genes are expressed in the brain, I wouldn't be surprised if an enormous number of SNP are involved.

Anne Buchanan said...

If this is in fact the report of negative results, it is certainly not clear to me. The title certainly doesn't suggest this ("GWAS of 126,559 Individuals Identifies Genetic Variants Associated with Educational Attainment"). I had read the FAQ, yes, and have read the Chronicle piece and they don't suggest it either. What I see in both of these is the usual "more research needed."

And, from the paper abstract: "Genes in the region of the loci have previously been associated with health, cognitive, and central nervous system phenotypes, and bioinformatics analyses suggest the involvement of the anterior caudate nucleus. These findings provide promising candidate SNPs for follow-up work, and our effect size estimates can anchor power analyses in social-science genetics."

This doesn't sound like a report of negative effects.

And, from the Supplemental Material:

"Exploring possible explanations for very small effect sizes

The effect sizes we find are much smaller than those found for other replicated SNP association results for complex physical traits such as body height (15), BMI (18), or metabolite profiles (47). In this section, we explore three possible explanations:

A. Measurement error attenuates the estimated effect;

B. The genetic effect is conditional on specific environmental circumstances, and hence a meta-analysis approach that averages across different environments partially masks the genetic effect;

C. “Biologically distal” phenotypes such as years of education have smaller effect sizes than more ”biologically proximal” phenotypes such as body height."

What this suggests, as I read it, are reasons why a true genetic effect was not found. It would have been if A, B or C hadn't been true. Again as I read it, there is nothing here that suggests that this paper is reporting a true negative result.

Ken Weiss said...

I'll make another few points about this study that critics of my views may find suprising, but that to me show even further why this kind of study, not just this study, has serious problems.

First, if the authors want to make the no-effects point, they could clearly say that, including in the title. That not the wording nor the tenor of the title.

Second, if this big a study can't account for more than a trivial fraction of the clear heritability (ignoring the amount of controversy about how much that is), then it is hardly very useful.

Third, related to the second point, there are tens (hundreds?) of genes that have been identified and in which mutations have substantial effect on cognitive ability (ignoring what that may mean), and essentially none of which were identified as having any effect in this study. This shows the problems with this sort of ascertainment and its interpretation. It tells us very little about genetics, even when we know from heritability that this trait, like any trait, is substantially affected by genes (in aggregate, even if no one gene has major effects).

Fourth, with this trait like others, it is almost certain that you can have major mutations including complete inactivation, individually, in some if not most of these genes and no cognitive problems. This is a general finding.

We are simply not taking what we know about genetics seriously. That's difficult and is not amenable to a steady flow of grant funds or of publications. We need to fix our system so we can think harder and learn from what we know.

Stephane said...

I will try to make a contribution to this debate,using a philosophical concept.
Lest imagine that the intelligence is determined by our genome. That would mean that the 'clever" or "stupid" decision we take are pre determined by definition.

Is this world only one good solution exist and preference and/or opinion do not.

In that case, no free will exist as our preference and opinion are determinate by our genome.

Does anyone can really believe we could analyze a gene which has been responsible for the best decision in the history of humanity ?

Should analyses the genome of Einstein or Churchill. Why not Hitler ? It was kind of intelligent as well ?

Anne Buchanan said...

At risk of torturing an analogy here, Stephane, I think your comment is interesting because the world you describe is basically the world as described by strict Darwinians. And I think your critique of ideas about intelligence is very apt. In the Darwinian's world, the one in which "survival of the fittest" rules, and there's one good solution to who survives to reproduce and evolve, the adaptive solution is chosen by natural selection. Less fit solutions don't exist, and there's no mechanism other than selection.

But a better description would be "failure of the frail" -- in this world there are many ways to survive and reproduce, and not all traits are here because of natural selection. Just as you note about many forms of intelligence and multiple good solutions, there's not simply a best, genetically determined way.

Stephane said...

Anne, thanks to answer my comments
The question is why scientific are fascinated by this social Darwinism ?
Is there any scientific argument against this elite legitimation through genetics.. Are we going to repeat the XVIII and those revolutions.
Is there not a scientific argument to totally separate the « biological » genetics which can serve to cure disease from this mad science which is social genetics ?
Anyway should explain to these scientists that a 2% explanation power is not a valid result in human science.

Anne Buchanan said...

I'm with you Stephane! Our post tomorrow, while not directly answering your questions, addresses the issues you raise.

Anonymous said...

That breastfeeding study? No correction for parental intelligence (which is likely the missing correlation given certain groups' issues with breastfeeding), and a stupid focus on 'social class', an increasingly useless variable. Worthless.

Anne Buchanan said...

So substitute studies of the effects of programs like Head Start, or even just the effects of having a good teacher. The point is only that many non-genetic factors surely have much more effect on success in school than shown by this GWAS.

RealityPlease said...

If people can divide into groups and exclude others based on genes pertaining to skin pigmentation, why couldn't they decide to exclude others based on other genetic markers defined as significant by the members of one such group? "Race" is a social fiction no matter how you define it in physical/genetic terms.

A real link between gene x and intelligence is not necessary to promote this view. A statistical association between the presence/absence of gene x and high intelligence may be enough to form a basis of discrimination. Later science may debunk the earlier research, but the damage would already be done.

Science has gone on a tear before hyping early results. People can react to that hype. Not all of the people reacting are scientists, & some have agendas that can be furthered by promoting it.

There are people who claim that everyone is looking out for their own self interest... except politicians, or everyone has an ideology... except themselves.