Intelligence is still heritable

By Razib Khan | June 12, 2013 1:23 am

Sir Francis Galton

Modern evolutionary genetics owes its origins to a series of intellectual debates around the turn of the 20th century. Much of this is outlined in Will Provines’ The Origins of Theoretical Population Genetics, though a biography of Francis Galton will do just as well. In short what happened is that during this period there were conflicts between the heirs of Charles Darwin as to the nature of inheritance (an issue Darwin left muddled from what I can tell). On the one side you had a young coterie around William Bateson, the champion of Gregor Mendel’s ideas about discrete and particulate inheritance via the abstraction of genes. Arrayed against them were the acolytes of Charles Darwin’s cousin Francis Galton, led by the mathematician Karl Pearson, and the biologist Walter Weldon. This school of “biometricians” focused on continuous characteristics and Darwinian gradualism, and are arguably the forerunners of quantitative genetics. There is some irony in their espousal of a “Galtonian” view, because Galton was himself not without sympathy for a discrete model of inheritance!

William Bateson

In the end science and truth won out. Young scholars trained in the biometric tradition repeatedly defected to the Mendelian camp (e.g. Charles Davenport). Eventually, R. A. Fisher, one of the founders of modern statistics and evolutionary biology, merged both traditions in his seminal paper The Correlation between Relatives on the Supposition of Mendelian Inheritance. The intuition for why Mendelism does not undermine classical Darwinian theory is simple (granted, some of the original Mendelians did seem to believe that it was a violation!). Many discrete genes of moderate to small effect upon a trait can produce a continuous distribution via the central limit theorem. In fact classical genetic methods often had difficulty perceiving traits with more than half dozen significant loci as anything but quantitative and continuous (consider pigmentation, which we know through genomic methods to vary across populations mostly due to half a dozen segregating genes or so).

Notice here I have not said a word about DNA. That is because 40 years before the understanding that DNA was the substrate of genetic inheritance scientists had a good grasp of the nature of inheritance through Mendelian processes. The gene is fundamentally an abstract unit, an analytic element subject to manipulation which allows us to intelligibly trace and predict patterns of variation across the generations. It so happens that the gene is instantiated in a material sense through sequences of the biomolecule DNA. This is very important. Because we know the material basis of modern genetics it is a much more fundamental science than economics (economics remains mired in its “biometric age!”).

The “post-genomic era” is predicated on industrial scale analysis of the material basis of genetics in the form of DNA sequence and structure. But we shouldn’t confuse DNA, concrete bases, with classical Mendelism. A focus on the material and concrete is not limited to genetics. In the mid-2000s there was a fad for cognitive neuroscience fMRI studies, which were perceived to be more scientific and convincing than classical cognitive scientific understandings of “how the mind works.” In the wake of the recession of fMRI “science” due to serious methodological problems we’re left to fall back on less sexy psychological abstractions, which may not be as simply reduced to material comprehension, but which have the redeeming quality of being informative nonetheless.

This brings me to the recent paper on SNPs associated with education in a massive cohort, GWAS of 126,559 Individuals Identifies Genetic Variants Associated with Educational Attainment. You should also read the accompanying FAQ. The bottom line is that the authors have convincingly identified three SNPs to explain 0.02% of the variation in educational attainment across their massive data set. Pooling all of the SNPs with some association they get ~2% of the variation explained. This is not particularly surprising. A few years back one of the authors on this paper wrote Most Reported Genetic Associations with General Intelligence Are Probably False Positives. Those with longer memories in human genetics warned me of this issue in the early 2000s. More statistically savvy friends began to warn me in 2007. At that point I began to caution people who assumed that genomics would reveal the variants which are responsible for normal variation on intelligence, because it seemed likely that we might have to wait a lot longer than I had anticipated. As suggested in the paper above previous work strongly implied that the genetic architecture of intelligence is one where the variation on the trait in the normal range is controlled by innumerable alleles of small effect segregating in the population. Otherwise classical genetic techniques may have been able to detect the number of loci with more surety. If you read Genetics of Human Populations you will note that using classical crossing techniques and pedigrees geneticists did in fact converge upon approximately the right number of loci segregating to explain the variation between European and African pigmentation 60 years ago!

Some of my friends have been arguing that the small effect sizes here validate the position that intelligence variation is mostly a function of environment. This is a complicated issue, and first I want to constrain the discussion to developed Western nations. It is an ironic aspect that arguably intelligence is most heritable among the most privileged. By heritable I mean the component of variation of the trait controlled by genes. When you remove environmental variation (i.e. deprivation) you are left with genetic variation. Within families there is a great deal of I.Q. difference across siblings. The correlation is about 0.5. Not bad, but not that high. Of course some of you may think that I’m going to talk about twin studies now. Not at all! Though contrary to what science journalists who seem to enjoy engaging in malpractice like Brian Palmer of Slate seem to think classical techniques have been to a great extent validated by genomics, it is by looking at unrelated individuals that some of the most persuasive evidence for the heritability of intelligence has been established. It is no coincidence that one of the major authors of the above study also is an author on the previous link. There is no contradiction in acknowledging difficulties of assessing the concrete material loci of a trait’s variation even if one can confidently infer that association. There was genetics before DNA. And there is heritability even without specific SNPs.

Additionally, I want to add one caveat into the “environmental” component of variation. For technical reasons this environmental component may actually include relatively fixed biological variables. Gene-gene interactions, or developmental stochasticity come to mind. Though these are difficult or impossible to predict from parent to offspring correlations they are not as simple as removing lead from the environment of deprived children. My own suspicion is that the large variation in intelligence across full siblings tell us a lot about the difficult to control and channel nature of “environmental” variation.

Finally, I want to point out that even small effect loci are not trivial. The authors mention this in their FAQ, but I want to be more clear, Small genetic effects do not preclude drug development:

Consider a trait like, say, cholesterol levels. Massive genome-wide association studies have been performed on this trait, identifying a large number of loci of small effect. One of these loci is HMGCR, coding for HMG-CoA reductase, an important molecule in cholesterol synthesis. The allele identified increases cholesterol levels by 0.1 standard deviations, meaning a genetic test would have essentially no ability to predict cholesterol levels. By the logic of the Newsweek piece, any drug targeted at HMGCR would have no chance of becoming a blockbuster.

Any doctor knows where I’m going with this: one of the best-selling groups of drugs in the world currently are statins, which inhibit the activity of (the gene product of) HMGCR. Of course, statins have already been invented, so this is something of a cherry-picked example, but my guess is that there are tens of additional examples like this waiting to be discovered in the wealth of genome-wide association study data. Figuring out which GWAS hits are promising drug targets will take time, effort, and a good deal of luck; in my opinion, this is the major lesson from Decode (which is not all that surprising a lesson)–drug development is really hard

Addendum: Most of my friends, who have undergraduate backgrounds in biology, and have taken at some quantitative genetics, seem to guess the heritability of I.Q. to be 0.0 to 0.20. This is just way too low. But is it even important to know this? I happen to think an accurate picture of genetic inheritance is probably useful when assessing prospects of mates….

Citation: Rietveld, Cornelius A., et al. “GWAS of 126,559 Individuals Identifies Genetic Variants Associated with Educational Attainment.” Science (New York, NY) (2013).


Comments (60)

  1. genobollocks

    Can you clarify for someone who, not coming from genetics, isn’t exactly sure which meaning is intended?

    > the genetic architecture of intelligence is one where the variation on the trait in the normal range is controlled by innumerable alleles of small effect segregating in the population

    Do you, by alleles, mean polymorphisms with a somewhat high MAF or do you also include rare variants, ie. private mutations etc.?

    Are you taking a stance, saying that common SNPs with very small effects are more important than very rare SNPs with sometimes large effects (like those we know from ID research).

    How do you relate these results to the GCTA results by Davies, Chabris (who you cited), most recently Plomin that found a lot of variance to be accounted for by common SNPs? I often heard the criticism that population stratification may account for these results; some did not think a big role for common SNPs was in concordance with other evidence on intelligence’s genetic architecture.

    Are you referring to pop. strat. when you say “segregating in the population”? I’m guessing you think that whereas pop. strat. clouds the picture, meaningful differences (not just related to cryptic ancestry and thus by proxy to SES etc) exist between pops?

    Anyway I’m curious why your takeaway is “innumerable alleles of small effect” when I thought these results could also stem from (even more) innumerable rare variants of sometimes large effect.

    • razibkhan

      Anyway I’m curious why your takeaway is “innumerable alleles of small effect” when I thought these results could also stem from (even more) innumerable rare variants of sometimes large effect.

      i don’t think they’re as likely to be large effect/rare. if so, they should show up in the old family based linkage studies, right? (they will at least run in families)

      not as clear whether the alleles are rare and many, many, vs. not so rare and just many.

    • razibkhan

      my working model for explaining intelligence variance:

      commonish very small effect alleles distributed across most of the genome > private alleles within family of modest but not tiny effect (perhaps due to mutational load) > population specific alleles of very small effect distributed across most of the genome

      only modest confidence in the hypothesis. though i’m more confident that there aren’t large effect alleles segregating in any great numbers to explain normal population wide variance.

      • genobollocks

        And your ultimate explanation? I.e. why no fixation of commonish small effect alleles? Ancestral neutrality? Too recent? Tradeoffs?

        And if you think higher intelligence has been selected for most of the time, do you think there’s something to Miller’s fitness indicator ideas (by consequence low robustness, high evolvability) or not?

        The linkage thing: don’t know if you read Plomin & Spinath ’04, there they say that “severe retardation doesn’t run in families”. I’d think that the cause for this obs. is likely de novo variants and infertility of ID individuals, not non-genetic causes (in most cases). And there’s 300+ known variants linked to ID, so some stuff did come up in the linkage harvest. GWAS are done mostly on a distribution that is arbitrarily cut-off at the lower end. I only skimmed the supplements to this one. They do have some population-based samples, but because the words ID and mental retardation don’t appear, I’m guessing they excluded them (hopefully they didn’t count years of special education as EduYears.

        But when I said rare variants of large effects I should have said modest (would’ve been the better imprecise word). Modest-effect deleterious de novos for complex, selected traits might accumulate for far more generations than for some complex diseases (ie. autism fertility 0.05, schizophrenia fertility 0.40, these are gone fast).

        I’m probably more 2 > 1 > 3 relative to your 1 > 2 > 3, even though I myself amassed some evidence to the contrary of 2, theory just seems to put 2 at the front. Which is why I’m confused about the GCTA results.

        • razibkhan

          i put up a response which was eaten by DISQUS 🙁

          in any case

          1) i’m talking *normal* variation. there are big QTLs at the low low end

          2) miller might be right. we’ll see. depends on the correlation between mutational load and the trait.

          3) i don’t think there has been strong unidirectional selection for intelligence or some sort for long periods of time since the last 100,000 years or so. encephalization probably took up a lot of the low hanging fruit, and now our heads can’t get too much bigger or we won’t be born (or, we’ll kill mom on the way out, if we can make it out).

          • genobollocks

            0) Happened frequently to me on smartphone. DISQUS is really pretty bad and widespread (pretty nasty FOUCs too).

            1) Yeah, yeah, that’s clear. It’s just not always smartly discussed in context, e.g. when wondering about possible stabilising selection (maybe mutations either devastate upstream or are buffered). For one, it’s clearly relevant for the mutational load question – there’s many complex diseases with negative IQ associations.
            Anyway, I think a compelling account of the mechanism is currently missing – most people just state the latest estimates, definition of rare or common is bent to fit with political proscriptions in in a lot of lay discourse (e.g. Miller’s endorsement of the BGI ideas which is of course ridiculous under the mechanism he proposed in MM. Maybe he changed his mind, maybe he doesn’t see the contradiction).

            2) Yes, will always be hard to isolate though. Maybe theory-derived, unexpected predictions are the more fruitful path here.

            3) How come? Pretty clear, strong mate preference for high IQ worldwide. Head circumference (the limiting factor for birth, I guess) and IQ have only a paltry .1-.3 correlation, without C sections that should be lower still.

          • razibkhan

            Pretty clear, strong mate preference for high IQ worldwide. Head circumference (the limiting factor for birth, I guess) and IQ have only a paltry .1-.3 correlation, without C sections that should be lower still.

            1) re: correlations of brain + IQ. i think a lot of the intelligence gains between 1 mill to 100,000 K were simply expansion of brain. brute force. the variation after has to be more subtle. not the low hanging fruit

            2) there may be a general averaged mate preference, but let’s look at the fecundity of the grandchilren cohort. i wouldn’t be surprised if there were trade offs. and, there are some problems with excessive non-typicality. a very smart person in a small tribe may just turn out to seem a weirdo who can’t find mates who don’t deem him too weird.

          • genobollocks

            1) Yeah, I tend to agree re brute force. But the fact that sheer size nowadays doesn’t count for that much (brain size of course accounts for more than head circ., but also has a less direct relationship to birth problems) should tell you something of how long and strong that subtle improvement thing has been going on, shaping individual differences we see today.

            2) It’s a bit better than “general average pref.”/”would-be-nice”, e.g look at mating market studies (Norman Li). Looking at fecundity makes little sense with easy contraception widespread. I think people (you too apparently) have a tendency to think more of the Rainmans than the Feynmans (or Schrödingers) when they consider the sex appeal of intellect, but anyway, I checked: there’s little data to support either notion. Much of what there is, was done by ideologically handicapped people and just doesn’t speak to the question very well.

          • Sandgroper

            I recall something about very high IQ people having difficulty holding down corporate jobs – maybe some analogy with HG group cooperation.

          • genobollocks

            Doubt this holds up to scrutiny. Usually IQ predicts job success very well. But if you can find it again, it’d be interesting.

          • Luis Aldamiz

            A quick search in the Internet reveals that, while there is a clear correlation between IQ and professional achievement, it’s not the only factor.

            For example an educator at Forbes says: “I have seen far too many intelligent students drop out of school and eventually fail to pursue a successful career. I have also seen less intelligent students finish school with high marks and eventually go on to pursue a superb career.”

            I also know many such cases from personal experience. It seems that other factors like emotional intelligence or just motivation are also very important.

            It may also depend on work environment. Particularly those companies or institutions (from the military to many European universities) that value more hierarchy than creativity may be hostile to certain types of “too smart” people, who may tend to think outside the box too much and therefore not be disciplined, obedient enough.

            So I’d add another factor (in many contexts): obedience, even a bit of a boot-licking attitude but at least the ability of knowing your social boundaries, what is more related to EQ than to IQ. In fact high IQ people tend to be often a tad arrogant (possibly with good reason), what, unless controlled, can become a serious barrier.

            EQ should help a lot here: getting bosses, colleagues and customers in your pocket, so to say, instead of annoying them.

          • genobollocks

            EQ is not a very highly-regarded concept in academic psychology, for good reasons.
            You can always find dissenting voices like an educator at Forbes or your own, but there really is a large amount of evidence for IQ and job success being linearly linked.

            There’s a lot of money in saying that you have discovered scale X (emotional intelligence, MBTI, whatever) and that it’s a good predictor of job success and that every individual can be good for the company in the right context. Doesn’t mean it’s true.

          • Karl Zimmerman

            I think part of the problem is there’s a large number of “hidden smart people” in the world.

            I would define a “hidden smart person” as being someone with a high IQ who doesn’t fit any nerdy/geeky stereotypes, and isn’t actually driven to learn new things except insofar as it helps them in school or in their job. Such people can come across as very dull, but in reality have both a high IQ as well as generally excellent job performance.

            In contrast, people who fit the stereotypes in our culture of smart people have more varied outcomes due to various issues; mostly some mixture of below-average social skills and lack of actual interest in personal advancement. They don’t see their intelligence as a tool to succeed, because they don’t want to succeed. They just want to learn.

          • genobollocks

            I don’t share that stereotype, may be a (sub)cultural thing.

            Actually I would include active information-seeking behaviour as pretty typical of intelligent individuals. You sometimes read about stuff like “the number of books in a household is heritable” or “crystalline IQ”/knowledge is heritable – obviously that doesn’t mean books or knowledge are in our genes, but seeking out books and knowledge is part of the higher IQ phenotype. IQ tests often measure knowledge – that wouldn’t work well for job prediction if a lot of “hidden smart people” were around, who could be expected to learn things on the job quickly with incentives, but fail at knowledge parts of the IQ test because they’re not driven.

            Because of the positive manifold findings, you should expect fewer people of high IQ and low social skills to exist than low-low-combinations. Disagreeableness (eg razibkhan) is something else.

            (Another DISQUS screwup: when you type someone’s name and are not at the end of the text you’re editing, the text after the name is lost).

          • Sandgroper

            Yes, up to a point, but then it drops off. I also recall Razib previously put up some American data that showed income peaks at around IQ = 135-140. I recall being mildly surprised that it peaks that low. I boggle at the prospect of finding that specific discussion again among Razib’s prolific outpourings, though.

          • genobollocks

            Still doubt it and haven’t read anything of the sort, always anecdote, no data. Beyond 130 you have huge CIs too, so wouldn’t trust a simple fitted curve too much there.

            Btw Luis Aldamiz, high correlation in the military:

          • Sandgroper

            OK. Think about how much you can earn with a PhD in Physics vs a Masters and a career in Civil Engineering. You need to be much smarter for Physics, and probably need to suffer fools a lot more gladly to practice Civil Engineering. I’m not going to go to war over it, though.

            Just idly fishing around I found this, which seems kind of interesting – esp. what she says about higher ed. women:


          • genobollocks

            IQ, even in this selective group [Terman sample, IQs of 140+], is still associated with higher earnings. This means that the lifetime eff ect of IQ is at least partly caused by a direct e ffect on earnings. This contradicts the claim by Gladwell (2008) that for the Terman men, IQ does not matter once family background and other observable personal characteristics are taken into account

            I really think this cliché about high-IQ people not being able to tolerate fools etc. comes from a false idea high IQ = disagreeable nerd on the autistic spectrum which is simply not what I think most high IQ people are (IQ is certainly uncorrelated with disagreeableness and correlated with openness). There are such individuals. They might have a harder time. The net effect of IQ on earnings still seems to be positive all the way, there’s not necessarily an interaction.

            The part about higher-ed. smart women is interesting. To summarise they marry less, thus earn less through their husbands, thus (because men earn more?) they are worse off. But they earn more on their own. Seems not very contradictory to my main point, that higher is always better for job achievement (seems like a somewhat sound life choice to me at that time, increases independence, money isn’t everything etc., so whatever life success is, they might have it). But who knows, that’s just story-telling.

          • Sandgroper

            Yeah. I think the point is more maybe that because they marry less, their lifetime income turns out to be less than the combined income of a married couple divided by two. Dunno – lifetime, men earn more because they don’t need to take time out for kids, plus work longer hours. Apparently. Whatever, evidently women are better off lifetime-financially just getting a first degree because they are more likely to marry a high earning male. That really is kind of interesting.

          • Sandgroper

            I mean, that’s exactly *not* the advice I give to my daughter, right?

          • Emil Kirkegaard

            There is no threshold effect for income and IQ. So your speculation is probably wrong. Numerous studies have shown this.

          • Sandgroper

            I only realised today when using DISQUS on a pad with a chip – you’re right, it sucks.

            I once idly posited that maybe head circumference was a limiting factor on successful Neanderthal-AMH births and got given a lecture by someone about how head size doesn’t give rise to birthing difficulties. I referenced the Californian data about the higher incidence of birthing difficulties in European male/East Asian female couples and was told that those data had not been replicated elsewhere. No idea, but intuitively you’d have to think that head size vs birth canal diameter would have to be an issue at some point.

        • highly_adequate

          “I.e. why no fixation of commonish small effect alleles? Ancestral neutrality? Too recent? Tradeoffs”

          Isn’t the length of time it takes for an allele to reach fixation dependent on how large its effect is on fitness? Why wouldn’t you expect an allele that has only an exceedingly small effect on IQ to take a very long time to go to fixation? Why wouldn’t one expect that all of the fairly common alleles that have a large or even a modest, but not exceedingly small, effect on IQ to have reached fixation quite quickly, leaving only those with very small effect?

          • genobollocks

            It’s less that I have much grounds for skepticism regarding SNPs of very small effect. Though very small is really very small here, when one goes with the current estimates for effective population size. I don’t know whether the current upper bounds from GWAS are inconsistent with this, it certainly depends on a few factors. Any suggested reading on this topic? Especially interested in: what if selection on the complex trait is strong, and we have assortative mating?
            I also asked because both balancing selection, neutrality or recency are compatible with common SNPs of small effect and I’d like to get an understanding that moves beyond the descriptive.

            It’s more that I’d expect de novo rares of modest to large effect to crop up often, given what I think about IQ being highly polygenic and thus offering a large target for mutations and about not being very robust given what I think about there not having been much stabilising selection. May be wrong on both counts.

  2. Luis Aldamiz

    The debate is served. I guess that one of the key issues here is that the brain is such a plastic organ that a large level of environmental influence in its development is almost necessarily true. The problem is how much can still be pinpointed to genetic influence, either simple SNPs/alleles or complex genetic builds, most of which we can only suspect at this stage of research.

    In any case, it was reported recently that an environmental influence like breastfeeding (vs. formula) accounts for a 20-30% difference in white matter generation at very young ages (2 y.o. to be precise). This is a single-cause major impact in intellectual development, even if white matter cannot directly be translated into IQ points.

    I am not knowledgeable enough but I would bet that other environmental elements like in-womb epigenetics (maybe partly triggered by the mother’s emotional state or nutrition) or later influences like nutrition of the baby itself or traumas (which can…

    Epigenetic differences have already been detected as significant influences in MZ twin IQ differences, dramatic effects in brain size caused by emotional trauma, etc.

    An intriguing issue is that at least some epigenetics seem to be inheritable, what may be behind part of the differences between apparent inheritance and known genetic influences. This issue of inheritable epigenetics (which to some extent reminds of Lamarckian evolution – but not quite the same thing anyhow) is a most important lesson of recent genetic research to be incorporated to our thought: not everything inherited is necessarily genetic senso stricto.

    • razibkhan

      I am not knowledgeable enough but I would bet that other environmental elements like in-womb epigenetics (maybe partly triggered by the mother’s emotional state or nutrition) or later influences like nutrition of the baby itself or traumas are most important in mental development.

      the issue is that when environmental effects like trauma or nutrition are strong you’d expect heritability to be low by definition. it’s when you have developed first world conditions that the heritability should increase, because genes are all that remains.

      my daughter is being breast fed, so i think this is good. but i am careful about the breast feeding = higher IQ, etc. literature after this sort of thing:

      also, on a personal note, i was the only one of my siblings not breast fed (my mother was ill after my birth). god knows how smart i would be otherwise!?!?! 😉

      • genobollocks

        I have private knowledge that the lead researchers were absolutely uninterested in continuing their research with comparions in Dutch women, many of whom pump their own breast milk, so you’d have a difference between formula in a bottle, milk in a bottle and milk + more intimacy/boob goodness (god knows how _nice_ you might be otherwise ;-).
        I don’t trust researchers who have no interest in testing their conclusions with better data.

        • Sandgroper

          On a slightly lighter personal anecdotal note, I was once carrying my baby daughter while shirtless in hot weather, and she suddenly fastened herself onto one of my nipples like a suction cup – it was not only embarrassing, it was very difficult to get her off. For the avoidance of doubt, I’m male. I figured she didn’t much care where lunch came from, as long as she got some – which in this case obviously she didn’t, but not for want of trying on her part.

          My wife couldn’t talk for laughing.

          • Luis Aldamiz

            Suckling is instinctive (and therefore emotional) for babies. For what I know, it’s also an issue of emotional satisfaction/support, not just a matter of food.

            As for pollution, it may indeed be a problem, but you would have to consider it in individual basis (not every mother eats the same and also a reason to ponder mother’s nutrition and health as a key part of the baby’s).

          • Sandgroper

            She didn’t look emotional when I managed to pull her off, just bad tempered/irritated by the lack of lunch. She still gets the same look when she doesn’t get lunch on time now, actually. It was a similar facial expression to the one she got the first time she was given solid food, just milli-seconds before her taste buds signalled to her that it was yummy.

            My daughter showed me some data on concentrations of DDT in human milk, broken down by country. Sweden was the lowest. Australia was kind upper middling. Countries with the highest concentrations included Malaysia and Indonesia – they didn’t posit why, but I assumed it was due to past efforts to eradicate malaria. I don’t know, but it seemed like a reasonable assumption for that kind of environmental exposure. I mean, otherwise, why Malaysia and Indonesia, specifically?

      • Sandgroper

        A human milk researcher with whom I’m on erm intimate terms genetically tells me they have identified 150 different pesticides in human milk. 150. Sounds like a lot. Don’t know what they do, but a blanket assumption that everything baby gets from mom is the good stuff might not be on the prudent side. In the Pearl River Estuary DDT concentrations in the breast milk of some Chinese white dolphins were high enough to kill the baby dolphins.

      • Luis Aldamiz

        “… when environmental effects like trauma or nutrition are strong you’d expect heritability to be low by definition”.

        There are two issues here, Razib:

        1. The effects can be strongly inheritable or genetic AND strongly environmentally caused at the same time. For example: 50% and 50% influence each.

        2. Heredity in twin studies is always measured after birth, what necessarily includes the nine months of (very intense) development in the womb inside the “inherited” area. All environmental or non-inherited epigenetic factors are wrongly dumped here into the “inherited” category, which is a clear source of error.

        When looking at the IQ twin studies data, we see that DZ twins get quite stronger heredity percentages than regular siblings, even if they have exactly the same genetic relation on average, this must be because of in-womb environmental factors, shared among DZ twins but not (or not so much) among regular siblings.

    • Chad

      “Epigenetic” differences, such as DNA methylation, are not necessarily due to the environment, although there is some influence there. In fact, most DNA methylation variants are typically linked to genetic variation. This has been one of the major findings of the only real population level bisulphate sequencing studies done to date, which have pretty much all been in plants. There are some “pure” epialleles that seem independent of genotype and may be environmental, but most variation is linked to genotype and gets inherited in a rather mendelian fashion.

  3. ” I happen to think an accurate picture of genetic inheritance is probably useful when assessing prospects of mates….”

    I’m always impressed at how precise people can be when selecting even without any knowledge or acknowledgement of the genetic underpinnings – many couples seem to “make sense.” Many *refute* that such a thing even exists but ideologues quickly turn practical when big life decisions are to be made.

  4. chris_T_T

    A major problem with low heritability estimates for human intelligence is it begs the question of how our species even evolved if environmental effects swamp the genetics component.

    • razibkhan

      well, if selection operates strongly on a trait eventually heritable variation can be exhausted. so it’s theoretically not implausible if you assume that the selection phase was 1 million BP to ~200 K BP when cranial capacity increases leveled off.

  5. highly_adequate

    “Some of my friends have been arguing that the small effect sizes here validate the position that intelligence variation is mostly a function of environment. ”

    Not sure why anyone would argue that small effect sizes suggests that IQ is low in heritability. What does one have to do with the other?

    • razibkhan

      for intelligence i think yeah, and large effect QTLs are going to be deleterious and purged.

  6. Karl Zimmerman

    One question.

    If intelligence genetically varies due to the influence of many small differences across the whole genome, doesn’t this cast doubt upon the idea of continental-wide differences in intelligence?

    Intelligence being subject to selection on the basis of a population group (either geographic or an endogamous minority) makes a great deal of sense. The problem is, across a whole continent both the natural environment (parasite load or climate, for example), and the human environment (extent of organized society or social value placed on intelligence) will vary. Hell, it will also vary over time – if there’s a societal collapse positive selection for intelligence may go into reversal, or if inbreeding becomes widespread. As a result, one might expect in recent history many intelligence trends cancel each other out, and they predominantly happen on the “nationality” level where an appreciable difference is found.

    • razibkhan

      doesn’t this cast doubt upon the idea of continental-wide differences in intelligence?

      quantitative traits can vary clinally.'s_rule

      i don’t see how fluctuating selection speaks to difference or lack thereof actually.

      • Karl Zimmerman

        Sure, there are lots of clinal traits. Height is mentioned in the wiki article, and obviously skin color is another one. Even the clinal nature of something like lactase persistence shows it can happen on “culturally selected” traits.

        Regardless, I’ll use an example from a few years back to show what I’m talking about: Ethiopia. Ethiopia has a mild climate, with few natural parasites. It has been “civilized” for a longer period of time than Northern Europe. And its population has considerable Eurasian admixture, which entered the population during several different periods. By many of the rationales commonly used to explain lower intelligence in Sub-Saharan Africa, it should be intermediate in IQ score between the rest of the continent and Eurasia. But it does not score appreciably different, AFAIK.

        If we do find, after a few decades of development, a big difference in IQ between Ethiopia and say South Sudan, I’ll be less incredulous. But as of now that says to me that even if anything can be said about the white-black IQ gap in the U.S., you shouldn’t judge that it says anything regarding Africa, except perhaps for those West African groups who contributed the most to the African American genepool. Again though, perhaps not, because it’s possible (despite the complications of European admixture) that there was either negative or positive selection for intelligence due to slavery itself.

        • razibkhan

          most adaptive explanations of intelligence are dumb. then again, we don’t know the real *reason* for many genomic signatures of selection in areas where it is unlikely to be stochastic (i.e. it really is likely to be selection for a host of reasons).

  7. gcochran

    “guess the heritability of I.Q. to be 0.0 to 0.20”

    That’s just weird.

  8. Ramez Naam

    Hey Razib,

    Just to clarify my view, and the mistake I made: I view *educational attainment* and IQ as fairly separable things. I got excited about the finding of SNPs correlated with educational attainment and (wrongly) concluded that overall heritability of educational attainment was low. I’ve learned that I was wrong since then. 🙂

    That said, the higher level point remains – you can imagine a society where educational attainment and IQ are weakly correlated due to (for instance) socio-economic factors that drive educational attainment more than IQ. And it does seem that in the US, IQ is substantially more heritable than educational attainment (perhaps twice as heritable, depending upon the estimates one uses) – and that gap says something.

    On IQ itself, I’m on pretty much the same page for you. The evidence for high heritability of IQ is pretty clear. I do think there are a couple caveats even there:

    1) Very large differences in environment (such as the difference between rich vs poor nations) clearly have a large effect on IQ. E.g., see the Ron Unz piece on this topic:

    2) Even with twins-separate-at-birth studies, there is a factor seldom discussed, which is pre-natal environment. We know that the condition of a mother during pregnancy (stress, nutrition, etc..) can have significant epigenetic impact on offspring. Current studies fail to account for this.

    • razibkhan

      i think epigenetics is a little over done. thought i would not be shocked if it was a big deal. it just seems people want epigenetics to be a rabbit coming out of the hat.

      • Chad

        Like I said above, epigenetic variation in a population, namely I am talking about methylation, appears to be closely linked to genotype. So called “pure” epialleles are relatively rare in comparison.

      • Isn’t the epigenome inherited anyway?

        • Luis Aldamiz

          Some parts of the epigenome (only) seem to be inherited but the extent of this phenomenon is not well understood yet, Robert. In plants at least, heritable epigenetics do play an important role in environmental adaption.

          • Chad

            I would be very cautious in making that statement. Your link quotes some parts of that paper out of context. It is also important to note that in epigenetics, particularly in plants, context is everything. Plants methylate cytosines in three different contexts (CG, CHG, and CHH) and those contexts differ both in origin and function. Independence from genetic structure is largely dependent on what context of methylation one is talking about. Furthermore, in soybean RIL lines, up to 91% (perhaps more) of differentially methylated regions segregate with the genotype, and so are inherited in a mendelian fashion. While there are examples of pure epialleles that act independent of the genotype, these are the exception, not the rule.

          • Luis Aldamiz

            I can’t review my notes from March because when I wrote them the paper was freely accessible and now it is pay per view. However most of what I copy-pasted was the abstract and conclusion, so hardly cherry-picking out of context (even if I wanted to, what I don’t).

            As for Mendelian inheritance, I can only imagine that the heritable epigenome is affected by selection in exactly the same way as regular genes, so, yes: Mendelian or quasi-Mendelian it should be. And it should also be interdependent with the genome, naturally. What heritable “epigenes” seem to be most useful for is to “cheat” or work around the limitations of purely genetic inheritance, so plants and animals do not always have to wait for the right mutation to happen randomly or to be introduced from another population.

            It adds a layer of extra flexibility to the genome which seems to increase short/mid term fitness, and, as usually happens in Nature, it is there because it does work. Much of adaption is not a matter of absolute fitness but of fitness relative to certain conditions which are often variable, in that aspect at least inheritable epigenetics seem to be quite important, as is genetic diversity within populations.

          • Chad

            Epialleles can be broken down into three types. Obligate epialleles are dependent upon the genetic variation, such as a nearby transposon insertion. Facilitated epialleles are initially caused by a genetic variant, but not dependent upon it for it to be maintained. For instance if a transposon inserts near a gene, resulting in methylation of the gene, then is excised, but the methylation persists. The final, and rarest is the pure epialllele, which is completely independent of genetic background. For a review of this see Schmitz and Ecker, 2012 in Trends in Plant Science.

            Pure epialleles are not inherited in a mendelian fashion. Paramutations are an interesting example of these. The other two, in particular the obligate epialleles do follow mendelian rules because they are not independent of genetic variation.

            It should also be noted that “epialleles” are not a way of “cheating”. As the obligate and facilitated are dependent on genetic variation and mutations, the organism does have to wait around for them. The exception, pure epialleles, also have their own mutation rate. See Schmitz’s 2011 Science paper where they looked at the inheritance of epialleles across 30 generations.

      • Luis Aldamiz

        Epigenetics is more like a big question mark, because we only know so much as of now. The fact that some of it is inheritable really breaches the purely genetic understanding of inheritance, while the non-inherited epigenetics shared by all kind of twins in the womb also raises some questions.

  9. Stephan Guyenet

    Personally I’m starting to wonder whether variability in common traits such as intelligence and BMI is really due to many genetic differences with small effects. It could be true, but there are other possible explanations. One possible explanation is that GWAS studies often miss important sources of genetic variability, such as copy number variability and triplet repeats. These could be having major effects that are largely overlooked by GWAS studies (you would think they would be detectable by haplotype at least to some extent, but from what I understand, they often aren’t). There are some data on BMI-related genetic variability that I hope will be published soon by Mario Falchi’s group, which should relate to this point.

    On a purely mathematical basis, it’s hard to believe that GWAS studies have identified the main sources of genetic variability related to BMI. All together, the top ~30 SNP loci that have been identified account for ~2% of BMI variability, while ~60% of BMI variability is heritable. The top locus, FTO, accounts for most of that 2%. After you get past the first few strongest loci, the contributions of the remaining loci are trivial– 0.1% or less. As we go down the list, the loci contribute less and less to variability, until they drop below our statistical detection threshold. This implies that for the “many small contributors” hypothesis to be correct, there must be thousands of relevant loci all acting at once. I suppose that’s possible, but another possibility is that we’re simply missing some big players that the GWAS studies are failing to detect.

    It seems hard to explain why fraternal twins can often be so BMI discordant if variability is distributed over thousands or even dozens of loci– this observation seems more consistent with there being a small number of more influential genes.

    • razibkhan

      had friends who did prelim look at CNV and intelligence. didn’t find anything that i recall….

      • genobollocks

        Yeo, Gangestad, … (2011) found this, nobody believed it (small sample of alcoholics). Two published nonreplications that I know of since: Bagshaw (2013); McRae (2013). Probably more unpublished ones.

      • Stephan Guyenet

        Here’s an example of what I’m talking about, hot off the presses. This paper involved Mario Falchi but was not the line of investigation I was referring to above. They identified variable number tandem repeats in the DOCK5 gene that explain a meaningful amount of obesity phenotype variability, and this was missed by GWAS methods.

  10. Chad

    I disagree that epigenetics is “not a source of phenotypic variation”. In plants, there are clear examples of phenotypic variation being linked to epigenetic variation rather than genetic variation. This has been shown in two epiRIL populations that possess identical genetic backgrounds, but were disrupted in their methylation status. It has also been shown in a doubled haploid line (thus being homozygous at all loci) which was then bred for variation in energy use, this being primarily tied to epigenetic changes.

    I agree that it does not change our understanding of genetics, in part because much of epigenetic variation is tied to genetic variation and partially because in the long run, its just not as stable.

    • td

      I was talking about fundamental sources of variation. There’s obviously a long and complex causal chain of events ultimately leading to phenotypes and any link in that chain could be construed as a source of variation, but there are genetic and environmental inputs driving the process at the bottom of it all.

      If that process is substantially probabilistic, which seems plausible (given the often substantial differences between isogenic organisms reared in highly similar environments), then maybe it’s appropriate to speak of epigenetic stochasticity as a source of variance on par with genetics and environmental influences, assuming epigenetic fluctuations are what’s driving the randomness.

      Regarding the isogenic plant lines, what is causing those epigenetic modifications? Methyl groups don’t just fall out of the sky. Is there some differential environmental factor affecting one line and not the other? Or perhaps epigenetic marks are being passed through the germline?

      • Chad

        Methylation can be passed through the germline. Both in plants and animals. This has been known for sometime. I suspect it is more common in plants than animals, but none the less, it does occur. Furthermore, epigenetic variation is subject to random variation. The only estimate I know of this was done by analyzing Arabidopsis plants that spanned a 30 generation time period.

  11. Emil Kirkegaard

    The newest study on heritability of IQ put it at 0.8-0.85.

    Cf. Genetic influence on human intelligence (Spearman’s g): How much?

    Curiously, few studies actually report on the heritability of g itself, which is intelligence, and not just IQ (fullscale or otherwise). IQ is more or less just a noisy estimate of g, which should attenuate heritabilities a bit.

    I wrote to Bouchard to ask him about studies that calculate the heritability of g specifically. He only knew of one study:

    Genetic and environmental influences on the Verbal-Perceptual-Image
    Rotation (VPR) model of the structure of mental abilities
    in the Minnesota study of twins reared apart

    His study found a heritability of 0.77.

    Jensen (1998:182) cites one that gives 0.81.


Discover's Newsletter

Sign up to get the latest science news delivered weekly right to your inbox!

Gene Expression

This blog is about evolution, genetics, genomics and their interstices. Please beware that comments are aggressively moderated. Uncivil or churlish comments will likely get you banned immediately, so make any contribution count!

About Razib Khan

I have degrees in biology and biochemistry, a passion for genetics, history, and philosophy, and shrimp is my favorite food. In relation to nationality I'm a American Northwesterner, in politics I'm a reactionary, and as for religion I have none (I'm an atheist). If you want to know more, see the links at


See More


RSS Razib’s Pinboard

Edifying books

Collapse bottom bar