Natural selection reduces years of schooling among Americans

As I’ve often said, the question I’m always asked after my public lectures on evolution is this: “Are humans still evolving?” And my answer is always the same: “Yes, but the evidence we have for evolution occurring right now involves traits that aren’t that interesting.” When people ask that question, what they really want to know is whether humans are getting better looking, more athletic, smarter—whether we’re turning into a race of superheroes. And all I can tell them is that, over the last 10,000 years, our species has evolved in some places to be more lactose-tolerant, in other places to be more resistant to malaria, and in still other places to adapt to the low-oxygen conditions of living at high altitude. But that’s still in the past (see a summary here).

As for evolution in the present day, we have “real time”, “horizontal” observations for things like selection in women for earlier age of first birth, later age of last birth and (also in women) increases in height in some places and decreases in others. Studies in the U.S., which haven’t been conducted elsewhere, show a recent evolution of reduced cholesterol levels and lower blood pressure, and an increased age of menopause (see here and here). But those results don’t excite people much. In general, though, as long as there is variation in some genes that causes variation in reproductive success, humans will continue to evolve. That hard part is documenting which genes and which traits are associated with reproductive success, and that means laborious studies correlating people’s genes and traits with their reproductive output.

One such study, by Jonathan Beauchamp, was just published in the Proceedings of the National Academy of Sciences (free download, link and title below). It used genetic, phenotypic (“trait”), and reproductive-success data from about 12,000 U.S. males and females of European ancestry, born between 1931 and 1953, all examined in a Health and Retirement Study (HRS). Using people of that age ensured that most of them had completed their reproduction, so the number of children they had (child mortality wasn’t counted) served as an index of their reproductive success (RS). And, of course, traits conferring a greater RS are those that selection will favor in populations. The currency of natural selection is reproductive output!

The authors did two types of analyses of the data.

a. Trait analysis. The authors used data on body mass index (BMI), educational attainment (EA), fasting glucose concentration (GLU), height (HGT), schizophrenia (SCZ), plasma concentration of total cholesterol (TC), and age at menarche (AAM), the last trait studied, of course, only in women. They then correlated the values of each trait with the reproductive success of their bearers:

The results? As the table below shows (asterisks denote statistical significance at the p < 0.01 level), there was evidence of phenotypic “selection” for stouter males and females, selection again educational attainment in both males and females, and selection for shorter women (but not men).

Screen Shot 2016-07-13 at 8.15.51 AM

Taken at face value, these data show that Americans in that year class are experiencing selection to get chubbier, to stay in school fewer years, and, in women, to get shorter. But of course there are problems, because they are looking at a correlation between a trait and the number of children produced by people with different trait values—and there’s no genetics here. Perhaps there are cultural or other reasons, for instance, for the correlation between staying in school for less time and having more children. One example would be if people leave school to have children, or put off having children while they’re in school. You’d get the appearance of selection on the trait, but there might not be any genes involved, so there would be no evolution. Because of these issues, the author did a study explicitly incorporating genes.

b. Genetic analysis. This involved looking at the DNA of every person at many sites (between 80,000 and 400,000 DNA positions, depending on the trait), and finding those combinations of gene positions best correlated with the values of the trait (I’m simplifying matters here, but it’s not important).  These combinations of DNA positions were then considered to be the genetic sites that could be influenced by selection on that trait. Having done that, Beauchamp could then see if those combinations of genes were themselves correlated with reproductive success, and thus could be under selection. After having done the appropriate statistical corrections for multiple tests, Beauchamp found the following results:

Screen Shot 2016-07-13 at 8.26.42 AM

So the negative correlation between genes associated with staying in school and the reproductive success of their carriers was still observed. In other words, both men and women were undergoing natural selection for fewer years of schooling. There was still a marginally significant (p < 0.1) association between age of menarche and reproductive success, in line with previous studies showing that American women are evolving to reach menopause later. There was no significant association between genetic constitution and reproductive output on the other traits.

Three questions remain:

  • How strong is selection against educational attainment? Answer: not very strong. The estimate given by Beauchamp is that we’re staying in school about one week to six weeks less per generation. If you take 25 years per generation, it would take about 35 generations, or 875 years, for selection to reduce our education by about a year.
  • So are we getting dumber? Answer: nope, because phenotypically we’re staying in school longer. That is, the cultural trends on Americans show that between 1876 and 1951 (75 years), Americans’ EA increased by 6.2 years—an increase of about two years of education per generation. That far outstrips the genetic change, so, even if the selection data are correct, we’re still going to keep increasing our educational attainment. What we’re seeing is the reproductive advantage of leaving school a bit earlier is being overcome by the cultural trend to stay in school longer.
  • What other problems are there? Beauchamp does a good job in pointing out the caveats of his study. I won’t recount them in detail, but they include the possibility that having more children doesn’t mean that you’re going to have better children—that is, that those extra kids may not themselves reproduce as well as kids from smaller families. That could weaken or even reverse the direction of selection. And, of course, this correlation was seen in only people from one or two generations, and there’s no guarantee that it will continue over the long term—or even that we’d see the same kind of selection in other countries.

In the end, we have a suggestive result, but one that needs a lot more work before it’s accepted as definitive. That work would somehow have to look at the reproductive output of the children themselves sired by the measured individuals, and nobody is going to do that. Further, the genetic result is being overcome by cultural trends, so we don’t really have to worry that we’re breeding a generation of dropouts!


Beauchamp, J. P. 2016. Genetic evidence for natural selection in humans in the contemporary United States. Proceedings of the National Academy of Sciences 113:7774-7779.


  1. Kevin
    Posted July 13, 2016 at 10:05 am | Permalink

    Great post. Having more children is also an economic burden, specifically with regard to higher education. More interestingly, children, not only the parents, are aware of the stigma associated with having more siblings and how that reflects, directly, upon their own opportunities. I do not see any genetic pressures with regard to the economics of more children, but there are salient cultural and societal ones at play.

  2. Posted July 13, 2016 at 10:33 am | Permalink

    Excellent. I always enjoy your critical analysis of scientific papers.

  3. peepuk
    Posted July 13, 2016 at 10:43 am | Permalink

    Quote from Stephen Hsu who works at the “Beijing Genomics Institute”, the world’s largest genomics research centre:

    “Genetic engineering will one day create the smartest humans who have ever lived.”

    from :

    “Are humans still evolving?”
    Probably yes, or at least we try 🙂

    • Zado
      Posted July 13, 2016 at 11:22 am | Permalink

      I also think genetic engineering will be the real game changer, as portrayed in the sci-fi flick Gattaca (good movie).

      As for natural selection, hasn’t that been out the window for some time now? The entire point of agriculture-based civilization is to remove Malthusian constraints on child bearing and ensure less people die young from war/disease. When you lower the threshold for reproductive success, you don’t get smarter/more athletic people on average. You just get more people.

      Which is fine by me. I’ll take civilization over a Hobbesian state of nature any day.

      • Michael Hart
        Posted July 13, 2016 at 12:59 pm | Permalink

        There are some other ways that selection could act on heritable human variation even after a decline in selection on traits associated with mortality before reproduction (sometimes called hard selection). This is a pretty good summary of those other selection mechanisms (Redneck 2015 doi: 10.1093/jhered/esv076). The general idea is that for some traits and genetic variables there may be no single absolute selection threshold or what you called a Malthusian constrain. Instead the fitness of a particular genetic variant depends on the population density of the organisms, or on the frequency of that genetic variant relative to other variants in the same population (sometimes called soft selection). So even if everybody gets a chance to survive to reproductive age and to reproduce, there may still be density- or frequency-dependent effects that lead to evolutionary changes under those forms of selection. Examples help to envision how this might work. Reznick’s paper has several examples.

        • Michael Hart
          Posted July 13, 2016 at 1:01 pm | Permalink

          Oops, that author is David Reznick…autofail

      • Diane G.
        Posted July 13, 2016 at 10:21 pm | Permalink

        “The entire point of agriculture-based civilization is to remove Malthusian constraints on child bearing and ensure less people die young from war/disease.”

        I thought agriculture led to more war? More land and resources coveted by other societies, thus in need of defense?

        • Posted July 14, 2016 at 1:53 pm | Permalink

          Also, a much higher capacity to wage war due to advances in technology and social organization. These advances are so impressive that some anthropologists even refuse to honor the conflicts of hunter-gatherers with the term “war”.

      • Gregory Kusnick
        Posted July 13, 2016 at 10:22 pm | Permalink

        As Jerry’s example of lactose tolerance shows, there are selection pressures that came into existence as a direct result of civilization.

        Similarly, if it could be shown that (for instance) literacy contributes to reproductive success, then we should expect to find selection pressure in favor of an innate aptitude for literacy, and against genetic impediments to literacy such as dyslexia.

        The bottom line is that while civilization may radically alter the contours of the selective landscape, it doesn’t flatten it out of existence.

  4. Posted July 13, 2016 at 10:45 am | Permalink

    To draw conclusions about how humans are evolving, I would think you’d need to look more broadly (as you say) than a sample of contemporary Americans of European ancestry, who comprise a small fraction of humanity and would hardly be a representative sample.

    A gene for staying in school? Had anyone but PCC said this, I would have laughed.

    • Posted July 13, 2016 at 11:47 am | Permalink

      I will need to re-read the post, but I feel pretty sure he didn’t say that there is “a gene for staying in school.”

      • Posted July 13, 2016 at 11:53 am | Permalink

        Yes, I should have said “genes associated with staying in school.”

        • Filippo
          Posted July 13, 2016 at 2:59 pm | Permalink

          I wonder if there are genes associated with intellectual curiosity. One can (be forced, or force oneself, to) persevere through through some sort of extensive training program, yet not be (much) intellectually curious. One can have precious few formal schooling opportunities, yet be a determined autodidact with great intellectual powers (e.g., Abraham Lincoln, Michael Faraday).

  5. Joseph Stans
    Posted July 13, 2016 at 10:48 am | Permalink

    New data in including millennials and evangelicals indicates that Americans are getting dumber at an alarming rate.

  6. frankschmidtmissouri
    Posted July 13, 2016 at 11:19 am | Permalink

    Your point that reproductive success has to go beyond a single generation is extremely important, given that life expectancy is going down for a significant portion of the American population.

    Are people facing decreased life expectancy likely to have more children? The differences between R-selected and K-selected species would suggest that this is a viable strategy to ensure that they would leave more descendants.

  7. Posted July 13, 2016 at 11:32 am | Permalink

    “selection for shorter women” – this is interesting because in many affluent places (including the US) there has been great increases in height since WWII, which are presumably largely “environmental”.

  8. Mark R.
    Posted July 13, 2016 at 11:45 am | Permalink

    This was a great post PCC(E), thanks.

    As far as white American women reaching menopause later (a pretty small sample as already stated), could this simply be a result of better health care and general fitness? I don’t know if there is a real correlation, but it seems logical that if women are healthier and live longer, they would continue to release eggs and release normal amounts of estrogen for a longer time; so nothing to do with genes.

  9. Avis James
    Posted July 13, 2016 at 12:20 pm | Permalink

    Thanks Jerry!

  10. Mark Sturtevant
    Posted July 13, 2016 at 12:38 pm | Permalink

    I am generally suspicious of studies that use a very broad net to find any correlation involving thousands of variables. Such an approach is guaranteed to find statistically significant correlations between something or other, and such correlations do not mean causation.
    But you say that the authors looked at this data in different and rather stringent ways, and so the data is tentatively believable. That carries a lot of weight for me, and so this is indeed interesting.

  11. µ
    Posted July 13, 2016 at 1:08 pm | Permalink

    Sexual-selection should have been operating in the population studied, perhaps far stronger than any of the natural-selection processes analyzed.

    Best to collect in such longitudinal studies therefore also data on phenotypic traits under sexual selection. Most of these sexually-selected phenotypes are more difficult to measure than hormone levels and height.

    • Diane G.
      Posted July 13, 2016 at 10:26 pm | Permalink

      If only men had dart frog colors or peacock tails…

  12. rickflick
    Posted July 13, 2016 at 4:58 pm | Permalink

    Fascinating (Mr. Spock raises an eyebrow).

    If it takes 875 years for a single year’s reduction in education, I would think we are pretty safe from most of the negative consequences one might worry about. A lot can change in that time. Control of our genetic progress is probably right around the corner. Just look at the new CRISPR tool being used now for so many areas of biology. I’d have to think that in the next 50 to 100 years we will be able to essentially design our next generations with pretty much the characteristics we want. What, me worry?

  13. Posted July 13, 2016 at 4:58 pm | Permalink

    Bit staying in school longer doesn’t mean people are smarter. Are universities being dumbed down as pressure rises to have more people attend them?

    • Diane G.
      Posted July 13, 2016 at 10:29 pm | Permalink

      They’re being dumbed down because they’re run like corporations trying to attract and maintain the customer base…


  14. Posted July 13, 2016 at 7:20 pm | Permalink

    Perhaps there are genes that influence childbearing more strongly than the ones that influence educational achievement. Maybe there are genetic components to finding babies cute and adorable. Or to being unwilling to take the time for contraception. Now that humans mostly don’t die before the reproductive years, I’d think these factors would become relatively more important.

    And then there are memeplexes that encourage childbearing

    • Gregory Kusnick
      Posted July 13, 2016 at 10:37 pm | Permalink

      As reproduction becomes further decoupled from sex (although Republicans are trying hard to reverse that trend), we should expect increasing selection pressure in favor of wanting to have children, and decreased pressure on wanting to have sex.

      But as designer babies become feasible and then mainstream, with parents able to choose the traits they want in their children, we should expect increasing selection pressure on wanting to have children who want to have children.

%d bloggers like this: