As I understand it, in some circles,  there is a burgeoning hope that practice in this generation will somehow improve performance in the next – based on a word they have heard but do not understand. That word is epigenetics.

Genes can certainly be modified in ways that persist.   For example, the cells in your skin produce more skin cells when they divide, rather than muscle cells or neurons.  Most of your cells have a copy of the entire human genome, but only certain elements are expressed in a particular type of cell, and that pattern persists  when that  kind of cell divides. We understand, to a degree, some of the chemical changes that cause these lasting changes in gene expression patterns.   One is methylation,  a method of suppressing gene activity.  It involves attaching a methyl group to a cytosine base. This methylation pattern is copied when somatic cells divide.

The question is whether A. such changes can persist into the next generation and B. if they do, is this some sort of adaptive process, rather than an occasional screwup?  We’re  interested in whether this happens in humans,  so we’ll only consider mammals.

It’s rare, but sometimes it happens.  It has only been found to happen at a few sites in the genome, and when it does happen,  only a fraction of the offspring are affected. Probably the best known example is the agouti yellow allele in mice.  Mice that carry this allele are fat, yellow, and prone to cancer and diabetes – some of them. Yellow mothers tend to have yellow babies,  while genetically identical brown mothers mostly have brown babies.  The agouti yellow allele is the product of a recent insertion in the genome, about 50 years ago.  For the overwhelming majority of genes, the epigenetic markers are reset in a new embryo, which means that epigenetic changes induced by the parent’s experiences disappear.  The embryo is back at square one.   This agouti yellow allele is screwed up – somehow the reset isn’t happening correctly.

In mice, the mammalian species in which most such investigations have been done,  the few other locations in the genome where anything like this happens are mainly retroposons and other repeated elements.

There is another way that you can get transmission across generations without genetic change.  Rats that are nurtured by stressed mothers are more likely to be stressed.  This isn’t transmitted perfectly, but it happens.  Presumably the uterine environment,  or maybe maternal behavior, is different in stressed mice in a way that stresses their offspring.   This reminds me of a science fiction story that abused this principle.  The  idea was that alligators (or maybe it was crocodiles) almost have a four-chambered heart, which is generally associated with higher metabolism and friskiness. Our protagonist operates on an alligator and soups up its heart: the now-more-vigorous animal has better blood circulation and lays healthier eggs that develop into babies that also have a working four-chambered heart. So ‘normal’ alligators were like stressed mice: fix the problem and you get to see what they’re really capable of. The problem was that the most interesting consequence was growing wings, flying around and eating people. Alligators turned out to be stunted dragons. Not so good.

Anyhow, what reason is there to believe that reading Gradshteyn and Ryzhik until your eyes bleed will plant the seeds of math to come in your descendants?  None. Oh, I can come up with a scenario, if you want: but it requires that civilization (in particular, the key part of civilization, heavy use of weird definite and indefinite integrals and vast reproductive rewards for those skilled in such things) has risen and fallen over and over again at fairly short (but irregular)  intervals, so that humans have faced this adaptive problem over and over and over again.  A little like the way in which generations of aphids do different things in the summer (parthenogenesis) than in the late fall (sexual reproduction) – although that probably depends on direct cues like length of day rather than epigenetic changes.  Something like Motie history, maybe. But  I don’t believe it.  Not even a little bit.

Nature hasn’t even figured out how to have Jewish boys be born circumcised yet.

So why are people talking about this? Why do people like Tyler Cowen invoke it to ward off evil facts?

Because they’re chuckleheads, what else?

This entry was posted in Uncategorized. Bookmark the permalink.

23 Responses to Epigenetics

  1. RS-prime says:

    Why do secular increases in height end so late in the West, generally around 1980?

    • RS-prime says:

      1980 birth cohort that is.
      Mystery’s all the greater in America since life has been relatively less malthusian here through the 1600s, 1700s, 1800s — odds of surviving to adulthood far higher IIRC.

  2. winestock says:

    Gradshteyn and Ryzhik? That’s cruel. Sure, you’re taking the cluelessness of the chuckleheads to its logical conclusion, but you should already know that all of the cool math nerds talk up Apostol, Spivak, and Rudin. The more snobbish will mention Loomis & Sternberg.

    There’s a guy at Tulane who’s writing down proofs for *all* of the formulas in Gradshteyn and Ryzhik. He’s a better man than me:

    In one of your earlier posts, you mentioned that—thanks to the study of inherited Ashkenazic neurological diseases—we should be able to identify what causes increased g-factor intelligence. Does epigenetics play no role in this at all?

    • gcochran says:

      I don’t think it does.

        • gcochran9 says:

          Because it’s stupid. The overwhelming majority of experimental evidence indicated that adaptive, Lamarckian-syle epigenetic change does not exist. Let alone get you ready for college. You’d have to have a mechanism that sensed genes that were disproportionately active (in the brain) during higher-than-typical levels of abstract thinking, or doing calculus, or whatever, and then send information down to the ovary or testes to selectively change the methylation state of those same genes in germ cells.

          Christ, I thought Lysenko was dead.

          Let me tell you, there’s one psychological trait that never changes: there’s a sucker born every minute.

          • gcochran9 says:

            Actually, it has to be more complicated than that. Those genes that are busier in the brains of people in college (ones other than those that detoxify alcohol) have to be set to have the higher activation later in life, not during embryogenesis – because they have other baby-specific jobs at that time. So this magic process would have to modify regulatory genes in a way that appropriately changes development. This isn’t just redesigning the genome to do something new, like the Tnuctipun making the Pak: this assumes that genome has the ability to redesign itself in response to brand-new circumstances. Our genome would have to be smarter than we are. I think this might be a tough job even for the Tnuctipun.

  3. Anonymous says:

    Correct nickname of the book
    “Table of Integrals, Series, and Products”
    should be just “Ryzhik”.
    It was I.M. Ryzhik ( Israil Moiseevich Ryzhik?), who compiled the first version of the book around 1930 (?), approximately 5 times thinner.
    Then Gradshteyn inherited the job to augument the book, then Geronimus and Tseitlin, then the editor of American edition Alan Jeffrey.

    • Florida resident says:

      We have our own “Handbook of Mathematical Functions”
      by Milton Abramovitz and Irene A. Stegun.
      Lots of graphs there, as opposed to Ryzhik and Gradshteyn.

  4. Anonymous says:


    “The Dutch researchers found de novo mutations which led to severe cognitive impairment. It is almost certain that there are many more mutations which lead to smaller impairment. This is evidence that mutational load is at least a partial factor in population variation in cognitive ability. See earlier post: Deleterious variants.”


  5. Glue says:

    “Oh, I can come up with a scenario, if you want: but it requires that civilization (in particular, the key part of civilization, heavy use of weird definite and indefinite integrals and vast reproductive rewards for those skilled in such things) has risen and fallen over and over again at fairly short (but irregular) intervals, so that humans have faced this adaptive problem over and over and over again.”


  6. DK says:

    Thank you, Greg! Between all the incessant epigenetics hype, someone gets it. Reassuring.

  7. Priceeqn says:

    An interesting talk last year was a fellow from a university whose name escapes me. His lab has shown that there are >100 commonly used chemicals that cause gene states (either genes will be expressed or suppressed when the F1 mother is exposed to the substance) to be inherited for at least 3 generations. Now sure, this isn’t the “good” sort of epigenetics… but that got me thinking: if natural selection can choose mothers who can vary the sex-ratio of their offspring, why not choose the ability for genes to respond to an external external environment? I’m thinking of the womb with a developing offspring as one environment, the mother’s insides as another environment, and that environment outside the mother’s body as of course the environment that we normally might think of. The potential epigenetic effect of our starvation on future generations, or “caloric restriction,” is a provocative one, no?

    • Priceeqn says:

      In addition, there are now 3 times more imprinted genes in mice than we thought there were about 1 year ago. Now that leaves some real potential for epigenetic effects!

      • gcochran9 says:

        It wouldn’t be a comment string without fools.

        • Priceeqn says:

          LOL. I agree that epigenetics has been overhyped. There was a horrible book published some years ago, “Evolution in 4 Dimensions,” and that summarizes all the b.s. Personally, I don’t see what all the excitement is… MUP genes are cool but they’re just a single type of so-called epigene. Imprinting is where the action is at.

    • DK says:

      His lab has shown that there are >100 commonly used chemicals that cause gene states (either genes will be expressed or suppressed when the F1 mother is exposed to the substance) to be inherited for at least 3 generations.

      Dosage? You know that this stuff is meaningless without specifying dosage, don’t you? Also, what’s the definition of “inherited for at least 3 generations”? 100% of the F3?

      • Anonymous says:

        Dosages aren’t very high for these sorts of studies… But you would want to read the paper to see what the (suspected) toxin doses were. The exposure doses are supposed to be too small to cause harm. What’s interesting though is that the mother is exposed, and there are no apparent fitness effects, then generation 2 is born when a daughter of the exposed mom was mated to an offspring of an unexposed male and there is a fitness decrease but no disease phenotype… Then those mice are mated with non-exposed male mice and yet their male offspring have the disease phenotype. It was funny because afterward, due to the study having been funded by the NIH, the speaker had to give a disclaimer that the effects observed apply only to mice, and not of course to humans.

      • DK says:

        If it’s one of these that you have in mind:

        then dosages *are* very high (Table S1A). BPA: 1% of oral LD50 given as intraperitoneal injection every day for the entire second week of pregnancy. This is about 5 mg daily. Compare with human typical exposures in the range 0.007-0.07 mg/day. Per body weight difference, add a factor of 400-500X.

  8. Kiwiguy says:

    A 2010 paper by the late James F Crow paper discussed ‘On epistasis: why it is unimportant in polygenic directional selection’.

    “… Students of development, evo-devo and human genetics often place great emphasis on epistasis. Usually they are identifying individual genes, and naturally the interactions among these are of the very essence of understanding. The individual gene effects are usually large enough for considerable epistasis to be expected.

    Quantitative genetics has a contrasting view. The foregoing analysis shows that, under typical conditions, the rate of change under selection is given by the additive genetic variance or covariance. Any attempt to include epistatic terms in prediction formulae is likely to do more harm than good. Animal and plant breeders who ignored epistasis, for whatever reasons, good or bad, were nevertheless on the right track. And prediction formulae based on simple heritability measurements are appropriate.”

    As Hsu notes:

    “…strong interactions at the level of individual genes do not preclude a linear (additive) analysis of population variation and natural selection.”


  9. Toddy Cat says:

    “Christ, I thought Lysenko was dead.

    Never. He’ll survive until the last liberal (and neo-con) is dead and buried.

  10. phenotypes are scored from 1 to 5 based on coat color. Fully yellow mice are scored as 1, and fully agouti mice are scored as 5. Phenotypes of mosaic mice range from mostly yellow ( 2 ) to mottled yellow/agouti ( 3 ) to mostly agouti ( 4 ).

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s