Smarter Children Have More Gray Matter 60 Years Later

By Neuroskeptic | June 8, 2013 8:56 am

Your IQ at the age of eleven predicts your brain anatomy sixty years later, according to a Canadian/Scottish team of neuroscientists: Childhood cognitive ability accounts for associations between cognitive ability and brain cortical thickness in old age.

The authors of the new paper, Karama et al, made use of a unique long-term study of Scottish volunteers, all of whom had IQ tests back in 1947, when they were 11. In 2009, aged 73, those who were still alive and willing to participate had an MRI brain scan – a total of 588 people.

Karama et al show that childhood IQ is correlated with the thickness of the brain’s cerebral cortex in old age. What’s more, IQ at age 70 was correlated with brain anatomy, but no more closely than the age 11 scores were – that is to say, cortical thickness in old age is correlated, not just with IQ in old age, but with IQ at any age.

As you can see on this image, the relationship was seen across most regions of the brain:

In other words, people with higher IQs just tend to have a thicker cerebral cortex across the lifespan.

The authors point out, however, that they didn’t include anyone with diagnosed or suspected dementia. It’s certainly true that some forms of dementia cause rapid cortical thinning, and that this is associated with cognitive decline. But dementia aside, it seems that the brain you have at 73 is a reflection of your lifelong IQ. There’s nothing special about old age which drives the relationship.

This result is pretty remarkable, although the correlations were small (coefficients of 0.1 to 0.3, depending upon the region.) Why this correlation exists is another question. One interpretation is that having a thicker cortex makes you more intelligent, and so the correlation between IQ and anatomy would have also been true at age 11.

However it could be that the smarter people took better care of their brains over 60 years, leaving them with more grey matter, even though this wasn’t what made them smart in the first place.

ResearchBlogging.orgKarama, S., Bastin, M., Murray, C., Royle, N., Penke, L., Muñoz Maniega, S., Gow, A., Corley, J., Valdés Hernández, M., Lewis, J., Rousseau, M., Lepage, C., Fonov, V., Collins, D., Booth, T., Rioux, P., Sherif, T., Adalat, R., Starr, J., Evans, A., Wardlaw, J., & Deary, I. (2013). Childhood cognitive ability accounts for associations between cognitive ability and brain cortical thickness in old age Molecular Psychiatry DOI: 10.1038/mp.2013.64

CATEGORIZED UNDER: genes, papers, select, Top Posts
  • psycritic

    It would be interesting to see what happens to cortical thickness after someone’s IQ has been decreased by chronic cannabis use, especially if those who started out with higher IQ’s have less of a decline in cortical thickness, which may reflect the higher IQ somehow being protective.

    • petrossa

      There is hardly a recreational substance to be found in literature which doesn’t claim permanent neurological damage which with me sets of a red light and an alarmbell. This just cannot be so, so evidently a portion of those ‘outcomes’ are resultdriven. Most likely studies for governments or similar groups with a negative attitude towards people enjoying themselves.

      There can be a case made for excessive use, but the again eating an excessive amount of salt also causes (fatal) neurological damage

      Personally i don’t use recreational drugs but no one is going to tell me that in a top 5 well off country such as the Netherlands where recreational drug use is common since the 1960’s is filled with stupid old people.

      As for the study. I guess if you have a more active brain due to higher intelligence it stands to reason you keep on using it and thereby retain/rebuild more then people who have a low active brain which sit watching Eastenders.

      • oregonlocal

        I much prefer watching mindless violence like the Walking Dead or Justified.

      • Gen

        If you read the abstract of the paper psycritic linked to, you’ll find that the focus is on adolescent onset cannibis use. Although I completely agree that a lot of research on recreational substance use is result driven and I always try to figure out where research funding is coming from when I read literature on the topic, it doesn’t seem far fetched for substance use during adolescence to interfere with development in some way. I do think that the balance of research in the U.S. (or at least research that gets attention) being focused on illegal recreational drugs is silly, and is a holdover from old propaganda campaigns. I think that we should perhaps pay more attention to legal drugs like alcohol, as well as finding a different strategy than “more strict substance regulation” to stop drug use. None of that, in itself, can invalidate research findings. We should be focused on flaws in methodology or reasoning when responding to claims like this, not discrepancies with our own political views.

        • petrossa

          It’s hard to argue with the fact that the most drug relaxed nation in the world, where drugs like XTC and stronger are freely available since the 80’s there is the least problems with drugrelated issues. The only real drugrelated problem is caused by the French, which nag the dutch government silly to do a ‘war on drugs’ causing criminalization of drug supplies. Also i find it extremely hard to believe that municipalities are now going to allot cannabis growing permits for supply if this drug is really that bad. It just isn’t logical that after 50 yrs of experience in the matter that should there be serious neurological effects they’d not noticed by now or one has to assume total incompetence. 50 years long enough for the supposed ill effects to appear. So imho any research that shows otherwise just is badly setup, and has many confounding variables. Anyway the only way to show causal relation between regular non-excessive cannabis use and neurological damage is to lock people up for their lives in a controlled environment. Afaik this hasn’t been done, making all research highly suspect.

          Indeed if there is one drug that needs to be taken more seriously it’s alcohol, on a population of 17 million (including infants) there are 300.000 alcohol addicts, and a stable 20.000 drug addicts. Not to talk about damage cause by alcohol related incidents. By supplying the stronger drugs like heroine on doctors prescription (paid by NHS) there was a spectacular drop in addict related crime.
          By supplying clean safe shooting environments, with a supply of needles and paraphernalia aids and other diseases also dropped.

          In other words cannabis is the highly preferable over alcohol or other recreational drugs.

  • Y.

    Why is it published in Molecular Psychiatry? Because the brain is made of molecules?

    • Neuroskeptic

      I guess because it has implications for genetics (genes that affect IQ and also cortical thickness).

  • Y.

    The use of so many covariates is worrisome: age, gender, and a third variable I couldn’t quite understand. Were covariates chosen so that a certain result is produced? The authors should also show, possibly as a supplement, the regressions without any covariates.

    • Neuroskeptic

      Three covariates is not unreasonable. Gender is a big one, and ought to be included; age ought to have had a very small effect since it will have been almost constant (due to the study design), and the third factor was a measure of total brain volume… although you’re right that the un-covaried results should also be shown.

      • Y.

        They are reasonable if a-priori. Since we don’t know that, we have to see uncovaried data, particularly with regard to the third covariate. This is true for every correlational study, not a specific complaint about this study.

  • Pingback: Smarter Children Have More Gray Matter 60 Years Later |

  • stefanopj

    What about income? If it turns out this is just a reflex of better economic environment overall, IQ is a red herring.

    • NaveedXVO

      IQ is highly correlated with income, so you wouldn’t be able to disentangle it. You’re assuming that because IQ and income are correlated IQ must be invalid as the explanation. This is only because this is the answer you want. If you look at what’s a stronger predictor of success in life, IQ or parents socioeconomic status, IQ is, according to The Bell Curve. I know the book has been vilified but it’s results on those points haven’t been disputed convincingly (of course I’d love to be wrong, link?).

      • stefanopj

        It’s hard to separate those variables, because they are entangled. If we assume, not unreasonably, that IQ tests are made to check for abilities that wealthier kids tend to develop more (math, logic, etc.), then IQ is a proxy. You seem to already know that the Bell Curve is highly criticized on almost every angle, but for a specific review on the socioeconomic status x IQ debate check this book: . A quote:

        “We review critical errors Herrnstein and Murray made in their analysis; we reanalyze the identical data; and we come–as other scholars have, also–to opposite conclusions: Social environment is more, not less, important than test scores in explaining poverty, likelihood of incarceration, and likelihood of having a child out of wedlock. For economic outcomes, gender, a trait Herrnstein and Murray ignored, matters most of all. Other social factors–education and community conditions–are at least as important as test scores. Stepping back from the specific data, we point out that these findings are not news to social scientists. We have long understood that a person’s economic fortunes are hostage to his or her gender, parents’ assets, schooling, marital status, community’s economy, stage in the business cycle, and so on; intelligence is just one item on such a list.”

        • JL

          Murray did a follow-up study showing that IQ is a strong predictor of outcome differences also between siblings. This shows that “Inequality by Design” type analyses where IQ is assumed to be a proxy of family environment are wrong. See here:
          It is of course well known from behavior genetic research that childhood family environment is a weak predictor of IQ. Genes and the non-shared environment are much more important.

        • JL

          Even without Murray’s behavior-genetic analysis, the arguments in Inequality by Design are unpersuasive, as Francois Nielsen wrote in his review:

          Their ostensible strategy is to show that characteristics of the environment explain larger proportion of the variance in poverty than the measure of cognitive ability. Even superficially, this strategy largely backfires as a means of demonstrating that cognitive ability is unimportant as a causal factor: in all the models of poverty the coefficient of cognitive ability remains stubbornly significant, even in a model in which they “throw” twenty-eight other variables at it… Readers are likely to be more impressed by the statistical resilience of the cognitive ability variable in the face of massive model overspecificiation than by the authors’ claim that it does not really matter.

          Cognitive ability remains a significant predictor of poverty, even though the authors use highly questionable modeling tricks to try to reduce its apparent effect. One such device is to include as “explanatory” factors variables, such as the number of children in the household, that are tautologically used in the very calculation of the poverty status of the respondent, the dependent variable to be explained.

          • stefanopj

            This review has been rebuked by the authors ( The argument you quote is disingenious: “it remains significant”, yes, but explains only 10% of the variation, in contrast to the 37% explained by social environment. Additionally, equating the army test with general intelligence is problematic. Skills like algebra are clearly the result of at least some exposure to an educational environment, which is no doubt affected by wealth.

            At any rate, let’s say you can isolate IQ from wealth in regression models. You will still find out, as your table shows, that people with higher IQ also have more wealth. Since the original study did not control for socioeconomic factors, they can be measuring IQ as a proxy for more wealth, and all the associated benefits that come with it. My original criticism stands, regardless of what you think the outcome of this particular side discussion is.

          • JL

            This review has been rebuked by the authors

            It’s a very weak rebuke. They cannot refute the point that they’re just throwing in variables that correlate with both IQ and poverty and then claiming, without any evidence, that the direction of causality is the one they prefer. Murray’s sibling analysis showed that “controlling” for family variables that correlate with IQ leads to downwardly biased estimates of the effect of IQ. Korenman and Winship replicated Murray’s sibling analysis and reached the same conclusion:

            Incredible as it may seem, our sibling analyses suggest that, even though Herrnstein and Murray’s parental SES index is poorly measured, it appears to be adequate for producing unbiased estimates of the effects of AFQT score on socioeconomic outcomes.

            There is one very simple fact that nurture enthusiasts always ignore. If socioeconomic background really had such an overwhelmingly strong effect on IQ and other variables, why would full siblings raised in the same family be so different from each other? For example, the average IQ difference between two siblings is only 30 percent lower than that between two strangers (12 versus 17 IQ points).

            Additionally, equating the army test with general intelligence is problematic.

            One can argue that using the AFQT to predict outcomes underestimates the effect of general intelligence because it is a test of crystallized abilities, but the correlation between “cognitive g” and “achievement g” is so high that this is a very small problem.

            My original criticism stands, regardless of what you think the outcome of this particular side discussion is.

            The correlation between IQ and childhood SES is about 0.3, or perhaps a bit higher if you use a more comprehensive index. Let’s put it liberally at 0.40.

            The (range restriction corrected) correlations between IQ at both ages (11 and 70) and various cortical regions in the study were 0.15-0.41. This means that for your theory to be true, the correlation between childhood SES and the thickness of several cortical regions at age 73 must be 1.0, i.e., childhood SES must fully determine old-age cortical thickness for those regions. That’s an absurd theory that refutes itself, so I don’t even need to go into how the high heritability and low “shared environmentality” of IQ and brain structure contradict your argument.

          • stefanopj

            The only reanalysis by Korenman and Winship I can find is here (, and they say right on the abstract that “In
            addition, Herrnstein and Murray’s measure of parental SES fails to
            capture the effects of important elements of family background (such as
            single-parent family structure at age 14). As a result, their analysis
            gives an exaggerated impression of the importance of IQ relative to
            parents’ SES, and relative to family background more generally.
            Estimates based on a variety of methods, including analyses of siblings,
            suggest that parental family background is at least as important, and
            may be more important than IQ in determining socioeconomic success in
            adulthood. ” I don’t see how that in any way resembles reaching the same conclusion.

            As usual with proponents of IQ, you confuse heritability with genetic determinism. Even if IQ is a largely independent variable, it simply does not mean it’s not affected by environment, as anything else in biology. What it means is that, given the same environment, some kids will develop better cognitive abilities than others. It does not mean environment is worthless, or less important. The heritability of a darker skin tone is as high as it gets, but whether that will give you problems in life depends on society.

            Coming back to the original point, if you test only IQ but not any of these other factors – wealth, years of education (which, as you know, is highly correlated with IQ), etc. -, as the original study failed to do, you end up with a correlation that is potentially deceptive. Every IQ study tries to address this by controlling for all of these factors, in fact that’s the only way to tell if IQ is a relevant measure at all. I suspect finding out that cortical thickness is correlated with wealth, years of education or occupation type is not worthy of a journal like Molecular Psychiatry these days. Even if you are a proponent of IQ and heritability or whatever, this study is flawed.

          • JL

            Korenman and Winship’s point is that IQ differences between siblings are about as potent determinants of socioeconomic outcomes as IQ differences between strangers. This means that models where the predictive validity of IQ is attributed to family-level environmental effects are severely misspecified.

            In addition, Korenman and Winship claim that those family-level environmental effects that are uncorrelated with IQ are, when taken together, as good or better predictors of outcomes as IQ. However, they assume that the full extent of genetic influence on sibling behavioral similarity is captured by IQ, and that similarity in variables other than IQ is entirely due to the shared family environment. This is a blatantly false assumption, because all behaviors are heritable to some extent. Their estimates of the effect of family environment are therefore biased.

            As usual with proponents of IQ, you confuse heritability with genetic determinism

            Save your strawmen. No one believes in “genetic determinism.” The fact is that you greatly overestimate the effect of childhood environment on adult IQ. You failed to address the point about large differences between siblings.

            I’m not saying that IQ differences have 100% heritability. No one argues that. I’m just saying that genes and the unshared environment are much more important than the shared environment. The behavioral genetic literature is absolutely clear on this.

            Coming back to the original point, if you test only IQ but not any of these other factors – wealth, years of education

            As I pointed out, for your theory to work, there’d have to be perfect correlations between early life SES and old age brain structure. There are no known perfect correlations between sociological variables and brain physiology, so the idea that there would be ones that persist for a lifetime is improbable in the extreme. You failed to address this point as well.

            Every IQ study tries to address this by controlling for all of these factors, in fact that’s the only way to tell if IQ is a relevant measure at all.

            That’s a completely fallacious argument. Just because variables like, say, the number of books at home correlate with child IQ does not mean that these variables cause IQ differences. Correlation is not causation. There are behavioral genetic study designs which enable the testing of hypotheses like these, and they incontrovertibly show that the effect of family SES on IQ is small to non-existent at the population level.

            I suspect finding out that cortical thickness is correlated with wealth, years of education or occupation type is not worthy of a journal like Molecular Psychiatry these days.

            If you had evidence that childhood SES completely determines cortical thickness in old age, journal publishers would beat a path to your door.

          • stefanopj

            Oh, many people believe in genetic determinism. The book you’re defending here, The Bell Curve, is the perfect example of that. I’m not interested in, and have not attempted to demonstrate that differences in IQ are caused by this or that. I pointed out that IQ correlates with several other factors that are relevant for future outcomes. As the book I quoted up there argues, if you include proper socioeconomic variables in your model, they can be almost 4 times more important than IQ to social outcomes. The heritability of IQ, and sibling studies, are not relevant to this sort of discussion.

            For this particular study, this is the only question that matters: the role of IQ in predicting future outcomes, in this case cortical thickness. It’s been known for decades that socioeconomic variables severely affects brain anatomy and function (see for a recent review). In this paper (, they show impoverished, undernourished kids have a smaller estimated brain size, and not surprisingly they also have a smaller IQ.

            What you fail to grasp is that you don’t need a perfect correlation between other proxy variables for them to be relevant. Wealth might have a mild correlation with IQ, but years of schooling has a bigger one, and there are many others we might want to explore (parents’ occupation, health at birth, reading habits, etc.) that pertain to socioeconomic factors and correlate with IQ. If the authors want to claim that IQ is a variable of interest, they need an appropriate model that attempts to disentangle obvious contributions from proxy variables that also correlate with IQ. Ironically, that was what IQ proponents had to do to prove their point.

  • DS

    Did they control for alcohol consumption?

    • NaveedXVO

      No, drink up!

  • Beth Morrison

    As a person soon to be 62, I’m just glad for once to hear something cheerful. I know they didn’t test functioning, but after years of being told that as I continue to age I’m going to be losing cognitive functioning, memory, reasoning power, word finding abilities, and everything else, I’ll take whatever good news I can get. Woohoo!

  • andrew oh-willeke

    A correlation that low makes a conclusion about the relationship largely a matter of data presentation. The low number means that there are many very smart people with thin cortexes and many low IQ people with thick cortexes. This defies simple, causal explanations (i.e. it does not show, and indeed almost rules out, that IQ is largely just a function of how many neurons you have in the right places). For example, I suspect that the correlation of male sperm count or stature (or just about any other phenotype correlated with overall selective fitness) with IQ is at least as great, even though no one would leap to the conclusion that the relationship is causal in those cases.

  • rameshraghuvanshi

    IQ test is really hoax created by some white bigot intellectual raceest.How can you compare child rared in Andaman with child rared in posh locality of London? IQ of child of London remain high compare to child of Andaman Another fallacy is if you remain physically and mentally active you brain have more gray . matter in age of 90 also,Where is .IQ in these observation?

  • Michael Henderson

    Although our SES wasn’t very high when I was growing up, my family and I read voraciously, as did my wife’s. We were, I’d say,lower middle class, five of us earning graduate degrees, two of them doctorates. Almost all of us made our living teaching. Now in our 70s, we find ourselves needing to write things down in order not to forget them within 5 minutes, and starting to say something, being interrupted, and forgetting what we were going to say. But we show no signs of dementia or loss of long-term memory, so I too am cheered by this article.



No brain. No gain.

About Neuroskeptic

Neuroskeptic is a British neuroscientist who takes a skeptical look at his own field, and beyond. His blog offers a look at the latest developments in neuroscience, psychiatry and psychology through a critical lens.


See More

@Neuro_Skeptic on Twitter


Discover's Newsletter

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

Collapse bottom bar