I think it depends on whether this mechanism exactly tracks genetic resemblance in monozygotic vs. dizygotic twins. If it doesn't, then twin studies continue to prove something about genes (and not this other thing), and molecular genetic studies obviously prove something about genes (and not this other thing), so the mismatch is still surprising.
If it does track genetic resemblance (ie much stronger in monozygotic than dizygotic twins) then I think it might work, although I'd have to think about it harder to be sure.
Epigenetic effects could be stronger in monozygotic twins because these twins split a few days after fertilization (and up to that point, they are the same group of cells).
It would be quite interesting to track the similarity of dichorionic vs. diamniotic vs. monoamniotic twins (which split at progressively later stages) on various traits and epigenetic marks. You mentioned this study: https://pubmed.ncbi.nlm.nih.gov/26410687/ which addresses some of this but l think there's more to be discovered here.
Yes! This was my first thought. The comment above merits discussion and critique. If anyone here can offer a strong argument against the opinion above, please offer it!
o3 suggest that epigenetics markers persist more through mitosis than through gametogenesis, so I would expect epigenetic correlations to be stronger in monozygotic vs dyzgotic twins.
I think in the end missing heritability will be resolved as "our computational genetic models made too many simplifications" and so genes will explain all the twins heritability, but require more sophisticated models involving more biological mechanisms.
Hypothetically, a mechanism that match this "dark inheritance" situation is the fine details of contents of the egg cell. e.g. Is protein A present in the egg cell in an amount over threshold At? How common is enzyme B relative to enzyme C? How close to optimally efficient are the ratios of organelles contributing to synthesis? and so on.
Given this hypothesis, we'd expect monozygotic twins to be most similar because they started from the same egg (c.f. twin studies). We'd expect dizygotic twins to be more like non-twin siblings of the same parents -- as they differ in fine details of the egg contents of the two eggs from the same mother, like two random draws from the same distribution (c.f. Sib-Regression). And we'd expect non-sibling relatives to be less alike that siblings but more alike than strangers, as with eggs from different but related mothers, it's like two random draws from two distributions that were themselves random draws from a common distribution (c.f. RDR).
This mechanism should operate Darwinianly, not Lamarckianly, because a woman's eggs are already present at birth and not filtered in any obvious way by her life events. If I'm thinking about it right, it also predicts that if a woman has two children by two unrelated men, versus if a man has two children by two unrelated women, that the former pair should be more similar to each other on average than the latter pair, i.e. the former should have higher RDR scores than the latter. ChatGPT tells me this is an empirically true finding about half-siblings. I suppose if this mechanism is real it's hard to separate from other environmental effects, and difficult to directly test in humans without harming the egg.
There is. Or rather, not exactly; it's still DNA, it just isn't human DNA. What about mitochondrial DNA- has anyone ever investigated whether this affects IQ?
Mitochondrial DNA is identical in both fraternal and identical twins, as it's inherited from the mother in a non-sexually-recombining way.
Wouldn't all twins (identical and fraternal) get the same mitochondrial DNA from their mother? It seems like this would cause us to underestimate heritability from these studies, not overestimate it.
Actually, this leads to a question: Do all children inherit the same mtDNA from their parents? Or is there some variation in this that might differ between identical and fraternal twins?
> Do all children inherit the same mtDNA from their parents?
Funny you should ask. I was curious about this subject and was digging into that question a while back. So, it's pretty fresh in my head.
Statistically, it's very likely that they do. The human mtDNA mutation has been estimated to occur approximately once every 30 generations (depending on the study methodology, this estimate may be slightly higher or lower). There are about 16.5K base pairs in mtDNA. The odds work out to something like one mtDNA mutation per 40 births.
A human egg cell has 100's of thousands of mitochondria though. I wonder how common it is that there might be multiple versions of mtDNA within those. In theory you could have multiple 'lineages' of mtDNA in your body that would pass down (possibly in different ratios through random chance when the egg cells formed) generationally. Probably not anything impactful, but I wonder to what extent it happens.
Good point. Hasn't the accumulation of mtDNA mutations been linked to diseases like Parkinson’s and certain cancers? But it would only matter across future generations if the mutation happens in the ovum.
I thought of the mitochondria too. Fraternal twins would get the mitochondrial effects only, identical twins would get genome plus mitochondrial effect
Transgenerational epigenetic inheritance doesn't seem to be important in humans. There are some epigenetic resets that occur early in embryonic development, plus the effects of epigenetic modifications acquired during life are much more important. The latter would show up as environment (some shared, some non-shared) in twin studies.
In your article, you wrote that, "Disruptions to the establishment of epigenetic marks in germ cells during fetal development can certainly affect the next generation..."
Wouldn’t that suggest the mechanism you describe could potentially account for part of the observed gap?
Yes, I know, but that doesn't answer my question. My point is that intergenerational—not transgenerational—effects might explain the gap. I should have been more clear. That's why I asked that question.
There *IS a non-DNA way to inherit traits. Many traits are inherited via the cytoplasm, e.g., IIUC, the structure of the ribosome. Also mitochondria are (essentially) only inherited via the cytoplasm, and that contributes to average energy level. There are certain to be lots of others.
AFAIK, the small and large subunits of the ribosomes are constructed in the nucleolus based on highly-conserved genes, and these are later put together in the cytoplasm. In fact, the genes are so highly conserved, that they are used to establish phylogenetic trees. What structure of the ribosome are you referring to as being inherited in the cytoplasm?
Yes, they are highly conserved, but the "highly conserved genes" part wasn't known when I read about it. And, IIRC, they were assumed to be metabolic inheritance via the maternal cytoplasm.
I'm sure lots of folks have, but there are lots of different studies being compared, and how frequently was this considered? Most of the time people don't even seem to think about cytoplasmic inheritance, even though when they *do* consider it they realize it has to be happening.
Well it would just show up as higher phenotypic correlations with the mother than with the father. AFAIK those doesn't exist for anything apart from the known sex-linked traits.
I'm not going to pretend that I can understand everything discussed above - or rather - that I can digest it meaningfully in my first read through.
But I'm curious as to whether genome wide polygenetic studies are good at identifying gene-environment interactions.
Most people think of the environment piece of gene-environment interactions as something obviously related. For IQ, you might expect it to be the parenting strategy of the parents that will evoke the best response for the IQ genes. But that's not necessarily the case. And a gene that manifests higher IQ, might not look like a gene for higher IQ unless its unfolding in the right environment - which can be a wide range of things.
I know that twin studies to some degree are meant to understand this, and there are also twin studies of twins reared apart, which should more strongly take care of this. But its still the case that, even when twins are reared a part, the environmental component that elicits the effect of the gene is so common that twins reared apart don't functionally constitute twins reared in different environments.
For example - and this is just a really random made up example - if full IQ development, and thus full expression of the heritability of IQ, depended on higher levels of choline intake, then you would see the heritability and full expression of IQ among people that eat a things higher in choline, like fish and meat. The heritability among different sets of twins would depend on their families, and perhaps individual preferences for, meat and fish. Twins reared apart could easily end up seeming very similar if they were likely to stay among households that had culturally very similar diet patterns.
I'm just really curious as to whether gene x environment interactions are extremely difficult to parse when you are also looking at polygenic traits with gene numbers in the 100s or 1000s?
Good point - things that are genuinely common in environments will never show up as "environmental influence". Of on statistics language, as a "treatment", the environment is neither random nor normally distributed.
Michael Levin's theory on bioelectricity seems like it would explain the discrepency, but I never studied biology and I don't know if his work is considered fringe/fake or genuine cutting edge.
Also Idk if he's actually the creator of that area, i just heard him do an interview once where he talked about making two-headed flatworms that gave birth to other two-headed flatworms, without ever touching the genes at all (because that's apparently possible, I guess???)
"Dynastic effects" and "genetic nurture" are very different processes and do not belong in the same category. Dynastic effects basically refers to the kind of phenomenon you mention of Norman surnames correlating with high status many years later, while genetic nurture refers to direct genetic effects in parents affecting their influence on children. Gemini's summary is as good as any:
"Genetic nurture is about what your parents do for you environmentally because of their genes.
Dynastic effects are more about the broader social and economic standing and inherited environmental context of your family lineage over generations, which is correlated with your genetic ancestry."
The most recent work (e.g. Nature Hum Behav Nivard 2024) suggests that for educational attainment, genetic nurture is very weak, while dynastic effects are quite considerable.
I'd be mildly concerned that correcting for population structure in GWAS analysis may be "the sociologist's fallacy" at work again, though if twin studies vs. sib-regression and RDR is yielding contradictory estimates for kidney function then something fishy really is going on.
As someone who has not yet seriously looked into adoption but who is open to the idea, your child psychiatric hospital anecdote is terrifying. Isn't the whole point of adoption (for many people) to offer a better life to kids who otherwise might have had a terrible time due to awful parents/environment? But actually adopting those kids puts you at a much higher chance of raising antisocial parasites, even if you treat them exactly like your other kids? Such a depressing thought. If this is a grave oversimplification, I'm happy to be proven wrong.
I think you can mostly avoid this if you're careful about who/where you adopt from. Some children are in orphanages because their parents died in a war or something, and they're pretty normal. If it's because the parents were involved in crime or drugs or something, then yeah, I think there's a significant risk that the child will be pretty tough to raise.
How much of that is explained by FASD and the like? (Or the developmental effects of extreme neglect in early childhood?) I was about to say “not that that makes much difference to adoption pessimism”, but FASD is actually not so hard for a physician (or even a trained layperson: kindergarten teachers in some countries can tell) to diagnose, no?
Well, sure; I asked how much. (I am not a medical professional, unlike you; my sense that there is more "FASD and the like" than many assume comes from talking to medical professionals and other people who know more than I do. People in the US now may drink less than they used to, but they also take more drugs.)
In the US the rate of FASD kids is somewhere between 1% and 5%, but 19% of foster kids have FASD (https://pubmed.ncbi.nlm.nih.gov/39031634/). It also looks like 30% of kids diagnosed with FASD end up in foster care. Foster care and adoption are not the same thing, but I would expect the populations to be more similar than they are different.
I looked into this once as when researching the whole "alcohol-during-pregnancy" topic. My takeaway: the answer is "yes" for FAS (Fetal Alcohol Syndrome) and very much "no" for FASD (Fetal Alcohol Spectrum Disorder). The former is characterised by the typical FAS face (small eyes, smooth philtrum, thin upper lip), low birth weight, microcephaly and various pretty severe developmental disorders.
The latter spectrum's diagnostic criteria are ridiculously loose leading some studies to conclude that as many as 10% of children have FASD (https://pmc.ncbi.nlm.nih.gov/articles/PMC5839298/). Add to that everyone who is "on the spectrum" or has ADHS and nobody normal is left.
The Christians I am familiar with are eager to adopt the children of homeless drug addicts. The children of recidivist criminals, I’m not so sure.
Either way: they’ve imbibed the conventional wisdom about heredity and life outcomes and while it only strengthens their compassionate determination to provide better-than-the-best in terms of nutrition, screens (none), reading, homeschooling if school is harsh and so forth - it also means their mantra is that the goal is simply a happy life.
>Some children are in orphanages because their parents died in a war or something, and they're pretty normal.
This is extremely unlikely in a modern developed country. Even in the improbable event that a child is unfortunate enough to lose both parents in a war, custody will almost always be taken over by a surviving close relative. You can try to adopt from an orphanage in the third world, but even then, they tend to prioritize prospective adopters from the same nation over foreigners.
If a child is available for adoption by a stranger in the US in 2025, you can pretty safely assume that "the parents were involved in crime or drugs or something".
> If a child is available for adoption by a stranger in the US in 2025, you can pretty safely assume that "the parents were involved in crime or drugs or something"
Perhaps this is covered by the qualifier "pretty", but there's also the possibility that it was simply an unwanted pregnancy. From what I can gather, in the US, it is simultaneously true that the majority of unwanted pregnancies are aborted (here counting anything after conception as a pregnancy***), and that the majority of children who are adopted were born to single mothers. But we're not talking about base rates here. The relevant question is "what percentage of children who end up adopted were unwanted pregnancies", which is surprisingly hard to find a straight answer to as it doesn't seem like this has been measured directly, but we can vaguely gesture at the reasonable assumption that most unwanted pregnancies are the result of circumstances that are conducive to the couple not staying together. Rape is the extreme example, but it could also be things like disagreement about the morality of abortion/contraception, one parent wanted a child when the other didn't, or both parents wanted to abort but neither could afford it.
Although, I must admit, being involved in crime or drugs would certainly make raising a child more difficult, as it would require reducing your involvement with one to make room for the other (keeping other cofounders like employment constant). So, in a sense, a child conceived by someone involved in crime or drugs is more likely to be an "unwanted pregnancy" than the base rate of unwanted pregnancies, to the extent that you trust someone involved in crime or (especially) drugs to know what they want.
Also, in the case of rape specifically, rape itself is a crime, so you could technically count it as "being involved in crime", with the caveat that the involvement is unwilling (as are many other forms of crime involvement, to varying degrees, so I don't think this association is totally unwarranted).
***I want to be explicitly clear that I'm not making any sort of political, moral, or religious statement by defining a pregnancy that way. In context I was merely pointing out that the availability of post-coital contraception prevents a large share of pregnancies that would have otherwise been aborted later or carried to full term.
With respect to children of addicts becoming criminals, do we know how much of this is genetic, and how much can be attributed to gestational effects, such as fetal alcohol syndrome?
> If it's because the parents were involved in crime or drugs or something, then yeah, I think there's a significant risk that the child will be pretty tough to raise.
Remember, your psych hospital sample will have a LOT of selection effects.
Maybe 90% of the kids of duggie/in jail parents who get adopted by those good christian families turn out totally fine? But you only saw the few cases that didn't.
I think the existence of these cases can tell us of the existence of an extreme tail to the distribution, but tells us less about the overall distribution shape.
I think you're right, and Scott and Cremiux are misinterpreting the posted study. The key line is:
"Note, however, that a genetic influence is not sufficient to produce criminal
convictions in the adoptee. Of those
adoptees whose biological parents have
three or more convictions, 75 percent
never received a court conviction. Another way of expressing this concentration of crime is that the chronic male
adoptee offenders with biological parents
having three or more offenses number
only 37. They make up 1 percent of the
3,718 male adoptees in Table 2 but are
responsible for 30 percent of the male
adoptee convictions."
So, 1% of male adoptees in the study have a parent with three or more convictions *and* themselves have been convicted of a crime. 3/4 of children whose parents have been convicted 3+ times are never convicted.
Oh I definitely agree with that. But Cremeiux (quoted by Scott) said: "The 1% of male adoptees whose parents had 3+ convictions were responsible for 30% of the sample's convictions." That's just not accurate: male adoptees whose parents had 3+ convictions made up 4% of the sample, not 1%. The 1% are the 1/4 of those adoptees with any criminal convictions. And Scott has come out explicitly in favor of nit-picking critiques of factual inaccuracies, so I feel pretty good about pointing this out :).
Bigger picture, I think this study is consistent with a world where, a) a large share of kids who end up committing crimes/having serious behavioral and emotional problems despite being raised by fairly normal families are adoptees (as Scott observed), and b) most adoptees raised by fairly normal families don't end up committing crimes/having serious behavioral and emotional problems.
Anecdata: I went to a fancy Christian private school (because my parents bankrupted themselves to find a somewhat-tolerable schooling environment for their weird son, not because I grew up fancy meself), and two of my classmates—out of about 100 total—were adopted through a similar "help the worst-off" sort of idea on the part of the parents. It was /very evident/ that they were... different... from the rest; you'd guess they'd been adopted even if they hadn't freely admitted it. The girl, as far as I know, didn't do great but didn't get in trouble; her brother did, indeed, end up arrested for assault.
Mostly, you might call it "difficulty fitting in", maybe.¹ The girl was nice enough, I think—not that I interacted with her much—but struggled with grades (always having to stay behind for extra help, retaking tests, etc.; in my memory, at least, more-so than any other female student in our grade) & occasionally made some odd comments in class or had weird dramatic outbursts. Nothing too bad, I mean, just stuff like suddenly telling a story about having bad diarrhea or the like (not an actual example, heh, I just mean /that sort of thing/).
The boy was... odder. He was caught masturbating /in class/ once, for example; always wore a large windbreaker thing, and apparently figured he could hide it underneath... He would become somewhat violent with people even then, e.g. when caught /very obviously/ cheating in a card-game (not even the sort for money, just a regular game of Uno or Pokemon something—so no one really cared very much, but it seemed to be the fact that no one believed him about the cheating being "accidental" that set him off); was twice caught with vodka in a water bottle (which is possibly more unusual at that place than in a regular high-school; if anyone else there did something like this, at least, I never heard about it); had even more difficulty with the educational material than did his sister; etc.
--------------------------
¹: (Not that this is a moral failing, or anything. I had struggled to fit in at all my previous schools, too, and for my first ~year at this one.)
Do you have positive knowledge that none of the other 98 were adopted in this manner? You might be providing strong anecdata that "strangeness" predicts "worse-off adoption" but weak anecdata that "worse-off adoption" predicts "strangeness".
I suspect babies put up for adoption in the postwar golden age of adoption, such as myself, tended to be more promising on average because there were a lot of social custom reasons for not raising the baby yourself that aren't as stringent anymore.
This is an important point! The social context for why children get put up for adoption matters a lot in what you'd expect the heritable traits of the kids to be, and the fact that this changes over time in different societies probably makes all kinds of adoption studies done at different times hard to compare. US adoptees in 1950 and in 2000 are probably very different in a lot of ways because of the changes in why people gave up their kids for adoption.
It matters for what you expected adopted children to be like, but not so much for whether adopted children are similar to their adoptive families or to their actual families.
Steve Jobs was put up for adoption in 1955 because his biological parents couldn't agree to get married. Then, they did get married and had a second child, the fine novelist Mona Simpson who gave the much-admired eulogy for Jobs at his funeral, even though they didn't meet until their 20s.
Which war marks the postwar golden age of adoption?
I am told that my grandfather was given up by his mother during the Great Depression on the theory that she couldn't afford to raise that many children. (She didn't give all of them up - just him.)
This permanently damaged the relationship between the two of them.
I'm thinking the prosperous 1950s-1960s, when society still frowned upon single mothers, saw a lot of women who were otherwise highly functional other than not having a ring giving up their babies for adoption.
Steve Jobs, born 1955, is an example of the quality of baby that could be adopted during the baby boom.
Personally, I always felt sorry for his adoptive little sister. She was just a random person who wanted to live her life, but she could never out-argue her big brother, who happened to be the World's Greatest Salesman.
A friend of mine said she'd deliberately chosen to adopt a kid with identifiable mental disorders (autism etc.). This is because, in the national adoption system (UK), healthy, happy kids from wonderful parents who tragically died together in a car crash, with no grandparents, aunties/uncles or family friends to absorb these kids, are vanishingly rare.
So, given that they're in the adoption system, all kids are problematic for *some* reason - whether their own or their parents. My friend judged that it would be less risky to choose a child whose problem is known and manageable.
Of course, waiting outside Ukrainian hospitals to whisk away the healthiest war orphans you can find is another option, but I suspect that it's a logistical nightmare.
Adopting from foreign countries seems to be the best way to improve a child’s environment. Twin studies are almost always comparing children in the same country, and cross country environments very significantly more.
There are almost no children left to adopt, unless you adopt severely disabled children, teenagers or do foster care (where the goal is reunification). There are like 30 times more prospective adoptive parents than there are children given up for adoption.
Read about the Baby Scoop Era and the investigations South Korea is now doing into their old adoption agencies. Even back then (~50's-90's), the demand for children was bigger than the supply. Most children "given up" were actually cases of hardcore pressuring young single mothers (by eg. only wanting to give them money or medical care if they gave up their baby) and even straight up kidnapping. Adoption was a business.
Not adopted but it still bothers me how uneducated the public is on the subject.
A number of European countries have recently stopped all international adoptions, because they couldn't accurately verify each case to see that it wasn't human trafficking.
The idea that there's millions of misty eyed orphans waiting for parents is just that, a cultural myth. In the 50's through 70's amplified by the prevailing idea that it was better to give up a baby than let an unmarried woman raise it, plus how shameful pregnancy out of wedlock was. I think it's extremely psychologically difficult to give up one's baby and almost any woman would choose an abortion. Cases where each parents are tragically dead and there are no other relatives (always first choice) left to raise it are vanishingly rare.
> Even back then (~50's-90's), the demand for children was bigger than the supply. Most children "given up" were actually cases of hardcore pressuring young single mothers (by eg. only wanting to give them money or medical care if they gave up their baby) and even straight up kidnapping.
I had a Chinese teacher (in China) who noted that when she was born, because she was a girl, her grandfather had advised her mother to give her up for adoption (making it legal to have another child), but her mother refused.
Maybe it's "woke," but as an educator with experience teaching "different" adopted children, and having adopted siblings myself, I find the idea of calling such children "antisocial parasites" both repulsive and dangerous.
"The children are always ours, every single one of them, all over the globe; and I am beginning to suspect that whoever is incapable of recognizing this may be incapable of morality."
I don't think it's woke, I think your reaction is normal and appropriate. I'd just like to mention that my antisocial parasite comment was strictly focused on this line from Scott's article: "Then they promptly proceeded to commit crime / get addicted to drugs / go to jail / abuse people". At face value, everything here except for "abuse people" is not that awful. But adoption, to me - and again I haven't given it much thought yet - is about choosing a child that you bring into your family. That choice will be made on your own terms, but I reckon a whole load of criteria that people routinely use when choosing a child to adopt are politically incorrect, or otherwise criteria they wouldn't feel super comfortable saying out loud. This can lead to certain subsets of kids being less adopted than others.
By using the word "antisocial parasite", I tried to capture the essence of the following feeling: If I am given a choice when it comes to adopting a kid, including the existential non-zero risk of the kid destroying my and my partner's lives, I will stay away from those that have (apparently) much higher risks of doing just that. It doesn't invalidate their existence and their right to a future and a loving family - but I won't be the one to take them on, and I suppose a lot of people wouldn't, if they knew about those risks upfront.
If I can butt in: I wonder if Jake's objection also comes from misunderstanding a subtle detail of what you said? It sounds to me like Jake is repulsed by the idea of calling a child an "antisocial parasite". Which I would agree with. But as I read it, I thought you were talking about how they'd be behaving once reaching, say, 16+. If you meant it as I understood it, you might want to clarify that.
I think in an atmosphere where many young people with the means to have their own children, forgo doing so (or even marrying) because it feels like a risk - it would be a little strange to expect a wholesale anything-goes attitude toward adopting those ill-parented children society is producing in numbers.
As usual, it will be down to those despised folks, the fundamentalist Christians, always hopeful, this world not entirely real to them.
You are ignoring the obvious difference that humans choose to adopt children, rather than being deceived into it. The analogue here is a woman cheating and not telling the man that the resultant child isn't his, not adoption.
One could argue that blank-slatists are in fact deceived into adopting children by antisocial parents, but it's typically not the parents doing the deception (though the children themselves might, to their however limited ability).
I'm always happy to see a "Much More Than You Wanted To Know"—and this one's even better than most, because "missing heritability" is something I've been quite puzzled about (& I trust our amigo Scott "Yvain" Alexander to make a good run at it if anyone can👌). Cheers.
On priors, I would suppose there's something we're missing with the newer methods—partly because that sort of thing happens all the time ("this thing actually worked this other way all along, oops!"), and partly because I agree that it's difficult to see exactly how major error could creep into the twin studies.
--------------------------
(Y'know, a little while ago I had an idea—possibly even a cromulent one!—for how to square some of the gap away; or, rather, for why it might not be as worrisome as it first appears... but I want to check me notes before I risk saying something totally wrong & dumb / something that turns out to be the first thing everyone thought of already. Now, where in my hundreds of untitled drafts /is/ it...)
"On priors, I would suppose there's something we're missing with the newer methods—partly because that sort of thing happens all the time ("this thing actually worked this other way all along, oops!"), and partly because I agree that it's difficult to see exactly how major error could creep in on the twin studies."
Yeah, these results should definitely make us less confident about the high heritability (though a large part of what is going on might be the focus on EA, like Scott said), but less confident in a case where we started out highly confident still can leave us confident. Though of course eventually the issue becomes harder and harder to ignore. I just don't think that the 'this definitely means something is wrong with adoption/ twin studies' can be said to have arrived until the system looking for genes is accounting for stuff other than common SNPs.
Perhaps something is wrong with the 100+ year history of twin and adoption studies. Alternatively, something might be not quite right with the 12 year history of GWAS studies, just as most of the candidate gene studies of 22 years ago flamed out.
"Cromulent" seems to be the 2025 World of the Year. I'm seeing it all the time in recent weeks. I had to look up what it means ("acceptable or adequate") a few weeks ago.
"Cromulent first appeared in the February 18, 1996 episode of The Simpsons called "Lisa the Iconoclast," in what could be considered a throw-away line given during the opening credits. The schoolchildren of Springfield are watching a film about the founding father of Springfield, Jebediah Springfield. The film ends with Jebediah intoning, “A noble spirit embiggens the smallest man.” One teacher at the back of the room leans over to another and says that she’d never heard the word embiggen before she moved to Springfield. “I don't know why,” the other teacher replies. “It's a perfectly cromulent word.”"
“Embiggen” was actually the word that entered my vocabulary from that scene, in near-earnest at this point.
Is that the one where the schoolkids have an essay or art contest or something and the top 67 out of 68 entries will be displayed in the administration building? I loved that more than I can say.
Even if we never understand DNA or the brain, we still had a golden age.
I’ve read a lot of behavioural genetics. When people cite the "missing heritability problem" I think they should add, "of course this also applies to height".
One reason for not citing it is it diminishes their "argument" (everyone knows that height is mostly genetic). I can’t think of another explanation, although they might be repeating someone else’s argument without being aware of the issue.
Stephen Hsu has a paper, "Determination of Nonlinear Genetic Architecture using Compressed Sensing" and another, "Accurate Genomic Prediction Of Human Height". Worth a read. My simplest summary, from reading these a while back - you need large sample sizes, and the relationship between accuracy of genomic prediction and sample size is not linear.
Everybody knows height is mostly genetic *except in environments in which nutritional deprivation is common*, in which case being short gets stigmatised in ways different from the way it gets stigmatised elsewhere (in some countries (most of them, in the recent past), it correlates strongly with the class you were born in, because of obvious, cause-and-effect non-genetic reasons).
I don't think this is drop-down obvious. We know that Japanese height was stunted in the early 20th century. But the Japanese weren't starving to death - they just didn't have enough protein or something. But if this can happen to seemingly-advanced populations meeting their calorie requirements, how do we know that we've finally got nutrition right or that remaining variance in nutrition isn't enough to make a big difference?
...I mean, I think we actually know because of twin studies, but if we didn't have them, or doubted them, I don't think we would know through common sense alone.
Going purely off reflexive intuition, sexual selection in intelligence will lead to mating pairs that are roughly aligned: smarter people want smarter mates, dumber want dumber. It's viscerally uncomfortable for both parties for one half to be vastly more/less intelligent than the other.
Meanwhile for height: the female desire for tall men is standard, and the male desire for a woman shorter than them is as well. But it's not obvious that, say, a 5'1" female prefers a 5'6" man to a 6'6" man, nor that a 6'6" man prefers a 6'0" woman to a 5'3" woman.
Sorry, I meant there's a lot of sexual selection for height in males. For females, it doesn't seem to matter nearly as much.
> It's viscerally uncomfortable for both parties for one half to be vastly more/less intelligent than the other.
I'm not so sure about that. There's the stereotype of eg scientists picking partners early on that are their intellectual equal (eg Albert Einstein's first wive), and later when they become successful they switch to a prettier wife (eg Albert Einstein's second wife).
There's some direct sexual selection for intelligence, but also lots of indirect selection, because (at least these days) intelligence correlates heavily with success and wealth.
Barbara Tuchman's prequel to "The Guns of August," "The Proud Tower" about aristocratic Europe before the Great War, starts off with a chapter about how tall Lord Salisbury's Tory cabinet of 1895 was: about 5 or 6 inches taller than the average British man.
I don't think, however, that there are major class differences in height anymore, at least among whites and blacks. A large fraction of NBA stars came from the underclass (e.g., 6'8" 260 pound LeBron James was born into the bottom of society, which, by the way, is why I'm so impressed with his incredibly consistent career), although the latest generation tends to more bourgeois.
I was struck by what you said about black people and lactose intolerance. I hadn’t heard that. I don’t know if you meant black Americans or black Africans. But on the subject of nutrition, milk for children was the holy grail for a long time. For growth and good bones. I was still getting a cup of milk, not water, with dinner when I was 12 years old.
I don’t know what to think about the importance of milk in connection with lactose intolerance.
Is milk good so lots of people still drink it despite discomfort?
Was milk pointless?
Or just valuable for easy availability of calories to those who can drink it?
Milk was important for kids when the diet of the average westerner consisted mostly of grains and fat. With the increase in meat, cheese and eggs consumption in the last century milk is not that important anymore.
Meat consumption hasn't changed much over the last ~60 years, and I seem to recall Taubes and company arguing that pre-1920 consumption of meat was quite substantial as well. I'm inclined to think the push for milk consumption was mostly a dairy industry propaganda campaign (for better or worse).
My understanding is that most children tolerate lactose but become intolerant as they get older, and the European lactose-tolerance gene is significant because it makes people keep the tolerance into adulthood and beyond.
It's a bit more complex than that. There are two common ways to be lactose-tolerant.
One is as you describe: when a normal human would lose the ability to make lactase, you just keep making it.
The other is finer-tuned. You keep making lactase as long as you have lactose to digest. If lactose leaves your diet, you stop making lactase, and you won't start again. This is closer to the human norm, but the point at which you stop making lactase can be delayed indefinitely as long as you keep drinking milk.
After I stopped drinking milk for a period I discovered that I'm no longer lactose-tolerant but I kept consuming milk and other fresh dairy and eventually I regained my ability to digest them.
Not sure if my body started to make lactase again or it's a change in gut biome but it works and later I've seen people mentioning this method on youtube so I'm not the only one.
Most people can't digest lactose as adults. A few can; northern Europeans are the most prominent such group. But the Maasai, traditional cattle herders, are another. I understand that the Mongols had a non-genetic method of digesting lactose; their traditional diet (including a good amount of horse milk) promoted, and possibly seeded, intestinal bacteria that would digest the lactose for them. I don't know to what extent modern Mongols still do this, but it seems like it must have persisted into the modern era for it to be something we know about at all.
Milk is healthy, because that's its purpose! It's healthy even if you can't digest the sugar; it still contains protein, fat, and various bonuses. But if you can't digest the sugar, drinking it will cause unpleasant problems.
> Is milk good so lots of people still drink it despite discomfort?
I'm always surprised that milk is so widely available in China, given that over 90% of Chinese can't digest lactose. I think they view it as a health food, much as you suggest, where it's ok if you don't like drinking it, because it's health food! I've wondered whether they decided it must be good by observing that Europeans drank it.
My experience with Native Americans suggests that once you've experienced ice cream you'll keep eating it no matter how much it hurts. I wonder how much that sort of effect is responsible for Chinese consumption.
> Mongols were still drinking a heck of a lot of milk as of ~20 years ago, per this book
Local? Imported? Horse? Cow? Sheep? Pasteurized? Raw? Raw local horse milk would be expected to provide a lot more of whatever it is than imported UHT milk.
> once you've experienced ice cream you'll keep eating it no matter how much it hurts. I wonder how much that sort of effect is responsible for Chinese consumption.
Well, none, because ice cream and milk are different things. There are ice cream bars in China too; I'm not surprised by them.
I was motivated to check on whether ice cream contains a similar amount of lactose to milk. The answer appears to be yes. (Although there are some low-lactose brands, apparently.) I also discovered that the reason lactose-intolerant people don't have problems with yogurt isn't that the yogurt is low in lactose! It must be the bacteria in the yogurt.
(Cheese and butter, unlike ice cream and yogurt, have had their lactose processed out.)
For ice cream, something that's relevant is that the effects of lactose intolerance may be muted by other food consumed at the same time. A Chinese teacher of mine was shocked when I planned to have milk for lunch (without other food, just milk), and warned me that drinking milk on an empty stomach causes diarrhea. She was confused by my response that that wouldn't happen to me because I'm white.
So the Chinese don't appear to realize that the ability to digest milk is a racial trait. They think it's unpleasant for everyone.
There have been a lot of identical twins in the NBA because height is so important to basketball success, but very few identical twins in major league baseball because baseball seems to be largely a subtle knack. Or something.
Gusev is being disingenuous. Even if intelligence is not like height, the fact that polygenic scores *can't even fully explain the known variance in height* should make everyone very suspicious of them.
(I made this reply comment above): I thought of nonlinearity too. The twin studies compare the effect of different sets of genes. The DNA studies are predictions using different individual genes as variables. Are these linear predictions? The post said interaction terms don't explain anything. How about polynomials of the individual genes ? This is easy to do.
I was going to mention Height too, but it seems there *isnt* much missing DNA explanatory power there-- the post has a bar graph showing heritability of about 10 biological traits, of which height is one. On the other hand, most of those 10 DO have the same missing DNA explanatory power as IQ.
> I was going to mention Height too, but it seems there *isnt* much missing DNA explanatory power there-- the post has a bar graph showing heritability of about 10 biological traits, of which height is one.
Looking at that bar graph, it shows the same problem for height as for everything else?
-----
There's something odd going on in Scott's essay where "heritability" and "variance explained" are often treated as synonymous. There are also times where he calls out the difference between the concepts! (Though not in those terms.)
Anyway, "variance" is an abstruse property of numerical distributions. People like working with it because, where you have a random variable [a "random variable" is a distribution, not a single value] that is composed of contributions from other random variables, you can divide up the variance of the composite random variable into contributions from each of the component ones.
(This is conceptually similar to how people like working with "least squares" because it's easier to differentiate the square function than it is to differentiate the absolute value function.)
So, variance is a concept with no clear meaning, but it *is* clear how much of the variance you're looking at is in some sense due to particular other sources.
"Heritability" is the percentage of variance in a trait that is explained, in this technical sense, by variance in genetics among the population of interest. Twin studies aim to measure this quantity directly.
But GWAS studies don't. They aim to find sections of the genome that affect the trait. Once you've concluded the GWAS, you can do a calculation of how much variance in the trait is explained by the regions you found. But that's not an estimate of "heritability".
You could also work directly with the genome data you're using for the GWAS, and do a calculation of heritability based on all genetic variation within your data. That would be an attempt to measure "heritability", but it's not compatible with the description Scott provides, that "the regions detected by the GWAS don't account for enough variance".
I was going to leave a comment observing that the empirical heritability from twin studies and the problem of missing heritability from GWAS studies are conceptually unlike in the same way that the concept of a "gene" as an abstract unit of inheritance and the concept of a "gene" as a region of DNA that codes for a protein are conceptually unlike, but then I clicked through to the East Hunter article and discovered that that's what it's about.
When you say "height is mainly genetic" you're making a lot of environmental assumptions. The average height of the Japanese increased dramatically after WWII.
It's well established that height is mainly genetic. Think of it as a ceiling for possible height, even if environmental effects can lead you to not reach that genetic ceiling. See also the difference in average height between some ethnic groups that both have adequate nutrition.
All %heritability statements encode a lot of assumptions about both environment and population. Consider for example in a study of pug dogs, what percentage of height variation is hereditary versus environmental. (Pugs are all genetically nearly-identical so the environmental percentage is near 100%.) Now do the same study but also include some Great Danes, suddenly you've got way more heritability. Dynomight wrote a good article on this stuff in general: https://dynomight.net/heritability/
For a similar environment, height is mainly genetic. The heritability of height is about 90%.
Heritability estimates attempt to break out the variance from genes, the shared environment and the unique environment.
Back around a decade ago, the "heritability gap" of height (difference between the estimate from twin studies, adoption studies, etc and the genomic estimate) was quite large. But it was smaller than that for intelligence. Better approaches, with much more data - see the Stephen Hsu papers - have reduced this gap in the last ten years.
Back then, when critics of behavioural genetics were writing about the "heritability gap" of intelligence, I didn’t find one that mentioned the same gap for height.
Because genetics isn’t popular, there are lots of critics. I did my best to find the best critics with the strongest arguments.
Here they are:
1. The heritability gap (no mention of height of course, because that would weaken their argument).
2. The "false assumption" of the "equal shared environment" in the ACE calculation (calculating the variance from genetics, shared environment, unique environment). No mention of 60+ studies on this because that would weaken their argument. And see "The Genome Factor" by Dalton Conley and Jason Fletchers. They also thought this was a naive assumption until they tested it in a unique way.
I didn't see any mention of the gut microbiome! It's not genetic, but it is heritable, especially from the mother (there are special mechanisms for carrying gut microbiota up into the breastmilk). Seems like a good place to look.
I think this wouldn't explain "missing heritability" per se, because it means "twin study heritability that is missing in molecular methods". But since gut microbiome is inherited equally between MZ and DZ twins, it wouldn't show up as heritable in twin studies. So it can't explain why twin studies find more heritability than molecular methods.
Gut microbiome is genetically influenced, as I noted in another post in this thread. MZ twins have more similar gut microbiomes than DZ twins, and increasing age increases divergence.
A difference in genetically mediated capacity to maintain an inherited gut biome could potentially explain observed outcomes very well.
> But since gut microbiome is inherited equally between MZ and DZ twins, it wouldn't show up as heritable in twin studies.
Well, there are different things that might theoretically happen.
If any two infants nursing from the same mother get the same gut flora, then there is no within-family variation and the concept of heritability isn't even defined.
On the other hand, if they're donated the same microbes, but those microbes flourish differently within different siblings for genetic reasons, then gut flora might show up as heritable.
Beat me to it! Only yesterday he posted an interesting article on how certain species of gut bacteria could release compounds which bring on or worsen symptoms of schizophrenia:
You are using "heritable" in a sense that is different from what it means in Scott's essay, and in genetics in general. Heritability is a measurement of how much variation in phenotype can be attributed to variation in genotype. If something is not attributed to variation in genotype, it is not heritable.
(In a technical sense, it most likely still is; a heritability calculation on zip code will show that heritability is high. You have to seek out a population to demonstrate that this is a spurious finding, which is easy to do since we know how zip code is really determined. But I don't think that's the kind of thing you meant?)
EEA criticisms are pretty weak, IMO. For one MZ and DZ twins really do have equally similar environments when it comes to all the things that environmentalists have been saying are the real cause of variation in cognitive traits: Parental income, wealth, and education, books in the home, neighborhood, schools, etc.
But also, if subtle excess variation in DZ twins' environments relative to MZ twins' really has such large effects, we'd expect to see sizable correlations among adopted siblings. It's hard to believe both that adoption studies don't work because range restriction makes adoptive households too similar to one another, and also that twin studies don't work between DZ twins' environments are too dissimilar.
This reminds me that I don't have a good sense of where Gusev and Turkheimer would say the missing variance is.
I think the strongest argument would be that it's non-shared environment - ie basically random, probably having to do with weird quirks of embryology - but I don't know if that's actually what they think. I agree that books and education are less plausible.
Yeah, Aporia did a pretty good takedown on the subject. The missing heritability in GWAS is a bit of a puzzle, alright, but the problems of missing environmentality are worse.
AFAIK Turkheimer also has stated repeatedly that high heritability would be disastrous for society, as it would vindicate the bad, old, obsolete theories of "bad apples", scientific racism, class-based societies etc.
This is something I've noticed repeatedly now. I'm working in genetics myself and have some contact with social scientists both professionally and privately (my wife, for one, studied psychology - thankfully she broadly is on my side), and not only is the definition of hereditarians often hilariously lopsided (especially for IQ, it's common to be called hereditarian as long as you consider genetics non-negligible), the anti-hereditarians almost always sooner or later start talking about how terrible it would be if IQ was genetic, how it's clearly part of a push to undermine education (by whom, for what purpose?), or how it's just a psychological defense mechanism by elitist, .... On the other hand, the hereditarians at most mention the possible positive applications such as improving IVF for couples struggling to conceive, but otherwise stick to scientific arguments.
Unlike most people who think about heredity, race, and IQ, I'm a sports fan. So I have a hard time worrying too much about Turkheimer's concern that if it turned out to be true that IQ tends to vary genetically by race, that would be the worst thing imaginable. Instead, it's pretty obvious from my 60 years of watching sports on TV that athletic abilities tend to vary by race, and yet ... we seem to be dealing with this revelation pretty well.
For example, in the NFL, only whites of the fleetest lineages seem to succeed at running back, such as Christian McCaffrey, whose father Ed was an All-Pro wide receiver and whose maternal grandfather, Dr. Dave Sime, won the silver medal in the 1960 Olympic 100 meter dash.
And yet, football fans (perhaps the broadest demographic in modern America) seem pretty cool with that.
Similarly, the NBA draft yesterday picked the first white American #1 draft pick, Cooper Flagg, since Kurt Benson in 1977. I wouldn't be surprised if the NBA would be 10% more popular if it weren't so racially disparate. But, still, it's awfully popular despite it's enormous racial tilt.
One of the striking things about the NBA is the degree to which athletes with different body types can be successful. I suspect that Victor Wembayama, Nicola Jokic, Steph Curry, and TJ McConnell all owe their basketball success mostly to their genes, just to different genes and groups of genes. Could EA be similar? Have any GWAS studies looked at specific populations as defined by ancestry?
Soccer really stands out for different body types. The 6'2" Ronaldo looks like a guy who could have been a quarterback, a tennis player, a pitcher, a golfer, a miler, anything, while short-legged Messi and Maradona don't look like conventional athletes.
>... and yet ... we seem to be dealing with this revelation pretty well.
If socioeconomic status depended on athletic ability as strongly as it does on IQ, we would find those particular racial disparities much harder to deal with.
Imagine high schoolers doing a 100 meter dash but with the stakes of the SAT. There would absolutely be resentment and wishful thinking.
I know the portion of variance explained by education is tiny, but how tiny? What are the percentages explainable by education, income, parental education, etc.? THere must be some best estimate for each, even if it's statistically insignificant.
Last time I read Gusev, I came away with the impression that it should be interaction effects, as he argues that molecular genetic methods control for those. But I'm no expert on the topic, and maybe I'm misremembering.
Extract: What could explain the gap? All of the twin study biases I described above are likely at play to some extent, but my personal view is that interactions between genes and the shared/familial environment play a major role. I alluded earlier to a study of gene-environment interactions for BMI; there is also striking evidence for an interaction between educational attainment and socioeconomic status (SES), where the heritability drops substantially for individuals in high SES versus low SES environments (curiously, the opposite has been observed in twin studies). My guess is that hundreds or even thousands of interacting environments accumulate over a person’s lifetime, many of which are not even measurable. When we then study unrelated individuals in GWAS, the participants are experiencing all of these diverse environments, and so genetic variation has a weak influence on their outcomes. But when we look at twins, who share their rearing environment extremely closely (are literally born at the same time), all of these interactions get assigned to and inflate the genetics/heritability bucket.
Edit 3: After having read into his stuff a bit again, my main question is if it actually possible to seperate direct effects and interaction effects like that. It seem unintuitive to me that the two are independent of each other, but I lack the statistical chops to have anything beyond an intuition on that.
Regarding your third edit, I think that's the main problem. In twin studies and (probably) in sibling regression we are getting the sum of interaction and direct linear effects. In GWAS and GREML and RDR we are getting the direct linear effect of some subset of the genome. The gap between RDR and GWAS can be explained by accounting for more genes, but the gap from there to sibling regression has to be some combination of environment, gene-environment interaction or gene-gene interaction. We can elimination direct linear effects of environment to some extent, but how do we separate gene-environment from gene-gene interaction? That's what everyone is struggling with.
I don't know anythig about RDR and sibling regression beyond what Scott and Gusev wrote, but as I understand it, both are methods for seperating direct from indirect (Scott says nondirect) effects.
Interaction effects are, in my understanding, seperable from that. At least Gusev seems to claim that direct effects can still be confounded by environment. But also, Gusev does seem to argue that there are ways of estimating GxE effects.
That is an important claim, but what I was getting at more is that seperating direct and nondirect effects should only work if they are independent. As I understand it, it's to control for the effects of parental genes creating environments with effects. But what if the same gene has a direct effect in the child and a nondirect effect in the parent? Seems to me like there is a potential danger of controlling away part of the effect.
Yes, I struggled with this when reading "Intelligence is not like height". I think Gusev really wants the missing variation to be in gene-environment interactions, but there doesn't seem to be a lot of evidence for this. The best conclusion I could get to from reading his essay is that its probably in gene-gene interactions, that mean identical twins maximize exactly the same set of gene-gene interactions so we see a non-linear drop-off in the impact of exactly the same genes in people who don't have identical genomes. Its nice to see this is one of the possibilities consistent with all the data even when viewed from "the other side".
Yeah this is why, although I don’t pretend to understand all the arguments inside and out, big picture… The weight of evidence seems to be solidly with hereditarianism. Sure, some of these recent findings are puzzling… But to actually shake my confidence you’d have to be able to tell me what these environmental factors are and provide evidence for them. And also explain how by staggering coincidence the results end up being what you would expect if hereditarianism were true. Without evoking all sorts of convoluted post hoc rationalizations about why the same cause can have opposite effects when it’s convenient for your argument etc.
I mean handedness is like this no? Most people think handedness is highly heritable (and it runs in families), but twin studies show very low heritability. Sometimes it just comes down to noisy variables that are hard to measure
Note that, although that the story you linked doesn't mention it, the study that found people with Norman names overrepresented at Oxford was by Greg Clark, who argues that very long-run persistence of social status is driven by genetics and highly assortative mating (0.8 on a latent measure of socioeconomic competence). So in this case it wouldn't be necessarily be confounding, because persistent differences between people with Norman and Saxon names might well be genetic.
The idea that truly exogenous economic endowments can persist over several generations has pretty weak empirical support, as I believe you mentioned in a post several years ago. Studies finding otherwise are likely confounded by genetics.
But there was also a St. Louis Fed study finding that the intergenerational correlation for residual wealth (i.e. wealth greater than or less than that predicted by earnings) is only about 0.2. Except perhaps in the most extreme cases, wealth has to be actively maintained and fortified by each successive generation, or even very wealthy families fall off the map within a few generations. Consider that less than a century after his death, there are no longer any Rockefellers on the Forbes 400 list.
Wouldn't your theory require that Normans started out smarter than Saxons? I'm not claiming they can't possibly be, just that I can't really figure out a reason to think this, and it would make more sense that the persistent Norman-Saxon difference is because Normans have always been the nobility (and therefore had more money) since the Norman Conquest.
What if social status gives you access to higher-quality mates? You're rich, so you marry other rich people, some of whom got rich through extraordinary ability. Through this mechanism, it seems like it should be possible to consolidate temporary social status into a permanent genetic advantage.
Note that this state of affairs is particularly favorable to a distinction between French-surnamed and English-surnamed people, because women don't pass on their surnames.
Oh, you're right. I don't remember the show very well, as I was pretty young when it was on the air. I just assumed that the surname that rhymed with her given name would have been her maiden name.
I always assumed this was part of the logic of noble birth. Old bloodlines with an accomplished past bring in new blood with an accomplished present, incentivied by the advantages of noble family, hopefully leading to stable excellence that adapts over time at a moderate rate.
Primary sources rarely talk about these things outright, but religion be damned, these people bred every animal they could for trait selection, and they knew what they were.
The Churchill lineage produced the great Duke of Marlborough, victor over Louis XIV's army at Blenheim, and his famous Duchess. Then they didn't do much for a couple of centuries, then many generations later they produced the brilliant but unsuccessful Randolph Churchill and the highly successful Winston Churchill.
Yes- this, exactly. The Normans who moved to England as a transplanted aristocracy were not the median Norman, and any genetic distinction could have been fortified in subsequent generations by strategic marriages and selection effects.
My name Sailer is usually spelled Seiler (ropemaker), but an ancestor became mayor of a small town in Switzerland and decided his lineage deserved a classier surname (rather like more ambitious Smiths in Britain tended to change their names to Smythe or even Psmith).
Gregory Clark found that Smythes are more likely to graduate from Oxford or Cambridge than Smiths. Presumably they were all Smiths at one point, but the people who changed their names to Smythe were obvious social climbers.
Nobles often inherited multiple names and titles so they had the option to choose which names and titles they used more prominently. Prestigious names and titles were kept in use even when the direct heirs went extinct while common ones were not.
Normans being by far the most effective in Europe and the Mediterranean in the medieval period seems some evidence there was something special about them. The Normans managed to kind of independently become the military elite in England, Ireland, Southern Italy and other places. Military skill is quite g loaded at least according to modern studies.
The problem with proposing that artificial social status persists after so many centuries, is that study after finds that extreme exogenous shocks to social hierarchies, including those specifically intended to disrupt them, have relatively low long-term effects on population hierarchies.
That strongly suggests that status persistence is due to the underlying genes that caused the status in the first place, not due to an inertial effect of status itself.
Similarly, in the USA, slavery represented a massive exogenous pressure on social status, but looking within counties at gaps in income and other status indicators between descendants of free vs. enslaved Blacks, shows that by 1940, the impacts of slavery had basically evaporated (https://archive.ph/PGwtu, https://www.cremieux.xyz/p/black-economic-progress-after-slavery).
Similarly (referencing Clark), in the Indian subcontinent, while caste hierarchies generally persist, artificial castes imposed by the British, didn't; which is what you'd expect, since they were arbitrary, representing exogenous shock.
Similarly, the Samurai had their status advantage actively eliminated through war, but their descendants ended up reverting to once again end up on top.
The same was the case for Jews. Before the Holocaust, they tended to do very well economically relative to local non-Jews. Then they faced widespread genocidal discrimination, but within just a couple of generations, their descendants ended up once again, far outpacing their non-Jewish neighbors.
I mostly agree with you here, but it seems like the Normans might be a special case, since they were literal titled nobility and in most cases you can just check and see that yes, the noble title remained within the family line from 1066 to the present.
I don't really understand the statistical analysis, but this paper (https://pubs.aeaweb.org/doi/pdfplus/10.1257/aer.20191422) examines the rebound of southern slaveholding families post US Civil War, and claims to have explicitly tested and rejected the hypothesis that status persistence was due to inherited ability, and that patronage networks are a better model.
I believe there was also a paper on the persistence of Chinese elites post-Cultural Revolution that made similar claims but I can't find it at the moment.
Yes this was the paper. IIRC the authors claimed the success of elite-children post-revolution was mediated by work ethic, and that the elite advantage in work ethic over non-elite children disappeared among elite children whose parents had died when they were young, suggesting that it was largely cultural transmission rather than genes. I could be misremembering though.
Norman aristocracy came over in boats and killed the upper crust of Anglo-Saxon society, so probably quite a big difference. I'm not really qualified to say, but I'm doubtful that much of that difference is preserved till today (and captured in surnames). Assortative mating might do it, but assortative mating might also have created these same class differences even if there hadn't been an invasion.
The Norman conquerors were a highly-selected hierarchical aristocracy. The invading army at Hastings was not composed of random farmers. Most were second and third sons from noble families (due to primogeniture), i.e. highly ambitious men with elite backgrounds but no inheritance. So you had a population bottleneck of highly-selected elites that became the top 1% of the English population and then maintained fairly consistent ingroup marriage patterns for centuries. I think there's every reason to believe that British Normans have a better genetic heritage.
The Normans were also more Germanic than the general British population, and across Europe there's a strong correlation between Germanic descent and economic development.
normans started out smarter cuz it wasnt the norman peasantry that invaded Britain but strong healthy men from the warrior class. When the founding population is above average the resulting people are (taiwan ,japan, england, are examples of this, you may even consider the founding populations of Eurasia that left Africa). Just like when Spain invaded South America the conquistadors were all of roughly the mercantile class well above the catholic peasantry. All the Spanish ancestry in Latin America, and all of the Norman ancestry in Britain is from above-average groups of those respective populations. And when they took Native and Saxon wives they took them from the best families ,so their brilliance was never diluted.
the normans that invaded were an above-average subgroup of normans. it wasnt a respresentative sample of normans, which may indeed not be smarter than the saxons. when the founding population is above average, the resulting people are (see taiwan and japan and some other island countries, or you can even consider the founding populations of eurasia that left africa). this exact thing happened in the new world, where above-average spanish (conquistadors who were from the merchantile class) conquered latin america. when they took native and saxon wives they took them from the best famlies, so their above-averageness never diminished over generations.
p.s. i couldn't find your reply to me about twin studies being needed to disentangle nature vs nurture ( i only see it in my email, im not sure why). the only thing i would add is i dont think the villagers would put much stock in things like parenting. they would see parenting as folllowing from the mediation between the innate qualities of the parent and the child who are closely linked. not separate from the hereditary principle, but something that follows from it. i suppose it would be possible a high quality person to be a poor quality parent (if they became alcoholic ?) , or that a good-natured child could prove rebellious but these would be notable exceptions.
I don't know about lately, but when I last checked about 10 or 15 years ago, there were still four Hearsts on the Forbes 400. They trace their fortune to the Nevada silver rush of 1859, although they hit it rich again in mining strikes in the late 19th Century as well.
But ... most really rich guys on the Forbes 400 were semi-self made like Bill Gates, Jeff Bezos, or Elon Musk. They enjoyed well above average upbringings but then took huge advantage relative to other upper middle class individuals of their generation.
The US might also just be unusual in this regard. I once (peripherally) knew some members of the Fugger family. They are not crazy-rich but still wealthy enough that most of their energy seems to be dedicated to resolving within-family inheritance squabbles.
I wonder if there is a within-family version of the resource curse going on there. If you are extremely able and you come from a family with a big fortune, maybe there's little you can do that is as likely to pay off as well as to make sure to inherit the big fortune and manage it and the family's reputation well. Whereas if you don't come from vast wealth, maybe striking out on your own and starting a business is a better bet.
Thomas Piketty is always complaining about all the Hidden Old Wealth. He must be referring to Europe, because America doesn't seem to have many, say, personal golf courses of obscure ownership. I can name the owners of most of the personal golf courses in Southern California, such as Larry Ellison.
I think it's a mistake to treat possession of surname as meaning genetic similarity, because only legtiimate male children would pass on a surname. And the inheritance system of the time favored them significantly.
Medieval nobles frequently had a lot of illegitimate children, and half their children would have been girls who lose the surname, and so there are likely lots of people in the population with Saxon or other surnames who have more of a genetic link to the Normans than the people with the surname.
But being a legitimate male child of an aristocrat would give significant material advantages
Thanks for such a wonderful write-up. One thing I’d love to see is more of a deep dive into the stats behind this. This seems like a very high dimensionality problem even without interactions, and I don’t see why interactions wouldn’t be as important as well. I could easily see k > n even with an n of three million, and it isn’t like k=n is a good situation. How the hell are they estimating this? LASSO? Something non-parametric? As an outsider, given the difficulty of the problem they’re facing, it seems like a no-brainier that regressions they’re running don’t add up.
You have to tune it correctly, but, no, what I expect is that you keep the variants whose effects you really know and you filter out the variants you can't distinguish from noise. I guess you could try ridge if you want to estimate heritability and don't care about having faith in the list of variants, but I'm sure people have tried it and I imagine I haven't heard of it because it doesn't give different results.
I would still suspect that there would be so many very small contributors that almost all of them could easily be mistaken for noise. Genomics just doesn't seem hospitable for methods used in less-complex domains.
ETA: in most regression studies, a reasonable prior is that most things *don't* have a meaningful effect on the outcome of interest, so it makes sense to try to pick the needles out of the haystack. In genomics, OTOH, it now seems a much more reasonable prior to expect that almost everything has a similarly-small-but-real effect on any given trait, in which case needle-picking throws the baby out with the bathwater (excuse the mixed metaphors).
Maybe you should think of GCTA, which finds most of the missing heritability, as ridge regression, but I think it's just that they focus on a small population so that each rare variant gets observed more.
For this "Alex Young thinks that once we get enough whole genomes sequenced (probably soon!) we might be able to use a technique called GREML-WGS to get more definitive answers about rare variants" is there any reason to think that even a sample size of 8 billion would be big enough to detect rare variant effects?
My guess is that yes, it requires a dramatically larger sample size to estimate interaction effects (at least, assuming that they are individually small, like direct individual SNP effects appear to be).
The reason for this is that in a regression of something like
y = beta_0 + beta_1 x_1 + beta_2 x_2 + beta_{12} x_1 x_2 + e
you often/typically end up with a lot less identifying variation in x_1 * x_2 (that is, variation in x_1 * x_2 net of a linear function of x_1 and x_2) than the amount of identifying variation in x1 in the original linear regression of
y = beta+0 + beta_1 x1 + beta_2 x2 + e
Now, add in the fact that you are testing some insanely large number of SNPs (hundreds of thousands? Millions? Icr).
The number of potential (simple) interaction terms scales quadratically, and because of this, to avoid false positives dominating true effects you need an exceptionally low type-1 error rate.
The combination of these two things requires extremely large samples.
I would note though...
Rare genes/mutations that might not be tested in standard DNA tests / GWAS (if I'm understanding this correctly) might theoretically explain a bunch of the missing variance, idk.
But it seems ~ impossible for interactions of fairly uncommon mutations to explain much of anything. Why not? Well, suppose they did. Suppose you need the rare mutation on gene 234 and 1267 to get a bump. Suppose you have this, and thus the bump (and most of your "true" polygenic score is similar interaction effects).
Will your kids have most of these interactions? No, they shouldn't. Isn't it the case that they have a 50% chance of get your rare mutation on gene 234 and 1267, and thus only a 25% chance of getting both? And it's rare, remember, they're extremely unlikely to pick up "the other bit" from your spouse.
So then you should have drastically lower heritability.
So if we already "know" that heritability is high, and believe it is mostly genetic, "rare interactions" won't help us. Rare single genes might. Interactions of common things might. But not interaction of rare things.
From an apriori point of view we should expect the relationship between genes and observed features to be incredibly complicated -- basically as complicated as the relationship between computer code and program operation -- and I'd argue the evidence we see is exactly what that model of highly complicated interactions predicts. Namely GWAS misses a bunch of twin study variation and so do simple non-linear models. Or to put the point differently the answer is the narrow/broad gap but broad hereditary is just really damn complicated.
Yes, people cite papers for the claim that non-linear effects don't seem to make the difference. But if you dig in the supposed evidence against non-linear effects (like you linked) are really only evidence against other very simple elaborations of the linear model. Showing that you don't predict much more of the variance by adding some simple non-linearity (eg dominance effects at a loci or quadratic terms etc) is exactly what you would expect if most effects are extremely complicated and neither model is even close to the complete causal story.
I mean, imagine that some OSS project like the Linux kernel gets bifrucated into a bunch of national variants which are developed independently with occasional merges between individual nation's versions (basically it acts like genes under recombination). Or better yet a Turing complete evolutionary programming experiment with complex behavior. Indeed this later one can be literally tested if people want.
We know in this case that code explains 100% of program variation and if you did the equivalent of a GWAS against it you would probably find some amount of correlations. I mean some perf improvements/behavior will be rarely discovered so they will be highly predicted by whether some specific code strings show up (they are all downstream of the initial mutation event) and other things like using certain approaches will have non-trivial correlation with certain keywords in the code.
But no linear model would actually even get close to capturing the true (indeed 100%) impact of code on those performance measurements. And it would look like there were no non-linear effects (in the sense of the papers claiming this in genetics) because adding some really simple addition to the linear model like "dominance loci" or whatever isn't going to do much better because you aren't anywhere close to the true causal model.
So the evidence we have is exactly what we should expect a priori -- genes have some really complicated (essentially Turing complete) relationship with observed behavior and simple models are going to guess at some of that but linear models will do about as well as simple non-linear ones until you break through some barrier.
I think certain twin/sibling studies have themselves been used to try and estimate the relative size of additive vs. non-additive genetic effects, and additive effects were maybe twice as large, though. Non-additive effects are also harder for natural selection to act on, which makes rapid evolution in a trait less likely.
Yes, but the problem is there are 2 types of studies that estimate non-additive effects.
1) Twin study type studies. Well that's exactly the high result which causes the gap.
2) Papers that test for non-additivity by elaborating the linear model in some way to allow for some simple non-linear effect. For instance, capturing the idea that maybe a gene can block the effect of other genes at the same loci.
But 2 is only evidence against the kind of non-linearity presumed in that model (more generally simple non-linearity). My point is that if the true relationship isn't something like mostly additive but maybe with some quadratic terms or multiplicative one but is actually more like a Turing complete interaction then you should expect that adding a bit of non-linearity doesn't improve the prediction much.
So as I said, we see exactly what we should expect consistent with the hypothesis of very complex genetic interaction s.
I'm probably not as immersed in the literature as you are, though I do work in the field of software so... I think I can partially intuit what you mean by non-linear program flow, although I think biological systems are both messier and generally include more built-in redundancy. However, I can certainly see the argument that complex interacting biological systems (such as brains, or even cellular metabolism) must have complex genetic correlates.
However, it's also the case that tiny regions of the genome can have counter-intuitively huge impacts on phenotype (e.g, humans vs. chimpanzees), so... I wouldn't assume in advance what percentage of gene-variants accounted for non-linear gene interactions, or what percentage of phenotypic variance those accounted for.
the hereditarians (the fact they call themselves this is funny, all geneticists are hereditarian) seem to have a weird idea that there's like a full genetic shuffle going on and evolution likes low-variance smooth fitness landscape. both are untrue afaict.
not a lot of shuffling is happening (1-3 crossovers per chromosome pair during human meiosis from memory) and with recombination hotspots the shuffle is somewhat predictable. so a lot of deep structure can be preserved and it's probably beneficial to have deep structure (you don't want an organism which rapidly loses genetic diversity to hit a local maximum and then all dies when the environment changes, similarly you don't want a function that's too full of 'holes' which the organism simply gets stuck in while maintaining diversity - you want an explorable and non-smooth fitness landscape).
note of course all these 'meta-parameters' like where the hotspots are, are themselves subject to selection which makes it harder to reason about.
i tend to agree with OP we should expect really complex interactions. the metaphor I'd use is that it's like we're analysing books by only doing word-counts. except in super high dimensional spaces it's often hard to intuit just how effective this is while still being completely wrong: for instance we can predict a lot from non-functional non-coding DNA simply because statistically it'll be colocated with genes that do matter. if you can make good predictions from junk DNA it's hard surprising you can do really well while wrongly assuming additivity.
i don't know much about the field but the impression i get is that the molecular geneticists largely take it for granted there's complex interactions because they've found really cool stuff in simple organisms. meanwhile you have more classical quantitative geneticists who have barely progressed since Fisher who are so far removed from the cutting edge that they barely interact with their much smarter brethren.
>the fact they call themselves this is funny, all geneticists are hereditarian
So one would think; and yet....
>hereditarians [...] seem to have a weird idea that there's like a full genetic shuffle going on and evolution likes low-variance smooth fitness landscape. both are untrue afaict [...] i tend to agree with OP we should expect really complex interactions
This is /better/ for the hereditarian position, not worse—insofar as complex, hard-to-pin-down interactions suggest that the "missing heritability" issue will be resolved with revisions upward rather than downward; I mean, not that you explicitly contradicted this, but the sense of the initial paragraph there seemed t'me to imply you felt this was a /problem/ for a "hereditarian" approach. (Or, possibly, I am confused and actually hereditarians DO hate this One Weird Genomic Trick?–)
It will be resolved in between the twin studies (which inflate heritability due to ACE assumption) and current GWAS etc.
Obviously genes will still be incredibly important but we might move away from 'parenting doesn't matter at all' which is close to where you get with twin studies.
they might argue parental investment mattered more when we faced chronic food scarcity and were being predated upon by sabre-toothed cats, but we have no twin studies from back then. heritability estimates are always for a given environment (ironically leftist attempts to level childhood environments, by e.g. ensuring no-one receives a good education, have increased the percentage of variance explained by genetics.)
This is true AT THE MARGIN for first-world middle-class parenting, but I'm pretty sure it's not true when you go far outside that range. Raise your kid in a cave and never show him a book, beat him with a stick regularly, etc., and you're well outside the data provided by adoption studies or twin studies.
> "the hereditarians (the fact they call themselves this is funny, all geneticists are hereditarian)"
Well... yes, but that only makes the environmentalist position more indefensible- it would literally be impossible for intelligence (or any other biologically-dependant human trait) to have evolved in the first place if it wasn't genetically influenced. So are "their much smarter brethren" deluded, or lying?
The question is about variability among humans. It's quite possible to have genetically determined traits within the species or some population where very little of the variability we observe is due to genes. The easy example is language--humans clearly evolved a bunch of mental machinery for handling language, that's got to be genetic, and yet nearly all the variability we see in what language people use is environmental, not genetic, and people seem quite capable of learning languages far from their ancestral lands--adopt a Chinese baby and raise him in Texas and he'll speak American English with a Texas accent.
Yes, but paradoxically, verbal IQ and vocabulary size are one of the most strongly genetically-influenced aspects of intelligence. The content of language is enormously culturally influenced, our aptitude for it is not.
Raw scores on IQ tests went up on average during the 20th Century: the Flynn Effect.
My guess is that one reason is because cultures evolved in the directions anticipated by early IQ researchers: e.g., Stanford professor Lewis Terman, who invented the first American IQ test in 1916, the Stanford-Binet, tended to make up questions that reflected the views of academics and professionals in Palo Alto about what constituted intelligence.
And 109 years later, those Palo Alto biases about intelligence appear to be right.
Palo Alto more or less conquered the world, with Lewis's son Fred Terman playing perhaps the central role in the creation of Silicon Valley as we know it.
TBF people can only do what is possible with the available tools and data. Fitting a linear model to genes is computationally possible. Even assuming there might be pairwise interactions that aren't limited to very nearby genes becomes very difficult to do from a computational and data gathering POV.
The problem isn't the work being done by scientists in this area but the people who want to argue that there can't be strong genetic effects that account for huge amounts of variance that are so far just beyond us.
Unfortunately, I fear that what you really need to make progress here is absolutely massive data sets that include substantial numbers of related individuals plus data about IQ educational attainment etc. The kind of massive everyone's genetic info database that makes everyone really nervous. That way you can start looking for the smallest changes that make big differences.
> But from an apriori point of view we should expect the relationship between genes and observed features to be incredibly complicated -- basically as complicated as the relationship between computer code and program operation -- so it's weird to think linear GWAS models should capture most of the variance.
If you only look at how genes get transcribed into proteins and what proteins actually do, and how long and convoluted the chain of cause and effect is to get anything we actually observe on a macro level, then your argument makes a lot of sense. It's ever crazier and weirder than human-readable computer programs.
But: remember that genes and chromosomes get reshuffled and stitched together and slightly mutated with every generation. That means for them to work, they need to be fairly robust.
If you want to think about a computational analogy, I wouldn't pick computer code ~ program execution, but perhaps like the weights in a neural network. Especially since methods 'dropout' (aka 'dilution') that randomly disable artificial neurons seem to work quite well, and not cripple them. See https://en.wikipedia.org/wiki/Dilution_(neural_networks)
(I guess it helps that 'dropout' in neural networks was inspired by an analogy with genes.)
Perhaps intelligence is the single thing that most requires intelligence to figure out? Hence, I'd hardly be surprised that we haven't figured out IQ completely yet.
Yes, paradoxically, nailing down the exact causal genetic pathways behind human intelligence will probably require a superhuman machine intelligence. Partly because I suspect the genetic code is not especially well-factored, and it certainly needs a lot more comment blocks.
I don't think the analogy with computer programs is on point.
The big difference is that the genetic code is recombinable. You can take two genomes, mash them together, add a few mutations, and it's still a working genome! You can't do that with computer code.(*) This is a pretty tough restriction, and it puts some severe restrictions on the complexity of interaction. Not that we would understand exactly how those restrictions look like. It doesn't just rule out "it's complicated" either, but it's enough that I find the comparison with computer code wrong.
(*)Actually you can, it's called Genetic Programming. The idea is that you evolve code via recombination and selective pressure by a fitness function, for example, to pass as many test cases s possible. It just doesn't work.(**) Since I work in a related field, I know a couple of people who work on Genetic Programming. The issue is precisely that it's so hard to recombine programs into another working program.
(**) My colleagues may beg to differ. But I would claim that it doesn't work unless you either work with absolutely tiny programs, or move so far away from the basic idea that you are no longer evolving code, but live in a totally different "genetic space".
Works better with neural networks and some other types of AI. I'd argue that various regularization techniques (dropout) etc. are fairly good analogues to what's happening biologically with recombination and the resultant networks can have quite complex interactions (indeed it's proven that they can, in theory, compute any function irrespective of how complex it is).
Well, sometimes it's still a working genome. Sometimes one little mutation breaks everything. And even for that , the two starting genomes need to be pretty similar. You can't mash one from a chimp and one from a lobster and get something that works. You can't even mash two from egg cells or sperm cells, they have to go through the correct processes on each side first.
But it does seem a bit less finangly than code execution, if only by having a (far) lower information density.
Back to the original question though, there's actually a lot of cases where interactions between genes are important. Like if a chemical is modified by a first protein and then a second protein and then a third protein, and you delete the first protein, it doesn't matter anymore what the genes are for the second protein. Or maybe you make the second twice as effective, but the third is really slow, so there's no difference without another gene upping the third protein as well.
That's a great observation. In this whole discussion, what's bugging me is that genes aren't studied in the way they actually work in order to produce phenotypes: as cascades of dynamical-system type interactions that end up with very strong attractors in aperiodic systems. If they were, then the other aspects of genes and the genetic code and of autocatalytic systems would also need to be considered: redundancy and canalizing functions for example would lead many variants to lead to the same outcome. Statistics is a blunt tool to study this kind of system. In twin studies however, you get the outcome of the entire system by default. So no wonder there is missing heritability in purely genetic association studies. It's like trying to predict the quality of the cake from the kind of eggs and flour used, when the sequence of mixing the ingredients (gene x gene interations) and the baking temperature (internal environment) mattered far more. Twin studies compare the cake, or pudding, directly, wherein lies the proof.
The point about computer programs is merely that the ways that genes can affect outcomes is essentially Turing complete -- basically as far from linear as you can get.
Yes, if you tried to randomly recombine C code it wouldn't work very well but you can maintain Turing completeness while allowing recombination to work much better as one tends to do when doing genetic programming. And yes it means it can be very hard for evolution to make big breakthroughs but that is why it can take millions of years.
And sure, most kinds of effects on intelligence aren't going to be as fiddly as say the fact you need to do transcription exactly right (which really is just a mini virtual machine in the code) but you don't need them to be that complicated not to get picked up by models that are just linear effects plus some minor tweak. The fact that we know genetic effects can be as complicated as possible is good reason to think that it's highly plausible that even in intelligence we are seeing at least effects from arbitrary pairs of genes (eg A and B together have an effect which is nothing like A + B...even when nowhere near each other on genome).
This is my intuition as well. And it's basically what Turkheimer thinks from what I can tell. Genetics interactions are extremely complicated and chaotic. You cannot really predict complex traits from genetics (unless you already have a clone or very close relative.) Just like you can't predict the output of a complex computer program without running it.
I'd hold out a little more hope for one day being able to make better predictions. But you'd need absolutely massive databases of sequences with tons of info about educational attainment and IQ testing that includes lots of related individuals so you can start by looking at the effects of more limited changes. And even then it's a massive project that could take generations if it is possible, even if people were cool with creating the needed database.
When I first heard about GWAS while taking a bio-informatics class in like 2011ish. The instructor, a theoretical computer scientist, just flat-out said, “the genotype to phenotype mapping is probably spectacularly non-linear and we should only expect meager and incomplete results from GWAS.” I’m impressed they’ve found as much of the heritability as they have. It’s crazy that people claim that because GWAS can’t find the genes, they must not exist. It reminds me of all of Stephen J. Gould’s sophistry in favor of ideologically preferred conclusions.
Educational attainment is a meaningless measurement, at least from the internet survey versions I've seen with categories like "some college" or "bachelor's degree". This misses a lot of education that I think also correlates with intelligence. A Journeyman electrician requires a five-year apprenticeship that includes both on the job and classroom education, where would a high school dropout who joined the IBEW and became a Journeyman electrician be classified?
Both of my parents have a bachelor's degree, I chose instead to enlist in the US Navys Nuclear Power program which at the time was compared to the most difficult programs at top universities. Am I a strike against the heritability of educational attainment since I earned a nuclear Naval Enlisted Classification instead of a degree?
Maybe current surveys could be better designed. But there's a difference between an imperfect proxy and a non-proxy when measuring a population.
Surveys of educational attainment don't classify *you* as an individual well or fairly. But they're still correlated with IQ, on the average, to the extent that general intelligence exists. Getting 60% of the answers right on a true false test is not a perfect score. But it's not a 50% either. Surveys using educational attainment can get the equivalent of a 60% correct score and still be potentially useful in telling us things about a population.
They might be confounded, I suppose, if certain forms of intelligence predicted a person pursuing trade schools rather than college education. Most surveys seem to rely on a notion of general intelligence. To the extent that college and 'advanced trades schools' represent distinct types of intelligence I could see the study being problematic. But if you're just an outlier then the format will just produce weaker results than it could.
I guess using educational attainment may be...what's the phrase, directionally correct? I don't think this tells you anything more than you could learn from dog breeds, yes, behaviors can be inherited. A retriever will retrieve from not long after birth, and a shepherd breed will herd anything that moves which is always fun when a family with one invites a bunch of their child's friends to a birthday party and the dog will do its best to keep them all together.
I would think these studies aim for more useful information to quantify the degree of hereditariness or something and EA isn't going to get there. A degree may have correlated with intelligence for a period of time since that was the smooth path to a decent paying career, but when the people spending billions of dollars on AI datacenters say the bottlenecks are electrical generation and electricians to run all the cables, then over the next ten years intelligence may be more closely corelated with people who join the IBEW after high school than with those who earned a degree.
Back in 1986 I put an ad in the Chicago Tribune looking to hire a personal computer technician. The HR person asked about academic requirements. I thought about it for 5 seconds and said, "Bachelor's degree required."
But then I got a long letter from an applicant saying, I don't have a college degree because I enlisted in the Navy's submarine nuclear reactor program out of high school, but here are ten problems you are probably dealing with and how I'd solve them for you.
I hired him, and by the afternoon of the first day, it was clear he was smarter than me. While he solved my immediate problems, he also caused me a lot of unexpected problems because because within a few months my boss's boss, who had an MIT advanced degree, was calling me up, irate, to find out why the computer repairman I'd hired was having lunch with the Chairman of the Board to discuss corporate technology strategy.
Great story! I'm just ordinary smart, but yeah, we had more than our share of ridiculously intelligent people through that program, and a few of them also had interesting problems. We had one when I was an instructor at the NY prototype who never missed a single exam question and knew all the answers but had zero life skills. He couldn't drive, bought a plane ticket to Albany airport then started walking down the highway towards Saratoga Springs with his seabag on his back. The police picked him up and drove him to the site. We had to assign him roommates who could drive him to work because he hadn't made any friends who would do it voluntarily. He slowly got a little better, but wow, I had a much easier time assisting the ones who struggled academically.
Thanks for the write-up. I suspect that seeing this kind of confusion and disagreement is a good indicator that we are about to witness some nice chunk of progress from clearing this up, soon.
Especially if (one of the) limiting factors seems to be genetic sequencing and computing power: two things that are still getting cheaper and better quickly.
I know of two secret results I'm not supposed to talk about, by people claiming they've found very large amounts of "missing heritability". Not yet peer-reviewed or confirmed by anything except rumor. I expect one to be out within six months, and the other maybe eventually.
It's a funny coincidence, I came across this other post from Stephen Skolnick just yesterday arguing that the gut microbiome strongly predicts schizophrenia risk.
I will admit that Skolnick's essay sets off my 'too good to be true' alarms, but it still seems an idea worth exploring and he rigorously argues for the mechanism.
To what extent does the gut microbiome account for the outcomes we're seeing here, also? Gut microbiome is inherited, to some extent. It is also genetically influenced, and monozygotic twins have more similar gut microbiomes than dizygotic twins, with the divergence apparently increasing as life goes on.
Goodrich et al., 2014 (Cell)
SA: "Finally, she realizes she’s been scooped: evolution has been working on the same project, and has a 100,000 year head start. In the context of intense, recent selection for intelligence, we should expect evolution to have already found (and eliminated) the most straightforward, easy-to-find genes for low intelligence."
Doesn't this assume that increasing intelligence always increases reproductive fitness? In modern industrial societies there's a negative correlation between high intelligence and reproductive fitness, especially in women. I recognize that the same trend might not apply, pre-birth control in more traditional societies. But it at least seems worthwhile remembering that various types of mental facility and reproductive success may not always be positively related in all environments. This suggests at least the possibility of low hanging fruit if those gains involve tradeoffs with reproductive fitness.
Humans evolved to be more intelligent than other species because it was an advantage to us (perhaps because of our large social groups). But brains are metabolically expensive, and there are limits to how big our heads can be while still fitting through a birth canal (hence humans being relatively helpless at birth). So it's advantageous to have higher IQs, but there are tradeoffs and other traits being selected.
If anyone really wants to listen to me drone on about an unpublished method I am unreasonably proud of, feel free to touch base. It’s been maybe six years since I’ve been thinking about this, but IIRC GREML ignores dominance and epistasis, and it’s not that hard to pick them back up. I’m not sure how to handle GxE, but probably using that sib method could negate the concern. And I don’t think ultra rare variants would be a problem, cause you don’t have to assign effects to any particular variant. And I think sibs should have a pretty good distinction of all relatedness modes, so it seems ideal.
Did you ask the polygenic embryo selection folks about collider bias?
In the section comparing Kemper’s sib-regression estimate (14%) and Young’s Icelandic estimate (~40%), you note that the UK Biobank sample may be skewed toward healthier, higher-SES volunteers (so-called healthy volunteer bias, which commonly creates selection effects in medical research). But the implications of such selection effects extend far beyond variability in heritability estimates.
This kind of bias can flip signs when people think they're being clever by adjusting for confounds (collider stratification bias). This is especially a relevant risk in highly selected samples like the UK Biobank, where the same factors that influence participation (e.g., health, SES, education) may also affect the outcomes of interest (mental and physical health). Conditioning on variables that causally contribute to both participation and outcome can introduce more bias than it corrects for.
I wrote a bit about this here: https://wildetruth.substack.com/p/designer-baby-maybe. Munafò et al.'s schizophrenia example illustrates the mechanism more clearly than I can easily argue it for IQ: people at higher genetic risk for psychosis may be less likely to volunteer for studies ("they're out to get me, so I'm not giving them my blood"). This warps the apparent genetic architecture in large datasets. Doing embryo selection against schizophrenia risk based on such a sample, from which high-risk individuals disproportionately self-selected out, could backfire. And parents who paid $50K for it might not know for 20 years (or ever).
Hopefully the real-life practical implications are trivial: if you pay $50K for a prospective 3-point IQ gain, and collider stratification bias turns it into a 3-point loss, no big deal? You probably won't notice over dinner, as long as your kid slept well last night.
But I remain curious how the corporations selling embryo selection address the possibility that they're accidentally giving parents the exact opposite of what they're paying for. My guess is: they just don't tell anybody, and there's something in the fine print saying they could be wrong and aren't liable. Causality probably can't be proven in any individual case, and anyway you're paying for a best effort process, not a result.
This is entirely dependent on whether the “warping” by not having X group in a sample actually warps the sample. Because there’s a high chance it doesn’t warp it at all and the sample is still valid, even without X group present.
Jury is still out on the powers of nature versus nurture, basically. It's slightly complex.
What struck me is the questionable logic of EA as a measure. Scott got into this slightly with grade inflation and economic prospects, but I'd go further. Many of the twin studies took place between 1970 and 2000, or at least the kids were in school during that period, some of them earlier. This corresponds with a high point in educational prestige, and an opening of college to more people on a merit basis, while college was still rigorous. Today, getting a college degree is obviously less meaningful than it used to be, so the value presumed of its presence is almost irrelevant, and I'm thinking this is generational, a reflection of boomer-GenX social philosophy. Expecting the same effects thirty years on is going to be very muddled by social change.
EA is really, really sloppy from a standpoint of correlation with IQ.
One way or another, I still see nature and nurture about the same way. A person's biological design is full of complexities and redundancies, while their experiences will prod adaptation in various directions within the limits of that design. I know people don't want to believe that adaptation can be bracketed like that, but the evidence is everywhere, we maladapt to all sorts of things in the modern world, some of us more than others. People can push their limits, but it requires strong incentives and going too far results in trauma. Or failure. None of this is particularly compatible with Western ideals regarding individual agency and equality, but that's why it's controversial, right?
My sympathies to adopting parents who thought they were getting a blank slate.
Saw recently, don't recall the source, that the average IQ of college students has gone from almost 120 in the 1960's to just above 100 today due to pushing college enrollment on broader populations.
It's not just about respecting values like individual agency. There's also a strong placebo effect associated with the growth mindset. In other words, an exaggerated belief that you can transform yourself into a better person makes you more capable of improving yourself.
Is this anything like the placebo effect of telling an ugly persona they’re beautiful? Because people can confidence game themselves into being a little more attractive if they get that kind of feedback, but ultimately, you are what you are. It seems like that would be made way worse when it comes to intellect. You can push your way up to a masters degree, but in the employment world, you will prove yourself incompetent at some point.
No, the logic behind these two phenomena are fundamentally different. I'm too tired to explain my thinking right now, though, sorry for that, because the logic is kinda intricate. The example you gave isn't even really the placebo effect, it's a completely different thing.
Well, now that you say that, I actually think the logic behind why placebos work (other than for pain relief placebos, which has a different logic) is something like that. So you're right that they sound similar. But TBH I haven't personally noticed "ugly people becoming slightly hotter when you call them beautiful," and that sounds a bit weird to me. But in the context of learning, yeah, I do think the logic is probably something like, "trying harder isn't hopeless, so it's worth increasing effort."
On second thought maybe the logic is closer to "people with a growth mindset think they can get more out of trying harder." Still similar logic but not the same.
Kind of surprised you don't see it in appearance. Someone tells you there's potential, so you lose weight, learn to dress well, get a haircut, and of course, walk and talk with confidence, you can maximize what you've got. You can go from a 4 to a 5, maybe even a 6. It's thinking you're a 10 that's going to end up being embarrassing.
With intellect, while I hate to bring this guy into it, Malcolm Gladwell said in Outliers that really, as long as you have an above average IQ, like 120, you can probably do about any job as long as you work at it, and a few extra points matter little. But trying to do organic chemistry with an average IQ is trading tuition money for frustration. If they let you graduate, it's worse.
To the point about EA being full of landmines, wouldn't one of those be the observation that so many modern jobs expect a Bachelor's degree? That seems like a shared environmental component that would push swathes of the population to hit certain thresholds they might not otherwise.
A small technical note that I think is crucial. In the first figure (Tan et al 2024)
1. The x-axis is "Proportion of variance in population effects *uncorrelated* with direct genetic effects".
2. Being on the left side means that the variance in population (given by GWAS) can be correlated with direct genetic effects.
3. So for example, properties that are highly due to direct genetic effects are on the left (height, myopia, eczema)
4. Educational attainment is at ~0.5, this means that the direct effect variance is 0.5 of the population effect. The population effect is ~40% and thus the direct effect is 20%, exactly the value that GWAS converges to.
5. We can also verify this by looking on Tan et al 2024. In another figure they show that EA's genome wide correlation between direct genetic effects and population effects is ~0.75, if you take the square, you get ~0.5
Yeah, when I read that his paragraph below, I didn't quite follow why he was squaring, and assumed the diagramme was perhaps mislabelled (ie it was intended to be the correlation, not proportion of variance).
Still though, you know why the graph has migraines as having a negative proportion of variance explained? I don't quite understand that part.
How much effort has been put into algorithms decoding complex geneXgene interactions? Also how much of these things could be down to epigenetic expression, and how could this hypothesis be tested?
I looked into this once; the answer is that although Australia was founded by (a small number of) prisoners, almost all modern Australians are descended from non-prisoner immigrants who came later.
Approx 20% of Australians are descendents of convicts, apparently. But there would be a lot of non-convict heritage mixed in as well. The gold rushes of the 1860s led to a huge population explosion, for example.
Also worth noting that the convicts in question might not have been as negatively selected as you're thinking. The classic claim is that most convicts were convicted of petty larceny, things like "steal a loaf of bread - 7 years".
I conceptualise all this as: we have a new method sibregression that does not give the same results as twin studies; this is strong evidence that broad != narrow heritability. Which is very interesting and worthy of a lot of discussion about what it means.
To me it suggests that a significant amount of heritability is chaotic: the genes matter, but small differences have big effects, being "similar" is less informative than we might have expected.
And as you say:
"If nonadditive interactions are so important, why have existing studies had such a hard time detecting them?"
Is it possible that the *fact of being a twin* makes you weird? Twins are more likely to be premature and to die in infancy. Maybe twinhood is "bad for you" (perhaps via having a worse uterine environment) such that twins are more likely to have all kinds of health & behavior problems than singletons? Perhaps having a "bad" gene is more likely to result in a "bad" trait if you're a twin, which will push up the heritability of lots of traits in twin studies specifically?
That wouldn't affect adoption & pedigree studies, though, so if those also show high heritabilities for lots of traits this hypothesis doesn't help.
Twins used to have lower IQ than singletons, but have now converged (probably with better prenatal care). I can't prove that they don't have some kind of unique disposition which results in equal IQ on average but makes it have a different genetic structure, but seems like that would be surprising.
Could you respond to Lyman Stone's "Why Twin Studies Are Garbage"?
I think that Stone makes two basic points:
1. Heritability is not worth measuring or thinking about, because it doesn't mean what most people think it means. A trait can have perfect 1.00 heritability and proven genetic basis, and yet still be 100% environmental in the wild, due to gene-environment interactions.
2. Gene-environment interactions can be both very strong and very difficult to detect. We didn't notice for decades that peanut allergies were >90% environmental in practice, despite having >50% heritability and known genes for susceptibility. Just a tiny difference in the timing of diet accounts for all of that. Because of this, it's impossible to say that fraternal and identical twins have similar environments, which violates the first assumption of twin studies (as you have numbered them above).
Something can be worth measuring and thinking about even if "most people" don't know what it actually means. Most people don't understand quantum physics.
Yes, gene-environment interactions exist and are a big deal. Basically the only conclusions that are consistent with all the research - assuming its all valid, which seems likely - are either gene-gene or gene-environment interaction or both are responsible for most of the missing heritability. Lyman Stone, like Sasha Gusev, wants most of the gap to be filled with gene-environment interaction.
That may very well be true for some traits, but it seems unlikely to be true for IQ or educational achievement. This now is just my personal take now from when I puzzled my why through Sasha's writings. Think of a simple nonlinear gene-environment interaction where a gene only confers an IQ gain in the presence of some external factor. Then we would expect to see an affect on IQ in siblings raised separately. But we don't generally see that - adoption studies, for example, generally measure broad heritability of adult IQ similar to twin studies. Now there might be gene-environment interactions in utero or even pre-fertilization, but just very broadly any postpartum gene-environment effect should show up in adoption studies and they mostly don't.
Which leaves gene-gene interaction. That is EA and IQ are highly heritably but mostly determined by non-linear gene-gene interactions.
§4.3.3 Possibility 3: Non-additive genetics (a.k.a. “a nonlinear map from genomes to outcomes”) (a.k.a. “epistasis”)
…Importantly, I think the nature of non-additive genetics is widely misunderstood. If you read the wikipedia article on epistasis, or Zuk et al. 2012, or any other discussion I’ve seen, you’ll get the idea that non-additive genetic effects happen for reasons that are very “organic”—things like genes for two different mutations of the same protein complex, or genes for two enzymes involved in the same metabolic pathway.
But here is a very different intuitive model, which I think is more important in practice for humans:
• Genome maps mostly-linearly to “traits” (strengths of different innate drives, synaptic learning rates, bone structure, etc.)
• “Traits” map nonlinearly to certain personality, behavior, and mental health “outcomes” (divorce, depression, etc.)
As some examples: …
• I think the antisocial personality disorder (ASPD) diagnosis gets applied in practice to two rather different clusters of people, one basically with an anger disorder, the other with low arousal. So the map from the space of “traits” to the outcome of “ASPD” is a very nonlinear function, with two separate “bumps”, so to speak. The same idea applies to any outcome that can result from two or more rather different (and disjoint) root causes, which I suspect is quite common across mental health, personality, and behavior. People can wind up divorced because they were sleeping around, and people can wind up divorced because their clinical depression was dragging down their spouse. People can seek out company because they want to be widely loved, and people can seek out company because they want to be widely feared. Etc.
• I dunno, maybe “thrill-seeking personality” and “weak bones” interact multiplicatively towards the outcome of “serious sports injuries”. If so, that would be another nonlinear map from “traits” to certain “outcomes”.
All of these and many more would mathematically manifest as “gene × gene interactions” or “gene × gene × gene interactions”, or other types of non-additive genetic effects. For example, in the latter case, the interactions would look like (some gene variant related to thrill-seeking) × (some gene variant related to bone strength).
But that’s a very different mental image from things like multiple genes affecting the same protein complex, or the Zuk et al. 2012 “limiting pathway model”. In particular, given a gene × gene interaction, you can’t, even in principle, peer into a cell with a microscope, and tell whether the two genes are “interacting” or not. In that last example above, the thrill-seeking-related genes really don’t “interact” with the bone-strength-related genes—at least, not in the normal, intuitive sense of the word “interact”. Indeed, those two genes might never be expressed at the same time in the same cell….
As far as I can tell, if you call this toy example “gene × gene interaction” or “epistasis”, then a typical genetics person will agree that that’s technically true, but they’ll only say that with hesitation, and while giving you funny looks. It’s just not the kind of thing that people normally have in mind when they talk about “epistasis”, or “non-additive genetic effects”, or “gene × gene interactions”, etc. And that’s my point: many people in the field have a tendency to think about those topics in an overly narrow way.
…
§4.4.3 My rebuttal to some papers arguing against non-additive genetics being a big factor in human outcomes:
The first thing to keep in mind is: for the kind of non-additive genetic effects I’m talking about (§4.3.3 above), there would be a massive number of “gene × gene interactions”, each with infinitesimal effects on the outcome.
If that’s not obvious, I’ll go through the toy example from above. Imagine a multiplicative interaction between thrill-seeking personality and fragile bone structure, which leads to the outcome of sports injuries. Let’s assume that there are 1000 gene variants, each with a tiny additive effect on thrill-seeking personality; and separately, let’s assume that there’s a different set of 1000 gene variants, each with a tiny additive effect on fragile bones. Then when you multiply everything together, you’d get 1000×1000=1,000,000 different gene × gene interactions involved in the “sports injury” outcome, each contributing a truly microscopic amount to the probability of injury.
In that model, if you go looking in your dataset for specific gene × gene interactions, you certainly won’t find them. They’re tiny—miles below the noise floor. So absence of (that kind of) evidence is not meaningful evidence of absence.
The second thing to keep in mind is: As above, I agree that there’s not much non-additive genetic effects for traits like height and blood pressure, and much more for things like neuroticism and divorce. And many papers on non-additive genetics are looking at things like height and blood pressure. So unsurprisingly, they don’t find much non-additive genetics.…
[Then I discuss three example anti-epistasis papers including both of the ones linked in OP.]
I thought of nonlinearity too. The twin studies compare the effect of different sets of genes. The DNA studies are predictions using different individual genes as variables. Are these linear predictions? The post said interaction terms don't explain anything. How about polynomials of the individual genes ?
Just wanted to say this is a very good comment. It does seem to me that the more likely source of discrepancy is how we are defining a "trait," which is typically a more or less fuzzy concept.
Yes, this is a very good comment and I have been thinking along similar lines. I think to find most of the missing heritability we would need to move from simple additive models to something more like a machine learning model, where rather than starting with the assumption that each SNP causes a simple linear change in a trait and trying to determine what that value is, we start with the assumption that each trait is a complex nonlinear convolution of a binary vector of SNPs, like a convolutional neural network, and try to determine its parameters.
Can confirm I was about to admit it's technically valid, but with hesitation and while giving you a funny look.
But for this particular example, thrill seeking and fragile bones, I think a simple linear model would do fine at identifying genes in both groups. Say you have one gene that increases thrill-seeking by a tiny bit and another that decreases bone strength a little bit. If you have neither, you get fewer sports injuries, both and you get more, just one and you're in between. This is the sort of thing a linear model can detect.
:) Well it's a made-up toy example, but sure, let’s roll with it. Let t be thrill-seeking, b be bone strength, tₐᵥ is population-average of t, etc. A person’s sports injuries = (tₐᵥ+Δt)×(bₐᵥ+Δb). Multiplying that out, we get a population average tₐᵥ×bₐᵥ, and linear (additive) terms (tₐᵥ×Δb + bₐᵥ×Δt), and a non-additive term Δt×Δb. The non-additive term would generally be negligible if the fractional variation is small (i.e. Δt/tₐᵥ and Δb/bₐᵥ << 1). But the fractional variation also might not be small, and thus the non-additive part might not be negligible. For behavioral things in particular—things where people are making decisions about how to spend their time—you can get order-of-magnitude variation in how much sports different people play, how much socializing they do, and so on.
Anyway, my claim is not that mental health, personality, and behavioral outcomes have no additive variation whatsoever, just that those outcomes are MOSTLY non-additive. IIRC, the PGSs for those things generally explain <5% of the population variation, but not zero.
For the record, I do think my ASPD, divorce, and extroversion example above is more realistic than the sports injury example. The sports injury thing was just something I made up to make the math easier to illustrate. :)
Ok, so I fleshed out that example a bit more and went through it. We have 1000 genes for each of thrill seeking and bone fragility, each gene has two incomplete dominant alleles one of which adds 1 point and one of which ads 0 points, and for each gene both alleles are equally common. So the population average is 1000, min is 0, max is 2000.
If we look at a person with values t=1500 and b=1500, then tₐᵥ×bₐᵥ = 1 million, the two deltas times averages both are 500 thousand, and Δt×Δb = 250 thousand. Not too shabby, and certainly large enough to be making a dent in the calculations.
Then I had python generate 3 random people from this population and took the one with the highest combined values for the two traits. This turns out to be t = 1036 and b = 1002. Δt×bₐᵥ = 36,000 and Δb×tₐᵥ = 2,000. Δt×Δb = 72. Small. If you arbitrarily bump up b to 1036 as well that term becomes 1296, still kind of small.
What would have to change for us to have individuals far from the average, like 1500, be more common? We'd need the genes to be fewer and with larger effect. At that point it becomes much easier to find the genes for each of those separate traits (unless the traits are something with extremely non-obvious effects) and then account for their effects on sports injuries, although I don't know of this being a methodology that is actually used.
Afaik, epistasis is defined as "change of fitness from two mutations is not equal to sum of change of fitness from having each one of them separately". Epistasis in this sense was definitely found in deep mutation scan (DMS) studies where one particular protein is mutated randomly. However, link between protein "fitness" and IQ is very much not obvious, and probably because of that also, we cannot detect epistasis reliably in biobanks
>The biggest losers are the epidemiologists. They had started using polygenic predictors as a novel randomization method; suppose, for example, you wanted to study whether smoking causes Alzheimers. If you just checked how many smokers vs. nonsmokers got Alzheimers, your result would be vulnerable to bias; maybe poor people smoke more and get more Alzheimers. But (they hoped) you might be able to check whether people with the genes for smoking get more Alzheimers. Poverty can’t make you have more or fewer genes! This was a neat idea, but if the polygenic predictors are wrong about which genes cause smoking and what effect size they have, then the less careful among these results will need to be re-examined.
I've always been rather skeptical of mendelian randomization (MR) studies. In particular for MR to work, you need:
1. the genes to actually affect the phenotype (often true, but as this shows the effects can be weaker than previously thought)
2. no confounding by population stratification / assortative mating (hard to achieve)
3. no pleiotropic effects: e.g., maybe the genes for smoking overlap with the genes for Alzheimers (unlikely, but possible)
Usually when I see bad mendelian randomization studies (e.g. of ALDH2 variants for alcohol drinking -> cancer) they fail assumptions #2 and/or #3. But now it looks like assumption #1 is potentially worse than expected too!
Any time new types of testing and analysis are administered, and they give *wildly* different results than both multiple types of long used analysis and common sense, we should be skeptical of the analysis and results before we throw away anything of value that has gotten us through millennia of human history and centuries of the scientific era.
In a way this debate reminds me of the intelligent design debate, where one side (evolution believers) have mountains of evidence, but often it was more common sense and circumstantial, based on careful observation. This opened the door to claims about intelligent design that were simple at first, but were transformed by rebuttals until they were in many ways just evolution but “not evolution because evolution is wrong”. Anti-hereditarians are getting closer and closer to the intelligent design group in that they are slowly conceding ground in every debate so as to at least throw into doubt the tiniest bit of hereditarian, trading accuracy for vibes as it were.
“Maybe genetics does affect intelligence, but akshually it’s not your genetics, it’s your parents genetics that you also have which confounds even the claim I just made! Let’s talk around in circles and never admit it’s almost all genetic!”
Right, any speck of evidence on the X side is evidence that X is right, any speck of evidence on the Y side is just confounding and circumstantial. The (high) influence of heredity is not scientifically bulletproof, but the evidence *against* heredity is even weaker and more tenuous, thus not effectively invalidating the longstanding idea of “the apple doesn’t fall far from the tree” or “like father, like son” concepts that go back to antiquity.
There’s an implicit assumption that heritability is a fixed property of traits, rather than a model-dependent statistic that varies with environment, measurement, and population structure. If twin studies, GWAS, and sib-regression all yield different numbers, is that necessarily a “problem to be solved,” or might it reflect that traits like intelligence and EA aren’t stable enough constructs to expect convergence?
I'm seeing a fair few comments about genetic screening and 'careful adoption choices' that are giving me some serious liberal eugenics vibes, nasty. Not sure using anecdotes about criminal kids to defend IQ heritability estimates is that valid; I believe criminal behavior shows less heritability than IQ?
I wonder if there's a sampling bias here? Like only seeing kids whose biological parents got caught and ended up in the system. What about white-collar criminals, financial fraudsters, and corrupt executives who are smart/wealthy enough to avoid prison? If antisocial behavior is really heritable, wouldn't their kids show it too? But those kids aren't in adoption; they've probably been shipped of to elite prep schools to help continue their gene pool.
Basically, it's considering ideas on their own merits, independently of their association with other concepts in popular media.
That being said, even with these ideas considered on their own merits, it _would_ be unfortunate if fewer people became willing to adopt kids of criminals, and so those kids had a worse time. I just don't know what to do about that to help.
It's mental masturbation of that kind that explains why we're sitting in air-conditioned rooms arguing on the internet, rather than staring at a mule's ass as it pulls a wooden plow down a field.
White collar criminals do tend to have a different cognitive and behavioral profile than violent or "compulsive" offenders. Even if perhaps morally the white collar ones are more evil( they have more "control" over their actions), it makes sense that the descendants of white collar criminals might have traits that lead to better outcomes( high conscientiousness, detail-oriented, ability to perform well socially in high-status places) than the descendants of, crudely, lower class criminals.
Of course, but its worth pointing out that white collar criminals like say Madoff or whoever still have a higher intelligence and capacity for work than violent criminals or petty thiefs. The difference between a capable sociopath and an incompetent sociopath matters a lot to social outcomes, even if both are amoral. Certainly sociopathy is a favored trait in positions of power, in virtually every society.
Sorry, I'm not sure what you mean, could you elaborate? Are you saying that I think personality traits aren't influenced by genetics at all, and it's all just about environment/context? Which I don't believe I said. Sorry, I'm not great with language, I'm quite blind to nuance and inference, amongst other things.
I'm responding to your "nasty eugenics" comment. My reading was that you seem to think that people who believe that socially-relevant traits (IQ, behaviors) are genetically-mediated are somehow bad or misinformed.
Oh, I understand now. Well, I was not critiquing the idea that genetics influence traits. I was critiquing the application of that belief to decide which kids are worthy of adoption or birth based on their genetic background. I mean I guess I'd be terminated for a start.
Are you opposed to being differentially attracted to beautiful, successful, or intelligent partners? Because that's just your biology steering your preferences with the goal of creating well-adapted children. I don't see why anyone should have any moral qualms with a more explicit version of the same process.
> I mean I guess I'd be terminated for a start.
I think a better framing is that you'd be recycled into a body that didn't have whatever limitations you seem to be alluding to.
I’m not sure that notion holds up. Not everyone prioritizes beauty, success, or intelligence in partners. people are drawn to kindness, humor, shared values, or just odour. And if biology is steering us toward ‘well-adapted children, how does that explain gay couples, who aren’t reproducing through their partnership?
The ‘recycling’ comment is interesting phrasing. I think you might be viewing neurodivergent as a limitation that needs fixing, but many of us see it more as neurological diversity, different wiring that comes with both challenges and strengths. The idea that I should have been ‘recycled’ into something ‘better’ assumes there’s an optimal human template we should all conform to.
I can understand wanting to give children the best possible start in life. But there’s a difference between choosing consciously or unconsciously compatible partners and systematically screening out entire categories of people from adoption or existence. The latter starts feeling uncomfortably close to deciding who deserves to be here.
I think the first step to thinking clearly about this kind of issue is to try to separate the factual questions from the moral/social/political/tribal ones. It may be that the claim that propensity to crime is heritable is associated with bad people or would have bad consequences if true or has bad "vibes," but that can't possibly inform you about whether or not it's true.
"In the degenerate case where the mother and father have exactly the same genes (“would you have sex with your clone?”) the fraternal twins will also share 100% of their genes."
Why is this so? If Mom and Dad both have genes (a, b) at one location, won't sometimes twin 1 get (a, a) while twin 2 will get (b, b)? Agree there's more commonality than normal because there's a possibility of 1 getting (a, b) while 2 gets (b, a), which isn't normally true.
Edit: I just saw that some of these points get covered in this post, thank you!
Someone with more human genomic knowledge please shoot me down.
Sequencing technology doesn't get discussed nearly enough in this area. Illumina short-read sequencing/SNP panels have been the major source of data for all of these studies, and they are absolutely delightful at finding SNPs but are crap at anything else. I think it will be appreciated that generally things that impact function aren't SNPs, they are broad changes, and so much human genomics seems to be hoping that the thing that is contributing to a change is able to spotted by being close to a SNP, instead of actually looking at the thing that is causing the change.
Genomes aren't lists of SNPs, they are mostly repeats and 2x150bp isn't going to get anywhere near close to capturing that variation, no matter how 'deep' you sequence. Long-read sequencing (PacBio & ONT, not Illumina's synthetic tech) is clearly better, and continues to demonstrate that there is massive variation that is easy to see when you have a bunch of 20kbp fragment, while almost impossible when you're just aligning little chunks of text to a 3gbp genome.
I work in non-model genomics and long-read sequencing is such a clear winner I keep getting surprised when Illumina gets contracts for these massive human studies. The Human Pangenome Consortium is going to be providing a dataset that is way more useful than anything that's come before. Anecdotally, I hear that for some non-European genomic data they know that ~10% of the data from an individual DOESN'T EVEN MAP to the human reference (but is human genomic data). This is all invisible to analysis, or even worse, just confounds things, as the 'true' causal SNP is somewhere in the data that doesn't get analysed, and so we're stuck looking at noise and trying to make sense of it.
I feel like this is such a blind-spot for human genomics, as it's always about the latest and greatest AI/ML method to try and get some information out, when it's the underlying data which just sucks and doesn't have a hope in actually being linked to function. There was even a point on an Open Thread a few weeks back (Open Thread 374 from tabulabio) asking for people to get in touch about Frontier Foundation Genomic models, with the focus being on fancy new ML architectures.
When I asked ChatGPT to write this comment for me ("Argue that sequencing technology could explain a lot of the Missing Heritability problem") it actually pushed back against me, trying to use the Wainschtein et al. 2022 paper as evidence that '...[this paper] used high-quality WGS (which includes better SVs than Illumina) and still found that adding rare and structural variants only modestly increased heritability estimates", which is NOT TRUE. Wainschtein uses the TOPMED dataset, which is from Illumina short reads. Yes, they do 'deep' sequencing, and yes it's analysed to the absolute hilt with the latest and greatest GatK pipeline and QC to the max. But that claim is false, it's just lists of SNPs, completely ignores huge chunks of the genome and just hopes that the thing contributing to a phenotype is is able to be fished out alongside a SNP.
Another anecdote - an older friend's wife died from a brain cancer. He was an old-school non-model genomicist and got all of the data from the oncologists and various tests and analysed things. All of it short-read, none of it turned anything up, either from his or the various doctors. Long-read sequencing run was eventually done after her death and indicated that it was a splice variant missed by short-reads. It was clear as day in the long-read data, since it didn't need to do fancy bioinformatic de bruijn graphs to figure out the splice isoform - it's just mapping the read against the genome and seeing it clear as day.
Thanks - can you explain more about whether things that are advertised as "whole genome sequencing" are the Illumina method you say is inadequate, or whether the WGS data is fine?
Hey Scott, I don't mean to sound hyperbolic but Illumina is kind of like the Illuminati. It's everywhere, monolithic and it's influenced genomics massively.
I had a quick look at a few "whole genome sequencing" retailers, and they’re usually using Next Generation Sequencing, which in most cases means Illumina. The phrase "sequence every base in the genome" sounds impressive, but it’s a bit misleading. Yes, they generate _reads_ from across the whole genome, but they’re in tiny fragments, and only make sense once you align them to a reference genome.
That's where reference bias comes in. You mostly detect things that are similar enough to map cleanly, and just a little different to be able to be called a variant. That’s fine for common variants, but bigger or more complex stuff tends to get missed or misinterpreted.
To give a sense of scale, the human genome is about 3 billion base pairs long. When you get your Illumina WGS results back from a provider, you don’t get a 3 Gbp text file with your actual genome. What you usually get is a variant (VCF) file with just the differences compared to the reference genome. And that makes sense to some extent. Why include everything that's the same? But there’s a lot of complex or unmappable variation that just isn’t detected with short reads.
If you used long-read sequencing and actually assembled your genome, you could get pretty close to your actual full genome, including regions that are repetitive, messy, or just structurally different. You’d see things that are completely invisible in an Illumina dataset. And you'd have much higher confidence in the things you see, since a lot of the artifacts come from using short-read data.
That’s why basically all genomics in non-model organisms is happening with long reads now. At the International Congress of Genetics in 2023 (major conference that only happens every five years) the keynote speaker Mark Blaxter opened the meeting by saying we can finally get real, complete genomes thanks to long-read sequencing. He was talking specifically about the Darwin Tree of Life project, which is trying to sequence all eukaryotic species in the UK.
So yeah, most consumer WGS is Illumina, and it’s fine if all you want is common SNPs. But I can't wait for human genomics to migrate to long reads and overturn some of the perceived wisdom from two decades of Illumina dominance.
I appreciate that this sounds like a lot, but every time I have to describe to someone how sequencing technology works and the degree of genetic variation out there and it starts hard to try and defend the continued use of short-reads.
Genetics _is_ complicated (this is only genomics! not even epigenomics or transcriptomics).
But there has been such a massive degree of complication from going from someone's actual genome info to how genomes are represented in data that, to me, it feels like only the most simple and clear things are able to be identified, and so many complex things are going to be unanswerable. That used to be a technology problem, but PacBio and ONT are almost on the same level as Illumina in terms of cost/genome, and they ACTUALLY give you the whole genome!
As a causal follower of the field, I'd never heard that there were huge questions about the implications of which brand of genome scanner you choose. But that sounds plausible.
>> "But there’s a lot of complex or unmappable variation that just isn’t detected with short reads."
Can you explain this a little more? The way I read what you said at first is that rather than storing each nucleotide in your genome, we instead just store the locations and the nucleotides at locations where where your genuine has a different nucleotide than a reference genome.
But under a naive reading of this, it sounds like as long as you know the reference genome, this is the same information as each nucleotide: at each location i your nucleotide is exactly given by the map "if i is in your genome file, take the nucleotide listed; otherwise take the reference genome at location i".
The simplest way I can imagine your description not matching my naive version is if there isn't a canonical way to match locations at two genomes: if the mapping is only defined on certain sections of the genome where we can define an unambiguous location.
Is that what's going on? Or is it something else I'm not considering?
I'm not OP, but I do work with whole exome data*. What you get back from an Illumina sequencer is a set of 150-250bp "reads", along with quality data. Each read is then mapped to the reference genome (along with some QC and clean-up before and after), and the final file is the reference genome with the reads aligned (uncompressed=sam, compressed=bam, super-compressed=cram). So some regions will have a lot of reads aligned, some may have fewer. And some reads will align wrongly - there are a lot of regions which look very much like other regions, and a short read will not be able to distinguish them.
Anyway, the VCF (variant call format) file is generated from this bam file by comparing the reads at each point with the reference, and it is typically just the differences which get returned. One read out of 20 with the wrong base pair is just a sequencing error, and won't return a variant. But if there are 20 reads covering a T base, say, and 10 of them are A, that's likely to be a heterozygous variant. Of course, it could also be a place very similar to another region where the A is the same as the reference, but both reads are mapping to this place instead. A good variant calling pipeline will use as much data as it can to minimise this - calling from multiple individuals at once helps - and for my work, I use three variant callers and combine their output with a set of criteria established by going back to the samples and resequencing (with old tech) various SNPs. But nothing is perfect. It is quite common to just get the VCF file from a commercial company, and this is why I do the whole thing myself, from the raw reads to the eventual VCF file - and then we also went back and checked a selection of the SNPs, both against the bam file and by old-tech sequencing.
* Exome sequence data gives you better coverage of the exome, which is just the 2% of the genome which codes for proteins. Depending on the kit, you sometimes get the untranslated regions at either end of the gene, and you always get 10-20bp into the intron too. You will of course miss a lot of intergenic and intronic variants, but nobody knows what to make of them anyway (this is improving all the time, though). But it is far cheaper, and in contrast to what another poster said, the kits are designed so that the entire exon is covered by actual reads.
But I would just draw attention to the difference in 'sequencing' the whole genome and doing all the things you are thinking about with genome alignments.
Sequence companies will say that they 'sequence' the whole genome to some depth (eg 30x coverage). But the majority of the genome is repetitive (transposable elements like SINEs, LINEs make up >25% of the genome), so lots of that data doesn't make any sense by itself, and you need to align it to a reference. You take this aligned data and try and call variants - but what you get in that VCF file _isn't_ your whole genome, it's the bits of your genetic data that are able to be analysed in context of the reference.
Anything that _doesn't_ map doesn't get called. And anything that maps awkwardly gets filtered away, or mis-called.
This is known and discussed as reference bias and there is work in human pan-genomics field trying to address this. But my main gripe is that Illumina has been the workhorse of genetics for so long that people forget that it's fundamentally limited in its ability to capture 'full' genomic information.
I work with lost of very smart AI and ML engineers and they hear what genomics people and companies say and have that image in their head - "If what I've bought is a Whole Genome Sequence, then all of the genome is there! Oooh and what's that? It comes as a table? Awesome, my ML model is going to chew this up and we'll have an answer in no time!". But that's not true, and I feel like often forces an answer from the data.
I'm mainly griping about this since PacBio and ONT (the main long-read providers) seem to not be able to break the market hold of Illumina because of this mis-perception. Those providers will give you an essentially 'full' Whole Genome Sequence, where all of your bases are captured, including the variable bits. And from a analysis perspective I just feel this is the correct way to start.
I guess if I had to justify this I would look at the Human Pan-genome data released for individuals (made from long-reads, full haplotype resolved chromosomes) and compare the information from that vs the same samples with a short-read libraries. Anyone feels like collabing on this that would be great! I've got free compute for general bioinformatic stuff but don't know my way around human genomics that well.
I'm probably not the first to think of this confounder, but it's been estimated that ~70% of human embryos turn out to be non-viable [https://pmc.ncbi.nlm.nih.gov/articles/PMC5443340/]. Do the sib-regression studies account for this? The 40-60% range of shared genes they measure is restricted to the siblings that survive to adulthood. But there may have been many 30% embryos that didn't survive the first month of gestation. Their educational achievement was 0.
In other words, there's a severe selection bias in any study that only looks at siblings who survived to adulthood. A genome that kills one of the siblings before they complete schooling or even get noticed on ultrasound is still an inherited trait.
I think this changes some extremely nitpicky form of heritability defined as "heritability across all embryos rather than just surviving ones", but I think as long as you define heritability as being across surviving humans, this definition is consistent and fine, and all of the existing studies remain comparable (and so it's still surprising when they don't match)
Would someone with strong quantitative chops be willing to explain whether GWAS (and related methods) have enough power to detect small effects—say, 0.5 or 1 percentage point—caused by combinations of 9, 11, or 53 SNPs? I’m particularly curious about scenarios where some of the SNPs are rare. In that case, the number of people with a given combination might be small, so the signal could be lost. If that sort of undetected combination effect happens multiple times—say, 30 or so—could that help explain the gap between GWAS results and what twin studies suggest? I’d love a mathematical explanation that someone with several semesters of college math/statistics could follow with effort.
I'm not aware of any method which will detect combinations of variants across multiple genes - if anyone knows of one, please tell me! There are a few methods which come close (eg ORVAL, https://orval.ibsquare.be/) but nothing I'm aware of for studying large cohorts and many combinations. I can see why it would be very computationally expensive.
I’ve been chatting about this with AI, and once you do combinations of 2 snps you are talking about billions or trillions of degrees of freedom and it’s really hard to get any signal. The biggest databases only have roughly a million genomes sequenced. you can’t do a matrix regression with an n of 1 million and 1 billion degrees of freedom.
Excellent post, I would just want to add that Sidorenko 2024 (https://pmc.ncbi.nlm.nih.gov/articles/PMC11835202/) that you cite for the discussion on rare variants also measure heritability using Sib-Regression although not for EA.
They show that correcting for assortative mating their estimates are coherent with twin studies :
"Therefore, if we account for assortative mating and assume that the resemblance between siblings is solely due to genetic effects (and not to common environmental), then our data are consistent with a heritability of 0.87 (s.e. 0.05) for height in the current population (Supplementary Note)."
they also show that shared environment estimates from twin studies are inflated by assortative mating :
"Our results for height and BMI agree with that conclusion in that we find no evidence of a residual sibling covariance (𝑐2) for BMI, while the significant 𝑐2 observed for height is largely explained by assortative mating."
And the WGS estimate is higher for height than the RDR estimates (55%), my take on this is that there is an assumption in RDR that posit that the environment of individuals are independent to each others and don't influence each others in correlation to the relatedness which since it's violated bias the estimate downward. (https://hereticalinsights.substack.com/i/146616013/disassortative-mating)
"Moreover, our estimates of 0.76 and 0.55 can also be compared to estimates from GWAS and WGS data. For height, the SNP-based estimate is about 0.55–0.60 (ref.41) and the WGS estimate ~0.70 (ref. 42; but with a large s.e. of ~0.10). These estimates imply that for height there is substantial genetic variation not captured by either SNP array and, to a lesser extent, sequence data, presumably ultra-rare variants (frequency <1/10,000 not included in ref.42)"
You are right, my bad, from the supplementary note 4 of Kemper 2021 (https://www.nature.com/articles/s41467-021-21283-4) they use the same equation to correct for AM and there is still a 0.1 gap in heritability for the two methods.
Sidorenko 2024 use a slightly different adjustment method to correct for the shared environment estimate becoming negative after correcting for assortative mating (another mystery to solve).
While writing this I just saw that the estimate twin heritability for height in Kemper 2021 is lower than the heritability for height in Sidorenko 2024 using sib-regression. Missing heritability solved I guess.
Gusev has a point when he says that when adjusted for assortative mating some of the high twin estimate for height or even sib regression get near or above 1 (when you read the supplementary note of Sidorenko 2024 the crude adjustement for a r=0.23 assortative matting in height push the estimate to 0.98 ! )
Turkheimer's quincunx metaphor is my personal mental model for the missing heritability phenomenon. A blog post is here: https://ericturkheimer.substack.com/p/a-speculative-explanation-of-the-e53 though you'll have to look at the few previous posts to fully understand it. It seemed like you dismissed this very briefly under non-additive effects and I don't have the technical background to grok those papers, but Turkheimer takes this hypothesis seriously so I figure it's serious.
Here's my summary:
Imagine a quincunx (aka a Galton board). You are a ball falling down the board. It bounces off a lot of pins, and it eventually falls into a bin at the bottom. That bin at the bottom is the outcome we're interested in (EA, IQ, etc). But the pins aren't random -- the pins are your genes, environmental variables, etc. So some pins are biased left or right, which influence the bin you end up in.
What Turkheimer shows is that in this model, if two people have the exact same quincunx (the exact same weights of pins), then they are very likely to have similar outcomes. Those are identical twins. If all your genes are the same, you are very likely to be similar. This reinforces the hereditarian position, and also makes a lot of intuitive sense. (One tricky part here is that it must be the case that mostly-the-same quincunxes (siblings, etc) cause similarity as well to a lesser degree. That's the biggest challenge to this mental model.)
But Turkheimer also shows that biases in pins are very very hard to detect in the non-identical-twin situation. Each pin only has an effect in the context of the entire system. For one person, a particular pin could have a large impact (maybe all the pins to the left bounce left, and all the pins to the right bounce right, so that pin matters a lot) but for another person that same pin has no influence at all (the pin to the left bounces right, the pin to the right bounces left, so it's like a funnel and that pin doesn't matter). So "in the wild," away from the twin context, it is very hard to come up with pins that matter in a GWAS type of way. But pins do matter! Genes do matter! They just happen to matter in this way where the whole package has a large influence on an individual, but individual pins are really hard to pick out in a practical way.
Turkheimer would emphasize that this is a simplified mental model. Our genes are not actually a quincunx -- but the interactions are much more complex, so if he can show that a quincunx has that sort of irreducible complexity, genetics must be even more complex.
I find this mental model really interesting and I actually wrote a blog about how I find it a helpful way to think about education and schools with some GIFs modeling the quincunx part if anyone is interested: https://fivetwelvethirteen.substack.com/p/the-quincunx
I think Turkheimer’s analogy here, given that he says that some of the quincunx pins are environmental, is dramatically failing to capture the fact that adult identical twins who were raised by different families in different cities are very obviously similar in all kinds of ways, indeed about as similar as if they were raised in the same family. (…With a few caveats & exceptions that I discuss in §2.2 here → https://www.lesswrong.com/posts/xXtDCeYLBR88QWebJ/heritability-five-battles )
That's a fair criticism of Turkheimer's position but I think you can rescue the broader idea by just saying that the quincunx is genetics only and the final bin is the sum of the genetic influence. To me, the main insight is that a complex system can have emergent properties that aren't easily correlated with the individual components of the system. And that's the core of the missing heritability problem: when we look at the whole system with twin studies we see high heritability, but when we try to find specific causal genes it becomes really messy.
I do think non-additive genetic effects are very important for adult personality, mental health and behavior (but not for other things like height and blood pressure). I talk about that a bunch in my post; see excerpts that I copied into a different comment → https://www.astralcodexten.com/p/missing-heritability-much-more-than/comment/129514879
Something that comes to my mind, having just read Pearl’s Book of Why, is whether collider bias or other common problems of regression in the hands of folks not very savvy about causal inference might explain the inconsistencies in the findings.
I mean, there is a lot of regression going on here, a lot of moderating and mediating variables, so it seems easy to make a mistake with the regression model.
Nice, I am a bioinformatics PHD working at least somewhat with GWAS and SNPs and from a Gell-Mann amnesia effect perspective this post was very reassuring.
My guess is also that the rare/ultra rare variants/combinations are responsible.
Lots of identical twins were raised to deny they were identical.
For example, in the 2004 Olympics Paul Hamm won the gold medal as the best all-around male gymnast while Morgan Hamm was 5th. But Paul and Morgan's parents doubted they were identical because their hair whorled in different directions.
Similarly, the nearly 7 foot tall Lopez twins of the NBA have done a great job over the decades of acting non-identical: they wear different hair styles and different clothes.
My impression is that some identical twins like being identical and work at being even more similar (like those two Australian ladies who were in the news recently), while others work at emphasizing their differences: e.g., the basketball Lopez twins display different personalities in social media.
The movie "Sinners" seems to do a pretty realistic job of displaying the identical twin characters as quite similar, but not wholly the same.
I think GxG is basically impossible to measure, so saying nobody found any GxG yet isn't really meaningful. It's harder to measure than rare variants; why believe in rare variants but doubt GxG?
Also, note that additiveness is sensitive to nonlinear functions of your measure. For example: if heritability of IQ is purely additive, then heritability of IQ^2 (the square of IQ) would not be. To posit no GxG at all is to say that the IQ scale is perfect and cannot be squared (and EA scale, and BMI, etc too).
A few final notes:
1. I think intuition from real life strongly suggests that everything is GxE (gene environment interaction); think, for example, about obesity. All the models assume GxE is zero in order to calculate the percent heritable.
2. You say assortative mating cannot bias downwards the heritability estimates from twin studies. That's only true if the twin studies don't already adjust for them! Many hereditarians cite 80% h^2 for IQ, but they get this by already adjusting for AM (and maybe even for attenuation). The can easily over-adjust! That would bias the estimates upwards!
3. Related to 2, I noticed that you quoted 60% for the typical twin-study heritability estimate, even though hereditarians (including you in the past) like to cite 80%. That's a substantial gap. Are you saying you now believe the 80% is wrong? Can you be more explicit about this disconnect?
Obesity obviously has a huge environmental component as can be seen in changes in obesity over time. For example, after being very scrawny from 1945 onward, West Germans from about 1955 onward suddenly got very fat (think of Augustus Glump in the 1960s book "Charlie and the Chocolate Factory.") But then they got skinnier again in the 1970s. Now I'd imagine they are fatter again than their parents were.
Beyond the fairly global Flynn Effect, I can think of only a few well documented cases at the national average IQ level of significant improvements relative to other countries: Japan and then South Korea. Both big gains in average IQ coincided with huge increases in average height over the generations.
Otherwise it's hard to see the kind of rapid national changes that were evident in West Germany's level of obesity.
I wouldn't be surprised if there have been more relative changes in average height than in IQ: for example, I was in my mid-30s before I heard that the Dutch were really, really tall. That hadn't been a stereotype in the 1970s yet, in part because the Dutch weren't that tall yet.
So, while I certainly believe that environment can have a big influence on IQ, it's hard to find all that many examples.
Well, I don't know why we are talking about countries improving relative to one another instead of in absolute terms. It seems like if you can improve IQ in absolute terms, then it is environmentally malleable, no?
Also, I don't know that we'd even know about it if there were changes in the rank order of the IQ of countries over the 20th century; IQ is much harder to measure than height or BMI (there are many different IQ tests, they need to be translated or you need to use a "culture fair" test, testing itself takes hours instead of seconds, and it is much more subject to selection effects such that it's unwise to take a convenience sample of (e.g.) immigrants or something, etc.)
we're talking about whether IQ is environmentally malleable. You may say that the answer is "yes but general intelligence is not". That's a fine answer. But "yes" part is what I'm asking about. I don't care about intelligence, I care about IQ. Is IQ environmentally malleable? If so, why do studies say the answer is no?
I'd say that IQ is environmentally malleable, but there isn't much evidence that it is easy to close racial disparities in average IQ scores (which is what people are mostly talking about and why so many science denialists get mad at people who know something about IQ.) Instead, what seems to happen most times and places is that improved environment leads to everybody's test scores going up.
Will try to respond to 1 in a Highlights From The Comments post.
Re 60% vs. 80%, I admit I haven't been very careful in keeping these numbers separate. My impression is that twin studies usually find somewhere between 50-80% with the differences being age (older = more heritability), assortative mating correction, which IQ test you're using, and whether you've already tried to correct for measurement error in the IQ test. I don't think there are "right" answers in what age or IQ test you should use, or in how many things you should try correcting for, so I don't particularly care about the differences between these estimates and mostly just say whichever one is on the top of my tongue at any given time.
Thanks. I think everyone usually talks about adult IQ and everyone wants to correct for AM (which is considered a source of bias for twin studies). As for measurement error, that one is admittedly trickier, because with other measures we don't usually adjust for it: measurement error exists for height or BMI too, but nobody tries to correct for it.
To make a fair head-to-head comparison with RDR or whatever, you'd surely want to adjust for AM, right? And the UK Biobank uses adult IQ tests, though arguably not very good ones. So we definitely want to compare to twin studies which use adult IQ and adjust for AM. Educational attainment is also about adults, though now that I think about it I don't know how they handle young adults who are not yet done their schooling.
(I'm suspicious of over-adjusting for measurement error, so I'd rather studies not try to do it, but it should definitely be noted that (from my understanding) the UK Biobank IQ test should have higher measurement error than more standard IQ tests.)
I find the idea of reading modern scientific findings or social innovations into ancient myths and religions very compelling. As if the ancients through cultural evolution were able to find fundamental truths about reality without having to actually understand them. In light of that interest, here we go:
The ancient Greeks had the concept of generational sin. When someone committed a truly horrendous sin, like feeding your brother his own children as an act of revenge (https://en.wikipedia.org/wiki/Atreus), the gods may not punish the man directly, but leave punishment for a future generation.
Atreus (Progenitor of the house Atreides. Dune anyone?) was grandson of Tantalus, infamous for feeding Zeus his own son, and being punished by eternally having food and water just out of reach (origin of the word: tantalizing). Atreus echoed his grandfather's sin by feeding his brother his nephew's corpse and usurping his throne. Atreus' son, Agamemnon (famous from the Iliad) killed his own daughter, Clytaemnestra, in order to be allowed to attack Troy. In punishment for that, he was killed by his own wife upon returning victorious from Troy. His son, Orestes (grandson of Atreus) was given the "divine madness" (basically Schizophrenia) as punishment and it was only through the gods intervention that the cycle of inherited sin was synbolically broken. It was only ultimately ended when a new lineage, the Heracleidae (descendants of Herakles) invaded and taking over their lands, loosely representing an invasion of Northern Dorian peoples taking over the Achaean society that was there.
The lesson I take from this is, the Greeks understood the heritability of preponderancy for antisocial behavior. If your father or grandfather did something unspeakable, you were by extension unclean, and shunned by society because of that. If there is some actual truth to this, and children of men who are so antisocial as to practice cannibalism or other terrible sins, were poisoned by a Miasma that would carry over between generations, then perhaps there was a collective social benefit towards shunning the children and grandchildren of heinous criminals. The societies that practiced this, while fundamentally less just on an individual level, would have been able to better weed out antisocial behavior, and in the long run, outperform societies that didn't have this cultural meme.
The lesson at a societal level is that if your rulers have 6 generations of inherited sin and insanity, you will be invaded by your neighbors and made into slaves.
Of course in the modern day we've moved past that, and if you think hard enough I think you can come up with a "reasonable" justification for literally any social practice, but I always find it interesting to think about.
It's in the Old Testament too: "for I, the Lord your God, am a jealous God, punishing the children for the sin of the parents to the third and fourth generation". I've been similarly impressed by the insight that Buddhism had into the functioning of the mind (anatta etc). Phenomenology and careful thinking can actually get you pretty far.
> Atreus (Progenitor of the house Atreides. Dune anyone?)
Not quite correct; -ides is the Attic (I think?) Greek patronymic and so Atreides only refers to a son of Atreus. (There are two of them, Agamemnon and Menelaos, both called "Atreides".)
The whole lineage, the metaphorical rather than literal sons of Atreus, would be called the Atreidai.
I was more referring to House Atreides from the book Dune, which are canonically descended from Agamemnon as a bit of rhetorical flare to make my comment slightly more interesting to the reader who has read the book or seen the movie.
Does it take any money to finish college? I thought we had a robust system of student loans that will happily finance any degree at any school no matter what the prospects of getting paid back might be.
Some people are responsible enough to realize that dropping out would be better for all concerned. Similarly, I can recall a public high school teacher during the Housing Boom in 2006 telling me that many of her more impressive male students were dropping out to work construction.
1. I wonder when we're going to have a good computational model of personality to just directly track these issues on a gear level, like what tradeoffs are solved in what proportion, what inefficiencies are present in the Turing machine that corresponds to one's intelligence, etc. (Sounds easy, right?..)
2. Is it possible that any single study (i. e. the Iceland study) introduced a trivial calculation error, or a different calculating method that at least partially accounts for the discrepancy? Are the datasets in question open-source, did people reproduce their calculations and make sure it's not that?
I wonder how much in vitro fertilization might impact more recent twin studies. Also the fact that we have mitochondrial DNA as well. So more and more data to feed into the maw of the ai models. De novo mutations also seem to impact certain genetic areas as well. Maybe we have a gene that predisposes to de novo mutations.
Consider syndrome vs. disease. It seems to me that the likely cause is that most of the things being measured have multiple independent ways of causing the measured result, and the genes that facilitate one of them don't necessarily facilitate another. Dawkins calls this teams of genes that are "good traveling companions". E.g. both Tibetan and some Peruvians have genetic adaptation to high altitude, but they do it in different ways.
So under this assumption (which I think is fairly reasonable) if you measure one gene, you're measuring one part of a collection that *IF PROPERLY ASSEMBLED* would facilitate one trait. But someone else presenting that trait with a different underlying approach would not find that gene helpful.
That's just non-additive effects. We ignore those for pretty good reasons; they cannot be picked up by natural selection in a system where reproduction occurs via meiosis. You'd need to evolve a way to ensure that they always passed on together. (The simplest way would be to drive them to total fixation, but that's hard to do.)
Note that it isn't true that providing Tibetan adaptations to a Peruvian, or Peruvian adaptations to a Tibetan, wouldn't be helpful in supplying additional oxygen. It would. The two populations are doing different things, but not things that conflict with each other.
(The Tibetan adaptations are much better; I believe the Andean ones have some potentially undesirable side effects, so a Tibetan might not find them helpful _on net_. An Andean would benefit from Tibetan admixture.)
But when you are trying to explain variation in a population, shouldn't sources of variation that natural selection doesn't work (efficiently) on be *over*represented? You would expect advantages that can be readily selected for to be present in everyone, and so contribute nothing to variation.
You don't expect natural selection to remove variation. The point of having the variation is that natural selection can act on it if the trait becomes more important than it is now.
For example, it isn't the case that there's one correct height for everyone.
To become fixed, a trait needs to be subject to extremely strong selection.
Note that the problem with your theory applies at both ends: just as the benefit of having variation is that selection can pick up on it in the future, the way that variation came to exist is that selection picked it up in the past. There isn't a mechanism by which adaptations that are invisible to selection can persist in the gene pool.
"An alternative way of validating twin studies... is to check them against their close cousins, adoption studies..."
Using adoption studies to "validate" twin studies seems completely backwards to me. Adoption studies are so obviously flawed that I do not see how they could possibly act as a validation set for anything!
The population of children going into adoption are wildly unrepresentative of the general population of children, unless that population is "children who go into the adoption system". The population of parents adopting children are obviously unrepresentative of the general population of children, unless that population is "parents who are adopting children". Being an adopted child to a set of parents is not like being a biological child to the same set of parents. Losing or being given up by birth parents is generally kind of a big deal.
If anything, the fact that twin studies give the same answers as adoption studies should make us more skeptical of twin studies!
In any case, I applaud Scott for both a) being totally transparent about his completely bonkers epistemic commitment here, and b) giving a good-faith description of the current evidence against his aforementioned bonkers epistemic commitment.
The claim isn't that adoptive kids are representative of the general population!
The claim is that you can compare the correlation between adoptive kids and their bio parents, vs. those same adoptive kids and their adoptive parents. As long as there's any variation at all among adoptees, you can use this to see whether it's the genes or the environment influencing them.
For example, if the child of a genius gets adopted by dumb people, and grows up to be a genius, that provides some evidence that intelligence is genetic rather than environmental.
You might additionally have a claim that adoptive people get traits in a totally different way than non-adoptive people, but I think the burden of proof is on you there and that has nothing to do with the question of whether adoptees are "representative".
Apart from the sometimes-questionable validity of the studied “trait” (e.g., “general intelligence,” “personality,” “schizophrenia”), however, others have shown that adoption studies are subject to numerous environmental confounds that some critics argue invalidate genetic interpretations of the results. That is, like family and twin studies, behavioral genetic adoption studies are unable to adequately separate (disentangle) potential genetic and environmental influences on behavior. These confounds include that most adopted children (1) shared a prenatal environment with their often-stressed birthmother during sensitive developmental periods; (2) were reared for a certain period by their biological parent(s); (3) suffered a rupture of attachment bonds with the biological parent(s) who gave them up for adoption (children grow up feeling abandoned); (4) may have been placed between separation and adoption into unstable orpsychologically/developmentally harmful environments, such as foster homes and orphanages; (5) share or potentially share with birthparents similar socioeconomic status, physical appearance, ethnicity, culture, religion, and so on; (6) were not randomly placed into available adoptive homes (agencies often selectively place adoptees into homes based on SES and the child’s perceived genetic background); and (7) were placed into adoptive homes of restricted socioeconomic range. In addition, adoption studies are subject to research/publication issues and confirmation biases coming increasingly to light in science’s replication crisis.
In addition to the abovementioned confounds, behavioral genetic adoption study researchers compute IQ correlations between different "Flynn effect" populations. Biological and adoptive parents are a generation older than adopted children, meaning they were born at different Flynn effect “massive IQ gain” starting points 25 or so years apart. Moreover, because birthparents are often unwed teenagers, they are typically closer in age to the child they gave up for adoption when compared with the older adoptive parents. In Plomin and colleagues’ Colorado Adoption Project IQ study, for example, birthparents “were younger on average (20 years) than the adoptive parents (33 years)” (Plomin et al., 1997, p. 443), suggesting additional Flynn effect confounding.
I go to one of those fundamentalist churches where people adopt kids and raise them piously. I think everybody knows by now that there's a big danger the kids will turn out badly compared to their siblings, even though there's considerable success in training them to behave well as kids. These adopted kids are often foster care kids, and that's going to add a lot of noise to any study. They don't have normal environmental differences; they have huge ones-- fetal alcohol syndrome, autism, abandonment as babies, abuse, . . . So for finding the environmental share for the US on average, they're not ideal.
I am alarmed at this statement about why we care if gene studies show that heritability may be misunderstood.
“Not doctors. So far this research has only just barely begun to reach the clinic. But also, all doctors want to do is predict things (like heart attack risk)”
We absolutely care if doctors over-treat or apply dubious preventative treatments based on unreliable associations.
Do I want surgery removing my prostate and rendering me incontinent because of a false read on my prostate cancer risk?
Correlation versus causation always matters when a reading is used as the basis for an intervention.
The next paragraph criticizes the use of these scores in epidemiology. Actually, even if all you have is variable with a correlation, it can be useful as an instrument in multiple regression analysis.
Suppose that black people have higher heart attack risk because they have worse diets, and genetic studies pick this up and (accidentally) identify the gene for black skin as a gene for heart attack risk.
This is bad if you're doing research or embryo selection. But a doctor would do a genetic test on their black patient, see that it said they had higher heart attack risk, put them in a high-risk category, and maybe give them more treatment. And that would be correct, because in fact black people do have higher heart attack risk! Even though the gene isn't causal, it's having the correct effect.
This might be a problem if the doctor already knew that black people had higher heart attack risk and so double-counted on the result of the screening (eg the doctor thought they're black AND they have bad genes so they must have two risk factors). But in real life it's rarely something as obvious as "they're black" and more some kind of hard-to-describe sub-sub-sub-group.
But if black people only have worse diets *on average* (say because they're economically disadvantaged *on average*), the doctor will wind up over-treating wealthy black patients with balanced diets for no reason.
Exactly. That is my concern. You maybe a black person who is socioeconomically “white” with a healthy diet andhabits. Such people could be systematically over treated for certain medical vulnerabilities based on a spurious genetic correlation
Obviously the best thing to do is completely understand the causal graph connecting race to diet to heart attack risk.
But if you can't do that, I don't think this is any worse than any other genetic situation. Every gene increases your risk of heart disease by some probability. Suppose people with some gene have an 80% chance of getting heart disease. If your doctor treats you, there's always a chance that you would have been in the lucky 20%, and this treatment was pointless. If the doctor could magically know whether you were the 80% or the 20%, they should do that. Otherwise they should just do the best risk-benefit calculation they can.
Likewise, suppose that 80% of black people get heart disease because of bad diet, and 20% eat a good diet and don't get heart disease. Since we're inventing this hypothetical and we know this, we can avoid treating black people with good diets. But if the doctor doesn't know this, them picking up on this "gene" (actually a gene associated with African ancestry) and detecting that it gives you 80/20 chances is no different than any other gene.
Wealthy black people have a much more elevated risk of heart disease compared to wealthy white people. It is a true genetic effect, not cultural, by the best of our scientific evidence.
I thought this little-read article (written by an astrophysicist) does an excellent job of explaining why we wouldn't expect GWAS studies in their current form to get anywhere close to "true" heritability: https://coel.substack.com/p/gwas-studies-underestimate-the-heritability. Notably, the author doesn't rest his argument on the "rare genes of large effect" factor.
To roughly summarize his article: We would expect the code for developing a human-level intelligence to be extremely complex. There are ~3 billion nucleotides in the human genome; these GWAS studies rest on the assumption that only a few hundred to a few thousand SNPs are entirely responsible for creating human intelligence. That seems wildly inaccurate. As the author argues, could you write code for developing a human-level intelligence with only a few thousand instructions? ChatGPT is based on hundreds of billions of neural-network weights (though admittedly this is the end-product, not the "recipe" for producing it like the human genome). If intelligence is based on tens of thousands (or even more) SNPs, then our current GWAS studies are simply statistically incapable of finding them.
On top of that, GWAS is only looking at a very simplistic model of gene-behavior causation. It assumes an "additive" model of heritability which simply sums the measured effect of each SNP, whereas in reality the recipe for intelligence is likely caused by subtle and complex interaction effects among genes as well. (The author also mentions that SNPs, which are all GWAS studies measure, are only one type of genetic variation. This seems like it could be important but I don't understand what it means.)
In contrast, twin studies take into account the complexity of developing intelligence by simply observing the output of the incredibly complex underlying genomic causes. It's therefore not surprising at all that they give much higher heritability scores, whereas GWAS studies, limited as they are, give nothing more than a lower bound for heritability.
How do they adjust EA for non-college education? In my job I interact with many elderly. I ask them what they did. I live in a high-tech area so frequently I hear 'engineer'. But when I ask them where they studied, I am more likely to hear 'I got out of the army and went to work for Sunstrand as a mechanic and they trained me' than to get the name of a university. Very smart folks, but formal education is high school.
Total non-expert speculation here. Has anyone looked into whatever the genetic equivalent of “sequence of returns risk” is? In the market it matters enormously if there is a crash immediately after you retire versus a decade after you retire and your investments have had extra time to grow.
I.e as the body is built, different genes will apply in sequence. It could be that early-in-the-sequence genes (about which we currently know little) will have an outsized effect. It may be that some genes only activate or have an impact conditional to other genes activating or having an impact. Etc. Or perhaps from another perspective the mistake is looking at the genome as if it is static and one dimensional when in fact it is self-referencing and applies across dimensions of time and space.
> In the market it matters enormously if there is a crash immediately after you retire versus a decade after you retire and your investments have had extra time to grow.
I'm always shocked at the widespread assumption by financial "health" "experts" that the goal of investing is to end your life with a net worth of zero, leaving nothing to your heirs.
I’m not sure whether this deserves a response or it just meant as a flippant comment. While there certainly is a “you can’t take it with you” attitude amongst some in the retirement savings space, I think many people don’t wish to burden their heirs with medical debt and still more are using consumption rules of thumb that are overly cautious. To bring this back to genetic fitness, genes that take you close to dying but don’t kill you before reproductive age are harder for nature to filter out out than genes that do lead to more death before reproductive age. The sequence of the effect matters greatly.
I'll be honest, I haven't read word for word your post, but I read well enough. I will continue to read as i have time. I think the answer doesn't exist yet but will exist soon considering ai. The true answer is having behavioral patterns linked to endocrinology. Many don't take psychology seriously, but it is what we have. I haven't seen any reliable studies regarding hormones and behavioral traits.
Regarding educational attainment, I think it's also important to consider diet. If people do not get enough protein and fats, they will not be at their peak intelligence. Addictions of any sort will lower iq.
I am of the mind that virtually everything is inherited. Even broad behaviors amongst the population. We can connect hormone levels to DNA and hormones to behavior.
Was surprised to find that Scott used o3 to help research this, and found it helpful, despite the continuing hallucination problem. Let's say it gives you certain hallucinations you know you need to check, like citations, and you check them and strike that, fine. What if it's negatively-hallucinating, insistent that something does NOT exist when it does? What if it has numerous unknown-unknowns that it doesn't realize are relevant to the research question?
It seems like you probably need at a *minimum* to have Scott's level of mastery and familiarity with the subject going into the project in order to have any chance at spotting the places that o3 screwed up, and knowing why, and knowing where to look to correct it. The entire field of research needs to have been reduced to known-unknowns for you the author before you can safely employ the assistance of AI tools.
Here he had extensive layers of human review/assistance by specialists in the field, and had previously researched the topic. But I just don't know why anyone would bother to use AI for research when the output is clearly unreliable absent such review and input, in ways that a researcher with incomplete knowledge won't be able to always detect.
To point out, the existence of unknown unknowns is unlikely to get *worse* with AI. If the AI explicitly denies something, that's a thread you can search on. If it doesn't mention it you are identically in the position of not using AI. The only way this can make it worse is if the AI somehow comes up with similar meaning terms that are false then unreasonably generalizes in a hard to fathom way, but then you'd have to explain why those hallucinations would be worse than for example, finding a bunch of confused politically motivated parts of research.
Why do we expect genes to have independent linear effects? From gene to protein to trait to life outcome, there are many nonlinear effects and interactions that we could imagine — is the problem really our imaginations? Hunting for genes that linearly independently correlate with causally remote life outcomes seems like looking for your keys under the lamppost. But I’m not really qualified to dig into the studies that you say disprove interactiond.
Not an expert, but I think there are a few arguments for this. One is that additive genetic models work pretty well: phenotype variability seems to depend linearly on degree of relatedness (MZ twins correlate twice as much as DZ twins, which correlate twice as much as half-siblings). Secondly, if nonlinear effects were large then phenotypes wouldn't be normally distributed: they'd have longer tails or be skewed or multimodal more than they appear to be. And finally from an evolutionary design perspective the system is going to be much more stable if it's linear: phenotypes that require a complex interaction of many independent alleles probably aren't going to do that well with recombination.
One easy possible confounder for twin studies would be if some aspect of your early childhood environment mattered a lot for some trait (IQ, kidney function, BMI, whatever) and also a lot of that environment was a function of your appearance. That would give you inflated heritability results for identical vs fraternal twins.
It doesn't seem so likely to me that this is important, though--it seems like adoption studies make a pretty strong argument that there's not much impact of variation in early childhood environment within normal middle-class-and-up norms and most measurable outcomes. (Raise a kid in a cave, beat him regularly, and never show him a book, and you'll probably depress his IQ. But letting him watch 20% more TV or taking him to amusement parks instead of museums on vacation isn't likely to have any noticeable effect.). But I guess if really cute kids get extra attention at some critical point in their development relative to ugly kids, it might have some effect?
Well, I never thought I would see this day. Finally, the smart people are coming around to what I have been claiming for decades. Inheritability is an interaction effect, thus the main effects (variation due entirely to genes or learning) are not directly measurable. Also, the twin studies are deeply flawed.
Let's start with twin studies. They have problems with their independent variables (IV's), their dependent variables (DV's) and their analysis.
IV's like personality traits and IQ test scores are highly controversial. We don't know exactly what IQ scores measure--except that it isn't "intelligence" the way most people mean that term (something like "an individual's inherent capacity to solve real world intellectual problems"). No one has proven that such a capacity exists as a coherent cognitive process, let alone that IQ scores measure it. Every other personality inventory suffers from similar problems.
The DV is correlation in such traits between pairs of siblings. This is measured two different ways: by degree of genetic similarity, compared to degree of environmental similarity. One of these two measures (genes) is much more precise than the other. These days, they can not only tell you how many genes two siblings have in common, but often which ones.
But not environment. The original twin studies used household to determine similarity of childhood environment. This is far too crude. Many twins are adopted into different households, but by members of the same extended family (aunts, uncles, grandparents, etc.). This means that the twins will know each other, often live in the same neighborhood, go to the same school, play with each other, etc. Since these things are largely undocumented, the data is contaminated, and we can't know by how much.
Finally, there are strong theoretical reasons to believe that evolutionary effects should be interactions between genes and the environment. The environment, after all, is the nature that selects. Therefore one would expect that any species will evolve traits that engage features of the environment, which is the mechanism by which an organism can gain an advantage. IQ is pretty useless if it doesn't give one a decisive advantage *somewhere*, and if it's mostly useless, why would it be selected?
In terms of analysis, in the presence of a strong interaction effect, the main effects (the effect of each key variable by itself, in this case genes and the environment) lose their meaning. It's easy to illustrate: Imagine that intelligence is 100% inherited. Two children with identical genes grow up in two different locations: one is a wealthy suburb in a household with plenty of resources, the other in a war zone, rife with disease and natural disasters. Do you expect these two to do equally well on IQ tests, Educational Achievement, or any other measure of life performance? One is pretty clearly at risk of leaving fewer descendants than the other, but remember, their genes are identical. There's nothing here for natural selection to grab onto. The main effect of genes is meaningless.
What this implies is that inheritability scores are probably meaningless. Intelligence isn't 60% inherited and 40% learned. It's probably something more like 80% interaction between the two.
Another random thought that I suspect people in the relevant field have probably already looked into: (I'm a cryptographer, what the heck do I know about behavioral genetics?)
Suppose a lot of outcome differences between people are due to pathogens. For example, if IQ regularly gets depressed a few points if you get a particular viral infection at a particular critical point in your development. (I think there's some reason to believe this does happen, because the month in which you're born does seem to have a small correlation with life outcomes, and how that overlaps with flu season is a reasonable guess about the mechanism by which that works. Also, this is likely to be some of why breastfeeding correlates with good outcomes for kids, though a lot is probably tangled up with social class/education of the mother.)
If this is true, maybe the critical genes to look at for a lot of heritability are the ones that determine your immune response (which MHC receptors you have, what genes you start from when you shuffle stuff around randomly to get T-cell receptors and antibody binding regions, variants in innate immune response) and maybe any rare variants that confer some immunity to common circulating viruses.
There are like a gazillion viruses that circulate all the time, often giving us infections we never even notice, sometimes giving us annoying but not serious symptoms (colds, stomach bugs, cold sores, warts, etc.). And we know that some of these (rubella, polio, zika) can do nasty developmental things to fetuses, babies, or small children.
That would track with at least some of the Flynn effect as sanitation and vaccination made people in-general healthier.
This problem should be getting as much attention as Hubble Tension, if not more.
Can we induce this measurement phenomenon in rats, or worms, something where we have control of the genome? A quick Google suggests identical twins are rare in nature, but we can possibly produce them artificially for some animals.
Nice overview! A few points I wanted to expand on.
1. I think the post conflates gene-gene and gene-environment interactions; the latter (specifically interactions between genes and the "shared" environment) also get counted by twin models as narrow sense heritability. While I agree there is very little evidence for gene-gene interactions (particularly dominance, as you cite [and, interestingly, twin/adoption studies actually forecast a huge amount of dominance -- another discrepancy we do not understand]) there is quote substantial evidence for gene-environment interactions including on educational attainment (see Cheesman et al: https://www.nature.com/articles/s41539-022-00145-8 ; Mostafavi et al: https://elifesciences.org/articles/48376), IQ, and BMI. In fact, Peter Visscher led a paper that came to the conclusion that twin estimates for the heritability of BMI are very likely to be overestimated by gene-environment interactions (https://pubmed.ncbi.nlm.nih.gov/28692066/). A large amount of GxE plus some amount of equal environment violation seems like a very plausible and parsimonious answer to the heritability gap.
2. You mention epidemiologists being the biggest losers of stratification in polygenic scores, but I think it is important to note a related group: the people who take polygenic scores trained in one population (with a ton of stratification) and directly apply them to other populations to make claims about innate abilities (see: (https://theinfinitesimal.substack.com/p/how-population-stratification-led). This is especially true for Edu/IQ GWAS, where every behavior geneticist has been screaming "do not do that!" since the very first study came out. People like Kirkegaard, Piffer, Lasker, etc. (and their boosters on social media like Steve Sailer and Cremieux) dedicated their careers to taking crappy GWAS data from and turning it into memes that show Africans on the bottom and Europeans on the top. These people also happen to be the court geneticists, so to speak, for SSC/ACX. I don't mean to come off as antagonistic and I'm sure some people will see this comment and immediately discount me as being an ideologue/Lysenkoist/etc so it does my broader position no favors, but this stuff has done and continues to do an enormous amount of damage to the field (including the now complete unwillingness of public companies like 23andme to collaborate on studies of sensitive traits).
3. I'm going to gently push back against the hereditarian/anti-hereditarian framing (which I understand is probably here as shorthand and scene setting). I am personally interested in accurate estimates that are free of assumptions. I believe twin study estimates are of low quality because the assumptions are untestable, not because they are high. I also think the public fixation on twin studies has created some real and damaging anti-genetics and anti-psychiatry backlash and wrong-headed Blank Slate views. People hear about twin studies, look up the literature and find that peanut allergy (or wearing sunglasses, or reading romance fiction) is estimated to be highly heritable and have minimal shared environment (https://lymanstone.substack.com/p/why-twin-studies-are-garbage), start thinking that the whole field is built on nonsense, and end up at quack theories about how schizophrenia is actually a non-genetic fungal condition or whatever. I've been very clear that there are direct genetic effects on essentially every trait out there, including behavioral traits and IQ. If someone were to run a large-scale RDR analysis of IQ tomorrow and got a heritability of 0.9 and it replicated and all that, I would say "okay, it looks like the heritability is 0.9 and we need to rethink our evolutionary models". If anything, large heritability estimates would make my actual day job much easier and more lucrative because I could confidently start writing a lot of grants about all the genome sequencing we should be doing.
4. Lastly, it's not clear to me where the conclusion that well-validated twin studies converge on "similar results" is coming from. To take one example: the leading lights of behavior genetics (Deary, McGue, Visscher, etc) ran a study looking at the relationship between intelligence and lifespan (https://pubmed.ncbi.nlm.nih.gov/26213105/). This is a nice study for us because they put together three large, modern, twin cohorts with IQ measurements, but the heritability of IQ was just a nuisance parameter for them, so they had no reason to scrutinize the findings or file-drawer them. If we look at their MZ/DZ correlations in Table S6 we find that the heritability of IQ was 0.36 in the US sample; 0.98 in the Swedish sample; 0.24 in the Danish sample; and ... 0.52 on average. In other words, all over the place (but averaging out to the nice "half nature half nurture" result you see in books); the authors themselves used an AE model in Table 2 and reported a range of 0.20 to 0.98. This is far greater than the variability we see with GWAS or Sib-Reg, so what are we to make of that?
"People like Kirkegaard, Piffer, Lasker, etc. (and their boosters on social media like Steve Sailer and Cremieux) dedicated their careers to taking crappy GWAS data from and turning it into memes that show Africans on the bottom and Europeans on the top. "
You mean East Asians on top?
Would you care to quantify in some way what proportion of their careers is dedicated to what you say?
Richard Lynn placed NE Asians as 5-10 pts higher than Europeans. But he placed SE Asians down in the mid-80s, though. But didn't Cremieux argue on X that Europeans were higher? Maybe I'm misremembering.
There are several academic journals that cater to studies on the heritability of IQ and other social traits. Some, like Mankind Quarterly, are funded by conservative foundations. Lynn had an academic position, but it was funded by the same foundation that funds Mankind Quarterly. Lasker, although he claims an association with Texas Tech University, the Guardian article about him said he wasn't on their faculty list. He seems to be making his living working for a foundation, whose funding is unclear. Happy to update my priors if anyone has better info, but it seems like the IQ-heritablity researchers work in an incestuous little bubble.
The general HBD consensus seems to be East Asians 5 points above Europeans, except in tests of verbal intelligence.
As to your second paragraph, it doesn't speak to the specific claim that they dedicated their careers to "taking crappy GWAS data from and turning it into memes that show Africans on the bottom and Europeans on the top."
The obvious guess is that the more resolution you get into smaller groups, the more occasional outliers you'll find, and the messier your data will become. At one end of the resolution scale, you can use the US census category of Asians, at another, maybe you're looking at people from a particular jati in the Southern tip of India.
The Lynn and Vanhanen datasets (later updated by Lynn and Becker), which claimed to provide average "national IQs" for countries around the world, have been proven to be a fraudulent concoction consisting of data sources that are statistically and methodologically faulty. To quote Rebecca Sear, "The majority of data included originates from samples which are wholly unrepresentative of their national populations. Many are convenience samples with small sample sizes, often including only children and often including individuals chosen because they had particular characteristics (i.e., samples which were deliberately chosen to be unrepresentative)." Even though some academic journals have retracted some of Lynn's papers, many haven't. And Lynn's National IQ datasets have taken on a life of their own, being quoted as gospel by the likes of Lasker and Cremieux, and have directly or indirectly contaminated the studies done by serious researchers who weren't aware of the fraud perpetrated by Lynn and his cohorts. My *favorite* fraud tidbit from the Lynn and Vanhanen datasets is that the source of the estimated average IQ of Equatorial Guinea turned out to be taken from a group of children in a home for developmentally disabled children in Spain (Wicharts et al).
Lynn himself is an avowed "scientific racist" who believes that the darkies are sapping the vital fluids of western civilization (#snarkasm).
"I am deeply pessimistic about the future of the European peoples because mass immigration of third world peoples will lead to these becoming majorities in the United States and westernmost Europe during the present century. I think this will mean the destruction of European civilization in these countries.” — from an interview with Alex Kurtagic, 2011
"I think the only solution lies in the breakup of the United States. Blacks and Hispanics are concentrated in the Southwest, the Southeast and the East, but the Northwest and the far Northeast, Maine, Vermont and upstate New York have a large predominance of whites. I believe these predominantly white states should declare independence and secede from the Union. They would then enforce strict border controls and provide minimum welfare, which would be limited to citizens. If this were done, white civilisation would survive within this handful of states.” — from an interview Right NOW! magazine
So we've got pseudoscientists like Lasker and Cremieux (mentioned above) recycling this claptrap, trying to correlate it to GWAS studies, to make a narrative that fits their ideological bent. That qualifies in my mind as "taking crappy GWAS data from and turning it into memes that show Africans on the bottom and Europeans on the top."
You forgot about the "dedicated their careers"-part of the claim, which is why I suggested Gusev "quantify in some way what proportion of their careers is dedicated to what you say." I never questioned that they had used that methodology to study that topic to some degree.
Lynn is neither here nor there, but this blog has explained why his results support the idea that environment matters for IQ:
Except that Scott never addressed the issue that Lynn either exaggerated or fabricated a significant portion of his National IQ datasets. Scroll down through the comments for my responses to Swami, Steve Sailer, and a person named Daniel (who did a good Glaucon imitation).
I look forward to Rebecca Sear publishing her own table of national average cognitive performance to refute the half-dozen different tables that have been published, including the World Bank's. I'm sure Dr. Sear is just finishing up on her project and we'll see her True Numbers any day now.
Why bother? IQ is mostly pseudoscientific bullshit. IQ tests were originally designed to identify individuals with cognitive disabilities, particularly children who required specialized education. No doubt they do a good job at that, but categorizing high-functioning individuals doesn't get you anywhere. Terman's famous longitudinal IQ study showed that geniuses don't do much better in life than people with IQs +/-1σ range (cue the Zagorsky comeback). And none of Terman's 1500+ geniuses did anything remarkable with their lives. They produced not titans of business, and none of them made any significant contribution to science or culture.
Also, if we're searching for Truth, proving the null hypothesis (H₀) is as important as proving Hₐ. Sears is under no obligation to run a international IQ-testing program because she and a bunch of other researchers have poked holes in the HBD balloon.
I've written relatively little over the years about GWAS and race, because, until recently, it mostly seemed optimized to work best on whites, because that's where the big biobanks were to provide the huge sample sizes needed.
Hopefully, we'll get bigger sample sizes in the future for diverse populations.
With regards to “Lysenko/blank slatist/liberal creationist” accusations, it’s always been pretty clear that hereditarians are every bit as politically motivated as their opponents, if not more so. Just about every high profile hereditarian, past and present, is a right wing political activist.
Is this actually true? My impression is that the consensus view is that, for example, intelligence (measured by IQ) is substantially heritable. I don't think that's a notably right-wing view among people who study the relevant fields.
HBD is overwhelmingly right wing, but that's a different package of beliefs.
I really meant public-facing researchers/popularizers but I should have been clearer about that. Comparing people like Sasha, Turkheimer, Kevin Bird who are all pretty open left wingers on the one side and Kirkegaard, Piffer, etc. on the other who are all pretty open right wingers.
It's almost as if publicly dissenting from the extremist conventional wisdom that the racial gap in IQ _must_ be 100% due to Nurture instead of Nature rather than to endorse the moderate but loathed view that both Nature and Nurture probably play a role requires a lot of commitment.
Who is on top seems to depend a lot on the ethnicity of the poster in question. Same with who is on the bottom. It's black people if the poster is American and MENAPT if the poster is European.
All the HBD people I've read much from seem to agree that Eastern European Jews and East Asians have higher average IQs than Gentile whites, and they are mostly Gentile whites.
For example, what would you find by reading Steve Sailer or Charles Murray?
I wouldn't know about those specifically. I was introduced to the HBD concepts through Scott (jewish), Cremieux (Jewish), and Hsu (east asian). The "white" people I have read include Kirkegaard, who argues that the Chinese data is heavily inflated, and Piffer, who goes back and forth on whether europeans or east Asians are on top.
I came up with the term "human biodiversity" around 1998, only to immediately discover that anthropologist Jonathan Marks had published a book with title in 1995. Still, if I have any claim to the term, please allow me to point out that I've always argued that it is likely that the human biodiversity perspective is that both Nature and Nurture tend to matter, in contrast to the mainstream but extremist position that only Nurture could possibly matter.
In the latest Piffer analysis (https://www.qeios.com/read/HDJK5P.2) the ordering is in fact Africans on the bottom and Europeans on top with the EA4 polygenic score (Figure 1b). I've said plenty of times that this kind of analysis is highly confounded and produces nonsense, and since multiple behavior geneticists have told Piffer the same, I can only conclude that the intent is malicious.
There were more others plots where Ashkenazi or East Asians on top, why single this one, except you want to imply that authors are white supremacists?
If this produces nonsense, why it produces relative rankings of populations similar to phenotypes, rather any any of possible permutations equally likely? Did Piffer arbitrarily chose a subsample of genes to compute the plot, or was training was "contaminated" with non-Europeans so linear regression learnt non-genetical effects for phenotypes?
I don't think it is "singling out" to cite the Figure 1 result which uses the latest/largest dataset for training (EA4). Population stratification is indeed expected to reproduce phenotypic rankings, as explained here: https://theinfinitesimal.substack.com/p/how-population-stratification-led
Thank you for engaging in a constructive discussion of these topics. How would you rate the accuracy of the description of GWAS provided here? I feel like it misses the importance of linkage disequilibrium to the design. A centimorgan is after all about a million BPs in the human genome. And this omission makes it seem like GWAS only assesses the importance of SNPs, which makes the method seem a lot less powerful than it actually is (in theory, anyway).
Thanks, I think the description of GWAS here is good enough for a lay article that is trying to explain a lot of different concepts. The changes I would make are: (1) [as you note] even though GWAS methods often estimate heritability from ~1M common SNPs, which may seem like few relative to the size of the genome, it's been pretty well established that this ~1M actually "tags" (through correlations) the rest of the common variation quite well. So when we talk about GWAS heritability we're really talking about "all common variant heritability". (2) Likewise, *common* structural variants are also "tagged" by these ~1M common variants very well, so that even if only SNPs are being tested, the influence of common structural variation will also be picked up. IMO the key missing piece for GWAS really is just rare variation and not any of the other stuff. But again these are relatively minor quibbles with a section that's mostly correct in a complicated post.
Thanks for your answer! (and your blogging in general)
1. I'm having trouble understanding what you mean by GxE interactions explaining much missing heritability.
For the Scarr-Rowe interaction, this would make heritability look higher in high-SES families. But wouldn't this affect twin and molecular estimates equally unless there's some reason that subjects for one type of study are consistently from a different economic stratum than the other?
If we're thinking more about, let's say, a pair of fraternal twins where one of them is ugly and so parents don't invest resources in their education, wouldn't this show up equally in twin studies and GWAS? That is, if this is a very uncommon effect, we shouldn't expect it to affect large twin studies much. But if it's a common effect, then shouldn't we expect that every ugly person is less intelligent, and so GWAS will find that a gene for ugliness is associated with lower intelligence (both within and between families)? Can you give an example of a case why this would show up in twin studies, but not GWAS, RDR, etc? Also, why would we privilege this circuitous explanation (ugliness is genetic and provokes strong parental response) over the more direct explanation (intelligence is genetic)?
Also, the papers you cite show effects on the order of 2-8%pp; do you think the real effect is higher?
2. I grant that this is extremely likely to be bad, though the only person on that list whose opinion I rely on heavily is Cremieux and I haven't seen him do that (I may have missed it). The people who I have seen do that claim they are doing things to mitigate portability problems; I am skeptical but will let them defend themselves if they see this.
3. Obviously everyone is just seeking the truth, but I think it's fair to describe people by where they think the truth lands; I think we both agree on this and I won't nitpick this further.
I don't understand what you mean by twin study assumptions being untestable; I tried to link many studies testing them here.
I'm not sure what you mean by claims that twin studies find romance novels are 99% heritable or whatever, and I can't read the Lyman Stone post because it's subscriber only. I am slightly confused on how you think *over-reliance on* twin studies is behind people doubting the heritability of schizophrenia - people use the same arguments about missing heritability, twin studies having unreliable assumptions, etc as the lynchpin of the schizophrenia-isn't-genetic case (see eg Awais at https://www.psychiatrymargins.com/p/schizophrenia-and-genetics-end-of , the Torrey paper that inspired him at https://www.sciencedirect.com/science/article/abs/pii/S0165178123006418, and the Mad In America people at https://www.madinamerica.com/2024/01/psychiatric-yeti-schizophrenia-genetic/). Obviously people should say true things and not try to fake their beliefs in order to avoid some ill-defined concept of "harm" or "misinformation", but unless I'm misunderstanding you I'm unclear why you think doubting twin studies everywhere else will make the problem of people doubting them in psychiatry better rather than worse.
4. I'm not sure exactly what claim of mine you're responding to, but I agree that most twin studies fall within a range of 40% - 70% heritability of intelligence, especially when you adjust for the test used, the age of participants, and the standard error (yes, I admit that all those adjustments give extra degrees of freedom) and that this is also true for adoption and pedigree studies (I linked my sources there, which I claim I didn't cherry-pick). I don't think finding one anomalous substudy (of a larger study that falls right in the middle of the usual range) is inconsistent with that claim. In the supplement of LSADT, the authors say that estimate might have been anomalously low because they used a less g-loaded IQ test (yes, I admit they probably would have found something to worry about in any case). I think this one outlier study in what's generally a pretty consistent field especially among the largest sample size studies is pretty different from eg Sib-Regression, where the two published studies say 40% and 9%.
Here’s a post working out Sasha’s point on GxE in some empirical cases where we actually have twins, GWAS, and also we know the exact GxE mechanisms, so we can confirm that twin studies really were capturing GxE: https://lymanstone.substack.com/p/why-twin-studies-are-garbage
You seem to analogize peanut allergy to intelligence. In the 1950's peanut allergy did not exist, which suggests to me that it is not genetic in origin, at least not directly (although the neuroticism that has produced peanut allergy may be). Intelligence, in contrast, is a real trait.
Appreciate the followup. I'll try to respond on areas where I won't be tedious.
1. >>Gene-Environment interactions.
Take the peanut allergy example. Let's say in order to develop an allergy you need a mutation in the PNUT gene AND ALSO grow up in a household with early exposure to nuts (no Bamba!); that's a gene-environment interaction. For MZ twins, they will always share PNUT mutant (or wildtype) and 100% of their household exposure, so they'll be perfectly correlated on allergy; for DZ twins, they will share PNUT mutations half the time and 100% of their household exposure, so their correlation drops in half. So the twin study will tell you allergy is a 100% heritable trait. Now we test the PNUT variant in a GWAS, the first thing you do is throw away all the relatives (i.e. take one of each twin). Some people will be PNUT mutants and grow up in a household with no exposure and be allergy free, some will be PNUT mutants with exposure and will have allergy (and vice versa for the non-carriers). The resulting correlation between PNUT mutation and allergy will be low, so the heritability estimate will be <100%. TLDR: in the ACE twin model (and sib-reg), AxA and AxC interactions get counted as A. In the GWAS (and RDR) model, AxA and AxC get counted as E. In my opinion AxA could plausible be considered "heritability" in the sense that it only relies on genes, but AxC cannot.
>>Regarding the magnitude.
In the Mostafavi et al. paper, the heritability of Education doubled between the low and high SES groups (and this has been observed for IQ as well in other studies). 2-8% for the few environmental exposures people have tested is IMO substantial and could easily stack up across many environments (I think this is suggested by "multivariate" environmental methods -- https://theinfinitesimal.substack.com/i/147322261/multivariate-environments -- but the application of these methods has been limited so far).
3. >>Twin study assumptions.
What I mean by this is that we do not have a global test for quantifying AxA/AxC or for whether the EEA is violated. We also do not have one flavor of twin study that requires the EEA and another flavor that doesn't so we can triangulate across them (well, we sort of have that and it's RDR/Sib-Reg). What we have is the ability to test *specific* environmental measures for EEA/interaction violations and then various crude approximations (like look-alike studies, incorrect zygosity, twins raised apart) which come with their own new assumptions about the assumptions. But the whole saga of behavior genetics is that we do not know which environments actually influence these outcomes. We know MZ twins do not turn out identical, yet we cannot explain why. So how do we know which environments to test for EEA violations or AxC to begin with? TLDR, I'll just quote Alex's recent review of social science genomics: "Before the modern molecular genetic era, a range of models seemed capable of “explaining” the available kinship correlations, despite major differences in their underlying assumptions about, for example, dominance, assortative mating, and gene-environment correlation (Loehlin, 1978). Despite some spirited debates, efforts to distinguish between these models were largely unsuccessful" [https://www.nber.org/papers/w32404]
>>Harm of over-reliance on twin studies.
This is a longer point but I'll just say that a common argument from the anti-psychiatry people is to raise concerns about twin studies that are directionally correct: many of the iconic twins raised apart actually had frequent contact; some data was outright manufactured; one of the major twins raised apart studies (MISTRA) never published half of its cohort for unstated reasons; the other major study (SATSA) produced large estimates of selective placement and bizarre estimates of dominance; we know for a fact traits like peanut allergies are influenced by the shared environment; etc. Then bad actors take these concerns and say "look, the whole field is built on a foundation of nonsense, don't trust your therapist, flush your medication!". If the response is, "no, twin studies are unassailable, oh and also molecular methods are weird and not to be trusted either", I think this ultimately plays into the anti-psych argument. Of course, we should care most about getting at the truth, since manipulators are always going to be out there doing their thing no matter what we say.
4. Here I was responding to the specific final claim ("They are strong designs, their assumptions are well-validated, and they all converge on similar results."). I do no think it is accurate to say that heritability estimates ranging from 20% to 98% across two Scandinavian countries are "converging on similar results". Nor do I think it is accurate to take these two estimates, which are highly significantly different from each other, and simply average them to 52% and say they "fall within the range". They clearly disagree, and when you have a lot of numbers between 0% and 100% that disagree, the fact that they average out to 50% is not a comfort. I also don't think this study is an outlier in any meaningful sense: a group of esteemed behavior geneticists pulled three high quality cohorts for some other goal and stumbled on three very different estimates. That's weird! And it mirrors the kind of weirdness and variability we are seeing with Sib-Reg and the kind of variability one would expect if these approaches are highly sensitive to environmental interactions or other assumptions.
For the GxC example I'm a bit confused as to how that could explain some of the missing heritability.
I understand that if the hypothetical twin study has sampling bias and only contains houses with peanut exposure, then that would inflate h^2. But if the sampling bias went in the other direction, only houses *without* peanut exposure, then that would *deflate* h^2.
With a proper sample this would cancel out, no? Or would this "cancelling out" only be partially/unreliably captured by the twin study model?
Sorry if the example was confusing. The twin study does NOT need to contain only houses with peanut exposure. It can be a totally random/representative sample. All you need is for exposure (or lack of exposure) to be a "shared/household environment" such that all twins match on their exposure/non-exposure. The statistical issue is that MZs share genes*household 100% and DZs share genes*household 50% (because both share household 100%) and that looks exactly like heritability.
If no houses have peanut exposure, then r = 1 for MZs and DZs alike, and that's an h^2 of 0, which seems correct (or technically r is undefined, but whatever).
And if all houses have peanut exposure, then r = 1 for MZs and 0.5 for DZs, which is an h^2 of 1, which also seems correct.
But I guess the issue arises between those two scenario? Is the bias like a non-linear thing? And is it always upwards?
> for DZ twins, they will share PNUT mutations half the time and 100% of their household exposure, so their correlation drops in half. So the twin study will tell you allergy is a 100% heritable trait.
Isn't this false unless the PNUT mutation is rare? For DZ twins the correlation is > 0.5, because there's a chance they both get the PNUT mutation from different sides. So if PNUT is common, you'll see some shared environment effect. (And if it's rare, then I would consider the allergy to be "mostly caused" by the gene.) I do see your point that you will see missing heritability here, though.
> In my opinion AxA could plausible be considered "heritability" in the sense that it only relies on genes, but AxC cannot.
"red hair isn't heritable since you can dye it black"
You might be interested to know that Cremieux blocked me on Twitter when I pointed out that his favorite "reared-apart" IQ twin study (Minnesota, Science Magazine 1990) arrived at a 70% IQ h2 conclusion only after the authors suppressed their DZ-apart control group IQ correlations.
Turning to reared-together MZ-DZ "twin method" studies, the "equal environments assumption" (EEA) has been thoroughly "tested." Over 100 years of research has conclusively confirmed what most people already know--that MZ (identical) twins grow up experiencing much more similar environments and much higher levels of identity confusion and attachment vs DZ (fraternal) twins. Behavioral twin method results are, therefore, confounded and provide no evidence in support of genetic influences (heritability).
I notice you say that "over 100 years of research has confirmed" that EEA is false, but you don't link any studies, whereas I link to many in the post demonstrating that it's true, which you haven't responded to.
"EEA test" studies typically involve twin researchers supposedly validating the research method on which their careers are based. It's not surprising that they conclude in favor of the twin method, and the task of replication crisis analyses is to root out p-hacked research where conclusions often match authors' confirmation biases.
The post seems to hinge on a table that is overwhelmingly things like parents having harder time telling identical twins apart. This doesn't respond to the EEA, which is that parents treat fraternal twins significantly differently from identical twins in ways that have significant effects on behavioral traits. It also doesn't respond to the studies linked in my post, which show that in cases where parents confuse one kind of twin for the other, their differences accord with genetics rather than parent-determined environment, or with the study showing that if you control for environmental differences it doesn't matter.
We can agree to disagree on whether twin studies are confounded. On another note, the first three paragraphs of your post align with the highly problematic history of genetic research Robert Plomin presented in “Blueprint.” Below, I tweak these three paragraphs a bit to better reflect the historical reality:
In the late 19th century and the first four decades of the 20th century, eugenic ideas dominated psychiatry and academic psychology. Based mainly on preconceived notions, family pedigree studies, and supposedly “degenerate families” such as the “Kallikaks” and the “Jukes,” it was axiomatic in psychiatry that conditions such as schizophrenia were largely the result of “hereditary taint,” and IQ hereditarianism was mainstream in academic psychology. In the 1920s and even earlier (Galton, Thorndike), eugenicists and “racial hygienists” developed twin studies of IQ, criminality, schizophrenia, and many other behavioral areas. Eugenicists promoted the passage of laws throughout the Western world enabling the forced sterilization of the “insane” and the “unfit.” In Germany, Ernst Rüdin and his psychiatric genetics colleagues carried out twin studies and promoted forced sterilization in National Socialist Germany. In the 1930s, their Munich Institute became a think tank of Nazi eugenic policies, which led to the killing of mental patients and others under a program euphemistically called “euthanasia.”
In the aftermath of World War II and revelations about Nazi atrocities in the name of eugenics and racial hygiene, eugenic policies and genetic explanations of behavioral differences lost some appeal. However, even though nurture influences received greater emphasis, the idea that heredity plays a role in causing differences in IQ and behavior never went out of favor. Fewer behavioral twin studies were conducted in the 1950s and 60s, although eugenic sterilization laws remained on the books in many U.S. states and in Europe. The American Journal of Psychiatry ran Franz Kallmann’s annual positive review of eugenics from 1944 until he died in 1965, and the word “eugenics” fell completely out of favor only after the social upheavals of the 1960s.
The socially and politically powerful have an interest in promoting behavioral genetic research, which led to an increase in twin studies and “heritability” research beginning in the 1970s. However, behavioral twin and adoption studies have always been based on false assumptions. They have also been subject to researcher data manipulation to confirm genetic confirmation biases, a widespread practice now known as “p-hacking” in science’s current and long-overdue “replication crisis.” The famous Danish schizophrenia adoption studies of the 1960s and 70s and the Minnesota Study of Twins Reared Apart 1990 Science Magazine IQ study exemplify instances where researchers found no evidence of genetic influences, but then p-hacked their data to transform negative genetic results into positive ones. Scandalously, both continue to be cited by psychology and psychiatry as “landmark” studies.
Beginning in the 1960s, researchers used DNA-based (molecular genetic) methods in attempts to identify genes that cause psychiatric conditions such as schizophrenia and bipolar disorder. By the early 1980s, the search was broadened to include other psychiatric conditions, IQ, personality, criminality, and other types of behavior. Despite decades of sensational claims in the media, and by researchers and the institutions they worked for, such claims turned out to be false alarms. This statement is true for earlier behavioral linkage and candidate gene studies, and most likely, current claims based on rare variants, genome-wide association studies (GWAS), and polygenic scores. The findings from these more recent methods are likely non-causal or spurious “hits.” “Association” (correlation) does not equal “cause,” and the half-century-long failure to identify causal genes for behavior leads to a conclusion that the critics of behavioral and psychiatric genetics were right all along about earlier family, twin, and adoption studies. The “missing heritability problem” is really a “family, twin, and adoption study misinterpretation problem.” Time to rewrite the textbooks.
Saying the critics were right all along does not necessarily mean that humans begin life as psychological “blank slates,” or that there are no inborn/genetic within-group individual differences in human intelligence and other behavioral areas, or that causal genes will never be discovered. But it does mean that perinatal, family, social, cultural, religious, educational, geographical, and political environments (including institutionalized oppression/privilege, economic inequality, and neocolonialism) play a decisive role in shaping human behavior, and that focusing on “individual differences” and problematic heritability estimates implies limited changeability and distracts our attention from the need to improve or radically change these environments. The same point holds for patterns of psychological distress and dysfunction that psychiatry and psychiatric genetics call “mental illnesses” or “diseases.” For all types of human behavioral differences, unlikely yet possible future discoveries of causal genes would, for the most part, be irrelevant distractions.
It seems like the issue is whether government programs that try to equalize environments can do all that much to equalize IQ among people who aren't identical in genes. For the government to equalize environments to the extent that environments are typical equalized by parents among their children would seem pretty ambitious, or downright utopian.
I can well believe that identical twins experience more equal environments than fraternal twins -- due to their near identical DNA -- but I can't see how the government can do much about that when trying to make the environments of children who don't have identical DNA more equal.
Thanks for your comment. I want to ask about gene-gene interactions. I understand that dominance effects have essentially been ruled out, but other than that, what evidence do we have against GxG? It seems implausible to me that GxG doesn't exist, because additivity is not invariant to our measurement scale (eg if IQ has additive genetics, then IQ^3 does not).
There are too many gene pairs to investigate GxG by directly adding interaction terms. Other than doing that, I'd think the main way to study GxG would be to look for a gap between narrow and broad heritability, right? But the gap seems to be there, doesn't it? Am I misunderstanding?
I think the primary argument against GxG is that lots of people have looked at it using a large number of methods and traits and nothing has ever replicated. The most recent study was from 23andme (https://www.biorxiv.org/content/10.1101/2024.08.15.608197v2) looking at height in millions of people and testing for interactions between variants we know influence height marginally and finding nothing. It's not the most satisfying answer but it's just rare to see not even a hint of signal like this.
One interesting counterpoint, however, is the "limiting pathways" model proposed by Zuk et al (https://www.pnas.org/doi/10.1073/pnas.1119675109) which hypothesizes that traits are a collection of pathways where any one of them breaking causes the entire trait to "break". Under this model, twins/siblings would "break" their traits in the same way, but unrelated individuals would all be breaking differently, which would create what they call "phantom heritability" that looks a lot like GxG in the close relatives. I think this is a plausible model but extremely difficult to actually quantify.
Thanks. But again, GxG just has to exist, no? Really, if height has no GxG, then square-of-height will have GxG and vice versa. I guess I should try to quantify that effect...
The 23andme paper says that typical detected alleles have 1.2mm effect on height, and they can rule out interacting pairs where the effect size is larger than 2.4mm. That does not seem very conclusive to me? One would expect each pairwise interactive effect to contribute a lot less than the contribution of each individual allele: if there are n genes, there are like n^2/2 pairs of genes, so we expect the contribution of each pair to be much smaller. Shouldn't we expect pairwise interactions to have effect sizes on the order of 0.1mm or less, even in the presence of GxG?
Yeah, so on the first point, if you are just thinking of GxG as the trait being the product of mutations instead of the sum of mutations, then taking the log of the trait should make it additive again (this is something Fisher pointed out). We generally see pretty similar heritability estimates for (trait) versus log(trait), which is maybe evidence against simple epistasis like that. On the other hand, if GxG includes more complex interactions (e.g. variant effects completely flip in the context of other variants) then scale transformation will not be enough.
To the second point. Yes, it is still possible that there is a very large number of very tiny epistatic effects that we cannot pick up. I think this is biologically unlikely but we cannot rule it out. One other way to look at this is to ask if trait correlations across relatives seem to follow an epistatic model. This doesn't require estimating any individual interactions, you just ask if people who are more closely genetically related are *much* more phenotypically related. This has also been done and doesn't seem to line up with epistasis (see Zaitlen et al. 2013 , https://journals.plos.org/plosgenetics/article?id=10.1371/journal.pgen.1003520 "We examine phenotypic correlations across a range of relationships, from siblings to first cousins, and find that the excess phenotypic correlation in these related individuals is predominantly due to shared environment as opposed to dominance or epistasis."). However, this is a bit circular to the broader heritability debate. If complex shared environments are at play, then they could be masking the effect of epistasis.
TLDR: we are mostly relying on the absence of evidence (but after many many attempts) rather than evidence of absence for GxG.
I disagree with the intuition that a very large number of very small effects is biologically unlikely. To me there's no biology here at all: if we assume a biologically additive effect on a phenotype P consisting of many small contributions by different genes, and then we try to measure the heritability of P^2, this will automatically introduce an effect of *all possible* pairwise interaction terms, each with tiny effect size. This is just a mathematical consequence, not a biological one. To say things are additive is to say our measures of phenotype are on the perfect "scale", in a sense. That seems unlikely.
To be clear, I'm not envisioning GxG effects as one gene affecting the expression of another, or anything like that. I'm just saying, if gene A gives you some kind of +1 for intelligence and gene B separately gives you another +1 (in some "true brain units" or whatever), there's absolutely no reason at all to assume that, as measured by number of correct answers on a test, the effect will still be additive.
I will check out Zaitlen et al. That sounds more along the lines of measuring GxG by comparing narrow to broad heritability, as I mentioned earlier.
“One other way to look at this is to ask if trait correlations across relatives seem to follow an epistatic model. This doesn't require estimating any individual interactions, you just ask if people who are more closely genetically related are much more phenotypically related.”
Isn’t this precisely the phenomenon that you’ve been writing about with twin studies?
I think maybe there’s also an issue arising from the different definitions of epistasis. With the one we’re using here, it’s a property of populations, not the genotype-phenotype map alone, so as I understand it there’s no contradiction in saying that GxG is small at the whole population level but large in twin samples. In fact, I think you expect this? Because in the whole population you have a U-shaped AFS with a lot of rare minor alleles, while in each twin pair minor alleles are all at 25% (and there are also some 50-50 loci).
Yes, agreed, that is what I meant by "this is a bit circular to the broader heritability debate". If there was no impact of the shared environment, we could easily fit an epistatic model to relative classes. But we know there are shared environment effects that can be complicated. And we don't even know if environments are equal between MZ and DZ twins, which would be one of the biggest sources of difference for identifying epistasis. So all Zaitlen et al. are able to conclude is that a purely epistasis/dominance model doesn't fit well, but some combination of shared environment AND epistasis could definitely be at play (though see the other more direct negative findings on epistasis I cited in the first comment).
"People like Kirkegaard, Piffer, Lasker, etc. (and their boosters on social media like Steve Sailer and Cremieux) dedicated their careers to taking crappy GWAS data from and turning it into memes that show Africans on the bottom and Europeans on the top."
The first of which is one of your readers writing to you rather than you using the name. In the other two you refer to "Racimo-Piffer". All three posts are from the 2010s.
It seems like there’s room for people more educated than I to do some simple “ceiling” calculations on the types of gene interaction models that could plausibly be at play here given the number of generations/iterations that evolution had to play with.
As an extreme example, assume you have 50 independent variables that can interact in any way (linearly or otherwise, eg, GBT/random forest style). Make some reasonable assumptions about their variance. Assume evolution is operating like some common gradient descent optimizer like NES. How many iterations would it take for evolution to effectively select for this or that gene. How many generations has evolution actually had?
This is a loose, hand-wavy description, but I think I’m making some sense here.
The number of causal variants for a typical common trait is on the order of thousands but can vary as much as 100k down to 100 from trait to trait. Evolution has also had half a million years to tune these variants in hominids. So it's a big space.
The anecdotal evidence of “troubled child raised in a good Christian home” seems a strong contender for selection bias. You wouldn’t have encountered the hypothetical well-behaved children from awful parents raised in supportive households. If that population outnumbers those you worked with, the point becomes moot.
Extremely interesting essay overall; thanks for sharing!
Agree with your general point! I would add that in my career as a primary care pediatrician I do see adopted kids who are raised in “good” families and who are not committing heinous crimes. But I do notice that their behavior pattern often seems more similar to that of their biological parents than their adoptive parents…
The point would be moot if the point was "the majority of children with troubled bio-parents become troubled themselves".
The actual point is "adopted children with troubled bio-parents do worse than their siblings even if raised in good families whose biological children were goody-goodies who never got so much as a school detention. "
It depends on what else Scott saw. If troubled children from good households tend to be disproportionately adopted, then that can't be attributed to the type of selection bias you're describing.
Note that Scott saw a population selected for its extraordinarily bad behavior--it's not so clear how much that very weird population tells us about variation within the usual bounds of "difficult kid."
Maybe there could be a self-reinforcing effect to sibling similarities? There's the stereotype of the twins that are always together and always do the same thing. Maybe small differences, especially in early childhood, separate siblings and make them more estranged so that in the future they affect each other less and less. So identical twins have some initial stronger bonds by being more alike, and then that develops into an environmentally stronger bond (i.e., they do most activities together and want to be equal), while fraternal twins are a little more distant from each other because of initial differences and then don't have that bond that makes them also have a more similar environment.
Theory: IQ is very heritable but also extremely subtly disabled (potentially even environmentally, think "more males born after wars"), because IQ is terrible for reproduction (now as ever) and you need some people to keep passing it on.
Crackpot theory: everything is physical appearance.
The causal mechanism connecting genes to outcomes could be that genes determine how you look, how you look determines how people treat you, and how people treat you determines how you behave.
If someone looks smart, they become smart because that's what everyone expects. If they look aggressive, they become aggressive. This is why people online can talk about stuff like "nerd physiognomy." It's real, but the causation is backwards from what you'd think.
I don't think this passes the smell test. As we know, physical appearance is not well-correlated with intelligence, even if some do claim that all "positive" genetic traits are correlated. Also, within an ethnic group, it's tough to really determine whether someone "looks smart". I'll grant you that people's development is influenced by how others see them, but it's not the physical appearance that would actually be biologically making the person smart or dumb. Of course, the popular nerd vs jock stereotype does have some truth to it, and could suggest two different "evolutionary strategies" for human males.
I’ll just throw in the small wink implied by the oversized banner I would see when my undergrad roaming at my rodent mascotted university took me to Eliot Hall where they were doing research in this area.
Scott’s personal experience during his psychiatry residency parallels my own experience as a primary care pediatrician. From my perspective, there is overwhelming clinical evidence that genetics must play a huge role in shaping behavior.
It’s actually somewhat baffling to me that pediatricians are generally opposed to “hereditarian” views, but I think there are many non-clinical reasons why that may be the case…
I dont get the darwinian argument cause, I'm always thinking like evolution really cared about calorie efficiency which we dgaf about. so why wouldn't there be a "more brain" knob in the genes? compare to height, or how we can increase muscularity with tren
Depends on what it was traded off against. You can get bigger brains with bigger heads, but that's not going to be selected for (even if it makes the kids much smarter) until you can do C-sections safely.
Genetic heritability of intelligence is precisely the eugenics-loaded, complex, wicked problem that is eminently vulnerable to all sorts of biases as well as prestige/social related competition altering scientific opinion.
Putting that aside: since we know that there are environmental factors in at least some genetic expression - does this not kill, dead, any idea of separating "genetic heritability of intelligence" vs. "EA generated intelligence"?
Nutrition seems like the obvious one. It seems inevitable that this influences the pattern of gene expression, since you're surely expressing a different pattern of genes when you grow than when you don't grow.
Is there some reason to think that heritability of intelligence is inherently harder to understand than heritability of height?
Nutrition is likely some effect on pre-modern diet humans and height, but I doubt nutrition plays any role in height in say, the Scandinavian countries today.
Intelligence, however, is not directly related to nutrition unless there is a serious lack of nutrition. While in theory, being smarter gets you more food - the reality is that modern humans don't generally lack for food nor is the availability of food in modern society, a function of intelligence. Welfare programs, for example, directly subsidize food to the economic losers in society in direct contrast to the theory of intelligence mitigating food scarcity.
But even for poor 3rd worlders that move into 1st world nations - there is no way to separate 1st world culture, education, etc from better nutrition for the immigrants' children. Is there even a visible improvement in intelligence for the already mature?
But nutrition is precisely not what I meant by environmental factors.
We know that genes turn on even in adults for various things - epigenetic expression is the term.
Intelligence seems like an emergent behavior, and therefore one which must arise from a variety of bases.
Are these bases relatively set in stone? Seems highly unlikely.
Are they relatively epigenetic? Seems possible but hard to prove given that we don't even know what actually creates intelligence. Note that these genetic studies don't know either - all they are doing is correlating genes vs. highly imperfect "intelligence" measures.
Among other things: my view is that there are different types of intelligence - horses for courses. I don't believe egghead PhDs are great at everything - in fact, they tend to be good only in very specific fields and are merely "somewhat" intelligent in others.
On the other hand, we don't even try to measure the type of intelligence needed to survive in the wild without society - which is realistically the only actual measure of evolutionary intelligence that should matter.
So what we really have is the usual ACX lotus eating: how can a bunch of professional/managerial types with 1st world society "high intelligence" figure out ways in which their type of "high intelligence" is developed. Most likely so that their kids can carry on the parents' legacy of PMC-dom.
IQ is interesting because it predicts success in two critical parts of the modern world: school and work. But yeah, I'm not at all confident that IQ correlates all that well with success in a hunter-gatherer tribe, say. Some kind of intelligence seems critical for that life, too, but who knows if it corresponds well to the stuff we measure on IQ tests?
Still, mostly when we talk about individual IQ or average IQ of a group, we're talking about how well we expect someone or some group of people to function in the modern US[1]. Like, if some guy has a 70 IQ, I expect he's going to have a really hard time at school, probably not be able to learn to read, and is going to have a very limited range of jobs he can do in a modern society. Maybe he'd be great at tracking antelope on a plain--I have no idea. But he's pretty unlikely to have a lot of success in modern 21st century countries.
I guess I come back to the same question: is your claim that heritability is in general very hard to untangle? Or that it is particularly so for intelligence?
[1] But honestly, people get wrapped around the axle thinking about group average IQs and reasoning about them like they're an individual IQ, even outside the political/social/tribal issues those discussions often raise.
1) Intelligence as measured by IQ is not "general" intelligence, it is PMC type intelligence
2) PMC intelligence is not in any way clearly correlated to evolutionary intelligence. I have bird seed and nectar feeders on my balcony; it is quite clear that the capability to remember seed and nectar locations is important for house finches and hummingbirds, respectively. But it is also clear that a hummingbird's version of intelligence is very different; observationally, it is fairly clear that they don't just remember locations but both develop habits i.e. they seem to always use the same feeders until changed and also have "modes" where they will test the other feeders (but still use the "preferred" one unless it runs out or goes bad).
You did not respond to my comments on the emergent nature of PMC intelligence or the known epigenetic expression of at least some genes. I think of these as the Scylla and Charybdis of any attempt to link PMC intelligence to genes.
Lastly, IQ measuring success in society. I don't know about you, but in the Bay Area, and I suspect in most PMC hives i.e. large cities, the most consistently successful types are the tradesmen. Do you need 140 IQ to be a good plumber? I think not. What about an auto mechanic? Or a repairer of $5000 appliances? I am sure not. Trades generally don't require reading or written testing capability so long as there is someone to train or demonstrate. And I would argue that the ability to make and repair plumbing, refrigerators, cars etc is far more critical to the function of modern society than the philosophy major now selling AI agent software to enterprises or even the vast majority of lawyers doing anything.
Using these polygenetic scores to evaluate a genome is like counting word frequency to evaluate the quality of a book. Sure, it probably correlates. But when you restrict yourself to evaluating a simple linear model on the input, you might not be able to capture the structure there. That doesn’t mean there is no structure present.
Maybe I missed it in the post, if so, I apologize.
But do we know the prevalence of de novo mutations? Even if they are rare, if they substantially impact fitness, it might create a significant difference. Is there a reason for dismissing this explanation out of hand?
I don't know what an actual smart geneticist would say. My hand-wavey response is that we're trying to use genetics to explain why children resemble their parents. De novo mutations don't explain that (if it's a generation up, it's not de novo, it's just another ultra-rare variant) , so since children do resemble their parents a lot de novo mutations can't be responsible for too much of the variance in traits among people (aside from the stuff we already know about where children *don't* necessarily resemble their parents, like certain out-of-the-blue rare developmental disorders)
Good point. I suppose your inclusion of the data on family pedigree/adoption studies answered this. I was thinking mostly about twins, as in, if some rare, but important de novo mutations skew the data they wouldn't show up in GWAS but they might presumably be shared in identical twins (assuming they occur during meiosis or before the embryo splits).
I think you're wrong here. Heritability as a concept is defined as a proportion of total phenotypic variance. It only exists with respect to some predefined reference population, and it matters a lot what particular range of genetic variance, environmental variance, and phenotypic variance you are considering. Even taking twin/adoption/etc studies at face value, most of the strong 'hereditarian' claims depend on ignoring or deliberately obscuring this issue.
Anyway if you took a baby from a family geniuses and you gave it to a family of idiots, and then it turned out that family of idiots belonged to the Concussionistic faith where they have ecstatic experiences by bonking each other on the head with hammers all day, then that baby would probably turn out to be an idiot. The result you get from adoption studies will depend very strongly on whether or not the adoption agencies are discriminating against concussionists.
Suppose I say "lighting a fire makes things hotter".
There are all sorts of possible objections, like "That's only true if you don't also fill the room with ice" or "This depends on environment - if there was a religion that forced people who lit fires to also take off their jackets, they might get colder". Whenever we state a fact, we're implicitly saying "And we're holding other things constant and not positing crazy religions devoted to making illogical things happen". You're allowed to say "shooting someone kills them" without adding "unless you magically hit a tumor and eliminate it, in which case shooting someone actually prevents them from dying!"
So yes, it's true that heritability only makes sense within something like the normal parameters that exist in basically every human society. But this is a stable enough concept that you can still talk about it just fine. For example, I think it's fair to say that having lots of genes for breast cancer causes you to be more likely to get breast cancer, and you should try to avoid having them, even though this wouldn't be true in a world dominated by a religion where people injected uranium into the breasts of anyone who *didn't* have genes for breast cancer.
Sorry if this comes off as snarky, I just don't know how else to respond to this.
I think this is where you are fundamentally wrong; or at least, where your beliefs are fundamentally unjustified in ways you are simply sweeping under the rug.
In the fire example and the bullet example there are clear, unambiguous, directly observable mechanistic processes (I'm sure you could add some more adjectives there but you get the idea) that connect the putative explanation to the predicted outcome. No such mechanisms are known for genetic explanations of complex, massively polygenic behavioral traits. Full stop, these mechanistic explanations have not been identified, and attempts to find them have largely been failures. The closest thing we have are rare genetic variants that disrupt core neurophysiological processes; these don't explain 'being a smart person' vs 'being a dumb person', they explain 'your brain basically works' vs 'your brain barely works', and they have little to do with the common variation that hereditarians want to explain.
Also, when we talking about fire melting ice, we are talking about a clearly and absolutely defined endpoint. We are not talking about percentages of total water volume that can be attributed to (poorly measured) similarity in thermal conditions. Heritability research is closer to the latter.
So yes, you get to blame the melting on the fire. When behavioral genetics has achieved even a fraction of the empirical and theoretical depth that thermal physics has, then maybe it will make sense to blame intelligence on the genes.
To put another way:
"So yes, it's true that heritability only makes sense within something like the normal parameters that exist in basically every human society. But this is a stable enough concept that you can still talk about it just fine."
No! This is a very unstable and poorly defined concept! And even if it were well defined, it would still matter tremendously how representative your study population is of whatever you mean by 'normal human society'!
In any case, you haven't been any snarkier than I have, and as someone who was once a behavioral geneticist (not in humans) but hasn't actively followed the latest developments, I really do appreciate the good-faith engagement with opposing views and evidence.
ISTM that what we have now is some observations that allow us to make good predictions.
We see that smarter parents tend to have smarter children, smarter kids tend to have smarter biological siblings, and smart kids with an identical twin tend to have their twin be smarter than if they have a fraternal twin.
We know enough that we can make a lot of useful predictions from this. For example, while the children of very smart people are usually not as brilliant as their parents due to regression to the mean, we know that the kids of a bunch of very smart parents will on average be smarter than the kids of a bunch of normal parents. This is messy--Einstein can have a dumb kid and the village idiot can have a brilliant one--but that's not the way to bet.
It seems like that's useful information even if we don't know exactly know all the mechanisms by which that heritability operates. It would be really nice if we understood those mechanisms, because we'd love to reach in and twist some of those knobs--figure out the micronutrient or preschool activity that would boost everyone's IQ by a couple points, give genetic counselors more interesting stuff to talk with their clients about, maybe even eventually find a way to select embryos for intelligence in a reliable way.
A good rule of thumb for evaluating any snarky dismissal of the heritability of mental traits is to ask whether it also snarkily dismisses the heritability of height.
Yours does. Height is a highly polygenic trait and, outside of a handful of rare variants with very large impact like achondroplasia, we have virtually no idea of the mechanics of individual genes that influence it. We also know that it has a significant environmental component, and we could at least imagine that some variants that affect it only do so via GXE interactions (e.g. a SNP that makes one’s olfactory receptors view fish as particularly unpleasant will probably have a negative effect on height in, say, premodern Polynesia but a much smaller effect in the modern West).
And yet, height, obviously is strongly heritable (source: c’mon). So snark that leads us to haughtily dismiss it should perhaps be reevaluated.
IIRC height is also very much subject to the 'missing heritability' problem. Twin/pedigree/adoption studies generally find something gargantuan like 0.8 or higher, and molecular studies AFAIK have topped out at something like 0.5. So whatever point we are arguing about—I'm not actually sure anymore—I don't think pointing to height gets you very far.
To clarify, my position is neither "children are not similar to parents" nor "children are similar to parents, but for reasons that have nothing to do with genetic similarity". I agree those are non-defensible positions; though maybe, on the basis of the work reviewed in Scott's post, they are becoming rapidly more defensible!
Regarding haughtiness and snark, I will just say that all internet arguments could benefit from a serious consideration by all parties of Matthew 7:3.
Adoption agencies generally do a decent job of filtering out the very worst potential parents. Consider as an example, Steve Jobs' adoptive parents: they were quite blue collar, but were excellent people. It was not Jobs custom to be uncritical of other people, but he was highly appreciative of his adoptive parents.
Yes, and this is a relevant fact about adoption in particular, and not about the whole of the human experience or genetics or any given complex phenotype!
Adoption studies are one tiny but not insignificant element in understanding the human experience, so knowing a little about trends in adoption is useful for evaluating adoption studies.
I never understood the finding that parenting and family environment have basically no effect on traits. Maybe it's believable for IQ... but on character and behaviour? Surely they must, right? It feels a stretch to argue that something that's so cultural, like for instance good manners, is hereditary as opposed to learned. Perhaps a level of amiability and disinclination to rebel are hereditary, which makes some children easier to raise to be polite, but you'd still need a parent teaching a child good manners for that child to adopt them. So how does that not show up in the studies?
I don't think anyone has ever done a twin study on good manners!
If someone did, I would expect some shared environmental effect, but also some genetic effect.
(before we even get to either of those, there would be a massive dominating culture effect - no genes will make a barbarian who never encounters civilization perfectly imitate the manners of Victorian Britain. But these genetic studies are generally thought of as partitioning variance within a single culture).
Within a single culture, manners partly depend on whether your parents teach you good manners. But they also partly depend on whether you listen when you're taught, whether you're conscientious, whether you care a lot about offending other people, etc. If you do, you'll probably figure out some way to pick up manners from someone other than your parents. If you don't, you'll probably forget the manners your parents taught you immediately after leaving home.
(one common pattern in a lot of traits, including IQ, is that shared environment matters most during childhood, and then genetics gradually takes over as you get older and more distant from your parents' teachings)
I'm using manners just as an example - what about work ethic, kindness, sense of duty, clarity of thought, appreciation of art, etc? Some of those definitely have a genetic component (I can see that my kids have different inclinations in all of these by nature), but parents do shape their kids to some extent, right? And these must be predictors of happiness and material success?
Back in the 1990s, I reviewed Judith Rich Harris's influential "The Nurture Assumption" about the limits of parenting for "National Review" mostly positively, but I noted some limitations to her focus:
"To show that peers outweigh parents, she repeatedly cites Darwinian linguist Pinker’s work on how young immigrant kids automatically develop the accents of their playmates, not their parents. True, but there’s more to life than language. Not until p. 191 does she admit — in a footnote — that immigrant parents do pass down home-based aspects of their culture like cuisine, since kids don’t learn to cook from their friends. (How about attitudes toward housekeeping, charity, courtesy, wife-beating, and child-rearing itself?) Not until p. 330 does she recall something else where peers don’t much matter: religion! Worse, she never notices what Thomas Sowell has voluminously documented in his accounts of ethnic economic specialization. It’s parents and relatives who pass on both specific occupations (e.g., Italians and marble-cutting or Cambodians and donut-making) and general attitudes toward hard work, thrift, and entrepreneurship.
“Nor can peers account for social change among young children, such as the current switch from football to soccer, since preteen peer groups are intensely conservative. (Some playground games have been passed down since Roman times). Even more so, the trend toward having little girls play soccer and other cootie-infested boys sports did not, rest assured, originate among peer groups of little girls. That was primarily their dads’ idea, especially sports-crazed dads without sons.”
I have twins. The one with ADHD was significantly smaller than the other at birth. I suspect epigenetic expression based on the finite amount of nutrients available in the womb, some of which is modulated by each fetus. In other words, I think twin studies might be a good gauge of heritability, but there's plenty more work to be done.
I agree with all of this except the epigenetic angle - why can't it just be that a nutrient deficiency caused their brain to develop less well? I don't think deficiency causes ADHD any more than it causes other things like low height or low IQ, and I can't think of any reason for careful epigenetic regulation of any of these.
You mention rare variants and structural variants as potentially not being well captured by SNPs panels behind most of these tests (and iirc some types of structural variation can be problematic even for whole genome sequencing). But a lot of the later arguments (around evolutionary pressure, etc.) seem to focus more or less entirely on the 'rare' part there. Is structural variation (e.g. copy number variations etc.) particularly rare in the human genome? Is there some reason to think that structural variation might not be a significant source of heritable variability? Some of that will be correlated with nearby SNPs but now we're talking a correlation to correlation and depending on assumption the analysis power of that can fall off quickly. Has anyone ascertained how much variation might be due to these sorts of things?
Have there been any studies of the connectomes of the brains of twins? I would guess the answer is no, because the technology does not exist yet. AFAIK, we are up to the connectome of a fruit fly. What I am wondering is the extent to which differences in intelligence are a result of different connectomes and how much the connectome is determined by genetics. Perhaps there is a large degree of randomness and the way the connectome develops during pregnancy cannot fully be predicted by genetics. Perhaps, as the brain develops, axons grow and connect based on factors independent of specific alleles, such as local concentrations of certain proteins or nutrients at the growing tip of an axon, environmental variables, cosmic rays, etc. Then there could simply be a random (i.e., untractable) variability in intelligence, which would have a high evolutionary advantage. Just as genetic evolution depends on the availability of variation in alleles that results from mostly random effects (e.g., point mutations, gene doubling events, etc.), so intelligence and, more broadly, behavior plasticity, in a population would then depend on such random variation in the development of the nervous system.
The idea that genes are having linearly additive effects always felt like a simplification for the sake of the model. Changing the way a complex system is made in many ways and expecting them not to meaningfully interact feels like video game logic. I hear that the interactions have not shown up when looked for, but might that be related to the point about computational tractability?
"Since the family unit is a perfect natural experiment that isolates the variable of interest (genes) while holding everything else (culture and parenting) constant"
As a parent of two children I find this assertion to be deeply flawed.
Most biobank participants are middle-aged, and many are older (UK Biobank median age is 58 according to a quick search). Could the somatic mutations that accumulate during aging lead to errors in the sequenced genomes, which then add noise to the genomic studies and decrease their ability to detect relevant rare variants?
Hi Scott, I think it all lies in conditions in the mother/womb.
Your data in section 4 on creatinine levels and such.
" Identical twins don’t have more similar kidney function environments than fraternal twins." - But perhaps they do.
For two identical twins the ideal concentration of zinc in the womb for optmial kidney development may be 0.000001 percent. But two fraternal twins may have slightly different requirements.
The womb will supply certain conditions that may be optimal for one twin but not the other. With identical twins their requirements are matched as well as the supply.
In fact I would expect the fraternal twins to show similar heritability to regular siblings, perhaps a little more similar since they share womb conditions, whereas with regular siblings the changes to the mother's diet/age and other factors will come into play.
- Apologies for all the extra posts, I tried to delete them. Something strange happened when I tried to fix a typo.
Here's a review that covers some of the same territory as Alexander. https://www.sciencedirect.com/science/article/pii/S1090513824000722 Author Zietsch argues "Recently, converging evidence has emerged that much of the remaining gap is due to imperfect correlation between the SNPs used in GWAS and the ungenotyped causal variants. ... To the extent that GWAS SNPs and causal variants are imperfectly correlated, SNP-based heritability based on unrelated individuals will underestimate total heritability. There is evidence that SNP-based heritability is an underestimate for this very reason." (He gives citations.) Zietsch considers evolutionary implications of the apparent fact that genes of large effect account for very little of phenotypic variance: selection is mostly purifying, not balancing. In other words, within most populations, it's not like there was a niche for small folks and a niche for tall folks (maybe at different times) and selection favored a mixture. Instead, either folks with more tall alleles than average, or folks with more small alleles, or both, were dealt a bad hand in the genetic lottery and consistently had lower fitness. Genetic variation within populations is mostly genetic load resulting from mutation and drift.
Are there any polygenic traits that are more fully mapped/understood that could serve as a control for other types of studies?
My recollection is that a given person's score on IQ tests can vary substantially time to time (like pretty much any test score, really). Do these studies get everyone to take multiple tests at different times to get a more solid number? If not, how do they control for that variability?
In other words, if identically intelligent twins take an IQ test, how much variance do we expect in their results, and how will that error propagate to the rest of these measurements?
"more home truths are to be learnt from listening to a noisy debate in an alehouse than from attending a formal one in the House of Commons. An elderly country gentlewoman will often know more of character, and be able to illustrate it by more amusing anecdotes taken from the history of what has been said, done, and gossiped in a country town for the last fifty years, than the best bluestocking of the age will be able to glean from that sort of learning which consists in an acquaintance with all the novels and satirical poems published in the same period. People in towns, indeed, are woefully deficient in a knowledge of character, which they see only in the bust, not as a whole-length. People in the country not only know all that has happened to a man, but trace his virtues or vices, as they do his features, in their descent through several generations, and solve some contradiction in his behaviour by a cross in the breed half a century ago."
people today would never be able to grasp and see the hereditary argument play out the way even a shrewd village person would who never left their town all their life did. by knowing personally each family of the village, and their qualities and characteristics, and seeing their kids develop and grow, and their kid's kids, and how the triats are passed down from each parent and renewed in their children in the subequent generation, the weight of the hereditary argument in palpable.
instead we have people arguing about heritability who have never observed anything firsthand or took those observations to heart, from pure statistics and data, and contriving all sort of confounding theories.
when you go back farther to the 19th century and before, it was almost taken as a given that hereditary effects were very large, by anyone who seriously thought about it, reasoning from experience and their powerful intuitions.
The problem is that those seriously thinking people with thier powerful intuitions thought the Irish were good for nothing layabouts, the Chinese were too stupid and drug addicted to be used for anything other than building railroads, and the Japanese were too dumb and squinty eyed to build and fly modern airplanes.
The intuitive hereditarian take has been wrong for several hundred years now and I'm not hopefull that future events will improve thier track record.
the chinese and japanese were actually considered really graceful and intelligent species. the japenese ink art was appreciated. the chinese boy servants were appreciated. i actually havent read too many negative things about asians from 19th century britons. the british society was way more advanced, but when they talked about the innate characteristics of the asians they were very glowing - the opposite to how they talked about blacks , indians, natives, for example.
the indians they understood contained much variety of ability and that underlying the castes was race, and that race would be approximated in a variety of ways like by taking measurements of the nose. an imprecise method ,yes, but the best they could do in their time. at least they were still acting like proper scientists, and dilineating and classifying and doing all that good stuff that scientists do to elucidate the underlying relationships. darwin knew evolution to be true from intuition, but he had to document a thousand small differences to give it scientific validity.
Risley "found a direct relation between the proportion of Aryan blood and the nasal index, along a gradient from the highest castes to the lowest. This assimilation of caste to race ... proved very influential."
Have any sources? Both the British and US passed laws to keep undesirable Chinese and Japanese out. The mongoliod race of Japanese and Chinese became synonymous with down syndrome in the western world. I'm not sure where your getting these ideas that they were widely respected from?
We have English statements from the 1100's about the Irish proclaiming "They use their fields mostly for pasture. Little is cultivated and even less is sown. The problem here is not the quality of the soil but rather the lack of industry on the part of those who should cultivate it. This laziness means that the different types of minerals with which hidden veins of the earth are full are neither mined nor exploited in any way. They do not devote themselves to the manufacture of flax or wool, nor to the practice of any mechanical or mercantile act. Dedicated only to leisure and laziness, this is a truly barbarous people. They depend on animals for their livelihood and they live like animals."
all that stuff about the irish sounds accurate to me . it is incredible the amount of eminent people scotland produced (robert burns, thomas carlyle, sir walter scott, john ruskin, to name a few) and eminent people with scottish heritage compared to irish, when they have similarly sized populations, and are similarly rural. likewise the scottish peasantry are talked about in a very different way to the irish , as being strong natured, hard workers, and so forth.
the reason underlying for this is that the irsh represent an archaic native population , while the scottish got a lot more recent influx of geneflow from western and nothern europe from notable groups. (this is simply a speculation, but this is what happened to england with the arrival of the french normans and other groups, who many of the most famous english are direct descendants from, like Byron).
i cant give you sources, but the britons that lived in the far east loved the qualities of the asians and wrote glowing things about them.
This is interesting because modern DNA studies have found that castes have been reproductively isolated for at least 3000 years and that India is not a big mixed hodgepodge of 1,5 billion people, but a huge set of very isolated subgroups of 0,5-1 million each, separated by various caste characteristics.
China, on the other hand, seems to be fairly mixed.
isn't this one of thise things that the modern bias that believes race can't explain anything is surprised by but which people who just lived a few centuries back could have easily guessed? its amazing what can be intuited from simple observation - like the inference of the atomic structure of matter from dust in the sunlight by that greek.
there was a lot of praise for chinese . they said the asian boy-servants were better than any girl, because all of their delicate manners of a girl but having the physical abilities of a boy. the british women would even undress in front of them without worrying of the impropriety because they were so amazingly well behaved lol.
my impression was sociology only went off track in the 20th century. everyone at that time was trying to solve all societies problems thru nurture. eugenics became taboo, blacks were trying to be integrated. they was a bouyant hope in nurture to fix everything because that was the only hope left. but it required turning over and ignoring all the old wisdom.
only the blinding lights of things like twin studies got people slightly back on track.
to try to discount these now is going backwards again, but im not surprised. the quality and soundess of the research being done in sociology has been very sketch every since great scientists like the founders of the field like Francis Galton disappeared.
one of harvards top-paid professor whose research was entirely discredited just had her tenure revoked.
the Irish were recognised by the English fromn their earliest encounters as brilliant, brave, witty, word masters and yes prone to drunkenness, laziness, and boasting.
They were recognized as some of the worst savages in Europe.
The Spectator 1882:
The Tragedy at Maamtrasna, investigated this week in Dublin, almost unique as it is in the annals of the United Kingdom, brings out in strong relief two facts which Englishmen are too apt to forget. One is the existence in particular districts of Ireland of a class of peasants who are scarcely civilised beings, and approach far nearer to savages than any other white men; and the other is their extraordinary and exceptional gloominess of temper. In remote places of Ireland, especially in Connaught, on a few of the islands, and in one or two mountain districts, dwell cultivators who are in knowledge, in habits, and in the discipline of life no higher than Maories or other Polynesians.
Theodore Roosevelt wrote in his diary that:
There are some twenty five Irish Democrats in the house. ... They are a stupid, sodden and vicious lot, most of them being equally deficient in brains and virtue. Three or four however ... seem to be pretty good men, and among the best members of the house are two Republican farmers named O'Neil and Sheehy, the grandsons of Irish immigrants. But the average catholic Irishman of first-generation as represented in this Assembly, is a low, venal, corrupt and unintelligent brute.
RDR is unbiased in theory. But in practice it's underpowered and highly sensitive to noise.
It gets worse because in small samples, the heritability is BIASED downwards. Because the % of IBD between sibs varies so little, any measurement error leads to attenuation and the signal is small relative to noise in this design. So great in simulations, bad in practice.
You should consider the simple hypothesis that RDR is wrong and that Gusev is dishonest
I confess that I have only skimmed Scott's article as of now. (I'm in the middle of a terrible project involving navigating 3 bureacracies at once.) So it may be that the point I'm making is not relevant to the question of why it's so hard to find genes that account for the heritability of IQ. But here it is, in case it is.
What about the impact of physical attractiveness on the developing mind? I have always wondered whether the impact of attractiveness is not taken seriously enough as something affecting life course, including ability to perform in intellectually demanding tasks. Last I knew, people, including teachers, rated physically attractive children higher on various other high-valued traits, including intelligence. Physically attractive kids certainly see more smiling faces than unattractive ones, and probably are given more opportunities, and probably have greater self-confidence and higher self-esteem.
To put another way, physical attractiveness affects the environment in which kids grow up. Imagine identical twins where one has some mildly disfiguring facial scars from an accident in early life and one does not. They have the same parents, toys, etc., attend the same schools, and let's say they take the exact same courses and play on the same school sports teams. If a researcher is thinking that these kids grew up in very similar environments, they're wrong.
I agree physical attractiveness is an important factor in life outcomes, but at least in my observation, lots of intellectual high-accomplishers are not necessarily good lookers. Think of many famous scientists, or even Bezos/Gates/Musk. Among the extremely intelligent, there doesn't seem to be a higher % of good looking people ( though of course, being good looking within this group could still be an advantage, especially today with ubiquitous marketing/media) , although in other things such as acting, politics, etc., looks do undoubtedly provide a larger advantage.
I was with 100 percent certainty going to blame measuring IQ being kinda bottle-shaky water-dowsey witch-doctory; ei It reliably works better than placebo but the mechanism of determination has nothing to do with the test itself.
I still believe this a bit, but everything having good prediction but missing predictors has me back in the pea soup fog level epistemic region with a solid carve out for "Personal Spiritual Experience that IQ tests suck and are deeply suspect".
Maybe I don't know enough about psychiatric practice, but why would you resist temptation urge to throw twin studies in the adoptive parents' faces? If parents are going through something terrible like that, telling them that it wasn't their fault and that they did the best that they could seems like a really kind and worthwhile thing to do.
Re the anecdotal evidence of kids you worked with: isn't it possible that the incorrigibility of these kids is due to being victims of horrific abuse before adoption by their new parents? Relatedly... isn't there a lot of stuff that shows just bad outcomes for adopted kids in general?
What are the studies/evidence that suggests interactions are small? My math/physics intuition would be when you do a linear approximation (assume no interactions) and then change the whole genome your baseline assumption should be that the linear approximation breaks down and has large errors.
You say, "Turkheimer is either misstating the relationship between polygenic scores and narrow-sense heritability," but in the passage you quote I am perfectly clear that I am talking about the DGE heritability, that is Column E from Supplemental Table 3. The value of .048 is the median of 17 (arbitrarily classified by me) "behavioral" DGE heritabilities taken from that column.
There is so much about these studies that I don't like, producing large errors in both directions.
First, there is something very wrong with twin studies. The assumption that separating twins at birth randomizes their environments is flatly false. When twins are given up for adoption in, say, Wisconsin, one may end up in Wisconsin, the other in Illinois. Very rarely does one end up Burundi. But Wisconsin and Illinois are extremely similar places, compared to Burundi!
I hate Educational Attainment as an target outcome. It is deeply immeshed in social practices, many of which are hugely impacted that things that are obviously not genetic. Consider affirmative action practices, as one example among hundreds. People sometimes claim to try to control for these, but it's not a matter of small number of confounders; it's hundreds of confounders all the way down.
I think genetics matter a very great deal, but in ways that thread through social practices in complex ways that make the conventional hereditarian vs. anti-hereditarian discourse meaningless.
The simplest explanation would seem to be that Sib-R and RDR just don't work very well.
Getting out of the weeds, we have one method that we have every reason to believe is producing quality results, and two other methods that maybe should work if we're right about a lot of extremely complex stuff. That the two latter methods don't reproduce the results of the first would seem to indicate that they're just wrong. This just doesn't seem very mysterious to me.
Okay Scott, now I am completely confused. In a somewhat tongue-in-cheek way, I do not know whether to be racist now or not. Can someone how is neither hateful not overly PC but just reasonable, give me some very very rough estimate whether that the say five million Syrian refugees we have here in Central Europe will have kids who will become say civil engineers and pay our social security / pension or not?
This is really what matters in practice.
Please note that I am really not trying to be prejudicial, I bought this very computer from a young Afghan man who built it, but then he sold it to me because he needed to buy a laptop for that civil engineering university and he looked all smart to me and he built this computer well. I am really open to all views.
OTOH it seems we in CE are so far not good at putting the refugees actually to work and making them tax suppliers, not tax consumers.
Can this calc be ran on the back of the envelope?
Is it ever possible to have a really dumbed down answer? Like muh millions of brown refugees, are they gonna cost us or will they contribute money to the state?
This is what the average Euro really wants to know.
No, don't count on them becoming engineers. Intelligence is highly heritable. Don't be racist but it was a mistake to let five million low-IQ people into your society.
The national IQ of Syria is in the 80's. While national IQ's aren't super-accurate I think they're directionally informative. The bigger issue is that these aren't normal economic immigrants (which are a better-than-average sample of the source population) but refugees who aren't selected for either ability or drive. Selection effects determine almost every social outcome. If you let in a bunch of low-human-capital people then you're going to wind up with a worse society in the long run.
the papers "Brain drain in Syria's ancient capital" and "Sex differences on the Progressive Matrices: Some data from Syria" have pretty good samples for getting an upper bound of national iq in Syria , and found 82.
The national IQ estimates are rarely inaccurate, the more valid complaints are whether that low IQ is due to their genes or their environment and are they just a few decades from going up 20 IQ points with the flynn effect or whether IQ test results don't mean quite the same thing that they do in western countries. thinking that these things will solve everything is cope they're better than trying to deny the national IQ samples entirely which is what leftoids normally do (not you)
Really? I'm nothing like an expert but I feel like one of the issues is getting a representative sample. Syria is pretty ethnically diverse, are the national IQ figures representative or did the testing systematically focus on a low-iq cohort? It's sort of like India: its national IQ is high-80s but that glosses over significant caste variation. For very poor countries I'm also somewhat skeptical that you can easily test the IQ of people who have never taken a written test before. Not that it's meaningless, I just mentally assign 5-10 point error bars and assume that reported figures are near the bottom end of that range.
they tested schools in damascus. capital cities usually have higher average IQs than the rest of the country due to brain drain. If you want to get a nationally representative sample you're right it would be harder but if you just want an upper bound , it's not so difficult.
They tested school children so it's not like they've never taken a written test before.
and yeah an IQ score of 82 in syria might mean something a bit different from an IQ score of 82 in a more developed country , but that still doesn't mean that the average national IQ scores is inaccruate. if you tested everyone in syria the average IQ really would be close to 82.
Arabs conquered the Classical Civilization, which gave them a lot of boost, but their intellectual trajectory since the destruction of Baghdad by the Mongols until the European colonial expansion was one of obvious descent.
Neither Egypt nor Iraq look like heirs to the most developed civilizations of Antiquity, and don't seem to be in recent upwind either. Italy and China, on the other hand, were able to reconstitute themselves into highly developed places, and India is probably on the road there as well.
Contrast also the Arabs with the Iranians, who are still more obviously intellectually developed as a culture, not even the Islamic Republic could extirpate the local intellectual life.
Even at Western universities, you will meet more Iranian than Arab scientists.
Gregory Clark in "for whom the bell curve tolls" also theorises that due to the cultural prevalence of cousin marriage in muslim countries there would be less assortative mating and so the selection pressure for IQ or general social competence at acquiring socioeconomic status would not act as intensely.
This would explain why the middle east cities of north africa were at the forefront of civilization before Islam and not long after Islam spread but the more centuries they were muslim , the more they decayed.
People think of 1683 as when islamic civilization started falling behind but in fact that's only when they stated being beaten back in territory. When you look at global discoveries in science and mathematics it's clear that they had started falling behind centuries earlier.
each descendant of immigrants from the middle east or north africa cost denmark on average 50000 Danish krona (6700 Euros) in just 2018 alone, even when standardising for the age distribution.
and in terms of crime they're even more criminal than their parents.
Specifically, 17% of middle eastern male immigrants in Sweden are crime suspects, but 24% of the male children of middle eastern immigrants born in Sweden are crime suspects.
And for North Africans it's even higher. 19% for the male immigrants and 31% for the male children of immigrants.
you would have to believe that simply being born and living in Europe all their lives doesn't change their human capital, but one or more further generations born in Europe will. The idea that they'll assimilate and become culturally indistinguishable from Europeans will happen a little bit but seems very unlikely to happen totally or even mostly, especially in the future as the % of indigenous children in european countries like Germany, France, UK, Austria, etc. plummets.
Why would the grandchildren of middle eastern and african immigrants in 20 years time assimilate to indigenous European culture when indigenous Europeans children will only make up 40% of the children growing up around them (much less in practice given that ethnic communities are separated to a large degree and spend much less time socialising than would be predicted by their overall percentages of the youth population)
So hoping that liberal, laissez-fair cultural assimilation will happen like a USA melting pot and that will raise the human capital of these immigrants and their descendants is very unlikely.
Once upon a time, there was a journalistic attempt to frame the German police as racist because they used a specific abbreviation "Nafri" (Nordafrikaner) for North Africans.
This attempt fizzled out immediately after the police showed just how much Nafris were overrepresented in all sorts of crimes. IIRC several hundred of them were responsible for half (?) of assaults in Saxony, with 4 million inhabitants, etc.
I hope the study corrects a changing baseline of educational attainment over the years. At some point it becomes more worthwhile to ask which college your parents went to than whether your parents went to college.
I can’t help but wonder what would happen if we analyzed the quality of novels by the usage of particular words or phrases. There is probably something there, as better novels, or better selling novels, or whatever, probably use or avoid certain words/phrases relative to crappy novels. But not much. Probably just enough to make it seem worthwhile to keep looking harder.
Of course we know that what makes a novel great is how all the words work together. That it isn’t so much which words, but how all the words chosen work together to create a higher level of “novel goodness.” Plot and readability and character insight are emergent qualities. Yes, the quality is in part connected back to the words, but only in a very holistic way (that also depends upon the cultural context the book exists in).
Not sure if any of this even helps… but a boy has to wonder.
"Typical estimates for adult IQ found it was about 60% genetic, 40% unpredictable, and barely related at all to parenting or family environment." A similar statement can be made regarding height where we find an even stronger genetic link, close to "80%".
But, this type of statement can be misleading and is only true in a limited scope of conditions.
To further clarify, here is a concrete example:
Average female height in Poland 1896-1914: 152.62 cm, std 5.74 cm
Average female height in Poland, children born 2000-2001: 167.85, std 6.91 cm
The average woman born in Poland 2000-2001 would have been over 2 standard deviations above the average in 1914, the average woman today would have been in the top 2.5% back then! Another way of stating the same fact, the 2 standard deviations increase in height is 0% genetic and 100% environmental.
The same trends can be observed in all developed countries all around the world. Less developed countries, for example in Africa, has experienced a much smaller increase in height. A clear separation can be seen been North and South Korea after the division of the countries with South Koreans gaining much more height than their North Korean relatives.
We understand this process fairly well, nutrition, pollution and disease greatly affect the growth and development of a child.
So, this type of analysis is fine, we should look for the genes that shape our bodies and lives, but we should remember that there are limitations to these types of studies.
> The average woman born in Poland 2000-2001 would have been over 2 standard deviations above the average in 1914, the average woman today would have been in the top 2.5% back then! Another way of stating the same fact, the 2 standard deviations increase in height is 0% genetic and 100% environmental.
3-5 generations passed between 1914 and 2000. Why would you assume no genetic change? There definitely was genetic change.
I am talking about population genetics where 3-5 generations for a population of millions is a very short time. You would have to have an extreme evolutionary pressure to select for such a drastic change in such a short time. The collective genome of the Polish population would have seen an extremely small change over such a small amount of time, a few changes due to genetic drift, maybe a few changes due to migration. The greatest source of change would have been ww2 when approximately 20% of the Polish population was killed, but even an event this big and tragical would be considered small in terms of population genetics.
In the lab we often select for new traits using heavy evolutionary pressure and even then selection often takes 100's of generations, and this is with genomes that are less complex than our own. For a polygenic trait like height that is decided by approximately 12,000 genes according to one recent publication, you would need an extreme evolutionary pressure and many generations.
No, I'm thinking that you stated explicitly that 0% of the increase in height is genetic. There is no reason to believe that. Your "rebuttal" makes the argument that more than 0% is environmental, which is unrelated to what you said earlier.
Apologies if this has already been mentioned - I've read the comments but couldn't see anything about it, but I could easily have missed it. GWAS use common SNPs as markers for putative causative variants (which may be anything from another SNP to large structural variants), but the thing is, that means that all those variants have to be ancient, both the causative variants and the GWAS markers. GWAS won't detect links to more recently-derived causative variants. Studies tend to focus either on common SNPs (as per GWAS) or ultra-rare ones (single family pedigrees and Mendelian segregation), but I wonder how much variation is down to semi-common variants which are too recently-derived to show up using a GWAS but too common and/or variably penetrant/mild in effect to be identified using traditional Mendelian segregation methods.
As an example, I would offer a 12bp in-frame deletion in the gene RIPOR2, which is associated with non-syndromic hearing loss and is very common in the Netherlands (and also exhibits variable penetrance - not every carrier develops hearing impairment). It's a bit too recent and rare to be a perfect example, but it's not far off.
This is one of those moments where it's very annoying that Substack doesn't allow you to gift-link specific articles. I understand why you wouldn't want to subscribe to a million Substacks of people who leave their articles in the comments. But I figure you also understand why I would keep interesting and original work paywalled. I would comp you a subscription but that requires an email or that you're already a free sub. Or you could do the trial-and-cancel thing. Alas, no simple workaround!
Also, your footnote on Akbari’s intelligence selection paper seems mistaken. Yes the IQ PGS score changed, but not by NEARLY as much as many other complex traits. Clearly there’s been less selection on IQ PGS than many other traits where we do find heritability still!
In computer science, there is something called a hash function. Or similarly a Pseudo Random Number Generator. This is an algorithm that, given the same input, produces the same output. But given any small change in the input, the output is completely different.
If these traits were partly determined by a hash function over the genome, then any 2 people who aren't identical twins would have independent rolls of the genetic dice.
So this would show up as twins being the same, and except for twins there would be no detectable pattern between genes and results.
Another option is a gene-environment interaction effects.
For example, suppose some kidney genes work well with a high salt diet, and others work better with less salt. Then there is a strong correlation between twins, which share both environmental salt levels and genes. And a weaker correlation between siblings who often have different genes.
Yet looking across all environments, there is no best genome for kidney function.
> This is an algorithm that, given the same input, produces the same output. But given any small change in the input, the output is completely different.
No, that's a cryptographic hash function. A hash function has much laxer requirements.
I'm not conjecturing that everything is a hash function. Only that there is some component that is a hash function. (Well any sufficiently complicated function that science hasn't spotted the pattern yet really)
Behavioral twin studies first appeared in the 1920s, not the 1970s. Your defense of twin studies is weak, relying on supposed validation from other flawed methods and studies (e.g., adoption studies). P-hacked studies based on false assumptions cannot be used to cross-validate similarly flawed studies. It’s time to abandon behavioral and psychiatric research using twin method MZ-DZ comparisons after a disastrous 100-year run, and to rewrite the social and behavioral science textbooks accordingly.
Tangential to all of this, but something I recall from Nevin (one of the proponents of lead-crime hypothesis) on Rationally Speaking a few years ago:
At first glance, lead is the biodeterminist's favorite example, but also seems to prove too much. Many lines of evidence suggest that it's a massive effect, yet shared environment doesn't seem to be a big deal.
But IIRC, Nevin went through many original adoption studies, and found that adoptions were often much later than we imagine - over one year old, rather than just a few months. This might mean that either maternal/womb or early-childhood environment is actually much larger than we give it credit for.
Suppose there was some non-DNA way to inherit traits. Perhaps something epigenetic. Would this explain part of the gap?
I think it depends on whether this mechanism exactly tracks genetic resemblance in monozygotic vs. dizygotic twins. If it doesn't, then twin studies continue to prove something about genes (and not this other thing), and molecular genetic studies obviously prove something about genes (and not this other thing), so the mismatch is still surprising.
If it does track genetic resemblance (ie much stronger in monozygotic than dizygotic twins) then I think it might work, although I'd have to think about it harder to be sure.
Epigenetic effects could be stronger in monozygotic twins because these twins split a few days after fertilization (and up to that point, they are the same group of cells).
It would be quite interesting to track the similarity of dichorionic vs. diamniotic vs. monoamniotic twins (which split at progressively later stages) on various traits and epigenetic marks. You mentioned this study: https://pubmed.ncbi.nlm.nih.gov/26410687/ which addresses some of this but l think there's more to be discovered here.
See also: https://denovo.substack.com/p/a-tale-of-twin-sisters (where one twin likely stole primordial germ cells from the other).
Yes! This was my first thought. The comment above merits discussion and critique. If anyone here can offer a strong argument against the opinion above, please offer it!
o3 suggest that epigenetics markers persist more through mitosis than through gametogenesis, so I would expect epigenetic correlations to be stronger in monozygotic vs dyzgotic twins.
https://chatgpt.com/share/686029e8-2474-8004-9dc1-3b1dc0a83673
I think in the end missing heritability will be resolved as "our computational genetic models made too many simplifications" and so genes will explain all the twins heritability, but require more sophisticated models involving more biological mechanisms.
Hypothetically, a mechanism that match this "dark inheritance" situation is the fine details of contents of the egg cell. e.g. Is protein A present in the egg cell in an amount over threshold At? How common is enzyme B relative to enzyme C? How close to optimally efficient are the ratios of organelles contributing to synthesis? and so on.
Given this hypothesis, we'd expect monozygotic twins to be most similar because they started from the same egg (c.f. twin studies). We'd expect dizygotic twins to be more like non-twin siblings of the same parents -- as they differ in fine details of the egg contents of the two eggs from the same mother, like two random draws from the same distribution (c.f. Sib-Regression). And we'd expect non-sibling relatives to be less alike that siblings but more alike than strangers, as with eggs from different but related mothers, it's like two random draws from two distributions that were themselves random draws from a common distribution (c.f. RDR).
This mechanism should operate Darwinianly, not Lamarckianly, because a woman's eggs are already present at birth and not filtered in any obvious way by her life events. If I'm thinking about it right, it also predicts that if a woman has two children by two unrelated men, versus if a man has two children by two unrelated women, that the former pair should be more similar to each other on average than the latter pair, i.e. the former should have higher RDR scores than the latter. ChatGPT tells me this is an empirically true finding about half-siblings. I suppose if this mechanism is real it's hard to separate from other environmental effects, and difficult to directly test in humans without harming the egg.
There is. Or rather, not exactly; it's still DNA, it just isn't human DNA. What about mitochondrial DNA- has anyone ever investigated whether this affects IQ?
Mitochondrial DNA is identical in both fraternal and identical twins, as it's inherited from the mother in a non-sexually-recombining way.
Wouldn't all twins (identical and fraternal) get the same mitochondrial DNA from their mother? It seems like this would cause us to underestimate heritability from these studies, not overestimate it.
Actually, this leads to a question: Do all children inherit the same mtDNA from their parents? Or is there some variation in this that might differ between identical and fraternal twins?
> Do all children inherit the same mtDNA from their parents?
Funny you should ask. I was curious about this subject and was digging into that question a while back. So, it's pretty fresh in my head.
Statistically, it's very likely that they do. The human mtDNA mutation has been estimated to occur approximately once every 30 generations (depending on the study methodology, this estimate may be slightly higher or lower). There are about 16.5K base pairs in mtDNA. The odds work out to something like one mtDNA mutation per 40 births.
A human egg cell has 100's of thousands of mitochondria though. I wonder how common it is that there might be multiple versions of mtDNA within those. In theory you could have multiple 'lineages' of mtDNA in your body that would pass down (possibly in different ratios through random chance when the egg cells formed) generationally. Probably not anything impactful, but I wonder to what extent it happens.
Good point. Hasn't the accumulation of mtDNA mutations been linked to diseases like Parkinson’s and certain cancers? But it would only matter across future generations if the mutation happens in the ovum.
I thought of the mitochondria too. Fraternal twins would get the mitochondrial effects only, identical twins would get genome plus mitochondrial effect
Or gut bacteria? I mean, it apparently affects everything else, so why not EA.
Transgenerational epigenetic inheritance doesn't seem to be important in humans. There are some epigenetic resets that occur early in embryonic development, plus the effects of epigenetic modifications acquired during life are much more important. The latter would show up as environment (some shared, some non-shared) in twin studies.
As a reproductive developmental biologist, I agree. I've written about this if people want to know more: https://denovo.substack.com/p/epigenetic-inheritance-in-mammals https://denovo.substack.com/p/epigenetics-of-the-mammalian-germline
In your article, you wrote that, "Disruptions to the establishment of epigenetic marks in germ cells during fetal development can certainly affect the next generation..."
Wouldn’t that suggest the mechanism you describe could potentially account for part of the observed gap?
That would be intergenerational but not transgenerational.
Yes, I know, but that doesn't answer my question. My point is that intergenerational—not transgenerational—effects might explain the gap. I should have been more clear. That's why I asked that question.
Do you think this is possible?
There *IS a non-DNA way to inherit traits. Many traits are inherited via the cytoplasm, e.g., IIUC, the structure of the ribosome. Also mitochondria are (essentially) only inherited via the cytoplasm, and that contributes to average energy level. There are certain to be lots of others.
AFAIK, the small and large subunits of the ribosomes are constructed in the nucleolus based on highly-conserved genes, and these are later put together in the cytoplasm. In fact, the genes are so highly conserved, that they are used to establish phylogenetic trees. What structure of the ribosome are you referring to as being inherited in the cytoplasm?
Got a source? (I don't.)
Yes, they are highly conserved, but the "highly conserved genes" part wasn't known when I read about it. And, IIRC, they were assumed to be metabolic inheritance via the maternal cytoplasm.
But that would be a purely maternal effect which I assume would be easily detectable with half-sibling studies.
Yes, these are things that would only be inherited in the maternal line. So they could be easily detected, but you'd need to look for them.
I'd be shocked if no one had.
I'm sure lots of folks have, but there are lots of different studies being compared, and how frequently was this considered? Most of the time people don't even seem to think about cytoplasmic inheritance, even though when they *do* consider it they realize it has to be happening.
Well it would just show up as higher phenotypic correlations with the mother than with the father. AFAIK those doesn't exist for anything apart from the known sex-linked traits.
Lares Familiares
I'm not going to pretend that I can understand everything discussed above - or rather - that I can digest it meaningfully in my first read through.
But I'm curious as to whether genome wide polygenetic studies are good at identifying gene-environment interactions.
Most people think of the environment piece of gene-environment interactions as something obviously related. For IQ, you might expect it to be the parenting strategy of the parents that will evoke the best response for the IQ genes. But that's not necessarily the case. And a gene that manifests higher IQ, might not look like a gene for higher IQ unless its unfolding in the right environment - which can be a wide range of things.
I know that twin studies to some degree are meant to understand this, and there are also twin studies of twins reared apart, which should more strongly take care of this. But its still the case that, even when twins are reared a part, the environmental component that elicits the effect of the gene is so common that twins reared apart don't functionally constitute twins reared in different environments.
Anyway, I'm just curious about that.
I second this. But it does not solve the issue of why the environment seems to matter so little in the twin studies, comparatively.
For example - and this is just a really random made up example - if full IQ development, and thus full expression of the heritability of IQ, depended on higher levels of choline intake, then you would see the heritability and full expression of IQ among people that eat a things higher in choline, like fish and meat. The heritability among different sets of twins would depend on their families, and perhaps individual preferences for, meat and fish. Twins reared apart could easily end up seeming very similar if they were likely to stay among households that had culturally very similar diet patterns.
I'm just really curious as to whether gene x environment interactions are extremely difficult to parse when you are also looking at polygenic traits with gene numbers in the 100s or 1000s?
Good point - things that are genuinely common in environments will never show up as "environmental influence". Of on statistics language, as a "treatment", the environment is neither random nor normally distributed.
Michael Levin's theory on bioelectricity seems like it would explain the discrepency, but I never studied biology and I don't know if his work is considered fringe/fake or genuine cutting edge.
Also Idk if he's actually the creator of that area, i just heard him do an interview once where he talked about making two-headed flatworms that gave birth to other two-headed flatworms, without ever touching the genes at all (because that's apparently possible, I guess???)
"Dynastic effects" and "genetic nurture" are very different processes and do not belong in the same category. Dynastic effects basically refers to the kind of phenomenon you mention of Norman surnames correlating with high status many years later, while genetic nurture refers to direct genetic effects in parents affecting their influence on children. Gemini's summary is as good as any:
"Genetic nurture is about what your parents do for you environmentally because of their genes.
Dynastic effects are more about the broader social and economic standing and inherited environmental context of your family lineage over generations, which is correlated with your genetic ancestry."
The most recent work (e.g. Nature Hum Behav Nivard 2024) suggests that for educational attainment, genetic nurture is very weak, while dynastic effects are quite considerable.
Thanks for catching that, I deleted the section on dynastic effects but forgot to delete it from the header.
Dynastic effects are also not necessarily non-genetic.
https://www.astralcodexten.com/p/missing-heritability-much-more-than/comment/129495557
I'd be mildly concerned that correcting for population structure in GWAS analysis may be "the sociologist's fallacy" at work again, though if twin studies vs. sib-regression and RDR is yielding contradictory estimates for kidney function then something fishy really is going on.
As someone who has not yet seriously looked into adoption but who is open to the idea, your child psychiatric hospital anecdote is terrifying. Isn't the whole point of adoption (for many people) to offer a better life to kids who otherwise might have had a terrible time due to awful parents/environment? But actually adopting those kids puts you at a much higher chance of raising antisocial parasites, even if you treat them exactly like your other kids? Such a depressing thought. If this is a grave oversimplification, I'm happy to be proven wrong.
I think you can mostly avoid this if you're careful about who/where you adopt from. Some children are in orphanages because their parents died in a war or something, and they're pretty normal. If it's because the parents were involved in crime or drugs or something, then yeah, I think there's a significant risk that the child will be pretty tough to raise.
How much of that is explained by FASD and the like? (Or the developmental effects of extreme neglect in early childhood?) I was about to say “not that that makes much difference to adoption pessimism”, but FASD is actually not so hard for a physician (or even a trained layperson: kindergarten teachers in some countries can tell) to diagnose, no?
I think some but not all.
Well, sure; I asked how much. (I am not a medical professional, unlike you; my sense that there is more "FASD and the like" than many assume comes from talking to medical professionals and other people who know more than I do. People in the US now may drink less than they used to, but they also take more drugs.)
In the US the rate of FASD kids is somewhere between 1% and 5%, but 19% of foster kids have FASD (https://pubmed.ncbi.nlm.nih.gov/39031634/). It also looks like 30% of kids diagnosed with FASD end up in foster care. Foster care and adoption are not the same thing, but I would expect the populations to be more similar than they are different.
Thanks; this is helpful. We are talking about a big effect - part of the main term, not the second-order effects.
I looked into this once as when researching the whole "alcohol-during-pregnancy" topic. My takeaway: the answer is "yes" for FAS (Fetal Alcohol Syndrome) and very much "no" for FASD (Fetal Alcohol Spectrum Disorder). The former is characterised by the typical FAS face (small eyes, smooth philtrum, thin upper lip), low birth weight, microcephaly and various pretty severe developmental disorders.
The latter spectrum's diagnostic criteria are ridiculously loose leading some studies to conclude that as many as 10% of children have FASD (https://pmc.ncbi.nlm.nih.gov/articles/PMC5839298/). Add to that everyone who is "on the spectrum" or has ADHS and nobody normal is left.
The Christians I am familiar with are eager to adopt the children of homeless drug addicts. The children of recidivist criminals, I’m not so sure.
Either way: they’ve imbibed the conventional wisdom about heredity and life outcomes and while it only strengthens their compassionate determination to provide better-than-the-best in terms of nutrition, screens (none), reading, homeschooling if school is harsh and so forth - it also means their mantra is that the goal is simply a happy life.
>Some children are in orphanages because their parents died in a war or something, and they're pretty normal.
This is extremely unlikely in a modern developed country. Even in the improbable event that a child is unfortunate enough to lose both parents in a war, custody will almost always be taken over by a surviving close relative. You can try to adopt from an orphanage in the third world, but even then, they tend to prioritize prospective adopters from the same nation over foreigners.
If a child is available for adoption by a stranger in the US in 2025, you can pretty safely assume that "the parents were involved in crime or drugs or something".
> If a child is available for adoption by a stranger in the US in 2025, you can pretty safely assume that "the parents were involved in crime or drugs or something"
Perhaps this is covered by the qualifier "pretty", but there's also the possibility that it was simply an unwanted pregnancy. From what I can gather, in the US, it is simultaneously true that the majority of unwanted pregnancies are aborted (here counting anything after conception as a pregnancy***), and that the majority of children who are adopted were born to single mothers. But we're not talking about base rates here. The relevant question is "what percentage of children who end up adopted were unwanted pregnancies", which is surprisingly hard to find a straight answer to as it doesn't seem like this has been measured directly, but we can vaguely gesture at the reasonable assumption that most unwanted pregnancies are the result of circumstances that are conducive to the couple not staying together. Rape is the extreme example, but it could also be things like disagreement about the morality of abortion/contraception, one parent wanted a child when the other didn't, or both parents wanted to abort but neither could afford it.
Although, I must admit, being involved in crime or drugs would certainly make raising a child more difficult, as it would require reducing your involvement with one to make room for the other (keeping other cofounders like employment constant). So, in a sense, a child conceived by someone involved in crime or drugs is more likely to be an "unwanted pregnancy" than the base rate of unwanted pregnancies, to the extent that you trust someone involved in crime or (especially) drugs to know what they want.
Also, in the case of rape specifically, rape itself is a crime, so you could technically count it as "being involved in crime", with the caveat that the involvement is unwilling (as are many other forms of crime involvement, to varying degrees, so I don't think this association is totally unwarranted).
***I want to be explicitly clear that I'm not making any sort of political, moral, or religious statement by defining a pregnancy that way. In context I was merely pointing out that the availability of post-coital contraception prevents a large share of pregnancies that would have otherwise been aborted later or carried to full term.
If a child is the result of a rape, both their parents were involved in a crime, but typically only one of them was unwilling.
With respect to children of addicts becoming criminals, do we know how much of this is genetic, and how much can be attributed to gestational effects, such as fetal alcohol syndrome?
> If it's because the parents were involved in crime or drugs or something, then yeah, I think there's a significant risk that the child will be pretty tough to raise.
Remember, your psych hospital sample will have a LOT of selection effects.
Maybe 90% of the kids of duggie/in jail parents who get adopted by those good christian families turn out totally fine? But you only saw the few cases that didn't.
I think the existence of these cases can tell us of the existence of an extreme tail to the distribution, but tells us less about the overall distribution shape.
I think you're right, and Scott and Cremiux are misinterpreting the posted study. The key line is:
"Note, however, that a genetic influence is not sufficient to produce criminal
convictions in the adoptee. Of those
adoptees whose biological parents have
three or more convictions, 75 percent
never received a court conviction. Another way of expressing this concentration of crime is that the chronic male
adoptee offenders with biological parents
having three or more offenses number
only 37. They make up 1 percent of the
3,718 male adoptees in Table 2 but are
responsible for 30 percent of the male
adoptee convictions."
So, 1% of male adoptees in the study have a parent with three or more convictions *and* themselves have been convicted of a crime. 3/4 of children whose parents have been convicted 3+ times are never convicted.
25% chance of conviction definitely qualifies as "a significant risk that the child will be pretty tough to raise".
Oh I definitely agree with that. But Cremeiux (quoted by Scott) said: "The 1% of male adoptees whose parents had 3+ convictions were responsible for 30% of the sample's convictions." That's just not accurate: male adoptees whose parents had 3+ convictions made up 4% of the sample, not 1%. The 1% are the 1/4 of those adoptees with any criminal convictions. And Scott has come out explicitly in favor of nit-picking critiques of factual inaccuracies, so I feel pretty good about pointing this out :).
Bigger picture, I think this study is consistent with a world where, a) a large share of kids who end up committing crimes/having serious behavioral and emotional problems despite being raised by fairly normal families are adoptees (as Scott observed), and b) most adoptees raised by fairly normal families don't end up committing crimes/having serious behavioral and emotional problems.
Anecdata: I went to a fancy Christian private school (because my parents bankrupted themselves to find a somewhat-tolerable schooling environment for their weird son, not because I grew up fancy meself), and two of my classmates—out of about 100 total—were adopted through a similar "help the worst-off" sort of idea on the part of the parents. It was /very evident/ that they were... different... from the rest; you'd guess they'd been adopted even if they hadn't freely admitted it. The girl, as far as I know, didn't do great but didn't get in trouble; her brother did, indeed, end up arrested for assault.
What differences did you observe in them?
Mostly, you might call it "difficulty fitting in", maybe.¹ The girl was nice enough, I think—not that I interacted with her much—but struggled with grades (always having to stay behind for extra help, retaking tests, etc.; in my memory, at least, more-so than any other female student in our grade) & occasionally made some odd comments in class or had weird dramatic outbursts. Nothing too bad, I mean, just stuff like suddenly telling a story about having bad diarrhea or the like (not an actual example, heh, I just mean /that sort of thing/).
The boy was... odder. He was caught masturbating /in class/ once, for example; always wore a large windbreaker thing, and apparently figured he could hide it underneath... He would become somewhat violent with people even then, e.g. when caught /very obviously/ cheating in a card-game (not even the sort for money, just a regular game of Uno or Pokemon something—so no one really cared very much, but it seemed to be the fact that no one believed him about the cheating being "accidental" that set him off); was twice caught with vodka in a water bottle (which is possibly more unusual at that place than in a regular high-school; if anyone else there did something like this, at least, I never heard about it); had even more difficulty with the educational material than did his sister; etc.
--------------------------
¹: (Not that this is a moral failing, or anything. I had struggled to fit in at all my previous schools, too, and for my first ~year at this one.)
Do you have positive knowledge that none of the other 98 were adopted in this manner? You might be providing strong anecdata that "strangeness" predicts "worse-off adoption" but weak anecdata that "worse-off adoption" predicts "strangeness".
I suspect babies put up for adoption in the postwar golden age of adoption, such as myself, tended to be more promising on average because there were a lot of social custom reasons for not raising the baby yourself that aren't as stringent anymore.
This is an important point! The social context for why children get put up for adoption matters a lot in what you'd expect the heritable traits of the kids to be, and the fact that this changes over time in different societies probably makes all kinds of adoption studies done at different times hard to compare. US adoptees in 1950 and in 2000 are probably very different in a lot of ways because of the changes in why people gave up their kids for adoption.
It matters for what you expected adopted children to be like, but not so much for whether adopted children are similar to their adoptive families or to their actual families.
Right. A big question is whether there is too much restriction of range for this analysis. That doesn't appear to have been true in the 1950s.
Steve Jobs was put up for adoption in 1955 because his biological parents couldn't agree to get married. Then, they did get married and had a second child, the fine novelist Mona Simpson who gave the much-admired eulogy for Jobs at his funeral, even though they didn't meet until their 20s.
Which war marks the postwar golden age of adoption?
I am told that my grandfather was given up by his mother during the Great Depression on the theory that she couldn't afford to raise that many children. (She didn't give all of them up - just him.)
This permanently damaged the relationship between the two of them.
I'm thinking the prosperous 1950s-1960s, when society still frowned upon single mothers, saw a lot of women who were otherwise highly functional other than not having a ring giving up their babies for adoption.
Steve Jobs, born 1955, is an example of the quality of baby that could be adopted during the baby boom.
Personally, I always felt sorry for his adoptive little sister. She was just a random person who wanted to live her life, but she could never out-argue her big brother, who happened to be the World's Greatest Salesman.
A friend of mine said she'd deliberately chosen to adopt a kid with identifiable mental disorders (autism etc.). This is because, in the national adoption system (UK), healthy, happy kids from wonderful parents who tragically died together in a car crash, with no grandparents, aunties/uncles or family friends to absorb these kids, are vanishingly rare.
So, given that they're in the adoption system, all kids are problematic for *some* reason - whether their own or their parents. My friend judged that it would be less risky to choose a child whose problem is known and manageable.
Of course, waiting outside Ukrainian hospitals to whisk away the healthiest war orphans you can find is another option, but I suspect that it's a logistical nightmare.
Adopting from foreign countries seems to be the best way to improve a child’s environment. Twin studies are almost always comparing children in the same country, and cross country environments very significantly more.
There are almost no children left to adopt, unless you adopt severely disabled children, teenagers or do foster care (where the goal is reunification). There are like 30 times more prospective adoptive parents than there are children given up for adoption.
Read about the Baby Scoop Era and the investigations South Korea is now doing into their old adoption agencies. Even back then (~50's-90's), the demand for children was bigger than the supply. Most children "given up" were actually cases of hardcore pressuring young single mothers (by eg. only wanting to give them money or medical care if they gave up their baby) and even straight up kidnapping. Adoption was a business.
Not adopted but it still bothers me how uneducated the public is on the subject.
Really? I thought it was the exact opposite! Is there a good overview article?
Not exactly an overview, but see for example this:
World’s ‘baby exporter’ South Korea violated human rights to meet demand, probe finds | CNN https://share.google/gYyGn1DaievLWBZOg
This book: The Baby Scoop Era: Unwed Mothers, Infant Adoption and Forced Surrender: Wilson-Buterbaugh, Karen: 9780692345795: Amazon.com: Books https://share.google/SIFXrdX9wYIBBmHnU
A number of European countries have recently stopped all international adoptions, because they couldn't accurately verify each case to see that it wasn't human trafficking.
The idea that there's millions of misty eyed orphans waiting for parents is just that, a cultural myth. In the 50's through 70's amplified by the prevailing idea that it was better to give up a baby than let an unmarried woman raise it, plus how shameful pregnancy out of wedlock was. I think it's extremely psychologically difficult to give up one's baby and almost any woman would choose an abortion. Cases where each parents are tragically dead and there are no other relatives (always first choice) left to raise it are vanishingly rare.
Sorry for the awkward links, I'm on mobile
Thanks
> Even back then (~50's-90's), the demand for children was bigger than the supply. Most children "given up" were actually cases of hardcore pressuring young single mothers (by eg. only wanting to give them money or medical care if they gave up their baby) and even straight up kidnapping.
I had a Chinese teacher (in China) who noted that when she was born, because she was a girl, her grandfather had advised her mother to give her up for adoption (making it legal to have another child), but her mother refused.
Maybe it's "woke," but as an educator with experience teaching "different" adopted children, and having adopted siblings myself, I find the idea of calling such children "antisocial parasites" both repulsive and dangerous.
"The children are always ours, every single one of them, all over the globe; and I am beginning to suspect that whoever is incapable of recognizing this may be incapable of morality."
- James Baldwin
I don't think it's woke, I think your reaction is normal and appropriate. I'd just like to mention that my antisocial parasite comment was strictly focused on this line from Scott's article: "Then they promptly proceeded to commit crime / get addicted to drugs / go to jail / abuse people". At face value, everything here except for "abuse people" is not that awful. But adoption, to me - and again I haven't given it much thought yet - is about choosing a child that you bring into your family. That choice will be made on your own terms, but I reckon a whole load of criteria that people routinely use when choosing a child to adopt are politically incorrect, or otherwise criteria they wouldn't feel super comfortable saying out loud. This can lead to certain subsets of kids being less adopted than others.
By using the word "antisocial parasite", I tried to capture the essence of the following feeling: If I am given a choice when it comes to adopting a kid, including the existential non-zero risk of the kid destroying my and my partner's lives, I will stay away from those that have (apparently) much higher risks of doing just that. It doesn't invalidate their existence and their right to a future and a loving family - but I won't be the one to take them on, and I suppose a lot of people wouldn't, if they knew about those risks upfront.
Thank you for your comment.
If I can butt in: I wonder if Jake's objection also comes from misunderstanding a subtle detail of what you said? It sounds to me like Jake is repulsed by the idea of calling a child an "antisocial parasite". Which I would agree with. But as I read it, I thought you were talking about how they'd be behaving once reaching, say, 16+. If you meant it as I understood it, you might want to clarify that.
I think in an atmosphere where many young people with the means to have their own children, forgo doing so (or even marrying) because it feels like a risk - it would be a little strange to expect a wholesale anything-goes attitude toward adopting those ill-parented children society is producing in numbers.
As usual, it will be down to those despised folks, the fundamentalist Christians, always hopeful, this world not entirely real to them.
Adopted children are parasites in the most literal sense possible. Check out https://en.wikipedia.org/wiki/Brood_parasitism .
This is one of those things that is obvious to everyone in any nonhuman context, and yet somehow difficult to see when humans are involved.
You are ignoring the obvious difference that humans choose to adopt children, rather than being deceived into it. The analogue here is a woman cheating and not telling the man that the resultant child isn't his, not adoption.
One could argue that blank-slatists are in fact deceived into adopting children by antisocial parents, but it's typically not the parents doing the deception (though the children themselves might, to their however limited ability).
In your opinion, are there any teenagers who could be morally classed as "antisocial parasites"?
What a great post
Thanks a lot
I'm always happy to see a "Much More Than You Wanted To Know"—and this one's even better than most, because "missing heritability" is something I've been quite puzzled about (& I trust our amigo Scott "Yvain" Alexander to make a good run at it if anyone can👌). Cheers.
On priors, I would suppose there's something we're missing with the newer methods—partly because that sort of thing happens all the time ("this thing actually worked this other way all along, oops!"), and partly because I agree that it's difficult to see exactly how major error could creep into the twin studies.
--------------------------
(Y'know, a little while ago I had an idea—possibly even a cromulent one!—for how to square some of the gap away; or, rather, for why it might not be as worrisome as it first appears... but I want to check me notes before I risk saying something totally wrong & dumb / something that turns out to be the first thing everyone thought of already. Now, where in my hundreds of untitled drafts /is/ it...)
"On priors, I would suppose there's something we're missing with the newer methods—partly because that sort of thing happens all the time ("this thing actually worked this other way all along, oops!"), and partly because I agree that it's difficult to see exactly how major error could creep in on the twin studies."
Yeah, these results should definitely make us less confident about the high heritability (though a large part of what is going on might be the focus on EA, like Scott said), but less confident in a case where we started out highly confident still can leave us confident. Though of course eventually the issue becomes harder and harder to ignore. I just don't think that the 'this definitely means something is wrong with adoption/ twin studies' can be said to have arrived until the system looking for genes is accounting for stuff other than common SNPs.
Perhaps something is wrong with the 100+ year history of twin and adoption studies. Alternatively, something might be not quite right with the 12 year history of GWAS studies, just as most of the candidate gene studies of 22 years ago flamed out.
"Cromulent" seems to be the 2025 World of the Year. I'm seeing it all the time in recent weeks. I had to look up what it means ("acceptable or adequate") a few weeks ago.
You should watch the old Simpsons episode that coined it.
I don't think there's anything 2025 specific to it. But might just be weirdly common in whatever media you consume?
"Cromulent first appeared in the February 18, 1996 episode of The Simpsons called "Lisa the Iconoclast," in what could be considered a throw-away line given during the opening credits. The schoolchildren of Springfield are watching a film about the founding father of Springfield, Jebediah Springfield. The film ends with Jebediah intoning, “A noble spirit embiggens the smallest man.” One teacher at the back of the room leans over to another and says that she’d never heard the word embiggen before she moved to Springfield. “I don't know why,” the other teacher replies. “It's a perfectly cromulent word.”"
https://www.merriam-webster.com/wordplay/what-does-cromulent-mean#:~:text=Though%20'cromulent'%20originated%20as%20a,future%20entry%20in%20the%20dictionary.
See also https://tvtropes.org/pmwiki/pmwiki.php/Main/PerfectlyCromulentWord
“Embiggen” was actually the word that entered my vocabulary from that scene, in near-earnest at this point.
Is that the one where the schoolkids have an essay or art contest or something and the top 67 out of 68 entries will be displayed in the administration building? I loved that more than I can say.
Even if we never understand DNA or the brain, we still had a golden age.
Embiggen suffers from the fact that "enlarge" is already a common word.
And there’s no really cromulent substitute for cromulent, I suppose.
I’ve read a lot of behavioural genetics. When people cite the "missing heritability problem" I think they should add, "of course this also applies to height".
One reason for not citing it is it diminishes their "argument" (everyone knows that height is mostly genetic). I can’t think of another explanation, although they might be repeating someone else’s argument without being aware of the issue.
Stephen Hsu has a paper, "Determination of Nonlinear Genetic Architecture using Compressed Sensing" and another, "Accurate Genomic Prediction Of Human Height". Worth a read. My simplest summary, from reading these a while back - you need large sample sizes, and the relationship between accuracy of genomic prediction and sample size is not linear.
Everybody knows height is mostly genetic *except in environments in which nutritional deprivation is common*, in which case being short gets stigmatised in ways different from the way it gets stigmatised elsewhere (in some countries (most of them, in the recent past), it correlates strongly with the class you were born in, because of obvious, cause-and-effect non-genetic reasons).
I don't think this is drop-down obvious. We know that Japanese height was stunted in the early 20th century. But the Japanese weren't starving to death - they just didn't have enough protein or something. But if this can happen to seemingly-advanced populations meeting their calorie requirements, how do we know that we've finally got nutrition right or that remaining variance in nutrition isn't enough to make a big difference?
...I mean, I think we actually know because of twin studies, but if we didn't have them, or doubted them, I don't think we would know through common sense alone.
Interestingly, there seems to be just as much (or perhaps even more?) sexual selection for height as for intelligence.
Is there?
Going purely off reflexive intuition, sexual selection in intelligence will lead to mating pairs that are roughly aligned: smarter people want smarter mates, dumber want dumber. It's viscerally uncomfortable for both parties for one half to be vastly more/less intelligent than the other.
Meanwhile for height: the female desire for tall men is standard, and the male desire for a woman shorter than them is as well. But it's not obvious that, say, a 5'1" female prefers a 5'6" man to a 6'6" man, nor that a 6'6" man prefers a 6'0" woman to a 5'3" woman.
Sorry, I meant there's a lot of sexual selection for height in males. For females, it doesn't seem to matter nearly as much.
> It's viscerally uncomfortable for both parties for one half to be vastly more/less intelligent than the other.
I'm not so sure about that. There's the stereotype of eg scientists picking partners early on that are their intellectual equal (eg Albert Einstein's first wive), and later when they become successful they switch to a prettier wife (eg Albert Einstein's second wife).
There's some direct sexual selection for intelligence, but also lots of indirect selection, because (at least these days) intelligence correlates heavily with success and wealth.
Barbara Tuchman's prequel to "The Guns of August," "The Proud Tower" about aristocratic Europe before the Great War, starts off with a chapter about how tall Lord Salisbury's Tory cabinet of 1895 was: about 5 or 6 inches taller than the average British man.
I don't think, however, that there are major class differences in height anymore, at least among whites and blacks. A large fraction of NBA stars came from the underclass (e.g., 6'8" 260 pound LeBron James was born into the bottom of society, which, by the way, is why I'm so impressed with his incredibly consistent career), although the latest generation tends to more bourgeois.
It's also fascinating how many NBA players have players who were also in the NBA.
Ha! This!
The number of second generation NBA players went up from 10 in 2009 to 35 in 2025.
I was struck by what you said about black people and lactose intolerance. I hadn’t heard that. I don’t know if you meant black Americans or black Africans. But on the subject of nutrition, milk for children was the holy grail for a long time. For growth and good bones. I was still getting a cup of milk, not water, with dinner when I was 12 years old.
I don’t know what to think about the importance of milk in connection with lactose intolerance.
Is milk good so lots of people still drink it despite discomfort?
Was milk pointless?
Or just valuable for easy availability of calories to those who can drink it?
Milk was important for kids when the diet of the average westerner consisted mostly of grains and fat. With the increase in meat, cheese and eggs consumption in the last century milk is not that important anymore.
Thanks
https://ourworldindata.org/grapher/per-capita-meat-usa
Meat consumption hasn't changed much over the last ~60 years, and I seem to recall Taubes and company arguing that pre-1920 consumption of meat was quite substantial as well. I'm inclined to think the push for milk consumption was mostly a dairy industry propaganda campaign (for better or worse).
My understanding is that most children tolerate lactose but become intolerant as they get older, and the European lactose-tolerance gene is significant because it makes people keep the tolerance into adulthood and beyond.
Oh, got it. I guess breast milk has to be tolerated!
It's a bit more complex than that. There are two common ways to be lactose-tolerant.
One is as you describe: when a normal human would lose the ability to make lactase, you just keep making it.
The other is finer-tuned. You keep making lactase as long as you have lactose to digest. If lactose leaves your diet, you stop making lactase, and you won't start again. This is closer to the human norm, but the point at which you stop making lactase can be delayed indefinitely as long as you keep drinking milk.
Ah, interesting. So is the famous recent European gene related to one or both of these?
After I stopped drinking milk for a period I discovered that I'm no longer lactose-tolerant but I kept consuming milk and other fresh dairy and eventually I regained my ability to digest them.
Not sure if my body started to make lactase again or it's a change in gut biome but it works and later I've seen people mentioning this method on youtube so I'm not the only one.
Most people can't digest lactose as adults. A few can; northern Europeans are the most prominent such group. But the Maasai, traditional cattle herders, are another. I understand that the Mongols had a non-genetic method of digesting lactose; their traditional diet (including a good amount of horse milk) promoted, and possibly seeded, intestinal bacteria that would digest the lactose for them. I don't know to what extent modern Mongols still do this, but it seems like it must have persisted into the modern era for it to be something we know about at all.
Milk is healthy, because that's its purpose! It's healthy even if you can't digest the sugar; it still contains protein, fat, and various bonuses. But if you can't digest the sugar, drinking it will cause unpleasant problems.
> Is milk good so lots of people still drink it despite discomfort?
I'm always surprised that milk is so widely available in China, given that over 90% of Chinese can't digest lactose. I think they view it as a health food, much as you suggest, where it's ok if you don't like drinking it, because it's health food! I've wondered whether they decided it must be good by observing that Europeans drank it.
Mongols were still drinking a heck of a lot of milk as of ~20 years ago, per this book: https://www.goodreads.com/book/show/2110698.Beyond_the_Great_Wall
My experience with Native Americans suggests that once you've experienced ice cream you'll keep eating it no matter how much it hurts. I wonder how much that sort of effect is responsible for Chinese consumption.
> Mongols were still drinking a heck of a lot of milk as of ~20 years ago, per this book
Local? Imported? Horse? Cow? Sheep? Pasteurized? Raw? Raw local horse milk would be expected to provide a lot more of whatever it is than imported UHT milk.
> once you've experienced ice cream you'll keep eating it no matter how much it hurts. I wonder how much that sort of effect is responsible for Chinese consumption.
Well, none, because ice cream and milk are different things. There are ice cream bars in China too; I'm not surprised by them.
I was motivated to check on whether ice cream contains a similar amount of lactose to milk. The answer appears to be yes. (Although there are some low-lactose brands, apparently.) I also discovered that the reason lactose-intolerant people don't have problems with yogurt isn't that the yogurt is low in lactose! It must be the bacteria in the yogurt.
(Cheese and butter, unlike ice cream and yogurt, have had their lactose processed out.)
For ice cream, something that's relevant is that the effects of lactose intolerance may be muted by other food consumed at the same time. A Chinese teacher of mine was shocked when I planned to have milk for lunch (without other food, just milk), and warned me that drinking milk on an empty stomach causes diarrhea. She was confused by my response that that wouldn't happen to me because I'm white.
So the Chinese don't appear to realize that the ability to digest milk is a racial trait. They think it's unpleasant for everyone.
The current state of this debate is that Gusev published a post called No, Intelligence Is Not Like Height ( https://theinfinitesimal.substack.com/p/no-intelligence-is-not-like-height ), but some other people say that the central finding in that post is based on a technical error that has been debunked by further research (see https://x.com/AlexTISYoung/status/1843288303325593923, search for "Our estimate of the SNP"). I haven't heard any response from Gusev and haven't dug into this further.
There have been a lot of identical twins in the NBA because height is so important to basketball success, but very few identical twins in major league baseball because baseball seems to be largely a subtle knack. Or something.
Nobody seems to know why there have been so few identical twins in major league baseball.
Except for Jose and Ozzie Canseco, which is likely due to Jose being "the Typhoid Mary of steroids" as a baseball agent told me over 30 years ago.
I am now envisioning Jose Canseco wandering around making everyone around him huge by contact, never feeling the effects of steroids himself...
The Onion has a lot of good Jose Canseco material:
https://theonion.com/baseball-hall-of-fame-getting-depraved-urge-to-induct-j-1819574099/
Gusev is being disingenuous. Even if intelligence is not like height, the fact that polygenic scores *can't even fully explain the known variance in height* should make everyone very suspicious of them.
(I made this reply comment above): I thought of nonlinearity too. The twin studies compare the effect of different sets of genes. The DNA studies are predictions using different individual genes as variables. Are these linear predictions? The post said interaction terms don't explain anything. How about polynomials of the individual genes ? This is easy to do.
I was going to mention Height too, but it seems there *isnt* much missing DNA explanatory power there-- the post has a bar graph showing heritability of about 10 biological traits, of which height is one. On the other hand, most of those 10 DO have the same missing DNA explanatory power as IQ.
> I was going to mention Height too, but it seems there *isnt* much missing DNA explanatory power there-- the post has a bar graph showing heritability of about 10 biological traits, of which height is one.
Looking at that bar graph, it shows the same problem for height as for everything else?
-----
There's something odd going on in Scott's essay where "heritability" and "variance explained" are often treated as synonymous. There are also times where he calls out the difference between the concepts! (Though not in those terms.)
Anyway, "variance" is an abstruse property of numerical distributions. People like working with it because, where you have a random variable [a "random variable" is a distribution, not a single value] that is composed of contributions from other random variables, you can divide up the variance of the composite random variable into contributions from each of the component ones.
(This is conceptually similar to how people like working with "least squares" because it's easier to differentiate the square function than it is to differentiate the absolute value function.)
So, variance is a concept with no clear meaning, but it *is* clear how much of the variance you're looking at is in some sense due to particular other sources.
"Heritability" is the percentage of variance in a trait that is explained, in this technical sense, by variance in genetics among the population of interest. Twin studies aim to measure this quantity directly.
But GWAS studies don't. They aim to find sections of the genome that affect the trait. Once you've concluded the GWAS, you can do a calculation of how much variance in the trait is explained by the regions you found. But that's not an estimate of "heritability".
You could also work directly with the genome data you're using for the GWAS, and do a calculation of heritability based on all genetic variation within your data. That would be an attempt to measure "heritability", but it's not compatible with the description Scott provides, that "the regions detected by the GWAS don't account for enough variance".
I was going to leave a comment observing that the empirical heritability from twin studies and the problem of missing heritability from GWAS studies are conceptually unlike in the same way that the concept of a "gene" as an abstract unit of inheritance and the concept of a "gene" as a region of DNA that codes for a protein are conceptually unlike, but then I clicked through to the East Hunter article and discovered that that's what it's about.
When you say "height is mainly genetic" you're making a lot of environmental assumptions. The average height of the Japanese increased dramatically after WWII.
It's well established that height is mainly genetic. Think of it as a ceiling for possible height, even if environmental effects can lead you to not reach that genetic ceiling. See also the difference in average height between some ethnic groups that both have adequate nutrition.
All %heritability statements encode a lot of assumptions about both environment and population. Consider for example in a study of pug dogs, what percentage of height variation is hereditary versus environmental. (Pugs are all genetically nearly-identical so the environmental percentage is near 100%.) Now do the same study but also include some Great Danes, suddenly you've got way more heritability. Dynomight wrote a good article on this stuff in general: https://dynomight.net/heritability/
In answer to a number of replies..
I didn’t want to write a long involved comment..
For a similar environment, height is mainly genetic. The heritability of height is about 90%.
Heritability estimates attempt to break out the variance from genes, the shared environment and the unique environment.
Back around a decade ago, the "heritability gap" of height (difference between the estimate from twin studies, adoption studies, etc and the genomic estimate) was quite large. But it was smaller than that for intelligence. Better approaches, with much more data - see the Stephen Hsu papers - have reduced this gap in the last ten years.
Back then, when critics of behavioural genetics were writing about the "heritability gap" of intelligence, I didn’t find one that mentioned the same gap for height.
Because genetics isn’t popular, there are lots of critics. I did my best to find the best critics with the strongest arguments.
Here they are:
1. The heritability gap (no mention of height of course, because that would weaken their argument).
2. The "false assumption" of the "equal shared environment" in the ACE calculation (calculating the variance from genetics, shared environment, unique environment). No mention of 60+ studies on this because that would weaken their argument. And see "The Genome Factor" by Dalton Conley and Jason Fletchers. They also thought this was a naive assumption until they tested it in a unique way.
3. Remember the Nazis.
I didn't see any mention of the gut microbiome! It's not genetic, but it is heritable, especially from the mother (there are special mechanisms for carrying gut microbiota up into the breastmilk). Seems like a good place to look.
(For anyone interested, researcher Stephen Skolnick has an excellent blog on this kind of thing: https://stephenskolnick.substack.com/)
I think this wouldn't explain "missing heritability" per se, because it means "twin study heritability that is missing in molecular methods". But since gut microbiome is inherited equally between MZ and DZ twins, it wouldn't show up as heritable in twin studies. So it can't explain why twin studies find more heritability than molecular methods.
Ah, right, thanks for the clarification.
Gut microbiome is genetically influenced, as I noted in another post in this thread. MZ twins have more similar gut microbiomes than DZ twins, and increasing age increases divergence.
A difference in genetically mediated capacity to maintain an inherited gut biome could potentially explain observed outcomes very well.
> But since gut microbiome is inherited equally between MZ and DZ twins, it wouldn't show up as heritable in twin studies.
Well, there are different things that might theoretically happen.
If any two infants nursing from the same mother get the same gut flora, then there is no within-family variation and the concept of heritability isn't even defined.
On the other hand, if they're donated the same microbes, but those microbes flourish differently within different siblings for genetic reasons, then gut flora might show up as heritable.
Beat me to it! Only yesterday he posted an interesting article on how certain species of gut bacteria could release compounds which bring on or worsen symptoms of schizophrenia:
https://stephenskolnick.substack.com/p/schizophrenia
> It's not genetic, but it is heritable
You are using "heritable" in a sense that is different from what it means in Scott's essay, and in genetics in general. Heritability is a measurement of how much variation in phenotype can be attributed to variation in genotype. If something is not attributed to variation in genotype, it is not heritable.
(In a technical sense, it most likely still is; a heritability calculation on zip code will show that heritability is high. You have to seek out a population to demonstrate that this is a spurious finding, which is easy to do since we know how zip code is really determined. But I don't think that's the kind of thing you meant?)
Sure, I meant heritable in the ordinary, non-statistical sense of, prone to be inherited.
I didn't mean to say you were wrong, but it's important to be aware that the essay means something different.
EEA criticisms are pretty weak, IMO. For one MZ and DZ twins really do have equally similar environments when it comes to all the things that environmentalists have been saying are the real cause of variation in cognitive traits: Parental income, wealth, and education, books in the home, neighborhood, schools, etc.
But also, if subtle excess variation in DZ twins' environments relative to MZ twins' really has such large effects, we'd expect to see sizable correlations among adopted siblings. It's hard to believe both that adoption studies don't work because range restriction makes adoptive households too similar to one another, and also that twin studies don't work between DZ twins' environments are too dissimilar.
This reminds me that I don't have a good sense of where Gusev and Turkheimer would say the missing variance is.
I think the strongest argument would be that it's non-shared environment - ie basically random, probably having to do with weird quirks of embryology - but I don't know if that's actually what they think. I agree that books and education are less plausible.
Gusev said on TwiX a week or two ago that it's obviously education. That really shattered the illusion that he might actually have a point.
Yeah, Aporia did a pretty good takedown on the subject. The missing heritability in GWAS is a bit of a puzzle, alright, but the problems of missing environmentality are worse.
https://www.aporiamagazine.com/p/sasha-gusev-is-wrong
AFAIK Turkheimer also has stated repeatedly that high heritability would be disastrous for society, as it would vindicate the bad, old, obsolete theories of "bad apples", scientific racism, class-based societies etc.
This is something I've noticed repeatedly now. I'm working in genetics myself and have some contact with social scientists both professionally and privately (my wife, for one, studied psychology - thankfully she broadly is on my side), and not only is the definition of hereditarians often hilariously lopsided (especially for IQ, it's common to be called hereditarian as long as you consider genetics non-negligible), the anti-hereditarians almost always sooner or later start talking about how terrible it would be if IQ was genetic, how it's clearly part of a push to undermine education (by whom, for what purpose?), or how it's just a psychological defense mechanism by elitist, .... On the other hand, the hereditarians at most mention the possible positive applications such as improving IVF for couples struggling to conceive, but otherwise stick to scientific arguments.
Unlike most people who think about heredity, race, and IQ, I'm a sports fan. So I have a hard time worrying too much about Turkheimer's concern that if it turned out to be true that IQ tends to vary genetically by race, that would be the worst thing imaginable. Instead, it's pretty obvious from my 60 years of watching sports on TV that athletic abilities tend to vary by race, and yet ... we seem to be dealing with this revelation pretty well.
For example, in the NFL, only whites of the fleetest lineages seem to succeed at running back, such as Christian McCaffrey, whose father Ed was an All-Pro wide receiver and whose maternal grandfather, Dr. Dave Sime, won the silver medal in the 1960 Olympic 100 meter dash.
And yet, football fans (perhaps the broadest demographic in modern America) seem pretty cool with that.
Similarly, the NBA draft yesterday picked the first white American #1 draft pick, Cooper Flagg, since Kurt Benson in 1977. I wouldn't be surprised if the NBA would be 10% more popular if it weren't so racially disparate. But, still, it's awfully popular despite it's enormous racial tilt.
One of the striking things about the NBA is the degree to which athletes with different body types can be successful. I suspect that Victor Wembayama, Nicola Jokic, Steph Curry, and TJ McConnell all owe their basketball success mostly to their genes, just to different genes and groups of genes. Could EA be similar? Have any GWAS studies looked at specific populations as defined by ancestry?
Soccer really stands out for different body types. The 6'2" Ronaldo looks like a guy who could have been a quarterback, a tennis player, a pitcher, a golfer, a miler, anything, while short-legged Messi and Maradona don't look like conventional athletes.
>... and yet ... we seem to be dealing with this revelation pretty well.
If socioeconomic status depended on athletic ability as strongly as it does on IQ, we would find those particular racial disparities much harder to deal with.
Imagine high schoolers doing a 100 meter dash but with the stakes of the SAT. There would absolutely be resentment and wishful thinking.
I think many of Scott's commenters underestimate how much American high schoolers care about sports.
I know the portion of variance explained by education is tiny, but how tiny? What are the percentages explainable by education, income, parental education, etc.? THere must be some best estimate for each, even if it's statistically insignificant.
Last time I read Gusev, I came away with the impression that it should be interaction effects, as he argues that molecular genetic methods control for those. But I'm no expert on the topic, and maybe I'm misremembering.
Edit: Source here:
https://theinfinitesimal.substack.com/p/gene-environment-interactions-ubiquitous
under "Twin estimates of G are inflated by GxE"
Edit 2: He goes into more depth on the issue in this interview:
https://www.psychiatrymargins.com/cp/153126659
Extract: What could explain the gap? All of the twin study biases I described above are likely at play to some extent, but my personal view is that interactions between genes and the shared/familial environment play a major role. I alluded earlier to a study of gene-environment interactions for BMI; there is also striking evidence for an interaction between educational attainment and socioeconomic status (SES), where the heritability drops substantially for individuals in high SES versus low SES environments (curiously, the opposite has been observed in twin studies). My guess is that hundreds or even thousands of interacting environments accumulate over a person’s lifetime, many of which are not even measurable. When we then study unrelated individuals in GWAS, the participants are experiencing all of these diverse environments, and so genetic variation has a weak influence on their outcomes. But when we look at twins, who share their rearing environment extremely closely (are literally born at the same time), all of these interactions get assigned to and inflate the genetics/heritability bucket.
Edit 3: After having read into his stuff a bit again, my main question is if it actually possible to seperate direct effects and interaction effects like that. It seem unintuitive to me that the two are independent of each other, but I lack the statistical chops to have anything beyond an intuition on that.
Regarding your third edit, I think that's the main problem. In twin studies and (probably) in sibling regression we are getting the sum of interaction and direct linear effects. In GWAS and GREML and RDR we are getting the direct linear effect of some subset of the genome. The gap between RDR and GWAS can be explained by accounting for more genes, but the gap from there to sibling regression has to be some combination of environment, gene-environment interaction or gene-gene interaction. We can elimination direct linear effects of environment to some extent, but how do we separate gene-environment from gene-gene interaction? That's what everyone is struggling with.
I don't know anythig about RDR and sibling regression beyond what Scott and Gusev wrote, but as I understand it, both are methods for seperating direct from indirect (Scott says nondirect) effects.
Interaction effects are, in my understanding, seperable from that. At least Gusev seems to claim that direct effects can still be confounded by environment. But also, Gusev does seem to argue that there are ways of estimating GxE effects.
That is an important claim, but what I was getting at more is that seperating direct and nondirect effects should only work if they are independent. As I understand it, it's to control for the effects of parental genes creating environments with effects. But what if the same gene has a direct effect in the child and a nondirect effect in the parent? Seems to me like there is a potential danger of controlling away part of the effect.
Yes, I struggled with this when reading "Intelligence is not like height". I think Gusev really wants the missing variation to be in gene-environment interactions, but there doesn't seem to be a lot of evidence for this. The best conclusion I could get to from reading his essay is that its probably in gene-gene interactions, that mean identical twins maximize exactly the same set of gene-gene interactions so we see a non-linear drop-off in the impact of exactly the same genes in people who don't have identical genomes. Its nice to see this is one of the possibilities consistent with all the data even when viewed from "the other side".
Yeah this is why, although I don’t pretend to understand all the arguments inside and out, big picture… The weight of evidence seems to be solidly with hereditarianism. Sure, some of these recent findings are puzzling… But to actually shake my confidence you’d have to be able to tell me what these environmental factors are and provide evidence for them. And also explain how by staggering coincidence the results end up being what you would expect if hereditarianism were true. Without evoking all sorts of convoluted post hoc rationalizations about why the same cause can have opposite effects when it’s convenient for your argument etc.
I mean handedness is like this no? Most people think handedness is highly heritable (and it runs in families), but twin studies show very low heritability. Sometimes it just comes down to noisy variables that are hard to measure
Note that, although that the story you linked doesn't mention it, the study that found people with Norman names overrepresented at Oxford was by Greg Clark, who argues that very long-run persistence of social status is driven by genetics and highly assortative mating (0.8 on a latent measure of socioeconomic competence). So in this case it wouldn't be necessarily be confounding, because persistent differences between people with Norman and Saxon names might well be genetic.
The idea that truly exogenous economic endowments can persist over several generations has pretty weak empirical support, as I believe you mentioned in a post several years ago. Studies finding otherwise are likely confounded by genetics.
But there was also a St. Louis Fed study finding that the intergenerational correlation for residual wealth (i.e. wealth greater than or less than that predicted by earnings) is only about 0.2. Except perhaps in the most extreme cases, wealth has to be actively maintained and fortified by each successive generation, or even very wealthy families fall off the map within a few generations. Consider that less than a century after his death, there are no longer any Rockefellers on the Forbes 400 list.
Wouldn't your theory require that Normans started out smarter than Saxons? I'm not claiming they can't possibly be, just that I can't really figure out a reason to think this, and it would make more sense that the persistent Norman-Saxon difference is because Normans have always been the nobility (and therefore had more money) since the Norman Conquest.
What if social status gives you access to higher-quality mates? You're rich, so you marry other rich people, some of whom got rich through extraordinary ability. Through this mechanism, it seems like it should be possible to consolidate temporary social status into a permanent genetic advantage.
Well, in a patriarchal society like Medieval England, it's more like you marry women whose fathers got rich/powerful through extraordinary ability.
Note that this state of affairs is particularly favorable to a distinction between French-surnamed and English-surnamed people, because women don't pass on their surnames.
Mr. Darcy, with his sexy Norman name, marries the smart Elizabeth Bennett in "Pride and Prejudice."
His great-great-great granddaughter, Marcy, did not marry as well.
https://marriedwithchildren.fandom.com/wiki/Marcy_D%27Arcy
Wouldn't it be Jefferson who was the great-great-great grandchild? D'Arcy wasn't her maiden name.
Oh, you're right. I don't remember the show very well, as I was pretty young when it was on the air. I just assumed that the surname that rhymed with her given name would have been her maiden name.
I always assumed this was part of the logic of noble birth. Old bloodlines with an accomplished past bring in new blood with an accomplished present, incentivied by the advantages of noble family, hopefully leading to stable excellence that adapts over time at a moderate rate.
Primary sources rarely talk about these things outright, but religion be damned, these people bred every animal they could for trait selection, and they knew what they were.
The Churchill lineage produced the great Duke of Marlborough, victor over Louis XIV's army at Blenheim, and his famous Duchess. Then they didn't do much for a couple of centuries, then many generations later they produced the brilliant but unsuccessful Randolph Churchill and the highly successful Winston Churchill.
What should we learn from this?
Beats me.
"Wouldn't your theory require that Normans started out smarter than Saxons? "
Could be that smarter Saxons preferentially took on Norman surnames, through marriage or otherwise?
But also, maybe selective immigration after the Norman conquest? Norman elites move in, but Norman peasants remain in Normandy.
Yes- this, exactly. The Normans who moved to England as a transplanted aristocracy were not the median Norman, and any genetic distinction could have been fortified in subsequent generations by strategic marriages and selection effects.
My name Sailer is usually spelled Seiler (ropemaker), but an ancestor became mayor of a small town in Switzerland and decided his lineage deserved a classier surname (rather like more ambitious Smiths in Britain tended to change their names to Smythe or even Psmith).
Gregory Clark found that Smythes are more likely to graduate from Oxford or Cambridge than Smiths. Presumably they were all Smiths at one point, but the people who changed their names to Smythe were obvious social climbers.
Yet, social climbing seems to work.
Nobles often inherited multiple names and titles so they had the option to choose which names and titles they used more prominently. Prestigious names and titles were kept in use even when the direct heirs went extinct while common ones were not.
Normans being by far the most effective in Europe and the Mediterranean in the medieval period seems some evidence there was something special about them. The Normans managed to kind of independently become the military elite in England, Ireland, Southern Italy and other places. Military skill is quite g loaded at least according to modern studies.
The Normans got all the way to the New World in 1000.
Are you referring to Leif Erikson? He wasn't from Normandy.
I think that was a different batch of Vikings.
The problem with proposing that artificial social status persists after so many centuries, is that study after finds that extreme exogenous shocks to social hierarchies, including those specifically intended to disrupt them, have relatively low long-term effects on population hierarchies.
That strongly suggests that status persistence is due to the underlying genes that caused the status in the first place, not due to an inertial effect of status itself.
E.g. areas in the Soviet Union near former gulags have higher GDPs, since former elites were disproportionately sent there. (https://cepr.org/voxeu/columns/enemies-people, https://x.com/cremieuxrecueil/status/1720885228514820125).
Similarly, in the USA, slavery represented a massive exogenous pressure on social status, but looking within counties at gaps in income and other status indicators between descendants of free vs. enslaved Blacks, shows that by 1940, the impacts of slavery had basically evaporated (https://archive.ph/PGwtu, https://www.cremieux.xyz/p/black-economic-progress-after-slavery).
Similarly (referencing Clark), in the Indian subcontinent, while caste hierarchies generally persist, artificial castes imposed by the British, didn't; which is what you'd expect, since they were arbitrary, representing exogenous shock.
Similarly, the Samurai had their status advantage actively eliminated through war, but their descendants ended up reverting to once again end up on top.
(https://www.cremieux.xyz/p/black-economic-progress-after-slavery which discusses additional groups).
The same was the case for Jews. Before the Holocaust, they tended to do very well economically relative to local non-Jews. Then they faced widespread genocidal discrimination, but within just a couple of generations, their descendants ended up once again, far outpacing their non-Jewish neighbors.
Similarly, variation in bombing by region during the Vietnam War wasn't correlated to status by region by 2002. By then, the massive exogenous shock of one of the largest bombing campaigns in history, had totally evaporated. (http://emiguel.econ.berkeley.edu/assets/miguel_research/18/_Paper__The_Long-Run_Impact_of_Bombing_Vietnam.pdf).
I mostly agree with you here, but it seems like the Normans might be a special case, since they were literal titled nobility and in most cases you can just check and see that yes, the noble title remained within the family line from 1066 to the present.
I don't really understand the statistical analysis, but this paper (https://pubs.aeaweb.org/doi/pdfplus/10.1257/aer.20191422) examines the rebound of southern slaveholding families post US Civil War, and claims to have explicitly tested and rejected the hypothesis that status persistence was due to inherited ability, and that patronage networks are a better model.
I believe there was also a paper on the persistence of Chinese elites post-Cultural Revolution that made similar claims but I can't find it at the moment.
Status rebound among descendants of Chinese elites after (roughly the period of) the Cultural Revolution is discussed in the Cremieux piece: https://www.cremieux.xyz/p/black-economic-progress-after-slavery, citing https://www.nber.org/papers/w27053.
Yes this was the paper. IIRC the authors claimed the success of elite-children post-revolution was mediated by work ethic, and that the elite advantage in work ethic over non-elite children disappeared among elite children whose parents had died when they were young, suggesting that it was largely cultural transmission rather than genes. I could be misremembering though.
It requires assuming that the Norman upper classes were smarter than the Saxons averaged for all classes, which is more plausible.
Norman aristocracy came over in boats and killed the upper crust of Anglo-Saxon society, so probably quite a big difference. I'm not really qualified to say, but I'm doubtful that much of that difference is preserved till today (and captured in surnames). Assortative mating might do it, but assortative mating might also have created these same class differences even if there hadn't been an invasion.
The Norman conquerors were a highly-selected hierarchical aristocracy. The invading army at Hastings was not composed of random farmers. Most were second and third sons from noble families (due to primogeniture), i.e. highly ambitious men with elite backgrounds but no inheritance. So you had a population bottleneck of highly-selected elites that became the top 1% of the English population and then maintained fairly consistent ingroup marriage patterns for centuries. I think there's every reason to believe that British Normans have a better genetic heritage.
The Normans were also more Germanic than the general British population, and across Europe there's a strong correlation between Germanic descent and economic development.
normans started out smarter cuz it wasnt the norman peasantry that invaded Britain but strong healthy men from the warrior class. When the founding population is above average the resulting people are (taiwan ,japan, england, are examples of this, you may even consider the founding populations of Eurasia that left Africa). Just like when Spain invaded South America the conquistadors were all of roughly the mercantile class well above the catholic peasantry. All the Spanish ancestry in Latin America, and all of the Norman ancestry in Britain is from above-average groups of those respective populations. And when they took Native and Saxon wives they took them from the best families ,so their brilliance was never diluted.
the normans that invaded were an above-average subgroup of normans. it wasnt a respresentative sample of normans, which may indeed not be smarter than the saxons. when the founding population is above average, the resulting people are (see taiwan and japan and some other island countries, or you can even consider the founding populations of eurasia that left africa). this exact thing happened in the new world, where above-average spanish (conquistadors who were from the merchantile class) conquered latin america. when they took native and saxon wives they took them from the best famlies, so their above-averageness never diminished over generations.
p.s. i couldn't find your reply to me about twin studies being needed to disentangle nature vs nurture ( i only see it in my email, im not sure why). the only thing i would add is i dont think the villagers would put much stock in things like parenting. they would see parenting as folllowing from the mediation between the innate qualities of the parent and the child who are closely linked. not separate from the hereditary principle, but something that follows from it. i suppose it would be possible a high quality person to be a poor quality parent (if they became alcoholic ?) , or that a good-natured child could prove rebellious but these would be notable exceptions.
I don't know about lately, but when I last checked about 10 or 15 years ago, there were still four Hearsts on the Forbes 400. They trace their fortune to the Nevada silver rush of 1859, although they hit it rich again in mining strikes in the late 19th Century as well.
But ... most really rich guys on the Forbes 400 were semi-self made like Bill Gates, Jeff Bezos, or Elon Musk. They enjoyed well above average upbringings but then took huge advantage relative to other upper middle class individuals of their generation.
The US might also just be unusual in this regard. I once (peripherally) knew some members of the Fugger family. They are not crazy-rich but still wealthy enough that most of their energy seems to be dedicated to resolving within-family inheritance squabbles.
I wonder if there is a within-family version of the resource curse going on there. If you are extremely able and you come from a family with a big fortune, maybe there's little you can do that is as likely to pay off as well as to make sure to inherit the big fortune and manage it and the family's reputation well. Whereas if you don't come from vast wealth, maybe striking out on your own and starting a business is a better bet.
Thomas Piketty is always complaining about all the Hidden Old Wealth. He must be referring to Europe, because America doesn't seem to have many, say, personal golf courses of obscure ownership. I can name the owners of most of the personal golf courses in Southern California, such as Larry Ellison.
I think it's a mistake to treat possession of surname as meaning genetic similarity, because only legtiimate male children would pass on a surname. And the inheritance system of the time favored them significantly.
Medieval nobles frequently had a lot of illegitimate children, and half their children would have been girls who lose the surname, and so there are likely lots of people in the population with Saxon or other surnames who have more of a genetic link to the Normans than the people with the surname.
But being a legitimate male child of an aristocrat would give significant material advantages
Thanks for such a wonderful write-up. One thing I’d love to see is more of a deep dive into the stats behind this. This seems like a very high dimensionality problem even without interactions, and I don’t see why interactions wouldn’t be as important as well. I could easily see k > n even with an n of three million, and it isn’t like k=n is a good situation. How the hell are they estimating this? LASSO? Something non-parametric? As an outsider, given the difficulty of the problem they’re facing, it seems like a no-brainier that regressions they’re running don’t add up.
Yeah, I am very bad at stats and not the right person to write this. Both Sasha Gusev and Alex Young are much better, and you might enjoy either of their work - https://x.com/SashaGusevPosts, https://theinfinitesimal.substack.com/, https://x.com/AlexTISYoung/, https://alextisyoung.github.io/
Yes, LASSO.
It's not that hard.
If a trait is influenced a tiny bit by a ton of genes, wouldn't one expect LASSO to (over)aggressively filter most of them out?
You have to tune it correctly, but, no, what I expect is that you keep the variants whose effects you really know and you filter out the variants you can't distinguish from noise. I guess you could try ridge if you want to estimate heritability and don't care about having faith in the list of variants, but I'm sure people have tried it and I imagine I haven't heard of it because it doesn't give different results.
I would still suspect that there would be so many very small contributors that almost all of them could easily be mistaken for noise. Genomics just doesn't seem hospitable for methods used in less-complex domains.
ETA: in most regression studies, a reasonable prior is that most things *don't* have a meaningful effect on the outcome of interest, so it makes sense to try to pick the needles out of the haystack. In genomics, OTOH, it now seems a much more reasonable prior to expect that almost everything has a similarly-small-but-real effect on any given trait, in which case needle-picking throws the baby out with the bathwater (excuse the mixed metaphors).
Maybe you should think of GCTA, which finds most of the missing heritability, as ridge regression, but I think it's just that they focus on a small population so that each rare variant gets observed more.
Sounds like you don’t know LASSO
For this "Alex Young thinks that once we get enough whole genomes sequenced (probably soon!) we might be able to use a technique called GREML-WGS to get more definitive answers about rare variants" is there any reason to think that even a sample size of 8 billion would be big enough to detect rare variant effects?
I don't know for sure, but right now I think the best whole genome sample sizes are in the 6 digits, so there's a lot of room to go up!
My guess is that yes, it requires a dramatically larger sample size to estimate interaction effects (at least, assuming that they are individually small, like direct individual SNP effects appear to be).
The reason for this is that in a regression of something like
y = beta_0 + beta_1 x_1 + beta_2 x_2 + beta_{12} x_1 x_2 + e
you often/typically end up with a lot less identifying variation in x_1 * x_2 (that is, variation in x_1 * x_2 net of a linear function of x_1 and x_2) than the amount of identifying variation in x1 in the original linear regression of
y = beta+0 + beta_1 x1 + beta_2 x2 + e
Now, add in the fact that you are testing some insanely large number of SNPs (hundreds of thousands? Millions? Icr).
The number of potential (simple) interaction terms scales quadratically, and because of this, to avoid false positives dominating true effects you need an exceptionally low type-1 error rate.
The combination of these two things requires extremely large samples.
I would note though...
Rare genes/mutations that might not be tested in standard DNA tests / GWAS (if I'm understanding this correctly) might theoretically explain a bunch of the missing variance, idk.
But it seems ~ impossible for interactions of fairly uncommon mutations to explain much of anything. Why not? Well, suppose they did. Suppose you need the rare mutation on gene 234 and 1267 to get a bump. Suppose you have this, and thus the bump (and most of your "true" polygenic score is similar interaction effects).
Will your kids have most of these interactions? No, they shouldn't. Isn't it the case that they have a 50% chance of get your rare mutation on gene 234 and 1267, and thus only a 25% chance of getting both? And it's rare, remember, they're extremely unlikely to pick up "the other bit" from your spouse.
So then you should have drastically lower heritability.
So if we already "know" that heritability is high, and believe it is mostly genetic, "rare interactions" won't help us. Rare single genes might. Interactions of common things might. But not interaction of rare things.
From an apriori point of view we should expect the relationship between genes and observed features to be incredibly complicated -- basically as complicated as the relationship between computer code and program operation -- and I'd argue the evidence we see is exactly what that model of highly complicated interactions predicts. Namely GWAS misses a bunch of twin study variation and so do simple non-linear models. Or to put the point differently the answer is the narrow/broad gap but broad hereditary is just really damn complicated.
Yes, people cite papers for the claim that non-linear effects don't seem to make the difference. But if you dig in the supposed evidence against non-linear effects (like you linked) are really only evidence against other very simple elaborations of the linear model. Showing that you don't predict much more of the variance by adding some simple non-linearity (eg dominance effects at a loci or quadratic terms etc) is exactly what you would expect if most effects are extremely complicated and neither model is even close to the complete causal story.
I mean, imagine that some OSS project like the Linux kernel gets bifrucated into a bunch of national variants which are developed independently with occasional merges between individual nation's versions (basically it acts like genes under recombination). Or better yet a Turing complete evolutionary programming experiment with complex behavior. Indeed this later one can be literally tested if people want.
We know in this case that code explains 100% of program variation and if you did the equivalent of a GWAS against it you would probably find some amount of correlations. I mean some perf improvements/behavior will be rarely discovered so they will be highly predicted by whether some specific code strings show up (they are all downstream of the initial mutation event) and other things like using certain approaches will have non-trivial correlation with certain keywords in the code.
But no linear model would actually even get close to capturing the true (indeed 100%) impact of code on those performance measurements. And it would look like there were no non-linear effects (in the sense of the papers claiming this in genetics) because adding some really simple addition to the linear model like "dominance loci" or whatever isn't going to do much better because you aren't anywhere close to the true causal model.
So the evidence we have is exactly what we should expect a priori -- genes have some really complicated (essentially Turing complete) relationship with observed behavior and simple models are going to guess at some of that but linear models will do about as well as simple non-linear ones until you break through some barrier.
I think certain twin/sibling studies have themselves been used to try and estimate the relative size of additive vs. non-additive genetic effects, and additive effects were maybe twice as large, though. Non-additive effects are also harder for natural selection to act on, which makes rapid evolution in a trait less likely.
Yes, but the problem is there are 2 types of studies that estimate non-additive effects.
1) Twin study type studies. Well that's exactly the high result which causes the gap.
2) Papers that test for non-additivity by elaborating the linear model in some way to allow for some simple non-linear effect. For instance, capturing the idea that maybe a gene can block the effect of other genes at the same loci.
But 2 is only evidence against the kind of non-linearity presumed in that model (more generally simple non-linearity). My point is that if the true relationship isn't something like mostly additive but maybe with some quadratic terms or multiplicative one but is actually more like a Turing complete interaction then you should expect that adding a bit of non-linearity doesn't improve the prediction much.
So as I said, we see exactly what we should expect consistent with the hypothesis of very complex genetic interaction s.
I'm probably not as immersed in the literature as you are, though I do work in the field of software so... I think I can partially intuit what you mean by non-linear program flow, although I think biological systems are both messier and generally include more built-in redundancy. However, I can certainly see the argument that complex interacting biological systems (such as brains, or even cellular metabolism) must have complex genetic correlates.
However, it's also the case that tiny regions of the genome can have counter-intuitively huge impacts on phenotype (e.g, humans vs. chimpanzees), so... I wouldn't assume in advance what percentage of gene-variants accounted for non-linear gene interactions, or what percentage of phenotypic variance those accounted for.
the hereditarians (the fact they call themselves this is funny, all geneticists are hereditarian) seem to have a weird idea that there's like a full genetic shuffle going on and evolution likes low-variance smooth fitness landscape. both are untrue afaict.
not a lot of shuffling is happening (1-3 crossovers per chromosome pair during human meiosis from memory) and with recombination hotspots the shuffle is somewhat predictable. so a lot of deep structure can be preserved and it's probably beneficial to have deep structure (you don't want an organism which rapidly loses genetic diversity to hit a local maximum and then all dies when the environment changes, similarly you don't want a function that's too full of 'holes' which the organism simply gets stuck in while maintaining diversity - you want an explorable and non-smooth fitness landscape).
note of course all these 'meta-parameters' like where the hotspots are, are themselves subject to selection which makes it harder to reason about.
i tend to agree with OP we should expect really complex interactions. the metaphor I'd use is that it's like we're analysing books by only doing word-counts. except in super high dimensional spaces it's often hard to intuit just how effective this is while still being completely wrong: for instance we can predict a lot from non-functional non-coding DNA simply because statistically it'll be colocated with genes that do matter. if you can make good predictions from junk DNA it's hard surprising you can do really well while wrongly assuming additivity.
i don't know much about the field but the impression i get is that the molecular geneticists largely take it for granted there's complex interactions because they've found really cool stuff in simple organisms. meanwhile you have more classical quantitative geneticists who have barely progressed since Fisher who are so far removed from the cutting edge that they barely interact with their much smarter brethren.
>the fact they call themselves this is funny, all geneticists are hereditarian
So one would think; and yet....
>hereditarians [...] seem to have a weird idea that there's like a full genetic shuffle going on and evolution likes low-variance smooth fitness landscape. both are untrue afaict [...] i tend to agree with OP we should expect really complex interactions
This is /better/ for the hereditarian position, not worse—insofar as complex, hard-to-pin-down interactions suggest that the "missing heritability" issue will be resolved with revisions upward rather than downward; I mean, not that you explicitly contradicted this, but the sense of the initial paragraph there seemed t'me to imply you felt this was a /problem/ for a "hereditarian" approach. (Or, possibly, I am confused and actually hereditarians DO hate this One Weird Genomic Trick?–)
It will be resolved in between the twin studies (which inflate heritability due to ACE assumption) and current GWAS etc.
Obviously genes will still be incredibly important but we might move away from 'parenting doesn't matter at all' which is close to where you get with twin studies.
"but we might move away from 'parenting doesn't matter at all' which is close to where you get with twin studies."
If parenting doesn't matter, I've always wondered why people evolved to feel the urge to act like parenting matters.
they might argue parental investment mattered more when we faced chronic food scarcity and were being predated upon by sabre-toothed cats, but we have no twin studies from back then. heritability estimates are always for a given environment (ironically leftist attempts to level childhood environments, by e.g. ensuring no-one receives a good education, have increased the percentage of variance explained by genetics.)
This is true AT THE MARGIN for first-world middle-class parenting, but I'm pretty sure it's not true when you go far outside that range. Raise your kid in a cave and never show him a book, beat him with a stick regularly, etc., and you're well outside the data provided by adoption studies or twin studies.
> "the hereditarians (the fact they call themselves this is funny, all geneticists are hereditarian)"
Well... yes, but that only makes the environmentalist position more indefensible- it would literally be impossible for intelligence (or any other biologically-dependant human trait) to have evolved in the first place if it wasn't genetically influenced. So are "their much smarter brethren" deluded, or lying?
The question is about variability among humans. It's quite possible to have genetically determined traits within the species or some population where very little of the variability we observe is due to genes. The easy example is language--humans clearly evolved a bunch of mental machinery for handling language, that's got to be genetic, and yet nearly all the variability we see in what language people use is environmental, not genetic, and people seem quite capable of learning languages far from their ancestral lands--adopt a Chinese baby and raise him in Texas and he'll speak American English with a Texas accent.
Yes, but paradoxically, verbal IQ and vocabulary size are one of the most strongly genetically-influenced aspects of intelligence. The content of language is enormously culturally influenced, our aptitude for it is not.
Raw scores on IQ tests went up on average during the 20th Century: the Flynn Effect.
My guess is that one reason is because cultures evolved in the directions anticipated by early IQ researchers: e.g., Stanford professor Lewis Terman, who invented the first American IQ test in 1916, the Stanford-Binet, tended to make up questions that reflected the views of academics and professionals in Palo Alto about what constituted intelligence.
And 109 years later, those Palo Alto biases about intelligence appear to be right.
Palo Alto more or less conquered the world, with Lewis's son Fred Terman playing perhaps the central role in the creation of Silicon Valley as we know it.
TBF people can only do what is possible with the available tools and data. Fitting a linear model to genes is computationally possible. Even assuming there might be pairwise interactions that aren't limited to very nearby genes becomes very difficult to do from a computational and data gathering POV.
The problem isn't the work being done by scientists in this area but the people who want to argue that there can't be strong genetic effects that account for huge amounts of variance that are so far just beyond us.
Unfortunately, I fear that what you really need to make progress here is absolutely massive data sets that include substantial numbers of related individuals plus data about IQ educational attainment etc. The kind of massive everyone's genetic info database that makes everyone really nervous. That way you can start looking for the smallest changes that make big differences.
> But from an apriori point of view we should expect the relationship between genes and observed features to be incredibly complicated -- basically as complicated as the relationship between computer code and program operation -- so it's weird to think linear GWAS models should capture most of the variance.
If you only look at how genes get transcribed into proteins and what proteins actually do, and how long and convoluted the chain of cause and effect is to get anything we actually observe on a macro level, then your argument makes a lot of sense. It's ever crazier and weirder than human-readable computer programs.
But: remember that genes and chromosomes get reshuffled and stitched together and slightly mutated with every generation. That means for them to work, they need to be fairly robust.
If you want to think about a computational analogy, I wouldn't pick computer code ~ program execution, but perhaps like the weights in a neural network. Especially since methods 'dropout' (aka 'dilution') that randomly disable artificial neurons seem to work quite well, and not cripple them. See https://en.wikipedia.org/wiki/Dilution_(neural_networks)
(I guess it helps that 'dropout' in neural networks was inspired by an analogy with genes.)
Thanks.
Perhaps intelligence is the single thing that most requires intelligence to figure out? Hence, I'd hardly be surprised that we haven't figured out IQ completely yet.
Yes, paradoxically, nailing down the exact causal genetic pathways behind human intelligence will probably require a superhuman machine intelligence. Partly because I suspect the genetic code is not especially well-factored, and it certainly needs a lot more comment blocks.
I don't think the analogy with computer programs is on point.
The big difference is that the genetic code is recombinable. You can take two genomes, mash them together, add a few mutations, and it's still a working genome! You can't do that with computer code.(*) This is a pretty tough restriction, and it puts some severe restrictions on the complexity of interaction. Not that we would understand exactly how those restrictions look like. It doesn't just rule out "it's complicated" either, but it's enough that I find the comparison with computer code wrong.
(*)Actually you can, it's called Genetic Programming. The idea is that you evolve code via recombination and selective pressure by a fitness function, for example, to pass as many test cases s possible. It just doesn't work.(**) Since I work in a related field, I know a couple of people who work on Genetic Programming. The issue is precisely that it's so hard to recombine programs into another working program.
(**) My colleagues may beg to differ. But I would claim that it doesn't work unless you either work with absolutely tiny programs, or move so far away from the basic idea that you are no longer evolving code, but live in a totally different "genetic space".
Works better with neural networks and some other types of AI. I'd argue that various regularization techniques (dropout) etc. are fairly good analogues to what's happening biologically with recombination and the resultant networks can have quite complex interactions (indeed it's proven that they can, in theory, compute any function irrespective of how complex it is).
Yes, agreed that this works better. There are many situations in which Evolutionary Algorithms work pretty well. But not on program code.
Well, sometimes it's still a working genome. Sometimes one little mutation breaks everything. And even for that , the two starting genomes need to be pretty similar. You can't mash one from a chimp and one from a lobster and get something that works. You can't even mash two from egg cells or sperm cells, they have to go through the correct processes on each side first.
But it does seem a bit less finangly than code execution, if only by having a (far) lower information density.
Back to the original question though, there's actually a lot of cases where interactions between genes are important. Like if a chemical is modified by a first protein and then a second protein and then a third protein, and you delete the first protein, it doesn't matter anymore what the genes are for the second protein. Or maybe you make the second twice as effective, but the third is really slow, so there's no difference without another gene upping the third protein as well.
That's a great observation. In this whole discussion, what's bugging me is that genes aren't studied in the way they actually work in order to produce phenotypes: as cascades of dynamical-system type interactions that end up with very strong attractors in aperiodic systems. If they were, then the other aspects of genes and the genetic code and of autocatalytic systems would also need to be considered: redundancy and canalizing functions for example would lead many variants to lead to the same outcome. Statistics is a blunt tool to study this kind of system. In twin studies however, you get the outcome of the entire system by default. So no wonder there is missing heritability in purely genetic association studies. It's like trying to predict the quality of the cake from the kind of eggs and flour used, when the sequence of mixing the ingredients (gene x gene interations) and the baking temperature (internal environment) mattered far more. Twin studies compare the cake, or pudding, directly, wherein lies the proof.
The point about computer programs is merely that the ways that genes can affect outcomes is essentially Turing complete -- basically as far from linear as you can get.
Yes, if you tried to randomly recombine C code it wouldn't work very well but you can maintain Turing completeness while allowing recombination to work much better as one tends to do when doing genetic programming. And yes it means it can be very hard for evolution to make big breakthroughs but that is why it can take millions of years.
And sure, most kinds of effects on intelligence aren't going to be as fiddly as say the fact you need to do transcription exactly right (which really is just a mini virtual machine in the code) but you don't need them to be that complicated not to get picked up by models that are just linear effects plus some minor tweak. The fact that we know genetic effects can be as complicated as possible is good reason to think that it's highly plausible that even in intelligence we are seeing at least effects from arbitrary pairs of genes (eg A and B together have an effect which is nothing like A + B...even when nowhere near each other on genome).
If epistatic effects were large then wouldn't that show up statistically? Meaning deviations from normality.
This is my intuition as well. And it's basically what Turkheimer thinks from what I can tell. Genetics interactions are extremely complicated and chaotic. You cannot really predict complex traits from genetics (unless you already have a clone or very close relative.) Just like you can't predict the output of a complex computer program without running it.
I'd hold out a little more hope for one day being able to make better predictions. But you'd need absolutely massive databases of sequences with tons of info about educational attainment and IQ testing that includes lots of related individuals so you can start by looking at the effects of more limited changes. And even then it's a massive project that could take generations if it is possible, even if people were cool with creating the needed database.
When I first heard about GWAS while taking a bio-informatics class in like 2011ish. The instructor, a theoretical computer scientist, just flat-out said, “the genotype to phenotype mapping is probably spectacularly non-linear and we should only expect meager and incomplete results from GWAS.” I’m impressed they’ve found as much of the heritability as they have. It’s crazy that people claim that because GWAS can’t find the genes, they must not exist. It reminds me of all of Stephen J. Gould’s sophistry in favor of ideologically preferred conclusions.
Educational attainment is a meaningless measurement, at least from the internet survey versions I've seen with categories like "some college" or "bachelor's degree". This misses a lot of education that I think also correlates with intelligence. A Journeyman electrician requires a five-year apprenticeship that includes both on the job and classroom education, where would a high school dropout who joined the IBEW and became a Journeyman electrician be classified?
Both of my parents have a bachelor's degree, I chose instead to enlist in the US Navys Nuclear Power program which at the time was compared to the most difficult programs at top universities. Am I a strike against the heritability of educational attainment since I earned a nuclear Naval Enlisted Classification instead of a degree?
Maybe current surveys could be better designed. But there's a difference between an imperfect proxy and a non-proxy when measuring a population.
Surveys of educational attainment don't classify *you* as an individual well or fairly. But they're still correlated with IQ, on the average, to the extent that general intelligence exists. Getting 60% of the answers right on a true false test is not a perfect score. But it's not a 50% either. Surveys using educational attainment can get the equivalent of a 60% correct score and still be potentially useful in telling us things about a population.
They might be confounded, I suppose, if certain forms of intelligence predicted a person pursuing trade schools rather than college education. Most surveys seem to rely on a notion of general intelligence. To the extent that college and 'advanced trades schools' represent distinct types of intelligence I could see the study being problematic. But if you're just an outlier then the format will just produce weaker results than it could.
I guess using educational attainment may be...what's the phrase, directionally correct? I don't think this tells you anything more than you could learn from dog breeds, yes, behaviors can be inherited. A retriever will retrieve from not long after birth, and a shepherd breed will herd anything that moves which is always fun when a family with one invites a bunch of their child's friends to a birthday party and the dog will do its best to keep them all together.
I would think these studies aim for more useful information to quantify the degree of hereditariness or something and EA isn't going to get there. A degree may have correlated with intelligence for a period of time since that was the smooth path to a decent paying career, but when the people spending billions of dollars on AI datacenters say the bottlenecks are electrical generation and electricians to run all the cables, then over the next ten years intelligence may be more closely corelated with people who join the IBEW after high school than with those who earned a degree.
Back in 1986 I put an ad in the Chicago Tribune looking to hire a personal computer technician. The HR person asked about academic requirements. I thought about it for 5 seconds and said, "Bachelor's degree required."
But then I got a long letter from an applicant saying, I don't have a college degree because I enlisted in the Navy's submarine nuclear reactor program out of high school, but here are ten problems you are probably dealing with and how I'd solve them for you.
I hired him, and by the afternoon of the first day, it was clear he was smarter than me. While he solved my immediate problems, he also caused me a lot of unexpected problems because because within a few months my boss's boss, who had an MIT advanced degree, was calling me up, irate, to find out why the computer repairman I'd hired was having lunch with the Chairman of the Board to discuss corporate technology strategy.
Great story! I'm just ordinary smart, but yeah, we had more than our share of ridiculously intelligent people through that program, and a few of them also had interesting problems. We had one when I was an instructor at the NY prototype who never missed a single exam question and knew all the answers but had zero life skills. He couldn't drive, bought a plane ticket to Albany airport then started walking down the highway towards Saratoga Springs with his seabag on his back. The police picked him up and drove him to the site. We had to assign him roommates who could drive him to work because he hadn't made any friends who would do it voluntarily. He slowly got a little better, but wow, I had a much easier time assisting the ones who struggled academically.
Thanks for the write-up. I suspect that seeing this kind of confusion and disagreement is a good indicator that we are about to witness some nice chunk of progress from clearing this up, soon.
Especially if (one of the) limiting factors seems to be genetic sequencing and computing power: two things that are still getting cheaper and better quickly.
I know of two secret results I'm not supposed to talk about, by people claiming they've found very large amounts of "missing heritability". Not yet peer-reviewed or confirmed by anything except rumor. I expect one to be out within six months, and the other maybe eventually.
Oh, neat! I had hoped for some progress, but hadn't expected anything quite so concrete and soon.
Interesting ...
It's a funny coincidence, I came across this other post from Stephen Skolnick just yesterday arguing that the gut microbiome strongly predicts schizophrenia risk.
https://stephenskolnick.substack.com/p/schizophrenia
I will admit that Skolnick's essay sets off my 'too good to be true' alarms, but it still seems an idea worth exploring and he rigorously argues for the mechanism.
To what extent does the gut microbiome account for the outcomes we're seeing here, also? Gut microbiome is inherited, to some extent. It is also genetically influenced, and monozygotic twins have more similar gut microbiomes than dizygotic twins, with the divergence apparently increasing as life goes on.
Goodrich et al., 2014 (Cell)
SA: "Finally, she realizes she’s been scooped: evolution has been working on the same project, and has a 100,000 year head start. In the context of intense, recent selection for intelligence, we should expect evolution to have already found (and eliminated) the most straightforward, easy-to-find genes for low intelligence."
Doesn't this assume that increasing intelligence always increases reproductive fitness? In modern industrial societies there's a negative correlation between high intelligence and reproductive fitness, especially in women. I recognize that the same trend might not apply, pre-birth control in more traditional societies. But it at least seems worthwhile remembering that various types of mental facility and reproductive success may not always be positively related in all environments. This suggests at least the possibility of low hanging fruit if those gains involve tradeoffs with reproductive fitness.
Apologies if I've missed your point, there.
Humans evolved to be more intelligent than other species because it was an advantage to us (perhaps because of our large social groups). But brains are metabolically expensive, and there are limits to how big our heads can be while still fitting through a birth canal (hence humans being relatively helpless at birth). So it's advantageous to have higher IQs, but there are tradeoffs and other traits being selected.
If anyone really wants to listen to me drone on about an unpublished method I am unreasonably proud of, feel free to touch base. It’s been maybe six years since I’ve been thinking about this, but IIRC GREML ignores dominance and epistasis, and it’s not that hard to pick them back up. I’m not sure how to handle GxE, but probably using that sib method could negate the concern. And I don’t think ultra rare variants would be a problem, cause you don’t have to assign effects to any particular variant. And I think sibs should have a pretty good distinction of all relatedness modes, so it seems ideal.
Did you ask the polygenic embryo selection folks about collider bias?
In the section comparing Kemper’s sib-regression estimate (14%) and Young’s Icelandic estimate (~40%), you note that the UK Biobank sample may be skewed toward healthier, higher-SES volunteers (so-called healthy volunteer bias, which commonly creates selection effects in medical research). But the implications of such selection effects extend far beyond variability in heritability estimates.
This kind of bias can flip signs when people think they're being clever by adjusting for confounds (collider stratification bias). This is especially a relevant risk in highly selected samples like the UK Biobank, where the same factors that influence participation (e.g., health, SES, education) may also affect the outcomes of interest (mental and physical health). Conditioning on variables that causally contribute to both participation and outcome can introduce more bias than it corrects for.
I wrote a bit about this here: https://wildetruth.substack.com/p/designer-baby-maybe. Munafò et al.'s schizophrenia example illustrates the mechanism more clearly than I can easily argue it for IQ: people at higher genetic risk for psychosis may be less likely to volunteer for studies ("they're out to get me, so I'm not giving them my blood"). This warps the apparent genetic architecture in large datasets. Doing embryo selection against schizophrenia risk based on such a sample, from which high-risk individuals disproportionately self-selected out, could backfire. And parents who paid $50K for it might not know for 20 years (or ever).
Hopefully the real-life practical implications are trivial: if you pay $50K for a prospective 3-point IQ gain, and collider stratification bias turns it into a 3-point loss, no big deal? You probably won't notice over dinner, as long as your kid slept well last night.
But I remain curious how the corporations selling embryo selection address the possibility that they're accidentally giving parents the exact opposite of what they're paying for. My guess is: they just don't tell anybody, and there's something in the fine print saying they could be wrong and aren't liable. Causality probably can't be proven in any individual case, and anyway you're paying for a best effort process, not a result.
This is entirely dependent on whether the “warping” by not having X group in a sample actually warps the sample. Because there’s a high chance it doesn’t warp it at all and the sample is still valid, even without X group present.
Jury is still out on the powers of nature versus nurture, basically. It's slightly complex.
What struck me is the questionable logic of EA as a measure. Scott got into this slightly with grade inflation and economic prospects, but I'd go further. Many of the twin studies took place between 1970 and 2000, or at least the kids were in school during that period, some of them earlier. This corresponds with a high point in educational prestige, and an opening of college to more people on a merit basis, while college was still rigorous. Today, getting a college degree is obviously less meaningful than it used to be, so the value presumed of its presence is almost irrelevant, and I'm thinking this is generational, a reflection of boomer-GenX social philosophy. Expecting the same effects thirty years on is going to be very muddled by social change.
EA is really, really sloppy from a standpoint of correlation with IQ.
One way or another, I still see nature and nurture about the same way. A person's biological design is full of complexities and redundancies, while their experiences will prod adaptation in various directions within the limits of that design. I know people don't want to believe that adaptation can be bracketed like that, but the evidence is everywhere, we maladapt to all sorts of things in the modern world, some of us more than others. People can push their limits, but it requires strong incentives and going too far results in trauma. Or failure. None of this is particularly compatible with Western ideals regarding individual agency and equality, but that's why it's controversial, right?
My sympathies to adopting parents who thought they were getting a blank slate.
Saw recently, don't recall the source, that the average IQ of college students has gone from almost 120 in the 1960's to just above 100 today due to pushing college enrollment on broader populations.
https://www.cremieux.xyz/p/education-isnt-what-it-used-to-be.
It's not just about respecting values like individual agency. There's also a strong placebo effect associated with the growth mindset. In other words, an exaggerated belief that you can transform yourself into a better person makes you more capable of improving yourself.
Is this anything like the placebo effect of telling an ugly persona they’re beautiful? Because people can confidence game themselves into being a little more attractive if they get that kind of feedback, but ultimately, you are what you are. It seems like that would be made way worse when it comes to intellect. You can push your way up to a masters degree, but in the employment world, you will prove yourself incompetent at some point.
No, the logic behind these two phenomena are fundamentally different. I'm too tired to explain my thinking right now, though, sorry for that, because the logic is kinda intricate. The example you gave isn't even really the placebo effect, it's a completely different thing.
I expect that’s right, it’s not a placebo effect, it’s more like “trying harder isn’t hopeless, so let’s up the effort.” Still seems similar.
Well, now that you say that, I actually think the logic behind why placebos work (other than for pain relief placebos, which has a different logic) is something like that. So you're right that they sound similar. But TBH I haven't personally noticed "ugly people becoming slightly hotter when you call them beautiful," and that sounds a bit weird to me. But in the context of learning, yeah, I do think the logic is probably something like, "trying harder isn't hopeless, so it's worth increasing effort."
On second thought maybe the logic is closer to "people with a growth mindset think they can get more out of trying harder." Still similar logic but not the same.
Kind of surprised you don't see it in appearance. Someone tells you there's potential, so you lose weight, learn to dress well, get a haircut, and of course, walk and talk with confidence, you can maximize what you've got. You can go from a 4 to a 5, maybe even a 6. It's thinking you're a 10 that's going to end up being embarrassing.
With intellect, while I hate to bring this guy into it, Malcolm Gladwell said in Outliers that really, as long as you have an above average IQ, like 120, you can probably do about any job as long as you work at it, and a few extra points matter little. But trying to do organic chemistry with an average IQ is trading tuition money for frustration. If they let you graduate, it's worse.
That sounds correct to me.
To the point about EA being full of landmines, wouldn't one of those be the observation that so many modern jobs expect a Bachelor's degree? That seems like a shared environmental component that would push swathes of the population to hit certain thresholds they might not otherwise.
A small technical note that I think is crucial. In the first figure (Tan et al 2024)
1. The x-axis is "Proportion of variance in population effects *uncorrelated* with direct genetic effects".
2. Being on the left side means that the variance in population (given by GWAS) can be correlated with direct genetic effects.
3. So for example, properties that are highly due to direct genetic effects are on the left (height, myopia, eczema)
4. Educational attainment is at ~0.5, this means that the direct effect variance is 0.5 of the population effect. The population effect is ~40% and thus the direct effect is 20%, exactly the value that GWAS converges to.
5. We can also verify this by looking on Tan et al 2024. In another figure they show that EA's genome wide correlation between direct genetic effects and population effects is ~0.75, if you take the square, you get ~0.5
Yeah, when I read that his paragraph below, I didn't quite follow why he was squaring, and assumed the diagramme was perhaps mislabelled (ie it was intended to be the correlation, not proportion of variance).
Still though, you know why the graph has migraines as having a negative proportion of variance explained? I don't quite understand that part.
How much effort has been put into algorithms decoding complex geneXgene interactions? Also how much of these things could be down to epigenetic expression, and how could this hypothesis be tested?
How do y'all explain Australia? Country literally founded by a prison population, now significantly safer than world average?
I looked into this once; the answer is that although Australia was founded by (a small number of) prisoners, almost all modern Australians are descended from non-prisoner immigrants who came later.
Approx 20% of Australians are descendents of convicts, apparently. But there would be a lot of non-convict heritage mixed in as well. The gold rushes of the 1860s led to a huge population explosion, for example.
Also worth noting that the convicts in question might not have been as negatively selected as you're thinking. The classic claim is that most convicts were convicted of petty larceny, things like "steal a loaf of bread - 7 years".
I conceptualise all this as: we have a new method sibregression that does not give the same results as twin studies; this is strong evidence that broad != narrow heritability. Which is very interesting and worthy of a lot of discussion about what it means.
To me it suggests that a significant amount of heritability is chaotic: the genes matter, but small differences have big effects, being "similar" is less informative than we might have expected.
And as you say:
"If nonadditive interactions are so important, why have existing studies had such a hard time detecting them?"
My assumption would be that the studies are underpowered to detect interactions. Sample sizes have to be much larger to reliably detect interactions.
Is it possible that the *fact of being a twin* makes you weird? Twins are more likely to be premature and to die in infancy. Maybe twinhood is "bad for you" (perhaps via having a worse uterine environment) such that twins are more likely to have all kinds of health & behavior problems than singletons? Perhaps having a "bad" gene is more likely to result in a "bad" trait if you're a twin, which will push up the heritability of lots of traits in twin studies specifically?
That wouldn't affect adoption & pedigree studies, though, so if those also show high heritabilities for lots of traits this hypothesis doesn't help.
Twins used to have lower IQ than singletons, but have now converged (probably with better prenatal care). I can't prove that they don't have some kind of unique disposition which results in equal IQ on average but makes it have a different genetic structure, but seems like that would be surprising.
I may be remembering wrong, but I think path is accounted for in the SEM by Hatemi in the "Not by twins alone" paper
https://matthewckeller.com/publications/16.Hatemi.et.al.2010.Nuc.fam.ajps.pdf
re: footnote 8, Graham Coop had a good critique in my opinion of the Reich paper when it was published: https://bsky.app/profile/gcbias.bsky.social/post/3l4oqx42hil2b
Could you respond to Lyman Stone's "Why Twin Studies Are Garbage"?
I think that Stone makes two basic points:
1. Heritability is not worth measuring or thinking about, because it doesn't mean what most people think it means. A trait can have perfect 1.00 heritability and proven genetic basis, and yet still be 100% environmental in the wild, due to gene-environment interactions.
2. Gene-environment interactions can be both very strong and very difficult to detect. We didn't notice for decades that peanut allergies were >90% environmental in practice, despite having >50% heritability and known genes for susceptibility. Just a tiny difference in the timing of diet accounts for all of that. Because of this, it's impossible to say that fraternal and identical twins have similar environments, which violates the first assumption of twin studies (as you have numbered them above).
What's your take?
Something can be worth measuring and thinking about even if "most people" don't know what it actually means. Most people don't understand quantum physics.
Environmental effects have been studied pretty exhaustively for things like IQ, if GxE was important we would know
Yes, gene-environment interactions exist and are a big deal. Basically the only conclusions that are consistent with all the research - assuming its all valid, which seems likely - are either gene-gene or gene-environment interaction or both are responsible for most of the missing heritability. Lyman Stone, like Sasha Gusev, wants most of the gap to be filled with gene-environment interaction.
That may very well be true for some traits, but it seems unlikely to be true for IQ or educational achievement. This now is just my personal take now from when I puzzled my why through Sasha's writings. Think of a simple nonlinear gene-environment interaction where a gene only confers an IQ gain in the presence of some external factor. Then we would expect to see an affect on IQ in siblings raised separately. But we don't generally see that - adoption studies, for example, generally measure broad heritability of adult IQ similar to twin studies. Now there might be gene-environment interactions in utero or even pre-fertilization, but just very broadly any postpartum gene-environment effect should show up in adoption studies and they mostly don't.
Which leaves gene-gene interaction. That is EA and IQ are highly heritably but mostly determined by non-linear gene-gene interactions.
But even if true, that doesn't necessarily mean that boosting spending on social programs would equalize the races in IQ.
> If nonadditive interactions are so important, why have existing studies had such a hard time detecting them?
Ooh, I have an extensive discussion of this in a recent post: https://www.lesswrong.com/posts/xXtDCeYLBR88QWebJ/heritability-five-battles Relevant excerpts follow:
§4.3.3 Possibility 3: Non-additive genetics (a.k.a. “a nonlinear map from genomes to outcomes”) (a.k.a. “epistasis”)
…Importantly, I think the nature of non-additive genetics is widely misunderstood. If you read the wikipedia article on epistasis, or Zuk et al. 2012, or any other discussion I’ve seen, you’ll get the idea that non-additive genetic effects happen for reasons that are very “organic”—things like genes for two different mutations of the same protein complex, or genes for two enzymes involved in the same metabolic pathway.
But here is a very different intuitive model, which I think is more important in practice for humans:
• Genome maps mostly-linearly to “traits” (strengths of different innate drives, synaptic learning rates, bone structure, etc.)
• “Traits” map nonlinearly to certain personality, behavior, and mental health “outcomes” (divorce, depression, etc.)
As some examples: …
• I think the antisocial personality disorder (ASPD) diagnosis gets applied in practice to two rather different clusters of people, one basically with an anger disorder, the other with low arousal. So the map from the space of “traits” to the outcome of “ASPD” is a very nonlinear function, with two separate “bumps”, so to speak. The same idea applies to any outcome that can result from two or more rather different (and disjoint) root causes, which I suspect is quite common across mental health, personality, and behavior. People can wind up divorced because they were sleeping around, and people can wind up divorced because their clinical depression was dragging down their spouse. People can seek out company because they want to be widely loved, and people can seek out company because they want to be widely feared. Etc.
• I dunno, maybe “thrill-seeking personality” and “weak bones” interact multiplicatively towards the outcome of “serious sports injuries”. If so, that would be another nonlinear map from “traits” to certain “outcomes”.
All of these and many more would mathematically manifest as “gene × gene interactions” or “gene × gene × gene interactions”, or other types of non-additive genetic effects. For example, in the latter case, the interactions would look like (some gene variant related to thrill-seeking) × (some gene variant related to bone strength).
But that’s a very different mental image from things like multiple genes affecting the same protein complex, or the Zuk et al. 2012 “limiting pathway model”. In particular, given a gene × gene interaction, you can’t, even in principle, peer into a cell with a microscope, and tell whether the two genes are “interacting” or not. In that last example above, the thrill-seeking-related genes really don’t “interact” with the bone-strength-related genes—at least, not in the normal, intuitive sense of the word “interact”. Indeed, those two genes might never be expressed at the same time in the same cell….
As far as I can tell, if you call this toy example “gene × gene interaction” or “epistasis”, then a typical genetics person will agree that that’s technically true, but they’ll only say that with hesitation, and while giving you funny looks. It’s just not the kind of thing that people normally have in mind when they talk about “epistasis”, or “non-additive genetic effects”, or “gene × gene interactions”, etc. And that’s my point: many people in the field have a tendency to think about those topics in an overly narrow way.
…
§4.4.3 My rebuttal to some papers arguing against non-additive genetics being a big factor in human outcomes:
The first thing to keep in mind is: for the kind of non-additive genetic effects I’m talking about (§4.3.3 above), there would be a massive number of “gene × gene interactions”, each with infinitesimal effects on the outcome.
If that’s not obvious, I’ll go through the toy example from above. Imagine a multiplicative interaction between thrill-seeking personality and fragile bone structure, which leads to the outcome of sports injuries. Let’s assume that there are 1000 gene variants, each with a tiny additive effect on thrill-seeking personality; and separately, let’s assume that there’s a different set of 1000 gene variants, each with a tiny additive effect on fragile bones. Then when you multiply everything together, you’d get 1000×1000=1,000,000 different gene × gene interactions involved in the “sports injury” outcome, each contributing a truly microscopic amount to the probability of injury.
In that model, if you go looking in your dataset for specific gene × gene interactions, you certainly won’t find them. They’re tiny—miles below the noise floor. So absence of (that kind of) evidence is not meaningful evidence of absence.
The second thing to keep in mind is: As above, I agree that there’s not much non-additive genetic effects for traits like height and blood pressure, and much more for things like neuroticism and divorce. And many papers on non-additive genetics are looking at things like height and blood pressure. So unsurprisingly, they don’t find much non-additive genetics.…
[Then I discuss three example anti-epistasis papers including both of the ones linked in OP.]
I thought of nonlinearity too. The twin studies compare the effect of different sets of genes. The DNA studies are predictions using different individual genes as variables. Are these linear predictions? The post said interaction terms don't explain anything. How about polynomials of the individual genes ?
Just wanted to say this is a very good comment. It does seem to me that the more likely source of discrepancy is how we are defining a "trait," which is typically a more or less fuzzy concept.
Yes, this is a very good comment and I have been thinking along similar lines. I think to find most of the missing heritability we would need to move from simple additive models to something more like a machine learning model, where rather than starting with the assumption that each SNP causes a simple linear change in a trait and trying to determine what that value is, we start with the assumption that each trait is a complex nonlinear convolution of a binary vector of SNPs, like a convolutional neural network, and try to determine its parameters.
Can confirm I was about to admit it's technically valid, but with hesitation and while giving you a funny look.
But for this particular example, thrill seeking and fragile bones, I think a simple linear model would do fine at identifying genes in both groups. Say you have one gene that increases thrill-seeking by a tiny bit and another that decreases bone strength a little bit. If you have neither, you get fewer sports injuries, both and you get more, just one and you're in between. This is the sort of thing a linear model can detect.
:) Well it's a made-up toy example, but sure, let’s roll with it. Let t be thrill-seeking, b be bone strength, tₐᵥ is population-average of t, etc. A person’s sports injuries = (tₐᵥ+Δt)×(bₐᵥ+Δb). Multiplying that out, we get a population average tₐᵥ×bₐᵥ, and linear (additive) terms (tₐᵥ×Δb + bₐᵥ×Δt), and a non-additive term Δt×Δb. The non-additive term would generally be negligible if the fractional variation is small (i.e. Δt/tₐᵥ and Δb/bₐᵥ << 1). But the fractional variation also might not be small, and thus the non-additive part might not be negligible. For behavioral things in particular—things where people are making decisions about how to spend their time—you can get order-of-magnitude variation in how much sports different people play, how much socializing they do, and so on.
Anyway, my claim is not that mental health, personality, and behavioral outcomes have no additive variation whatsoever, just that those outcomes are MOSTLY non-additive. IIRC, the PGSs for those things generally explain <5% of the population variation, but not zero.
For the record, I do think my ASPD, divorce, and extroversion example above is more realistic than the sports injury example. The sports injury thing was just something I made up to make the math easier to illustrate. :)
Ok, so I fleshed out that example a bit more and went through it. We have 1000 genes for each of thrill seeking and bone fragility, each gene has two incomplete dominant alleles one of which adds 1 point and one of which ads 0 points, and for each gene both alleles are equally common. So the population average is 1000, min is 0, max is 2000.
If we look at a person with values t=1500 and b=1500, then tₐᵥ×bₐᵥ = 1 million, the two deltas times averages both are 500 thousand, and Δt×Δb = 250 thousand. Not too shabby, and certainly large enough to be making a dent in the calculations.
Then I had python generate 3 random people from this population and took the one with the highest combined values for the two traits. This turns out to be t = 1036 and b = 1002. Δt×bₐᵥ = 36,000 and Δb×tₐᵥ = 2,000. Δt×Δb = 72. Small. If you arbitrarily bump up b to 1036 as well that term becomes 1296, still kind of small.
What would have to change for us to have individuals far from the average, like 1500, be more common? We'd need the genes to be fewer and with larger effect. At that point it becomes much easier to find the genes for each of those separate traits (unless the traits are something with extremely non-obvious effects) and then account for their effects on sports injuries, although I don't know of this being a methodology that is actually used.
Anyway, very interesting thoughts, thank you.
Afaik, epistasis is defined as "change of fitness from two mutations is not equal to sum of change of fitness from having each one of them separately". Epistasis in this sense was definitely found in deep mutation scan (DMS) studies where one particular protein is mutated randomly. However, link between protein "fitness" and IQ is very much not obvious, and probably because of that also, we cannot detect epistasis reliably in biobanks
>The biggest losers are the epidemiologists. They had started using polygenic predictors as a novel randomization method; suppose, for example, you wanted to study whether smoking causes Alzheimers. If you just checked how many smokers vs. nonsmokers got Alzheimers, your result would be vulnerable to bias; maybe poor people smoke more and get more Alzheimers. But (they hoped) you might be able to check whether people with the genes for smoking get more Alzheimers. Poverty can’t make you have more or fewer genes! This was a neat idea, but if the polygenic predictors are wrong about which genes cause smoking and what effect size they have, then the less careful among these results will need to be re-examined.
I've always been rather skeptical of mendelian randomization (MR) studies. In particular for MR to work, you need:
1. the genes to actually affect the phenotype (often true, but as this shows the effects can be weaker than previously thought)
2. no confounding by population stratification / assortative mating (hard to achieve)
3. no pleiotropic effects: e.g., maybe the genes for smoking overlap with the genes for Alzheimers (unlikely, but possible)
Usually when I see bad mendelian randomization studies (e.g. of ALDH2 variants for alcohol drinking -> cancer) they fail assumptions #2 and/or #3. But now it looks like assumption #1 is potentially worse than expected too!
Any time new types of testing and analysis are administered, and they give *wildly* different results than both multiple types of long used analysis and common sense, we should be skeptical of the analysis and results before we throw away anything of value that has gotten us through millennia of human history and centuries of the scientific era.
In a way this debate reminds me of the intelligent design debate, where one side (evolution believers) have mountains of evidence, but often it was more common sense and circumstantial, based on careful observation. This opened the door to claims about intelligent design that were simple at first, but were transformed by rebuttals until they were in many ways just evolution but “not evolution because evolution is wrong”. Anti-hereditarians are getting closer and closer to the intelligent design group in that they are slowly conceding ground in every debate so as to at least throw into doubt the tiniest bit of hereditarian, trading accuracy for vibes as it were.
“Maybe genetics does affect intelligence, but akshually it’s not your genetics, it’s your parents genetics that you also have which confounds even the claim I just made! Let’s talk around in circles and never admit it’s almost all genetic!”
Kinda like how Creationists are very expert on the subject of petrified wood.
Right, any speck of evidence on the X side is evidence that X is right, any speck of evidence on the Y side is just confounding and circumstantial. The (high) influence of heredity is not scientifically bulletproof, but the evidence *against* heredity is even weaker and more tenuous, thus not effectively invalidating the longstanding idea of “the apple doesn’t fall far from the tree” or “like father, like son” concepts that go back to antiquity.
There’s an implicit assumption that heritability is a fixed property of traits, rather than a model-dependent statistic that varies with environment, measurement, and population structure. If twin studies, GWAS, and sib-regression all yield different numbers, is that necessarily a “problem to be solved,” or might it reflect that traits like intelligence and EA aren’t stable enough constructs to expect convergence?
I'm seeing a fair few comments about genetic screening and 'careful adoption choices' that are giving me some serious liberal eugenics vibes, nasty. Not sure using anecdotes about criminal kids to defend IQ heritability estimates is that valid; I believe criminal behavior shows less heritability than IQ?
I wonder if there's a sampling bias here? Like only seeing kids whose biological parents got caught and ended up in the system. What about white-collar criminals, financial fraudsters, and corrupt executives who are smart/wealthy enough to avoid prison? If antisocial behavior is really heritable, wouldn't their kids show it too? But those kids aren't in adoption; they've probably been shipped of to elite prep schools to help continue their gene pool.
If your thinking is based on "vibes", then you might not like blogs by decouplers.
whereas, I do seem to. Go figure? I seem to doom scroll on condescending passive aggressive flexing vocabulary intellectual gatekeeping vibes.
For those unfamiliar with "decoupling" in this context, see here: https://everythingstudies.com/2018/05/25/decoupling-revisited/
Basically, it's considering ideas on their own merits, independently of their association with other concepts in popular media.
That being said, even with these ideas considered on their own merits, it _would_ be unfortunate if fewer people became willing to adopt kids of criminals, and so those kids had a worse time. I just don't know what to do about that to help.
Aka mental masturbation.
But yes, it would be unfortunate. Though some on her seem to suggest euthanasia. This post has attracted an odd crowd .
It's mental masturbation of that kind that explains why we're sitting in air-conditioned rooms arguing on the internet, rather than staring at a mule's ass as it pulls a wooden plow down a field.
Iirc callous-unemotional traits (those responsible for most psychopathy and often criminal behaviour) shows high heritability
Interesting a trait that's useful in one environment gets pathologized in another. 'leadership material' i guess it is called.
White collar criminals do tend to have a different cognitive and behavioral profile than violent or "compulsive" offenders. Even if perhaps morally the white collar ones are more evil( they have more "control" over their actions), it makes sense that the descendants of white collar criminals might have traits that lead to better outcomes( high conscientiousness, detail-oriented, ability to perform well socially in high-status places) than the descendants of, crudely, lower class criminals.
Yes, and also a better support system to help them avoid consequences and exploit those qualities than less affluent families.
Of course, but its worth pointing out that white collar criminals like say Madoff or whoever still have a higher intelligence and capacity for work than violent criminals or petty thiefs. The difference between a capable sociopath and an incompetent sociopath matters a lot to social outcomes, even if both are amoral. Certainly sociopathy is a favored trait in positions of power, in virtually every society.
or even Pablo Escobar.
What's your position, that genetics doesn't matter at all for socially-relevant traits?
Sorry, I'm not sure what you mean, could you elaborate? Are you saying that I think personality traits aren't influenced by genetics at all, and it's all just about environment/context? Which I don't believe I said. Sorry, I'm not great with language, I'm quite blind to nuance and inference, amongst other things.
I'm responding to your "nasty eugenics" comment. My reading was that you seem to think that people who believe that socially-relevant traits (IQ, behaviors) are genetically-mediated are somehow bad or misinformed.
Oh, I understand now. Well, I was not critiquing the idea that genetics influence traits. I was critiquing the application of that belief to decide which kids are worthy of adoption or birth based on their genetic background. I mean I guess I'd be terminated for a start.
Are you opposed to being differentially attracted to beautiful, successful, or intelligent partners? Because that's just your biology steering your preferences with the goal of creating well-adapted children. I don't see why anyone should have any moral qualms with a more explicit version of the same process.
> I mean I guess I'd be terminated for a start.
I think a better framing is that you'd be recycled into a body that didn't have whatever limitations you seem to be alluding to.
I’m not sure that notion holds up. Not everyone prioritizes beauty, success, or intelligence in partners. people are drawn to kindness, humor, shared values, or just odour. And if biology is steering us toward ‘well-adapted children, how does that explain gay couples, who aren’t reproducing through their partnership?
The ‘recycling’ comment is interesting phrasing. I think you might be viewing neurodivergent as a limitation that needs fixing, but many of us see it more as neurological diversity, different wiring that comes with both challenges and strengths. The idea that I should have been ‘recycled’ into something ‘better’ assumes there’s an optimal human template we should all conform to.
I can understand wanting to give children the best possible start in life. But there’s a difference between choosing consciously or unconsciously compatible partners and systematically screening out entire categories of people from adoption or existence. The latter starts feeling uncomfortably close to deciding who deserves to be here.
I think the first step to thinking clearly about this kind of issue is to try to separate the factual questions from the moral/social/political/tribal ones. It may be that the claim that propensity to crime is heritable is associated with bad people or would have bad consequences if true or has bad "vibes," but that can't possibly inform you about whether or not it's true.
This is my favourite ACX post in a hot minute
Thank you Scott!
"In the degenerate case where the mother and father have exactly the same genes (“would you have sex with your clone?”) the fraternal twins will also share 100% of their genes."
Why is this so? If Mom and Dad both have genes (a, b) at one location, won't sometimes twin 1 get (a, a) while twin 2 will get (b, b)? Agree there's more commonality than normal because there's a possibility of 1 getting (a, b) while 2 gets (b, a), which isn't normally true.
Thanks for the correction, fixed.
Edit: I just saw that some of these points get covered in this post, thank you!
Someone with more human genomic knowledge please shoot me down.
Sequencing technology doesn't get discussed nearly enough in this area. Illumina short-read sequencing/SNP panels have been the major source of data for all of these studies, and they are absolutely delightful at finding SNPs but are crap at anything else. I think it will be appreciated that generally things that impact function aren't SNPs, they are broad changes, and so much human genomics seems to be hoping that the thing that is contributing to a change is able to spotted by being close to a SNP, instead of actually looking at the thing that is causing the change.
Genomes aren't lists of SNPs, they are mostly repeats and 2x150bp isn't going to get anywhere near close to capturing that variation, no matter how 'deep' you sequence. Long-read sequencing (PacBio & ONT, not Illumina's synthetic tech) is clearly better, and continues to demonstrate that there is massive variation that is easy to see when you have a bunch of 20kbp fragment, while almost impossible when you're just aligning little chunks of text to a 3gbp genome.
I work in non-model genomics and long-read sequencing is such a clear winner I keep getting surprised when Illumina gets contracts for these massive human studies. The Human Pangenome Consortium is going to be providing a dataset that is way more useful than anything that's come before. Anecdotally, I hear that for some non-European genomic data they know that ~10% of the data from an individual DOESN'T EVEN MAP to the human reference (but is human genomic data). This is all invisible to analysis, or even worse, just confounds things, as the 'true' causal SNP is somewhere in the data that doesn't get analysed, and so we're stuck looking at noise and trying to make sense of it.
I feel like this is such a blind-spot for human genomics, as it's always about the latest and greatest AI/ML method to try and get some information out, when it's the underlying data which just sucks and doesn't have a hope in actually being linked to function. There was even a point on an Open Thread a few weeks back (Open Thread 374 from tabulabio) asking for people to get in touch about Frontier Foundation Genomic models, with the focus being on fancy new ML architectures.
When I asked ChatGPT to write this comment for me ("Argue that sequencing technology could explain a lot of the Missing Heritability problem") it actually pushed back against me, trying to use the Wainschtein et al. 2022 paper as evidence that '...[this paper] used high-quality WGS (which includes better SVs than Illumina) and still found that adding rare and structural variants only modestly increased heritability estimates", which is NOT TRUE. Wainschtein uses the TOPMED dataset, which is from Illumina short reads. Yes, they do 'deep' sequencing, and yes it's analysed to the absolute hilt with the latest and greatest GatK pipeline and QC to the max. But that claim is false, it's just lists of SNPs, completely ignores huge chunks of the genome and just hopes that the thing contributing to a phenotype is is able to be fished out alongside a SNP.
Another anecdote - an older friend's wife died from a brain cancer. He was an old-school non-model genomicist and got all of the data from the oncologists and various tests and analysed things. All of it short-read, none of it turned anything up, either from his or the various doctors. Long-read sequencing run was eventually done after her death and indicated that it was a splice variant missed by short-reads. It was clear as day in the long-read data, since it didn't need to do fancy bioinformatic de bruijn graphs to figure out the splice isoform - it's just mapping the read against the genome and seeing it clear as day.
Thanks - can you explain more about whether things that are advertised as "whole genome sequencing" are the Illumina method you say is inadequate, or whether the WGS data is fine?
Hey Scott, I don't mean to sound hyperbolic but Illumina is kind of like the Illuminati. It's everywhere, monolithic and it's influenced genomics massively.
I had a quick look at a few "whole genome sequencing" retailers, and they’re usually using Next Generation Sequencing, which in most cases means Illumina. The phrase "sequence every base in the genome" sounds impressive, but it’s a bit misleading. Yes, they generate _reads_ from across the whole genome, but they’re in tiny fragments, and only make sense once you align them to a reference genome.
That's where reference bias comes in. You mostly detect things that are similar enough to map cleanly, and just a little different to be able to be called a variant. That’s fine for common variants, but bigger or more complex stuff tends to get missed or misinterpreted.
To give a sense of scale, the human genome is about 3 billion base pairs long. When you get your Illumina WGS results back from a provider, you don’t get a 3 Gbp text file with your actual genome. What you usually get is a variant (VCF) file with just the differences compared to the reference genome. And that makes sense to some extent. Why include everything that's the same? But there’s a lot of complex or unmappable variation that just isn’t detected with short reads.
If you used long-read sequencing and actually assembled your genome, you could get pretty close to your actual full genome, including regions that are repetitive, messy, or just structurally different. You’d see things that are completely invisible in an Illumina dataset. And you'd have much higher confidence in the things you see, since a lot of the artifacts come from using short-read data.
That’s why basically all genomics in non-model organisms is happening with long reads now. At the International Congress of Genetics in 2023 (major conference that only happens every five years) the keynote speaker Mark Blaxter opened the meeting by saying we can finally get real, complete genomes thanks to long-read sequencing. He was talking specifically about the Darwin Tree of Life project, which is trying to sequence all eukaryotic species in the UK.
So yeah, most consumer WGS is Illumina, and it’s fine if all you want is common SNPs. But I can't wait for human genomics to migrate to long reads and overturn some of the perceived wisdom from two decades of Illumina dominance.
Thanks.
It sounds like 21st Century genetics is really, really complicated, and we shouldn't be in too much of a hurry to proclaim conclusions.
I appreciate that this sounds like a lot, but every time I have to describe to someone how sequencing technology works and the degree of genetic variation out there and it starts hard to try and defend the continued use of short-reads.
Genetics _is_ complicated (this is only genomics! not even epigenomics or transcriptomics).
But there has been such a massive degree of complication from going from someone's actual genome info to how genomes are represented in data that, to me, it feels like only the most simple and clear things are able to be identified, and so many complex things are going to be unanswerable. That used to be a technology problem, but PacBio and ONT are almost on the same level as Illumina in terms of cost/genome, and they ACTUALLY give you the whole genome!
As a causal follower of the field, I'd never heard that there were huge questions about the implications of which brand of genome scanner you choose. But that sounds plausible.
>> "But there’s a lot of complex or unmappable variation that just isn’t detected with short reads."
Can you explain this a little more? The way I read what you said at first is that rather than storing each nucleotide in your genome, we instead just store the locations and the nucleotides at locations where where your genuine has a different nucleotide than a reference genome.
But under a naive reading of this, it sounds like as long as you know the reference genome, this is the same information as each nucleotide: at each location i your nucleotide is exactly given by the map "if i is in your genome file, take the nucleotide listed; otherwise take the reference genome at location i".
The simplest way I can imagine your description not matching my naive version is if there isn't a canonical way to match locations at two genomes: if the mapping is only defined on certain sections of the genome where we can define an unambiguous location.
Is that what's going on? Or is it something else I'm not considering?
I'm not OP, but I do work with whole exome data*. What you get back from an Illumina sequencer is a set of 150-250bp "reads", along with quality data. Each read is then mapped to the reference genome (along with some QC and clean-up before and after), and the final file is the reference genome with the reads aligned (uncompressed=sam, compressed=bam, super-compressed=cram). So some regions will have a lot of reads aligned, some may have fewer. And some reads will align wrongly - there are a lot of regions which look very much like other regions, and a short read will not be able to distinguish them.
Anyway, the VCF (variant call format) file is generated from this bam file by comparing the reads at each point with the reference, and it is typically just the differences which get returned. One read out of 20 with the wrong base pair is just a sequencing error, and won't return a variant. But if there are 20 reads covering a T base, say, and 10 of them are A, that's likely to be a heterozygous variant. Of course, it could also be a place very similar to another region where the A is the same as the reference, but both reads are mapping to this place instead. A good variant calling pipeline will use as much data as it can to minimise this - calling from multiple individuals at once helps - and for my work, I use three variant callers and combine their output with a set of criteria established by going back to the samples and resequencing (with old tech) various SNPs. But nothing is perfect. It is quite common to just get the VCF file from a commercial company, and this is why I do the whole thing myself, from the raw reads to the eventual VCF file - and then we also went back and checked a selection of the SNPs, both against the bam file and by old-tech sequencing.
* Exome sequence data gives you better coverage of the exome, which is just the 2% of the genome which codes for proteins. Depending on the kit, you sometimes get the untranslated regions at either end of the gene, and you always get 10-20bp into the intron too. You will of course miss a lot of intergenic and intronic variants, but nobody knows what to make of them anyway (this is improving all the time, though). But it is far cheaper, and in contrast to what another poster said, the kits are designed so that the entire exon is covered by actual reads.
Thank you human genetics person!!
@Jerl: This is a great summary and correct.
But I would just draw attention to the difference in 'sequencing' the whole genome and doing all the things you are thinking about with genome alignments.
Sequence companies will say that they 'sequence' the whole genome to some depth (eg 30x coverage). But the majority of the genome is repetitive (transposable elements like SINEs, LINEs make up >25% of the genome), so lots of that data doesn't make any sense by itself, and you need to align it to a reference. You take this aligned data and try and call variants - but what you get in that VCF file _isn't_ your whole genome, it's the bits of your genetic data that are able to be analysed in context of the reference.
Anything that _doesn't_ map doesn't get called. And anything that maps awkwardly gets filtered away, or mis-called.
This is known and discussed as reference bias and there is work in human pan-genomics field trying to address this. But my main gripe is that Illumina has been the workhorse of genetics for so long that people forget that it's fundamentally limited in its ability to capture 'full' genomic information.
I work with lost of very smart AI and ML engineers and they hear what genomics people and companies say and have that image in their head - "If what I've bought is a Whole Genome Sequence, then all of the genome is there! Oooh and what's that? It comes as a table? Awesome, my ML model is going to chew this up and we'll have an answer in no time!". But that's not true, and I feel like often forces an answer from the data.
I'm mainly griping about this since PacBio and ONT (the main long-read providers) seem to not be able to break the market hold of Illumina because of this mis-perception. Those providers will give you an essentially 'full' Whole Genome Sequence, where all of your bases are captured, including the variable bits. And from a analysis perspective I just feel this is the correct way to start.
I guess if I had to justify this I would look at the Human Pan-genome data released for individuals (made from long-reads, full haplotype resolved chromosomes) and compare the information from that vs the same samples with a short-read libraries. Anyone feels like collabing on this that would be great! I've got free compute for general bioinformatic stuff but don't know my way around human genomics that well.
Is there any organization selling long-read sequencing to consumers? A brief internet search showed only very "contact us" things.
I'm probably not the first to think of this confounder, but it's been estimated that ~70% of human embryos turn out to be non-viable [https://pmc.ncbi.nlm.nih.gov/articles/PMC5443340/]. Do the sib-regression studies account for this? The 40-60% range of shared genes they measure is restricted to the siblings that survive to adulthood. But there may have been many 30% embryos that didn't survive the first month of gestation. Their educational achievement was 0.
In other words, there's a severe selection bias in any study that only looks at siblings who survived to adulthood. A genome that kills one of the siblings before they complete schooling or even get noticed on ultrasound is still an inherited trait.
I think this changes some extremely nitpicky form of heritability defined as "heritability across all embryos rather than just surviving ones", but I think as long as you define heritability as being across surviving humans, this definition is consistent and fine, and all of the existing studies remain comparable (and so it's still surprising when they don't match)
Would someone with strong quantitative chops be willing to explain whether GWAS (and related methods) have enough power to detect small effects—say, 0.5 or 1 percentage point—caused by combinations of 9, 11, or 53 SNPs? I’m particularly curious about scenarios where some of the SNPs are rare. In that case, the number of people with a given combination might be small, so the signal could be lost. If that sort of undetected combination effect happens multiple times—say, 30 or so—could that help explain the gap between GWAS results and what twin studies suggest? I’d love a mathematical explanation that someone with several semesters of college math/statistics could follow with effort.
GWAS doesn't look at effects of multiple SNPs at all, except by accident.
I'm not aware of any method which will detect combinations of variants across multiple genes - if anyone knows of one, please tell me! There are a few methods which come close (eg ORVAL, https://orval.ibsquare.be/) but nothing I'm aware of for studying large cohorts and many combinations. I can see why it would be very computationally expensive.
I’ve been chatting about this with AI, and once you do combinations of 2 snps you are talking about billions or trillions of degrees of freedom and it’s really hard to get any signal. The biggest databases only have roughly a million genomes sequenced. you can’t do a matrix regression with an n of 1 million and 1 billion degrees of freedom.
Excellent post, I would just want to add that Sidorenko 2024 (https://pmc.ncbi.nlm.nih.gov/articles/PMC11835202/) that you cite for the discussion on rare variants also measure heritability using Sib-Regression although not for EA.
They show that correcting for assortative mating their estimates are coherent with twin studies :
"Therefore, if we account for assortative mating and assume that the resemblance between siblings is solely due to genetic effects (and not to common environmental), then our data are consistent with a heritability of 0.87 (s.e. 0.05) for height in the current population (Supplementary Note)."
they also show that shared environment estimates from twin studies are inflated by assortative mating :
"Our results for height and BMI agree with that conclusion in that we find no evidence of a residual sibling covariance (𝑐2) for BMI, while the significant 𝑐2 observed for height is largely explained by assortative mating."
And the WGS estimate is higher for height than the RDR estimates (55%), my take on this is that there is an assumption in RDR that posit that the environment of individuals are independent to each others and don't influence each others in correlation to the relatedness which since it's violated bias the estimate downward. (https://hereticalinsights.substack.com/i/146616013/disassortative-mating)
"Moreover, our estimates of 0.76 and 0.55 can also be compared to estimates from GWAS and WGS data. For height, the SNP-based estimate is about 0.55–0.60 (ref.41) and the WGS estimate ~0.70 (ref. 42; but with a large s.e. of ~0.10). These estimates imply that for height there is substantial genetic variation not captured by either SNP array and, to a lesser extent, sequence data, presumably ultra-rare variants (frequency <1/10,000 not included in ref.42)"
Doesn't assortative mating hit twin studies and Sib-Regression about the same amount, so the uncorrected results should be comparable?
You are right, my bad, from the supplementary note 4 of Kemper 2021 (https://www.nature.com/articles/s41467-021-21283-4) they use the same equation to correct for AM and there is still a 0.1 gap in heritability for the two methods.
Sidorenko 2024 use a slightly different adjustment method to correct for the shared environment estimate becoming negative after correcting for assortative mating (another mystery to solve).
While writing this I just saw that the estimate twin heritability for height in Kemper 2021 is lower than the heritability for height in Sidorenko 2024 using sib-regression. Missing heritability solved I guess.
Gusev has a point when he says that when adjusted for assortative mating some of the high twin estimate for height or even sib regression get near or above 1 (when you read the supplementary note of Sidorenko 2024 the crude adjustement for a r=0.23 assortative matting in height push the estimate to 0.98 ! )
Turkheimer's quincunx metaphor is my personal mental model for the missing heritability phenomenon. A blog post is here: https://ericturkheimer.substack.com/p/a-speculative-explanation-of-the-e53 though you'll have to look at the few previous posts to fully understand it. It seemed like you dismissed this very briefly under non-additive effects and I don't have the technical background to grok those papers, but Turkheimer takes this hypothesis seriously so I figure it's serious.
Here's my summary:
Imagine a quincunx (aka a Galton board). You are a ball falling down the board. It bounces off a lot of pins, and it eventually falls into a bin at the bottom. That bin at the bottom is the outcome we're interested in (EA, IQ, etc). But the pins aren't random -- the pins are your genes, environmental variables, etc. So some pins are biased left or right, which influence the bin you end up in.
What Turkheimer shows is that in this model, if two people have the exact same quincunx (the exact same weights of pins), then they are very likely to have similar outcomes. Those are identical twins. If all your genes are the same, you are very likely to be similar. This reinforces the hereditarian position, and also makes a lot of intuitive sense. (One tricky part here is that it must be the case that mostly-the-same quincunxes (siblings, etc) cause similarity as well to a lesser degree. That's the biggest challenge to this mental model.)
But Turkheimer also shows that biases in pins are very very hard to detect in the non-identical-twin situation. Each pin only has an effect in the context of the entire system. For one person, a particular pin could have a large impact (maybe all the pins to the left bounce left, and all the pins to the right bounce right, so that pin matters a lot) but for another person that same pin has no influence at all (the pin to the left bounces right, the pin to the right bounces left, so it's like a funnel and that pin doesn't matter). So "in the wild," away from the twin context, it is very hard to come up with pins that matter in a GWAS type of way. But pins do matter! Genes do matter! They just happen to matter in this way where the whole package has a large influence on an individual, but individual pins are really hard to pick out in a practical way.
Turkheimer would emphasize that this is a simplified mental model. Our genes are not actually a quincunx -- but the interactions are much more complex, so if he can show that a quincunx has that sort of irreducible complexity, genetics must be even more complex.
I find this mental model really interesting and I actually wrote a blog about how I find it a helpful way to think about education and schools with some GIFs modeling the quincunx part if anyone is interested: https://fivetwelvethirteen.substack.com/p/the-quincunx
I think Turkheimer’s analogy here, given that he says that some of the quincunx pins are environmental, is dramatically failing to capture the fact that adult identical twins who were raised by different families in different cities are very obviously similar in all kinds of ways, indeed about as similar as if they were raised in the same family. (…With a few caveats & exceptions that I discuss in §2.2 here → https://www.lesswrong.com/posts/xXtDCeYLBR88QWebJ/heritability-five-battles )
That's a fair criticism of Turkheimer's position but I think you can rescue the broader idea by just saying that the quincunx is genetics only and the final bin is the sum of the genetic influence. To me, the main insight is that a complex system can have emergent properties that aren't easily correlated with the individual components of the system. And that's the core of the missing heritability problem: when we look at the whole system with twin studies we see high heritability, but when we try to find specific causal genes it becomes really messy.
I do think non-additive genetic effects are very important for adult personality, mental health and behavior (but not for other things like height and blood pressure). I talk about that a bunch in my post; see excerpts that I copied into a different comment → https://www.astralcodexten.com/p/missing-heritability-much-more-than/comment/129514879
Amazing post, thank you.
Something that comes to my mind, having just read Pearl’s Book of Why, is whether collider bias or other common problems of regression in the hands of folks not very savvy about causal inference might explain the inconsistencies in the findings.
I mean, there is a lot of regression going on here, a lot of moderating and mediating variables, so it seems easy to make a mistake with the regression model.
Nice, I am a bioinformatics PHD working at least somewhat with GWAS and SNPs and from a Gell-Mann amnesia effect perspective this post was very reassuring.
My guess is also that the rare/ultra rare variants/combinations are responsible.
Why is "principal components" in scare quotes?
Lots of identical twins were raised to deny they were identical.
For example, in the 2004 Olympics Paul Hamm won the gold medal as the best all-around male gymnast while Morgan Hamm was 5th. But Paul and Morgan's parents doubted they were identical because their hair whorled in different directions.
Similarly, the nearly 7 foot tall Lopez twins of the NBA have done a great job over the decades of acting non-identical: they wear different hair styles and different clothes.
My impression is that some identical twins like being identical and work at being even more similar (like those two Australian ladies who were in the news recently), while others work at emphasizing their differences: e.g., the basketball Lopez twins display different personalities in social media.
The movie "Sinners" seems to do a pretty realistic job of displaying the identical twin characters as quite similar, but not wholly the same.
Great post!
I think GxG is basically impossible to measure, so saying nobody found any GxG yet isn't really meaningful. It's harder to measure than rare variants; why believe in rare variants but doubt GxG?
Also, note that additiveness is sensitive to nonlinear functions of your measure. For example: if heritability of IQ is purely additive, then heritability of IQ^2 (the square of IQ) would not be. To posit no GxG at all is to say that the IQ scale is perfect and cannot be squared (and EA scale, and BMI, etc too).
A few final notes:
1. I think intuition from real life strongly suggests that everything is GxE (gene environment interaction); think, for example, about obesity. All the models assume GxE is zero in order to calculate the percent heritable.
2. You say assortative mating cannot bias downwards the heritability estimates from twin studies. That's only true if the twin studies don't already adjust for them! Many hereditarians cite 80% h^2 for IQ, but they get this by already adjusting for AM (and maybe even for attenuation). The can easily over-adjust! That would bias the estimates upwards!
3. Related to 2, I noticed that you quoted 60% for the typical twin-study heritability estimate, even though hereditarians (including you in the past) like to cite 80%. That's a substantial gap. Are you saying you now believe the 80% is wrong? Can you be more explicit about this disconnect?
Obesity obviously has a huge environmental component as can be seen in changes in obesity over time. For example, after being very scrawny from 1945 onward, West Germans from about 1955 onward suddenly got very fat (think of Augustus Glump in the 1960s book "Charlie and the Chocolate Factory.") But then they got skinnier again in the 1970s. Now I'd imagine they are fatter again than their parents were.
I agree, but the same logic applies to IQ, as can be seen in changes in IQ over time.
Beyond the fairly global Flynn Effect, I can think of only a few well documented cases at the national average IQ level of significant improvements relative to other countries: Japan and then South Korea. Both big gains in average IQ coincided with huge increases in average height over the generations.
Otherwise it's hard to see the kind of rapid national changes that were evident in West Germany's level of obesity.
I wouldn't be surprised if there have been more relative changes in average height than in IQ: for example, I was in my mid-30s before I heard that the Dutch were really, really tall. That hadn't been a stereotype in the 1970s yet, in part because the Dutch weren't that tall yet.
So, while I certainly believe that environment can have a big influence on IQ, it's hard to find all that many examples.
Well, I don't know why we are talking about countries improving relative to one another instead of in absolute terms. It seems like if you can improve IQ in absolute terms, then it is environmentally malleable, no?
Also, I don't know that we'd even know about it if there were changes in the rank order of the IQ of countries over the 20th century; IQ is much harder to measure than height or BMI (there are many different IQ tests, they need to be translated or you need to use a "culture fair" test, testing itself takes hours instead of seconds, and it is much more subject to selection effects such that it's unwise to take a convenience sample of (e.g.) immigrants or something, etc.)
Large changes in IQ over time, as in the Flynn Effect, don't appear to reflect actual changes in general intelligence. See e.g. https://www.cremieux.xyz/p/the-demise-of-the-flynn-effect, https://www.cremieux.xyz/p/a-requiem-for-nutrition.
That's irrelevant
we're talking about whether IQ is environmentally malleable. You may say that the answer is "yes but general intelligence is not". That's a fine answer. But "yes" part is what I'm asking about. I don't care about intelligence, I care about IQ. Is IQ environmentally malleable? If so, why do studies say the answer is no?
I'd say that IQ is environmentally malleable, but there isn't much evidence that it is easy to close racial disparities in average IQ scores (which is what people are mostly talking about and why so many science denialists get mad at people who know something about IQ.) Instead, what seems to happen most times and places is that improved environment leads to everybody's test scores going up.
Will try to respond to 1 in a Highlights From The Comments post.
Re 60% vs. 80%, I admit I haven't been very careful in keeping these numbers separate. My impression is that twin studies usually find somewhere between 50-80% with the differences being age (older = more heritability), assortative mating correction, which IQ test you're using, and whether you've already tried to correct for measurement error in the IQ test. I don't think there are "right" answers in what age or IQ test you should use, or in how many things you should try correcting for, so I don't particularly care about the differences between these estimates and mostly just say whichever one is on the top of my tongue at any given time.
Thanks. I think everyone usually talks about adult IQ and everyone wants to correct for AM (which is considered a source of bias for twin studies). As for measurement error, that one is admittedly trickier, because with other measures we don't usually adjust for it: measurement error exists for height or BMI too, but nobody tries to correct for it.
To make a fair head-to-head comparison with RDR or whatever, you'd surely want to adjust for AM, right? And the UK Biobank uses adult IQ tests, though arguably not very good ones. So we definitely want to compare to twin studies which use adult IQ and adjust for AM. Educational attainment is also about adults, though now that I think about it I don't know how they handle young adults who are not yet done their schooling.
(I'm suspicious of over-adjusting for measurement error, so I'd rather studies not try to do it, but it should definitely be noted that (from my understanding) the UK Biobank IQ test should have higher measurement error than more standard IQ tests.)
I find the idea of reading modern scientific findings or social innovations into ancient myths and religions very compelling. As if the ancients through cultural evolution were able to find fundamental truths about reality without having to actually understand them. In light of that interest, here we go:
The ancient Greeks had the concept of generational sin. When someone committed a truly horrendous sin, like feeding your brother his own children as an act of revenge (https://en.wikipedia.org/wiki/Atreus), the gods may not punish the man directly, but leave punishment for a future generation.
Atreus (Progenitor of the house Atreides. Dune anyone?) was grandson of Tantalus, infamous for feeding Zeus his own son, and being punished by eternally having food and water just out of reach (origin of the word: tantalizing). Atreus echoed his grandfather's sin by feeding his brother his nephew's corpse and usurping his throne. Atreus' son, Agamemnon (famous from the Iliad) killed his own daughter, Clytaemnestra, in order to be allowed to attack Troy. In punishment for that, he was killed by his own wife upon returning victorious from Troy. His son, Orestes (grandson of Atreus) was given the "divine madness" (basically Schizophrenia) as punishment and it was only through the gods intervention that the cycle of inherited sin was synbolically broken. It was only ultimately ended when a new lineage, the Heracleidae (descendants of Herakles) invaded and taking over their lands, loosely representing an invasion of Northern Dorian peoples taking over the Achaean society that was there.
The lesson I take from this is, the Greeks understood the heritability of preponderancy for antisocial behavior. If your father or grandfather did something unspeakable, you were by extension unclean, and shunned by society because of that. If there is some actual truth to this, and children of men who are so antisocial as to practice cannibalism or other terrible sins, were poisoned by a Miasma that would carry over between generations, then perhaps there was a collective social benefit towards shunning the children and grandchildren of heinous criminals. The societies that practiced this, while fundamentally less just on an individual level, would have been able to better weed out antisocial behavior, and in the long run, outperform societies that didn't have this cultural meme.
The lesson at a societal level is that if your rulers have 6 generations of inherited sin and insanity, you will be invaded by your neighbors and made into slaves.
Of course in the modern day we've moved past that, and if you think hard enough I think you can come up with a "reasonable" justification for literally any social practice, but I always find it interesting to think about.
It's in the Old Testament too: "for I, the Lord your God, am a jealous God, punishing the children for the sin of the parents to the third and fourth generation". I've been similarly impressed by the insight that Buddhism had into the functioning of the mind (anatta etc). Phenomenology and careful thinking can actually get you pretty far.
> Atreus (Progenitor of the house Atreides. Dune anyone?)
Not quite correct; -ides is the Attic (I think?) Greek patronymic and so Atreides only refers to a son of Atreus. (There are two of them, Agamemnon and Menelaos, both called "Atreides".)
The whole lineage, the metaphorical rather than literal sons of Atreus, would be called the Atreidai.
True.
I was more referring to House Atreides from the book Dune, which are canonically descended from Agamemnon as a bit of rhetorical flare to make my comment slightly more interesting to the reader who has read the book or seen the movie.
Educational Attainment is obviously both part Nature and part Nurture (e.g., did your ancestors pony up enough money for you to finish college?)
Does it take any money to finish college? I thought we had a robust system of student loans that will happily finance any degree at any school no matter what the prospects of getting paid back might be.
Opportunity cost.
Some people are responsible enough to realize that dropping out would be better for all concerned. Similarly, I can recall a public high school teacher during the Housing Boom in 2006 telling me that many of her more impressive male students were dropping out to work construction.
1. I wonder when we're going to have a good computational model of personality to just directly track these issues on a gear level, like what tradeoffs are solved in what proportion, what inefficiencies are present in the Turing machine that corresponds to one's intelligence, etc. (Sounds easy, right?..)
2. Is it possible that any single study (i. e. the Iceland study) introduced a trivial calculation error, or a different calculating method that at least partially accounts for the discrepancy? Are the datasets in question open-source, did people reproduce their calculations and make sure it's not that?
I wonder how much in vitro fertilization might impact more recent twin studies. Also the fact that we have mitochondrial DNA as well. So more and more data to feed into the maw of the ai models. De novo mutations also seem to impact certain genetic areas as well. Maybe we have a gene that predisposes to de novo mutations.
Banned for this comment - controversial/insulting assertion with no evidence provided.
Consider syndrome vs. disease. It seems to me that the likely cause is that most of the things being measured have multiple independent ways of causing the measured result, and the genes that facilitate one of them don't necessarily facilitate another. Dawkins calls this teams of genes that are "good traveling companions". E.g. both Tibetan and some Peruvians have genetic adaptation to high altitude, but they do it in different ways.
So under this assumption (which I think is fairly reasonable) if you measure one gene, you're measuring one part of a collection that *IF PROPERLY ASSEMBLED* would facilitate one trait. But someone else presenting that trait with a different underlying approach would not find that gene helpful.
That's just non-additive effects. We ignore those for pretty good reasons; they cannot be picked up by natural selection in a system where reproduction occurs via meiosis. You'd need to evolve a way to ensure that they always passed on together. (The simplest way would be to drive them to total fixation, but that's hard to do.)
Note that it isn't true that providing Tibetan adaptations to a Peruvian, or Peruvian adaptations to a Tibetan, wouldn't be helpful in supplying additional oxygen. It would. The two populations are doing different things, but not things that conflict with each other.
(The Tibetan adaptations are much better; I believe the Andean ones have some potentially undesirable side effects, so a Tibetan might not find them helpful _on net_. An Andean would benefit from Tibetan admixture.)
But when you are trying to explain variation in a population, shouldn't sources of variation that natural selection doesn't work (efficiently) on be *over*represented? You would expect advantages that can be readily selected for to be present in everyone, and so contribute nothing to variation.
No. That is not what you would expect, at all.
You don't expect natural selection to remove variation. The point of having the variation is that natural selection can act on it if the trait becomes more important than it is now.
For example, it isn't the case that there's one correct height for everyone.
To become fixed, a trait needs to be subject to extremely strong selection.
Note that the problem with your theory applies at both ends: just as the benefit of having variation is that selection can pick up on it in the future, the way that variation came to exist is that selection picked it up in the past. There isn't a mechanism by which adaptations that are invisible to selection can persist in the gene pool.
"An alternative way of validating twin studies... is to check them against their close cousins, adoption studies..."
Using adoption studies to "validate" twin studies seems completely backwards to me. Adoption studies are so obviously flawed that I do not see how they could possibly act as a validation set for anything!
The population of children going into adoption are wildly unrepresentative of the general population of children, unless that population is "children who go into the adoption system". The population of parents adopting children are obviously unrepresentative of the general population of children, unless that population is "parents who are adopting children". Being an adopted child to a set of parents is not like being a biological child to the same set of parents. Losing or being given up by birth parents is generally kind of a big deal.
If anything, the fact that twin studies give the same answers as adoption studies should make us more skeptical of twin studies!
In any case, I applaud Scott for both a) being totally transparent about his completely bonkers epistemic commitment here, and b) giving a good-faith description of the current evidence against his aforementioned bonkers epistemic commitment.
The claim isn't that adoptive kids are representative of the general population!
The claim is that you can compare the correlation between adoptive kids and their bio parents, vs. those same adoptive kids and their adoptive parents. As long as there's any variation at all among adoptees, you can use this to see whether it's the genes or the environment influencing them.
For example, if the child of a genius gets adopted by dumb people, and grows up to be a genius, that provides some evidence that intelligence is genetic rather than environmental.
You might additionally have a claim that adoptive people get traits in a totally different way than non-adoptive people, but I think the burden of proof is on you there and that has nothing to do with the question of whether adoptees are "representative".
Apart from the sometimes-questionable validity of the studied “trait” (e.g., “general intelligence,” “personality,” “schizophrenia”), however, others have shown that adoption studies are subject to numerous environmental confounds that some critics argue invalidate genetic interpretations of the results. That is, like family and twin studies, behavioral genetic adoption studies are unable to adequately separate (disentangle) potential genetic and environmental influences on behavior. These confounds include that most adopted children (1) shared a prenatal environment with their often-stressed birthmother during sensitive developmental periods; (2) were reared for a certain period by their biological parent(s); (3) suffered a rupture of attachment bonds with the biological parent(s) who gave them up for adoption (children grow up feeling abandoned); (4) may have been placed between separation and adoption into unstable orpsychologically/developmentally harmful environments, such as foster homes and orphanages; (5) share or potentially share with birthparents similar socioeconomic status, physical appearance, ethnicity, culture, religion, and so on; (6) were not randomly placed into available adoptive homes (agencies often selectively place adoptees into homes based on SES and the child’s perceived genetic background); and (7) were placed into adoptive homes of restricted socioeconomic range. In addition, adoption studies are subject to research/publication issues and confirmation biases coming increasingly to light in science’s replication crisis.
In addition to the abovementioned confounds, behavioral genetic adoption study researchers compute IQ correlations between different "Flynn effect" populations. Biological and adoptive parents are a generation older than adopted children, meaning they were born at different Flynn effect “massive IQ gain” starting points 25 or so years apart. Moreover, because birthparents are often unwed teenagers, they are typically closer in age to the child they gave up for adoption when compared with the older adoptive parents. In Plomin and colleagues’ Colorado Adoption Project IQ study, for example, birthparents “were younger on average (20 years) than the adoptive parents (33 years)” (Plomin et al., 1997, p. 443), suggesting additional Flynn effect confounding.
I go to one of those fundamentalist churches where people adopt kids and raise them piously. I think everybody knows by now that there's a big danger the kids will turn out badly compared to their siblings, even though there's considerable success in training them to behave well as kids. These adopted kids are often foster care kids, and that's going to add a lot of noise to any study. They don't have normal environmental differences; they have huge ones-- fetal alcohol syndrome, autism, abandonment as babies, abuse, . . . So for finding the environmental share for the US on average, they're not ideal.
Throughout this entire article, the “computational intractability” buzzer was ringing in my head, yet you only got there at the very end.
Isn’t it obvious that the interaction of genes quickly reaches a combinatorial explosion?
Is there any psychometric data set that comes even close to covering even the tiniest fraction of theoretical gene combinations?
Don’t we have good ideas for addressing the theoretical limits of these gene studies?
I am alarmed at this statement about why we care if gene studies show that heritability may be misunderstood.
“Not doctors. So far this research has only just barely begun to reach the clinic. But also, all doctors want to do is predict things (like heart attack risk)”
We absolutely care if doctors over-treat or apply dubious preventative treatments based on unreliable associations.
Do I want surgery removing my prostate and rendering me incontinent because of a false read on my prostate cancer risk?
Correlation versus causation always matters when a reading is used as the basis for an intervention.
The next paragraph criticizes the use of these scores in epidemiology. Actually, even if all you have is variable with a correlation, it can be useful as an instrument in multiple regression analysis.
I'm not sure I understand.
Suppose that black people have higher heart attack risk because they have worse diets, and genetic studies pick this up and (accidentally) identify the gene for black skin as a gene for heart attack risk.
This is bad if you're doing research or embryo selection. But a doctor would do a genetic test on their black patient, see that it said they had higher heart attack risk, put them in a high-risk category, and maybe give them more treatment. And that would be correct, because in fact black people do have higher heart attack risk! Even though the gene isn't causal, it's having the correct effect.
This might be a problem if the doctor already knew that black people had higher heart attack risk and so double-counted on the result of the screening (eg the doctor thought they're black AND they have bad genes so they must have two risk factors). But in real life it's rarely something as obvious as "they're black" and more some kind of hard-to-describe sub-sub-sub-group.
But if black people only have worse diets *on average* (say because they're economically disadvantaged *on average*), the doctor will wind up over-treating wealthy black patients with balanced diets for no reason.
Exactly. That is my concern. You maybe a black person who is socioeconomically “white” with a healthy diet andhabits. Such people could be systematically over treated for certain medical vulnerabilities based on a spurious genetic correlation
Obviously the best thing to do is completely understand the causal graph connecting race to diet to heart attack risk.
But if you can't do that, I don't think this is any worse than any other genetic situation. Every gene increases your risk of heart disease by some probability. Suppose people with some gene have an 80% chance of getting heart disease. If your doctor treats you, there's always a chance that you would have been in the lucky 20%, and this treatment was pointless. If the doctor could magically know whether you were the 80% or the 20%, they should do that. Otherwise they should just do the best risk-benefit calculation they can.
Likewise, suppose that 80% of black people get heart disease because of bad diet, and 20% eat a good diet and don't get heart disease. Since we're inventing this hypothetical and we know this, we can avoid treating black people with good diets. But if the doctor doesn't know this, them picking up on this "gene" (actually a gene associated with African ancestry) and detecting that it gives you 80/20 chances is no different than any other gene.
Wealthy black people have a much more elevated risk of heart disease compared to wealthy white people. It is a true genetic effect, not cultural, by the best of our scientific evidence.
Is there a source? High blood pressure for sure. But is there an elevated risk taking into account all the metrics doctors normally look at?
Many sources, heres one.
https://spssi.onlinelibrary.wiley.com/doi/abs/10.1111/sipr.12042
I thought this little-read article (written by an astrophysicist) does an excellent job of explaining why we wouldn't expect GWAS studies in their current form to get anywhere close to "true" heritability: https://coel.substack.com/p/gwas-studies-underestimate-the-heritability. Notably, the author doesn't rest his argument on the "rare genes of large effect" factor.
To roughly summarize his article: We would expect the code for developing a human-level intelligence to be extremely complex. There are ~3 billion nucleotides in the human genome; these GWAS studies rest on the assumption that only a few hundred to a few thousand SNPs are entirely responsible for creating human intelligence. That seems wildly inaccurate. As the author argues, could you write code for developing a human-level intelligence with only a few thousand instructions? ChatGPT is based on hundreds of billions of neural-network weights (though admittedly this is the end-product, not the "recipe" for producing it like the human genome). If intelligence is based on tens of thousands (or even more) SNPs, then our current GWAS studies are simply statistically incapable of finding them.
On top of that, GWAS is only looking at a very simplistic model of gene-behavior causation. It assumes an "additive" model of heritability which simply sums the measured effect of each SNP, whereas in reality the recipe for intelligence is likely caused by subtle and complex interaction effects among genes as well. (The author also mentions that SNPs, which are all GWAS studies measure, are only one type of genetic variation. This seems like it could be important but I don't understand what it means.)
In contrast, twin studies take into account the complexity of developing intelligence by simply observing the output of the incredibly complex underlying genomic causes. It's therefore not surprising at all that they give much higher heritability scores, whereas GWAS studies, limited as they are, give nothing more than a lower bound for heritability.
How do they adjust EA for non-college education? In my job I interact with many elderly. I ask them what they did. I live in a high-tech area so frequently I hear 'engineer'. But when I ask them where they studied, I am more likely to hear 'I got out of the army and went to work for Sunstrand as a mechanic and they trained me' than to get the name of a university. Very smart folks, but formal education is high school.
Total non-expert speculation here. Has anyone looked into whatever the genetic equivalent of “sequence of returns risk” is? In the market it matters enormously if there is a crash immediately after you retire versus a decade after you retire and your investments have had extra time to grow.
I.e as the body is built, different genes will apply in sequence. It could be that early-in-the-sequence genes (about which we currently know little) will have an outsized effect. It may be that some genes only activate or have an impact conditional to other genes activating or having an impact. Etc. Or perhaps from another perspective the mistake is looking at the genome as if it is static and one dimensional when in fact it is self-referencing and applies across dimensions of time and space.
> In the market it matters enormously if there is a crash immediately after you retire versus a decade after you retire and your investments have had extra time to grow.
I'm always shocked at the widespread assumption by financial "health" "experts" that the goal of investing is to end your life with a net worth of zero, leaving nothing to your heirs.
I’m not sure whether this deserves a response or it just meant as a flippant comment. While there certainly is a “you can’t take it with you” attitude amongst some in the retirement savings space, I think many people don’t wish to burden their heirs with medical debt and still more are using consumption rules of thumb that are overly cautious. To bring this back to genetic fitness, genes that take you close to dying but don’t kill you before reproductive age are harder for nature to filter out out than genes that do lead to more death before reproductive age. The sequence of the effect matters greatly.
I'll be honest, I haven't read word for word your post, but I read well enough. I will continue to read as i have time. I think the answer doesn't exist yet but will exist soon considering ai. The true answer is having behavioral patterns linked to endocrinology. Many don't take psychology seriously, but it is what we have. I haven't seen any reliable studies regarding hormones and behavioral traits.
Regarding educational attainment, I think it's also important to consider diet. If people do not get enough protein and fats, they will not be at their peak intelligence. Addictions of any sort will lower iq.
I am of the mind that virtually everything is inherited. Even broad behaviors amongst the population. We can connect hormone levels to DNA and hormones to behavior.
Great post.
Was surprised to find that Scott used o3 to help research this, and found it helpful, despite the continuing hallucination problem. Let's say it gives you certain hallucinations you know you need to check, like citations, and you check them and strike that, fine. What if it's negatively-hallucinating, insistent that something does NOT exist when it does? What if it has numerous unknown-unknowns that it doesn't realize are relevant to the research question?
It seems like you probably need at a *minimum* to have Scott's level of mastery and familiarity with the subject going into the project in order to have any chance at spotting the places that o3 screwed up, and knowing why, and knowing where to look to correct it. The entire field of research needs to have been reduced to known-unknowns for you the author before you can safely employ the assistance of AI tools.
Here he had extensive layers of human review/assistance by specialists in the field, and had previously researched the topic. But I just don't know why anyone would bother to use AI for research when the output is clearly unreliable absent such review and input, in ways that a researcher with incomplete knowledge won't be able to always detect.
To point out, the existence of unknown unknowns is unlikely to get *worse* with AI. If the AI explicitly denies something, that's a thread you can search on. If it doesn't mention it you are identically in the position of not using AI. The only way this can make it worse is if the AI somehow comes up with similar meaning terms that are false then unreasonably generalizes in a hard to fathom way, but then you'd have to explain why those hallucinations would be worse than for example, finding a bunch of confused politically motivated parts of research.
Twin study consistency in EA might be because EA used to be a better measure in the past and twin studies are disproportionately from the past?
Why do we expect genes to have independent linear effects? From gene to protein to trait to life outcome, there are many nonlinear effects and interactions that we could imagine — is the problem really our imaginations? Hunting for genes that linearly independently correlate with causally remote life outcomes seems like looking for your keys under the lamppost. But I’m not really qualified to dig into the studies that you say disprove interactiond.
Not an expert, but I think there are a few arguments for this. One is that additive genetic models work pretty well: phenotype variability seems to depend linearly on degree of relatedness (MZ twins correlate twice as much as DZ twins, which correlate twice as much as half-siblings). Secondly, if nonlinear effects were large then phenotypes wouldn't be normally distributed: they'd have longer tails or be skewed or multimodal more than they appear to be. And finally from an evolutionary design perspective the system is going to be much more stable if it's linear: phenotypes that require a complex interaction of many independent alleles probably aren't going to do that well with recombination.
One easy possible confounder for twin studies would be if some aspect of your early childhood environment mattered a lot for some trait (IQ, kidney function, BMI, whatever) and also a lot of that environment was a function of your appearance. That would give you inflated heritability results for identical vs fraternal twins.
It doesn't seem so likely to me that this is important, though--it seems like adoption studies make a pretty strong argument that there's not much impact of variation in early childhood environment within normal middle-class-and-up norms and most measurable outcomes. (Raise a kid in a cave, beat him regularly, and never show him a book, and you'll probably depress his IQ. But letting him watch 20% more TV or taking him to amusement parks instead of museums on vacation isn't likely to have any noticeable effect.). But I guess if really cute kids get extra attention at some critical point in their development relative to ugly kids, it might have some effect?
Well, I never thought I would see this day. Finally, the smart people are coming around to what I have been claiming for decades. Inheritability is an interaction effect, thus the main effects (variation due entirely to genes or learning) are not directly measurable. Also, the twin studies are deeply flawed.
Let's start with twin studies. They have problems with their independent variables (IV's), their dependent variables (DV's) and their analysis.
IV's like personality traits and IQ test scores are highly controversial. We don't know exactly what IQ scores measure--except that it isn't "intelligence" the way most people mean that term (something like "an individual's inherent capacity to solve real world intellectual problems"). No one has proven that such a capacity exists as a coherent cognitive process, let alone that IQ scores measure it. Every other personality inventory suffers from similar problems.
The DV is correlation in such traits between pairs of siblings. This is measured two different ways: by degree of genetic similarity, compared to degree of environmental similarity. One of these two measures (genes) is much more precise than the other. These days, they can not only tell you how many genes two siblings have in common, but often which ones.
But not environment. The original twin studies used household to determine similarity of childhood environment. This is far too crude. Many twins are adopted into different households, but by members of the same extended family (aunts, uncles, grandparents, etc.). This means that the twins will know each other, often live in the same neighborhood, go to the same school, play with each other, etc. Since these things are largely undocumented, the data is contaminated, and we can't know by how much.
Finally, there are strong theoretical reasons to believe that evolutionary effects should be interactions between genes and the environment. The environment, after all, is the nature that selects. Therefore one would expect that any species will evolve traits that engage features of the environment, which is the mechanism by which an organism can gain an advantage. IQ is pretty useless if it doesn't give one a decisive advantage *somewhere*, and if it's mostly useless, why would it be selected?
In terms of analysis, in the presence of a strong interaction effect, the main effects (the effect of each key variable by itself, in this case genes and the environment) lose their meaning. It's easy to illustrate: Imagine that intelligence is 100% inherited. Two children with identical genes grow up in two different locations: one is a wealthy suburb in a household with plenty of resources, the other in a war zone, rife with disease and natural disasters. Do you expect these two to do equally well on IQ tests, Educational Achievement, or any other measure of life performance? One is pretty clearly at risk of leaving fewer descendants than the other, but remember, their genes are identical. There's nothing here for natural selection to grab onto. The main effect of genes is meaningless.
What this implies is that inheritability scores are probably meaningless. Intelligence isn't 60% inherited and 40% learned. It's probably something more like 80% interaction between the two.
Good lord, have a gander at the Sib Regression error bars in part 4! I believe the 95% error bars for menarche contain ~95% of all possible values.
Another random thought that I suspect people in the relevant field have probably already looked into: (I'm a cryptographer, what the heck do I know about behavioral genetics?)
Suppose a lot of outcome differences between people are due to pathogens. For example, if IQ regularly gets depressed a few points if you get a particular viral infection at a particular critical point in your development. (I think there's some reason to believe this does happen, because the month in which you're born does seem to have a small correlation with life outcomes, and how that overlaps with flu season is a reasonable guess about the mechanism by which that works. Also, this is likely to be some of why breastfeeding correlates with good outcomes for kids, though a lot is probably tangled up with social class/education of the mother.)
If this is true, maybe the critical genes to look at for a lot of heritability are the ones that determine your immune response (which MHC receptors you have, what genes you start from when you shuffle stuff around randomly to get T-cell receptors and antibody binding regions, variants in innate immune response) and maybe any rare variants that confer some immunity to common circulating viruses.
There are like a gazillion viruses that circulate all the time, often giving us infections we never even notice, sometimes giving us annoying but not serious symptoms (colds, stomach bugs, cold sores, warts, etc.). And we know that some of these (rubella, polio, zika) can do nasty developmental things to fetuses, babies, or small children.
That would track with at least some of the Flynn effect as sanitation and vaccination made people in-general healthier.
This problem should be getting as much attention as Hubble Tension, if not more.
Can we induce this measurement phenomenon in rats, or worms, something where we have control of the genome? A quick Google suggests identical twins are rare in nature, but we can possibly produce them artificially for some animals.
Nice overview! A few points I wanted to expand on.
1. I think the post conflates gene-gene and gene-environment interactions; the latter (specifically interactions between genes and the "shared" environment) also get counted by twin models as narrow sense heritability. While I agree there is very little evidence for gene-gene interactions (particularly dominance, as you cite [and, interestingly, twin/adoption studies actually forecast a huge amount of dominance -- another discrepancy we do not understand]) there is quote substantial evidence for gene-environment interactions including on educational attainment (see Cheesman et al: https://www.nature.com/articles/s41539-022-00145-8 ; Mostafavi et al: https://elifesciences.org/articles/48376), IQ, and BMI. In fact, Peter Visscher led a paper that came to the conclusion that twin estimates for the heritability of BMI are very likely to be overestimated by gene-environment interactions (https://pubmed.ncbi.nlm.nih.gov/28692066/). A large amount of GxE plus some amount of equal environment violation seems like a very plausible and parsimonious answer to the heritability gap.
2. You mention epidemiologists being the biggest losers of stratification in polygenic scores, but I think it is important to note a related group: the people who take polygenic scores trained in one population (with a ton of stratification) and directly apply them to other populations to make claims about innate abilities (see: (https://theinfinitesimal.substack.com/p/how-population-stratification-led). This is especially true for Edu/IQ GWAS, where every behavior geneticist has been screaming "do not do that!" since the very first study came out. People like Kirkegaard, Piffer, Lasker, etc. (and their boosters on social media like Steve Sailer and Cremieux) dedicated their careers to taking crappy GWAS data from and turning it into memes that show Africans on the bottom and Europeans on the top. These people also happen to be the court geneticists, so to speak, for SSC/ACX. I don't mean to come off as antagonistic and I'm sure some people will see this comment and immediately discount me as being an ideologue/Lysenkoist/etc so it does my broader position no favors, but this stuff has done and continues to do an enormous amount of damage to the field (including the now complete unwillingness of public companies like 23andme to collaborate on studies of sensitive traits).
3. I'm going to gently push back against the hereditarian/anti-hereditarian framing (which I understand is probably here as shorthand and scene setting). I am personally interested in accurate estimates that are free of assumptions. I believe twin study estimates are of low quality because the assumptions are untestable, not because they are high. I also think the public fixation on twin studies has created some real and damaging anti-genetics and anti-psychiatry backlash and wrong-headed Blank Slate views. People hear about twin studies, look up the literature and find that peanut allergy (or wearing sunglasses, or reading romance fiction) is estimated to be highly heritable and have minimal shared environment (https://lymanstone.substack.com/p/why-twin-studies-are-garbage), start thinking that the whole field is built on nonsense, and end up at quack theories about how schizophrenia is actually a non-genetic fungal condition or whatever. I've been very clear that there are direct genetic effects on essentially every trait out there, including behavioral traits and IQ. If someone were to run a large-scale RDR analysis of IQ tomorrow and got a heritability of 0.9 and it replicated and all that, I would say "okay, it looks like the heritability is 0.9 and we need to rethink our evolutionary models". If anything, large heritability estimates would make my actual day job much easier and more lucrative because I could confidently start writing a lot of grants about all the genome sequencing we should be doing.
4. Lastly, it's not clear to me where the conclusion that well-validated twin studies converge on "similar results" is coming from. To take one example: the leading lights of behavior genetics (Deary, McGue, Visscher, etc) ran a study looking at the relationship between intelligence and lifespan (https://pubmed.ncbi.nlm.nih.gov/26213105/). This is a nice study for us because they put together three large, modern, twin cohorts with IQ measurements, but the heritability of IQ was just a nuisance parameter for them, so they had no reason to scrutinize the findings or file-drawer them. If we look at their MZ/DZ correlations in Table S6 we find that the heritability of IQ was 0.36 in the US sample; 0.98 in the Swedish sample; 0.24 in the Danish sample; and ... 0.52 on average. In other words, all over the place (but averaging out to the nice "half nature half nurture" result you see in books); the authors themselves used an AE model in Table 2 and reported a range of 0.20 to 0.98. This is far greater than the variability we see with GWAS or Sib-Reg, so what are we to make of that?
"People like Kirkegaard, Piffer, Lasker, etc. (and their boosters on social media like Steve Sailer and Cremieux) dedicated their careers to taking crappy GWAS data from and turning it into memes that show Africans on the bottom and Europeans on the top. "
You mean East Asians on top?
Would you care to quantify in some way what proportion of their careers is dedicated to what you say?
Richard Lynn placed NE Asians as 5-10 pts higher than Europeans. But he placed SE Asians down in the mid-80s, though. But didn't Cremieux argue on X that Europeans were higher? Maybe I'm misremembering.
There are several academic journals that cater to studies on the heritability of IQ and other social traits. Some, like Mankind Quarterly, are funded by conservative foundations. Lynn had an academic position, but it was funded by the same foundation that funds Mankind Quarterly. Lasker, although he claims an association with Texas Tech University, the Guardian article about him said he wasn't on their faculty list. He seems to be making his living working for a foundation, whose funding is unclear. Happy to update my priors if anyone has better info, but it seems like the IQ-heritablity researchers work in an incestuous little bubble.
https://en.wikipedia.org/wiki/East_Asia
The general HBD consensus seems to be East Asians 5 points above Europeans, except in tests of verbal intelligence.
As to your second paragraph, it doesn't speak to the specific claim that they dedicated their careers to "taking crappy GWAS data from and turning it into memes that show Africans on the bottom and Europeans on the top."
The obvious guess is that the more resolution you get into smaller groups, the more occasional outliers you'll find, and the messier your data will become. At one end of the resolution scale, you can use the US census category of Asians, at another, maybe you're looking at people from a particular jati in the Southern tip of India.
The Lynn and Vanhanen datasets (later updated by Lynn and Becker), which claimed to provide average "national IQs" for countries around the world, have been proven to be a fraudulent concoction consisting of data sources that are statistically and methodologically faulty. To quote Rebecca Sear, "The majority of data included originates from samples which are wholly unrepresentative of their national populations. Many are convenience samples with small sample sizes, often including only children and often including individuals chosen because they had particular characteristics (i.e., samples which were deliberately chosen to be unrepresentative)." Even though some academic journals have retracted some of Lynn's papers, many haven't. And Lynn's National IQ datasets have taken on a life of their own, being quoted as gospel by the likes of Lasker and Cremieux, and have directly or indirectly contaminated the studies done by serious researchers who weren't aware of the fraud perpetrated by Lynn and his cohorts. My *favorite* fraud tidbit from the Lynn and Vanhanen datasets is that the source of the estimated average IQ of Equatorial Guinea turned out to be taken from a group of children in a home for developmentally disabled children in Spain (Wicharts et al).
Lynn himself is an avowed "scientific racist" who believes that the darkies are sapping the vital fluids of western civilization (#snarkasm).
"I am deeply pessimistic about the future of the European peoples because mass immigration of third world peoples will lead to these becoming majorities in the United States and westernmost Europe during the present century. I think this will mean the destruction of European civilization in these countries.” — from an interview with Alex Kurtagic, 2011
"I think the only solution lies in the breakup of the United States. Blacks and Hispanics are concentrated in the Southwest, the Southeast and the East, but the Northwest and the far Northeast, Maine, Vermont and upstate New York have a large predominance of whites. I believe these predominantly white states should declare independence and secede from the Union. They would then enforce strict border controls and provide minimum welfare, which would be limited to citizens. If this were done, white civilisation would survive within this handful of states.” — from an interview Right NOW! magazine
So we've got pseudoscientists like Lasker and Cremieux (mentioned above) recycling this claptrap, trying to correlate it to GWAS studies, to make a narrative that fits their ideological bent. That qualifies in my mind as "taking crappy GWAS data from and turning it into memes that show Africans on the bottom and Europeans on the top."
You forgot about the "dedicated their careers"-part of the claim, which is why I suggested Gusev "quantify in some way what proportion of their careers is dedicated to what you say." I never questioned that they had used that methodology to study that topic to some degree.
Lynn is neither here nor there, but this blog has explained why his results support the idea that environment matters for IQ:
https://www.astralcodexten.com/p/how-to-stop-worrying-and-learn-to
I'd say about 0.1% of my career has been dedicated to writing about GWAS, and usually quite gingerly.
Except that Scott never addressed the issue that Lynn either exaggerated or fabricated a significant portion of his National IQ datasets. Scroll down through the comments for my responses to Swami, Steve Sailer, and a person named Daniel (who did a good Glaucon imitation).
I look forward to Rebecca Sear publishing her own table of national average cognitive performance to refute the half-dozen different tables that have been published, including the World Bank's. I'm sure Dr. Sear is just finishing up on her project and we'll see her True Numbers any day now.
Why bother? IQ is mostly pseudoscientific bullshit. IQ tests were originally designed to identify individuals with cognitive disabilities, particularly children who required specialized education. No doubt they do a good job at that, but categorizing high-functioning individuals doesn't get you anywhere. Terman's famous longitudinal IQ study showed that geniuses don't do much better in life than people with IQs +/-1σ range (cue the Zagorsky comeback). And none of Terman's 1500+ geniuses did anything remarkable with their lives. They produced not titans of business, and none of them made any significant contribution to science or culture.
Also, if we're searching for Truth, proving the null hypothesis (H₀) is as important as proving Hₐ. Sears is under no obligation to run a international IQ-testing program because she and a bunch of other researchers have poked holes in the HBD balloon.
I've written relatively little over the years about GWAS and race, because, until recently, it mostly seemed optimized to work best on whites, because that's where the big biobanks were to provide the huge sample sizes needed.
Hopefully, we'll get bigger sample sizes in the future for diverse populations.
With regards to “Lysenko/blank slatist/liberal creationist” accusations, it’s always been pretty clear that hereditarians are every bit as politically motivated as their opponents, if not more so. Just about every high profile hereditarian, past and present, is a right wing political activist.
Is this actually true? My impression is that the consensus view is that, for example, intelligence (measured by IQ) is substantially heritable. I don't think that's a notably right-wing view among people who study the relevant fields.
HBD is overwhelmingly right wing, but that's a different package of beliefs.
I really meant public-facing researchers/popularizers but I should have been clearer about that. Comparing people like Sasha, Turkheimer, Kevin Bird who are all pretty open left wingers on the one side and Kirkegaard, Piffer, etc. on the other who are all pretty open right wingers.
It's almost as if publicly dissenting from the extremist conventional wisdom that the racial gap in IQ _must_ be 100% due to Nurture instead of Nature rather than to endorse the moderate but loathed view that both Nature and Nurture probably play a role requires a lot of commitment.
Or maybe people drawn to right wing politics simply find the idea of genetically fixed class/race hierarchies very congenial. We may never know…
Great response but
“turning it into memes that show Africans on the bottom and Europeans on the top”
Don't their studies show East Asians and Ashkenazi Jews are at the top?
I’d edit this out if I were you, otherwise they'll use it as ammunition for their claims that you are a “liar”.
Who is on top seems to depend a lot on the ethnicity of the poster in question. Same with who is on the bottom. It's black people if the poster is American and MENAPT if the poster is European.
All the HBD people I've read much from seem to agree that Eastern European Jews and East Asians have higher average IQs than Gentile whites, and they are mostly Gentile whites.
For example, what would you find by reading Steve Sailer or Charles Murray?
I wouldn't know about those specifically. I was introduced to the HBD concepts through Scott (jewish), Cremieux (Jewish), and Hsu (east asian). The "white" people I have read include Kirkegaard, who argues that the Chinese data is heavily inflated, and Piffer, who goes back and forth on whether europeans or east Asians are on top.
I came up with the term "human biodiversity" around 1998, only to immediately discover that anthropologist Jonathan Marks had published a book with title in 1995. Still, if I have any claim to the term, please allow me to point out that I've always argued that it is likely that the human biodiversity perspective is that both Nature and Nurture tend to matter, in contrast to the mainstream but extremist position that only Nurture could possibly matter.
In the latest Piffer analysis (https://www.qeios.com/read/HDJK5P.2) the ordering is in fact Africans on the bottom and Europeans on top with the EA4 polygenic score (Figure 1b). I've said plenty of times that this kind of analysis is highly confounded and produces nonsense, and since multiple behavior geneticists have told Piffer the same, I can only conclude that the intent is malicious.
Ah, I missed that paper, thanks for clarifying.
There were more others plots where Ashkenazi or East Asians on top, why single this one, except you want to imply that authors are white supremacists?
If this produces nonsense, why it produces relative rankings of populations similar to phenotypes, rather any any of possible permutations equally likely? Did Piffer arbitrarily chose a subsample of genes to compute the plot, or was training was "contaminated" with non-Europeans so linear regression learnt non-genetical effects for phenotypes?
I don't think it is "singling out" to cite the Figure 1 result which uses the latest/largest dataset for training (EA4). Population stratification is indeed expected to reproduce phenotypic rankings, as explained here: https://theinfinitesimal.substack.com/p/how-population-stratification-led
Thank you for engaging in a constructive discussion of these topics. How would you rate the accuracy of the description of GWAS provided here? I feel like it misses the importance of linkage disequilibrium to the design. A centimorgan is after all about a million BPs in the human genome. And this omission makes it seem like GWAS only assesses the importance of SNPs, which makes the method seem a lot less powerful than it actually is (in theory, anyway).
Thanks, I think the description of GWAS here is good enough for a lay article that is trying to explain a lot of different concepts. The changes I would make are: (1) [as you note] even though GWAS methods often estimate heritability from ~1M common SNPs, which may seem like few relative to the size of the genome, it's been pretty well established that this ~1M actually "tags" (through correlations) the rest of the common variation quite well. So when we talk about GWAS heritability we're really talking about "all common variant heritability". (2) Likewise, *common* structural variants are also "tagged" by these ~1M common variants very well, so that even if only SNPs are being tested, the influence of common structural variation will also be picked up. IMO the key missing piece for GWAS really is just rare variation and not any of the other stuff. But again these are relatively minor quibbles with a section that's mostly correct in a complicated post.
Thanks for your answer! (and your blogging in general)
1. I'm having trouble understanding what you mean by GxE interactions explaining much missing heritability.
For the Scarr-Rowe interaction, this would make heritability look higher in high-SES families. But wouldn't this affect twin and molecular estimates equally unless there's some reason that subjects for one type of study are consistently from a different economic stratum than the other?
If we're thinking more about, let's say, a pair of fraternal twins where one of them is ugly and so parents don't invest resources in their education, wouldn't this show up equally in twin studies and GWAS? That is, if this is a very uncommon effect, we shouldn't expect it to affect large twin studies much. But if it's a common effect, then shouldn't we expect that every ugly person is less intelligent, and so GWAS will find that a gene for ugliness is associated with lower intelligence (both within and between families)? Can you give an example of a case why this would show up in twin studies, but not GWAS, RDR, etc? Also, why would we privilege this circuitous explanation (ugliness is genetic and provokes strong parental response) over the more direct explanation (intelligence is genetic)?
Also, the papers you cite show effects on the order of 2-8%pp; do you think the real effect is higher?
2. I grant that this is extremely likely to be bad, though the only person on that list whose opinion I rely on heavily is Cremieux and I haven't seen him do that (I may have missed it). The people who I have seen do that claim they are doing things to mitigate portability problems; I am skeptical but will let them defend themselves if they see this.
3. Obviously everyone is just seeking the truth, but I think it's fair to describe people by where they think the truth lands; I think we both agree on this and I won't nitpick this further.
I don't understand what you mean by twin study assumptions being untestable; I tried to link many studies testing them here.
I'm not sure what you mean by claims that twin studies find romance novels are 99% heritable or whatever, and I can't read the Lyman Stone post because it's subscriber only. I am slightly confused on how you think *over-reliance on* twin studies is behind people doubting the heritability of schizophrenia - people use the same arguments about missing heritability, twin studies having unreliable assumptions, etc as the lynchpin of the schizophrenia-isn't-genetic case (see eg Awais at https://www.psychiatrymargins.com/p/schizophrenia-and-genetics-end-of , the Torrey paper that inspired him at https://www.sciencedirect.com/science/article/abs/pii/S0165178123006418, and the Mad In America people at https://www.madinamerica.com/2024/01/psychiatric-yeti-schizophrenia-genetic/). Obviously people should say true things and not try to fake their beliefs in order to avoid some ill-defined concept of "harm" or "misinformation", but unless I'm misunderstanding you I'm unclear why you think doubting twin studies everywhere else will make the problem of people doubting them in psychiatry better rather than worse.
4. I'm not sure exactly what claim of mine you're responding to, but I agree that most twin studies fall within a range of 40% - 70% heritability of intelligence, especially when you adjust for the test used, the age of participants, and the standard error (yes, I admit that all those adjustments give extra degrees of freedom) and that this is also true for adoption and pedigree studies (I linked my sources there, which I claim I didn't cherry-pick). I don't think finding one anomalous substudy (of a larger study that falls right in the middle of the usual range) is inconsistent with that claim. In the supplement of LSADT, the authors say that estimate might have been anomalously low because they used a less g-loaded IQ test (yes, I admit they probably would have found something to worry about in any case). I think this one outlier study in what's generally a pretty consistent field especially among the largest sample size studies is pretty different from eg Sib-Regression, where the two published studies say 40% and 9%.
Here’s a post working out Sasha’s point on GxE in some empirical cases where we actually have twins, GWAS, and also we know the exact GxE mechanisms, so we can confirm that twin studies really were capturing GxE: https://lymanstone.substack.com/p/why-twin-studies-are-garbage
You seem to analogize peanut allergy to intelligence. In the 1950's peanut allergy did not exist, which suggests to me that it is not genetic in origin, at least not directly (although the neuroticism that has produced peanut allergy may be). Intelligence, in contrast, is a real trait.
Appreciate the followup. I'll try to respond on areas where I won't be tedious.
1. >>Gene-Environment interactions.
Take the peanut allergy example. Let's say in order to develop an allergy you need a mutation in the PNUT gene AND ALSO grow up in a household with early exposure to nuts (no Bamba!); that's a gene-environment interaction. For MZ twins, they will always share PNUT mutant (or wildtype) and 100% of their household exposure, so they'll be perfectly correlated on allergy; for DZ twins, they will share PNUT mutations half the time and 100% of their household exposure, so their correlation drops in half. So the twin study will tell you allergy is a 100% heritable trait. Now we test the PNUT variant in a GWAS, the first thing you do is throw away all the relatives (i.e. take one of each twin). Some people will be PNUT mutants and grow up in a household with no exposure and be allergy free, some will be PNUT mutants with exposure and will have allergy (and vice versa for the non-carriers). The resulting correlation between PNUT mutation and allergy will be low, so the heritability estimate will be <100%. TLDR: in the ACE twin model (and sib-reg), AxA and AxC interactions get counted as A. In the GWAS (and RDR) model, AxA and AxC get counted as E. In my opinion AxA could plausible be considered "heritability" in the sense that it only relies on genes, but AxC cannot.
>>Regarding the magnitude.
In the Mostafavi et al. paper, the heritability of Education doubled between the low and high SES groups (and this has been observed for IQ as well in other studies). 2-8% for the few environmental exposures people have tested is IMO substantial and could easily stack up across many environments (I think this is suggested by "multivariate" environmental methods -- https://theinfinitesimal.substack.com/i/147322261/multivariate-environments -- but the application of these methods has been limited so far).
3. >>Twin study assumptions.
What I mean by this is that we do not have a global test for quantifying AxA/AxC or for whether the EEA is violated. We also do not have one flavor of twin study that requires the EEA and another flavor that doesn't so we can triangulate across them (well, we sort of have that and it's RDR/Sib-Reg). What we have is the ability to test *specific* environmental measures for EEA/interaction violations and then various crude approximations (like look-alike studies, incorrect zygosity, twins raised apart) which come with their own new assumptions about the assumptions. But the whole saga of behavior genetics is that we do not know which environments actually influence these outcomes. We know MZ twins do not turn out identical, yet we cannot explain why. So how do we know which environments to test for EEA violations or AxC to begin with? TLDR, I'll just quote Alex's recent review of social science genomics: "Before the modern molecular genetic era, a range of models seemed capable of “explaining” the available kinship correlations, despite major differences in their underlying assumptions about, for example, dominance, assortative mating, and gene-environment correlation (Loehlin, 1978). Despite some spirited debates, efforts to distinguish between these models were largely unsuccessful" [https://www.nber.org/papers/w32404]
>>Harm of over-reliance on twin studies.
This is a longer point but I'll just say that a common argument from the anti-psychiatry people is to raise concerns about twin studies that are directionally correct: many of the iconic twins raised apart actually had frequent contact; some data was outright manufactured; one of the major twins raised apart studies (MISTRA) never published half of its cohort for unstated reasons; the other major study (SATSA) produced large estimates of selective placement and bizarre estimates of dominance; we know for a fact traits like peanut allergies are influenced by the shared environment; etc. Then bad actors take these concerns and say "look, the whole field is built on a foundation of nonsense, don't trust your therapist, flush your medication!". If the response is, "no, twin studies are unassailable, oh and also molecular methods are weird and not to be trusted either", I think this ultimately plays into the anti-psych argument. Of course, we should care most about getting at the truth, since manipulators are always going to be out there doing their thing no matter what we say.
4. Here I was responding to the specific final claim ("They are strong designs, their assumptions are well-validated, and they all converge on similar results."). I do no think it is accurate to say that heritability estimates ranging from 20% to 98% across two Scandinavian countries are "converging on similar results". Nor do I think it is accurate to take these two estimates, which are highly significantly different from each other, and simply average them to 52% and say they "fall within the range". They clearly disagree, and when you have a lot of numbers between 0% and 100% that disagree, the fact that they average out to 50% is not a comfort. I also don't think this study is an outlier in any meaningful sense: a group of esteemed behavior geneticists pulled three high quality cohorts for some other goal and stumbled on three very different estimates. That's weird! And it mirrors the kind of weirdness and variability we are seeing with Sib-Reg and the kind of variability one would expect if these approaches are highly sensitive to environmental interactions or other assumptions.
For the GxC example I'm a bit confused as to how that could explain some of the missing heritability.
I understand that if the hypothetical twin study has sampling bias and only contains houses with peanut exposure, then that would inflate h^2. But if the sampling bias went in the other direction, only houses *without* peanut exposure, then that would *deflate* h^2.
With a proper sample this would cancel out, no? Or would this "cancelling out" only be partially/unreliably captured by the twin study model?
Sorry if the example was confusing. The twin study does NOT need to contain only houses with peanut exposure. It can be a totally random/representative sample. All you need is for exposure (or lack of exposure) to be a "shared/household environment" such that all twins match on their exposure/non-exposure. The statistical issue is that MZs share genes*household 100% and DZs share genes*household 50% (because both share household 100%) and that looks exactly like heritability.
And what would the bias look like?
If no houses have peanut exposure, then r = 1 for MZs and DZs alike, and that's an h^2 of 0, which seems correct (or technically r is undefined, but whatever).
And if all houses have peanut exposure, then r = 1 for MZs and 0.5 for DZs, which is an h^2 of 1, which also seems correct.
But I guess the issue arises between those two scenario? Is the bias like a non-linear thing? And is it always upwards?
Yeah, the bias is when you have heterogeneity in the shared environment (C), every genetic interaction with the shared environment (AxC) makes MZs look more similar to DZs and therefore adds to heritability (A). You can see the derivation in Purcell 2002 (https://zzz.bwh.harvard.edu/library/purcell-2002-twin-gxe.pdf - search for "In short") or see the result in this simulation (https://theinfinitesimal.substack.com/i/147322261/twin-estimates-of-g-are-inflated-by-gxe).
> for DZ twins, they will share PNUT mutations half the time and 100% of their household exposure, so their correlation drops in half. So the twin study will tell you allergy is a 100% heritable trait.
Isn't this false unless the PNUT mutation is rare? For DZ twins the correlation is > 0.5, because there's a chance they both get the PNUT mutation from different sides. So if PNUT is common, you'll see some shared environment effect. (And if it's rare, then I would consider the allergy to be "mostly caused" by the gene.) I do see your point that you will see missing heritability here, though.
> In my opinion AxA could plausible be considered "heritability" in the sense that it only relies on genes, but AxC cannot.
"red hair isn't heritable since you can dye it black"
You might be interested to know that Cremieux blocked me on Twitter when I pointed out that his favorite "reared-apart" IQ twin study (Minnesota, Science Magazine 1990) arrived at a 70% IQ h2 conclusion only after the authors suppressed their DZ-apart control group IQ correlations.
Turning to reared-together MZ-DZ "twin method" studies, the "equal environments assumption" (EEA) has been thoroughly "tested." Over 100 years of research has conclusively confirmed what most people already know--that MZ (identical) twins grow up experiencing much more similar environments and much higher levels of identity confusion and attachment vs DZ (fraternal) twins. Behavioral twin method results are, therefore, confounded and provide no evidence in support of genetic influences (heritability).
I notice you say that "over 100 years of research has confirmed" that EEA is false, but you don't link any studies, whereas I link to many in the post demonstrating that it's true, which you haven't responded to.
"EEA test" studies typically involve twin researchers supposedly validating the research method on which their careers are based. It's not surprising that they conclude in favor of the twin method, and the task of replication crisis analyses is to root out p-hacked research where conclusions often match authors' confirmation biases.
Here's a link to a table I put together on all studies I was aware of that assessed MZ vs DZ environmental similarity and levels of identity confusion and attachment to each other: https://jayjoseph22.substack.com/p/levels-of-identity-confusion-and
Any thoughts on the table?
The post seems to hinge on a table that is overwhelmingly things like parents having harder time telling identical twins apart. This doesn't respond to the EEA, which is that parents treat fraternal twins significantly differently from identical twins in ways that have significant effects on behavioral traits. It also doesn't respond to the studies linked in my post, which show that in cases where parents confuse one kind of twin for the other, their differences accord with genetics rather than parent-determined environment, or with the study showing that if you control for environmental differences it doesn't matter.
We can agree to disagree on whether twin studies are confounded. On another note, the first three paragraphs of your post align with the highly problematic history of genetic research Robert Plomin presented in “Blueprint.” Below, I tweak these three paragraphs a bit to better reflect the historical reality:
In the late 19th century and the first four decades of the 20th century, eugenic ideas dominated psychiatry and academic psychology. Based mainly on preconceived notions, family pedigree studies, and supposedly “degenerate families” such as the “Kallikaks” and the “Jukes,” it was axiomatic in psychiatry that conditions such as schizophrenia were largely the result of “hereditary taint,” and IQ hereditarianism was mainstream in academic psychology. In the 1920s and even earlier (Galton, Thorndike), eugenicists and “racial hygienists” developed twin studies of IQ, criminality, schizophrenia, and many other behavioral areas. Eugenicists promoted the passage of laws throughout the Western world enabling the forced sterilization of the “insane” and the “unfit.” In Germany, Ernst Rüdin and his psychiatric genetics colleagues carried out twin studies and promoted forced sterilization in National Socialist Germany. In the 1930s, their Munich Institute became a think tank of Nazi eugenic policies, which led to the killing of mental patients and others under a program euphemistically called “euthanasia.”
In the aftermath of World War II and revelations about Nazi atrocities in the name of eugenics and racial hygiene, eugenic policies and genetic explanations of behavioral differences lost some appeal. However, even though nurture influences received greater emphasis, the idea that heredity plays a role in causing differences in IQ and behavior never went out of favor. Fewer behavioral twin studies were conducted in the 1950s and 60s, although eugenic sterilization laws remained on the books in many U.S. states and in Europe. The American Journal of Psychiatry ran Franz Kallmann’s annual positive review of eugenics from 1944 until he died in 1965, and the word “eugenics” fell completely out of favor only after the social upheavals of the 1960s.
The socially and politically powerful have an interest in promoting behavioral genetic research, which led to an increase in twin studies and “heritability” research beginning in the 1970s. However, behavioral twin and adoption studies have always been based on false assumptions. They have also been subject to researcher data manipulation to confirm genetic confirmation biases, a widespread practice now known as “p-hacking” in science’s current and long-overdue “replication crisis.” The famous Danish schizophrenia adoption studies of the 1960s and 70s and the Minnesota Study of Twins Reared Apart 1990 Science Magazine IQ study exemplify instances where researchers found no evidence of genetic influences, but then p-hacked their data to transform negative genetic results into positive ones. Scandalously, both continue to be cited by psychology and psychiatry as “landmark” studies.
Beginning in the 1960s, researchers used DNA-based (molecular genetic) methods in attempts to identify genes that cause psychiatric conditions such as schizophrenia and bipolar disorder. By the early 1980s, the search was broadened to include other psychiatric conditions, IQ, personality, criminality, and other types of behavior. Despite decades of sensational claims in the media, and by researchers and the institutions they worked for, such claims turned out to be false alarms. This statement is true for earlier behavioral linkage and candidate gene studies, and most likely, current claims based on rare variants, genome-wide association studies (GWAS), and polygenic scores. The findings from these more recent methods are likely non-causal or spurious “hits.” “Association” (correlation) does not equal “cause,” and the half-century-long failure to identify causal genes for behavior leads to a conclusion that the critics of behavioral and psychiatric genetics were right all along about earlier family, twin, and adoption studies. The “missing heritability problem” is really a “family, twin, and adoption study misinterpretation problem.” Time to rewrite the textbooks.
Saying the critics were right all along does not necessarily mean that humans begin life as psychological “blank slates,” or that there are no inborn/genetic within-group individual differences in human intelligence and other behavioral areas, or that causal genes will never be discovered. But it does mean that perinatal, family, social, cultural, religious, educational, geographical, and political environments (including institutionalized oppression/privilege, economic inequality, and neocolonialism) play a decisive role in shaping human behavior, and that focusing on “individual differences” and problematic heritability estimates implies limited changeability and distracts our attention from the need to improve or radically change these environments. The same point holds for patterns of psychological distress and dysfunction that psychiatry and psychiatric genetics call “mental illnesses” or “diseases.” For all types of human behavioral differences, unlikely yet possible future discoveries of causal genes would, for the most part, be irrelevant distractions.
It seems like the issue is whether government programs that try to equalize environments can do all that much to equalize IQ among people who aren't identical in genes. For the government to equalize environments to the extent that environments are typical equalized by parents among their children would seem pretty ambitious, or downright utopian.
I can well believe that identical twins experience more equal environments than fraternal twins -- due to their near identical DNA -- but I can't see how the government can do much about that when trying to make the environments of children who don't have identical DNA more equal.
Hi Sasha,
Thanks for your comment. I want to ask about gene-gene interactions. I understand that dominance effects have essentially been ruled out, but other than that, what evidence do we have against GxG? It seems implausible to me that GxG doesn't exist, because additivity is not invariant to our measurement scale (eg if IQ has additive genetics, then IQ^3 does not).
There are too many gene pairs to investigate GxG by directly adding interaction terms. Other than doing that, I'd think the main way to study GxG would be to look for a gap between narrow and broad heritability, right? But the gap seems to be there, doesn't it? Am I misunderstanding?
I think the primary argument against GxG is that lots of people have looked at it using a large number of methods and traits and nothing has ever replicated. The most recent study was from 23andme (https://www.biorxiv.org/content/10.1101/2024.08.15.608197v2) looking at height in millions of people and testing for interactions between variants we know influence height marginally and finding nothing. It's not the most satisfying answer but it's just rare to see not even a hint of signal like this.
One interesting counterpoint, however, is the "limiting pathways" model proposed by Zuk et al (https://www.pnas.org/doi/10.1073/pnas.1119675109) which hypothesizes that traits are a collection of pathways where any one of them breaking causes the entire trait to "break". Under this model, twins/siblings would "break" their traits in the same way, but unrelated individuals would all be breaking differently, which would create what they call "phantom heritability" that looks a lot like GxG in the close relatives. I think this is a plausible model but extremely difficult to actually quantify.
Thanks. But again, GxG just has to exist, no? Really, if height has no GxG, then square-of-height will have GxG and vice versa. I guess I should try to quantify that effect...
The 23andme paper says that typical detected alleles have 1.2mm effect on height, and they can rule out interacting pairs where the effect size is larger than 2.4mm. That does not seem very conclusive to me? One would expect each pairwise interactive effect to contribute a lot less than the contribution of each individual allele: if there are n genes, there are like n^2/2 pairs of genes, so we expect the contribution of each pair to be much smaller. Shouldn't we expect pairwise interactions to have effect sizes on the order of 0.1mm or less, even in the presence of GxG?
Yeah, so on the first point, if you are just thinking of GxG as the trait being the product of mutations instead of the sum of mutations, then taking the log of the trait should make it additive again (this is something Fisher pointed out). We generally see pretty similar heritability estimates for (trait) versus log(trait), which is maybe evidence against simple epistasis like that. On the other hand, if GxG includes more complex interactions (e.g. variant effects completely flip in the context of other variants) then scale transformation will not be enough.
To the second point. Yes, it is still possible that there is a very large number of very tiny epistatic effects that we cannot pick up. I think this is biologically unlikely but we cannot rule it out. One other way to look at this is to ask if trait correlations across relatives seem to follow an epistatic model. This doesn't require estimating any individual interactions, you just ask if people who are more closely genetically related are *much* more phenotypically related. This has also been done and doesn't seem to line up with epistasis (see Zaitlen et al. 2013 , https://journals.plos.org/plosgenetics/article?id=10.1371/journal.pgen.1003520 "We examine phenotypic correlations across a range of relationships, from siblings to first cousins, and find that the excess phenotypic correlation in these related individuals is predominantly due to shared environment as opposed to dominance or epistasis."). However, this is a bit circular to the broader heritability debate. If complex shared environments are at play, then they could be masking the effect of epistasis.
TLDR: we are mostly relying on the absence of evidence (but after many many attempts) rather than evidence of absence for GxG.
Thanks again.
I disagree with the intuition that a very large number of very small effects is biologically unlikely. To me there's no biology here at all: if we assume a biologically additive effect on a phenotype P consisting of many small contributions by different genes, and then we try to measure the heritability of P^2, this will automatically introduce an effect of *all possible* pairwise interaction terms, each with tiny effect size. This is just a mathematical consequence, not a biological one. To say things are additive is to say our measures of phenotype are on the perfect "scale", in a sense. That seems unlikely.
To be clear, I'm not envisioning GxG effects as one gene affecting the expression of another, or anything like that. I'm just saying, if gene A gives you some kind of +1 for intelligence and gene B separately gives you another +1 (in some "true brain units" or whatever), there's absolutely no reason at all to assume that, as measured by number of correct answers on a test, the effect will still be additive.
I will check out Zaitlen et al. That sounds more along the lines of measuring GxG by comparing narrow to broad heritability, as I mentioned earlier.
One way that the phenotype measures are tuned is that they are designed to be approximately normally distributed.
“One other way to look at this is to ask if trait correlations across relatives seem to follow an epistatic model. This doesn't require estimating any individual interactions, you just ask if people who are more closely genetically related are much more phenotypically related.”
Isn’t this precisely the phenomenon that you’ve been writing about with twin studies?
I think maybe there’s also an issue arising from the different definitions of epistasis. With the one we’re using here, it’s a property of populations, not the genotype-phenotype map alone, so as I understand it there’s no contradiction in saying that GxG is small at the whole population level but large in twin samples. In fact, I think you expect this? Because in the whole population you have a U-shaped AFS with a lot of rare minor alleles, while in each twin pair minor alleles are all at 25% (and there are also some 50-50 loci).
Yes, agreed, that is what I meant by "this is a bit circular to the broader heritability debate". If there was no impact of the shared environment, we could easily fit an epistatic model to relative classes. But we know there are shared environment effects that can be complicated. And we don't even know if environments are equal between MZ and DZ twins, which would be one of the biggest sources of difference for identifying epistasis. So all Zaitlen et al. are able to conclude is that a purely epistasis/dominance model doesn't fit well, but some combination of shared environment AND epistasis could definitely be at play (though see the other more direct negative findings on epistasis I cited in the first comment).
"People like Kirkegaard, Piffer, Lasker, etc. (and their boosters on social media like Steve Sailer and Cremieux) dedicated their careers to taking crappy GWAS data from and turning it into memes that show Africans on the bottom and Europeans on the top."
Citation please.
There are three blog posts containing "Piffer" at your Unz blog:
https://www.unz.com/?s=Piffer&Action=Search&authors=steve-sailer&ptype=isteve
The first of which is one of your readers writing to you rather than you using the name. In the other two you refer to "Racimo-Piffer". All three posts are from the 2010s.
Contrary to his claim about "social media" though, a Twitter search turns up zero references to Piffer in your feed:
https://x.com/search?q=Piffer%20(from%3ASteve_Sailer)&src=typed_query
It seems like there’s room for people more educated than I to do some simple “ceiling” calculations on the types of gene interaction models that could plausibly be at play here given the number of generations/iterations that evolution had to play with.
As an extreme example, assume you have 50 independent variables that can interact in any way (linearly or otherwise, eg, GBT/random forest style). Make some reasonable assumptions about their variance. Assume evolution is operating like some common gradient descent optimizer like NES. How many iterations would it take for evolution to effectively select for this or that gene. How many generations has evolution actually had?
This is a loose, hand-wavy description, but I think I’m making some sense here.
The number of causal variants for a typical common trait is on the order of thousands but can vary as much as 100k down to 100 from trait to trait. Evolution has also had half a million years to tune these variants in hominids. So it's a big space.
The anecdotal evidence of “troubled child raised in a good Christian home” seems a strong contender for selection bias. You wouldn’t have encountered the hypothetical well-behaved children from awful parents raised in supportive households. If that population outnumbers those you worked with, the point becomes moot.
Extremely interesting essay overall; thanks for sharing!
Agree with your general point! I would add that in my career as a primary care pediatrician I do see adopted kids who are raised in “good” families and who are not committing heinous crimes. But I do notice that their behavior pattern often seems more similar to that of their biological parents than their adoptive parents…
The point would be moot if the point was "the majority of children with troubled bio-parents become troubled themselves".
The actual point is "adopted children with troubled bio-parents do worse than their siblings even if raised in good families whose biological children were goody-goodies who never got so much as a school detention. "
It depends on what else Scott saw. If troubled children from good households tend to be disproportionately adopted, then that can't be attributed to the type of selection bias you're describing.
Note that Scott saw a population selected for its extraordinarily bad behavior--it's not so clear how much that very weird population tells us about variation within the usual bounds of "difficult kid."
Maybe there could be a self-reinforcing effect to sibling similarities? There's the stereotype of the twins that are always together and always do the same thing. Maybe small differences, especially in early childhood, separate siblings and make them more estranged so that in the future they affect each other less and less. So identical twins have some initial stronger bonds by being more alike, and then that develops into an environmentally stronger bond (i.e., they do most activities together and want to be equal), while fraternal twins are a little more distant from each other because of initial differences and then don't have that bond that makes them also have a more similar environment.
Theory: IQ is very heritable but also extremely subtly disabled (potentially even environmentally, think "more males born after wars"), because IQ is terrible for reproduction (now as ever) and you need some people to keep passing it on.
Crackpot theory: everything is physical appearance.
The causal mechanism connecting genes to outcomes could be that genes determine how you look, how you look determines how people treat you, and how people treat you determines how you behave.
If someone looks smart, they become smart because that's what everyone expects. If they look aggressive, they become aggressive. This is why people online can talk about stuff like "nerd physiognomy." It's real, but the causation is backwards from what you'd think.
I don't think this passes the smell test. As we know, physical appearance is not well-correlated with intelligence, even if some do claim that all "positive" genetic traits are correlated. Also, within an ethnic group, it's tough to really determine whether someone "looks smart". I'll grant you that people's development is influenced by how others see them, but it's not the physical appearance that would actually be biologically making the person smart or dumb. Of course, the popular nerd vs jock stereotype does have some truth to it, and could suggest two different "evolutionary strategies" for human males.
I’ll just throw in the small wink implied by the oversized banner I would see when my undergrad roaming at my rodent mascotted university took me to Eliot Hall where they were doing research in this area.
“Minnesota Twins Study”
The flaw in this starts early: "For teacher reported ADHD.......".
That's all I'll say.
Scott’s personal experience during his psychiatry residency parallels my own experience as a primary care pediatrician. From my perspective, there is overwhelming clinical evidence that genetics must play a huge role in shaping behavior.
It’s actually somewhat baffling to me that pediatricians are generally opposed to “hereditarian” views, but I think there are many non-clinical reasons why that may be the case…
I dont get the darwinian argument cause, I'm always thinking like evolution really cared about calorie efficiency which we dgaf about. so why wouldn't there be a "more brain" knob in the genes? compare to height, or how we can increase muscularity with tren
Depends on what it was traded off against. You can get bigger brains with bigger heads, but that's not going to be selected for (even if it makes the kids much smarter) until you can do C-sections safely.
I suspect wider hips for easier birthing of babies means slower running speeds. And speed was often important in the action-packed world of past.
Genetic heritability of intelligence is precisely the eugenics-loaded, complex, wicked problem that is eminently vulnerable to all sorts of biases as well as prestige/social related competition altering scientific opinion.
Putting that aside: since we know that there are environmental factors in at least some genetic expression - does this not kill, dead, any idea of separating "genetic heritability of intelligence" vs. "EA generated intelligence"?
Why would that be more true for intelligence than for height?
Not clear to me what environmental or nurture factors encourage height.
Nutrition seems like the obvious one. It seems inevitable that this influences the pattern of gene expression, since you're surely expressing a different pattern of genes when you grow than when you don't grow.
Is there some reason to think that heritability of intelligence is inherently harder to understand than heritability of height?
Nutrition is likely some effect on pre-modern diet humans and height, but I doubt nutrition plays any role in height in say, the Scandinavian countries today.
Intelligence, however, is not directly related to nutrition unless there is a serious lack of nutrition. While in theory, being smarter gets you more food - the reality is that modern humans don't generally lack for food nor is the availability of food in modern society, a function of intelligence. Welfare programs, for example, directly subsidize food to the economic losers in society in direct contrast to the theory of intelligence mitigating food scarcity.
But even for poor 3rd worlders that move into 1st world nations - there is no way to separate 1st world culture, education, etc from better nutrition for the immigrants' children. Is there even a visible improvement in intelligence for the already mature?
But nutrition is precisely not what I meant by environmental factors.
We know that genes turn on even in adults for various things - epigenetic expression is the term.
Intelligence seems like an emergent behavior, and therefore one which must arise from a variety of bases.
Are these bases relatively set in stone? Seems highly unlikely.
Are they relatively epigenetic? Seems possible but hard to prove given that we don't even know what actually creates intelligence. Note that these genetic studies don't know either - all they are doing is correlating genes vs. highly imperfect "intelligence" measures.
Among other things: my view is that there are different types of intelligence - horses for courses. I don't believe egghead PhDs are great at everything - in fact, they tend to be good only in very specific fields and are merely "somewhat" intelligent in others.
On the other hand, we don't even try to measure the type of intelligence needed to survive in the wild without society - which is realistically the only actual measure of evolutionary intelligence that should matter.
So what we really have is the usual ACX lotus eating: how can a bunch of professional/managerial types with 1st world society "high intelligence" figure out ways in which their type of "high intelligence" is developed. Most likely so that their kids can carry on the parents' legacy of PMC-dom.
IQ is interesting because it predicts success in two critical parts of the modern world: school and work. But yeah, I'm not at all confident that IQ correlates all that well with success in a hunter-gatherer tribe, say. Some kind of intelligence seems critical for that life, too, but who knows if it corresponds well to the stuff we measure on IQ tests?
Still, mostly when we talk about individual IQ or average IQ of a group, we're talking about how well we expect someone or some group of people to function in the modern US[1]. Like, if some guy has a 70 IQ, I expect he's going to have a really hard time at school, probably not be able to learn to read, and is going to have a very limited range of jobs he can do in a modern society. Maybe he'd be great at tracking antelope on a plain--I have no idea. But he's pretty unlikely to have a lot of success in modern 21st century countries.
I guess I come back to the same question: is your claim that heritability is in general very hard to untangle? Or that it is particularly so for intelligence?
[1] But honestly, people get wrapped around the axle thinking about group average IQs and reasoning about them like they're an individual IQ, even outside the political/social/tribal issues those discussions often raise.
I think we are in pretty broad agreement:
1) Intelligence as measured by IQ is not "general" intelligence, it is PMC type intelligence
2) PMC intelligence is not in any way clearly correlated to evolutionary intelligence. I have bird seed and nectar feeders on my balcony; it is quite clear that the capability to remember seed and nectar locations is important for house finches and hummingbirds, respectively. But it is also clear that a hummingbird's version of intelligence is very different; observationally, it is fairly clear that they don't just remember locations but both develop habits i.e. they seem to always use the same feeders until changed and also have "modes" where they will test the other feeders (but still use the "preferred" one unless it runs out or goes bad).
You did not respond to my comments on the emergent nature of PMC intelligence or the known epigenetic expression of at least some genes. I think of these as the Scylla and Charybdis of any attempt to link PMC intelligence to genes.
Lastly, IQ measuring success in society. I don't know about you, but in the Bay Area, and I suspect in most PMC hives i.e. large cities, the most consistently successful types are the tradesmen. Do you need 140 IQ to be a good plumber? I think not. What about an auto mechanic? Or a repairer of $5000 appliances? I am sure not. Trades generally don't require reading or written testing capability so long as there is someone to train or demonstrate. And I would argue that the ability to make and repair plumbing, refrigerators, cars etc is far more critical to the function of modern society than the philosophy major now selling AI agent software to enterprises or even the vast majority of lawyers doing anything.
Using these polygenetic scores to evaluate a genome is like counting word frequency to evaluate the quality of a book. Sure, it probably correlates. But when you restrict yourself to evaluating a simple linear model on the input, you might not be able to capture the structure there. That doesn’t mean there is no structure present.
Maybe I missed it in the post, if so, I apologize.
But do we know the prevalence of de novo mutations? Even if they are rare, if they substantially impact fitness, it might create a significant difference. Is there a reason for dismissing this explanation out of hand?
I don't know what an actual smart geneticist would say. My hand-wavey response is that we're trying to use genetics to explain why children resemble their parents. De novo mutations don't explain that (if it's a generation up, it's not de novo, it's just another ultra-rare variant) , so since children do resemble their parents a lot de novo mutations can't be responsible for too much of the variance in traits among people (aside from the stuff we already know about where children *don't* necessarily resemble their parents, like certain out-of-the-blue rare developmental disorders)
Good point. I suppose your inclusion of the data on family pedigree/adoption studies answered this. I was thinking mostly about twins, as in, if some rare, but important de novo mutations skew the data they wouldn't show up in GWAS but they might presumably be shared in identical twins (assuming they occur during meiosis or before the embryo splits).
Genetics seems a lot like LLMs. A big black box that intuitively makes sense but its almost impossible to dig into the details of why.
I think you're wrong here. Heritability as a concept is defined as a proportion of total phenotypic variance. It only exists with respect to some predefined reference population, and it matters a lot what particular range of genetic variance, environmental variance, and phenotypic variance you are considering. Even taking twin/adoption/etc studies at face value, most of the strong 'hereditarian' claims depend on ignoring or deliberately obscuring this issue.
Anyway if you took a baby from a family geniuses and you gave it to a family of idiots, and then it turned out that family of idiots belonged to the Concussionistic faith where they have ecstatic experiences by bonking each other on the head with hammers all day, then that baby would probably turn out to be an idiot. The result you get from adoption studies will depend very strongly on whether or not the adoption agencies are discriminating against concussionists.
Suppose I say "lighting a fire makes things hotter".
There are all sorts of possible objections, like "That's only true if you don't also fill the room with ice" or "This depends on environment - if there was a religion that forced people who lit fires to also take off their jackets, they might get colder". Whenever we state a fact, we're implicitly saying "And we're holding other things constant and not positing crazy religions devoted to making illogical things happen". You're allowed to say "shooting someone kills them" without adding "unless you magically hit a tumor and eliminate it, in which case shooting someone actually prevents them from dying!"
So yes, it's true that heritability only makes sense within something like the normal parameters that exist in basically every human society. But this is a stable enough concept that you can still talk about it just fine. For example, I think it's fair to say that having lots of genes for breast cancer causes you to be more likely to get breast cancer, and you should try to avoid having them, even though this wouldn't be true in a world dominated by a religion where people injected uranium into the breasts of anyone who *didn't* have genes for breast cancer.
Sorry if this comes off as snarky, I just don't know how else to respond to this.
I think this is where you are fundamentally wrong; or at least, where your beliefs are fundamentally unjustified in ways you are simply sweeping under the rug.
In the fire example and the bullet example there are clear, unambiguous, directly observable mechanistic processes (I'm sure you could add some more adjectives there but you get the idea) that connect the putative explanation to the predicted outcome. No such mechanisms are known for genetic explanations of complex, massively polygenic behavioral traits. Full stop, these mechanistic explanations have not been identified, and attempts to find them have largely been failures. The closest thing we have are rare genetic variants that disrupt core neurophysiological processes; these don't explain 'being a smart person' vs 'being a dumb person', they explain 'your brain basically works' vs 'your brain barely works', and they have little to do with the common variation that hereditarians want to explain.
Also, when we talking about fire melting ice, we are talking about a clearly and absolutely defined endpoint. We are not talking about percentages of total water volume that can be attributed to (poorly measured) similarity in thermal conditions. Heritability research is closer to the latter.
So yes, you get to blame the melting on the fire. When behavioral genetics has achieved even a fraction of the empirical and theoretical depth that thermal physics has, then maybe it will make sense to blame intelligence on the genes.
To put another way:
"So yes, it's true that heritability only makes sense within something like the normal parameters that exist in basically every human society. But this is a stable enough concept that you can still talk about it just fine."
No! This is a very unstable and poorly defined concept! And even if it were well defined, it would still matter tremendously how representative your study population is of whatever you mean by 'normal human society'!
In any case, you haven't been any snarkier than I have, and as someone who was once a behavioral geneticist (not in humans) but hasn't actively followed the latest developments, I really do appreciate the good-faith engagement with opposing views and evidence.
ISTM that what we have now is some observations that allow us to make good predictions.
We see that smarter parents tend to have smarter children, smarter kids tend to have smarter biological siblings, and smart kids with an identical twin tend to have their twin be smarter than if they have a fraternal twin.
We know enough that we can make a lot of useful predictions from this. For example, while the children of very smart people are usually not as brilliant as their parents due to regression to the mean, we know that the kids of a bunch of very smart parents will on average be smarter than the kids of a bunch of normal parents. This is messy--Einstein can have a dumb kid and the village idiot can have a brilliant one--but that's not the way to bet.
It seems like that's useful information even if we don't know exactly know all the mechanisms by which that heritability operates. It would be really nice if we understood those mechanisms, because we'd love to reach in and twist some of those knobs--figure out the micronutrient or preschool activity that would boost everyone's IQ by a couple points, give genetic counselors more interesting stuff to talk with their clients about, maybe even eventually find a way to select embryos for intelligence in a reliable way.
A good rule of thumb for evaluating any snarky dismissal of the heritability of mental traits is to ask whether it also snarkily dismisses the heritability of height.
Yours does. Height is a highly polygenic trait and, outside of a handful of rare variants with very large impact like achondroplasia, we have virtually no idea of the mechanics of individual genes that influence it. We also know that it has a significant environmental component, and we could at least imagine that some variants that affect it only do so via GXE interactions (e.g. a SNP that makes one’s olfactory receptors view fish as particularly unpleasant will probably have a negative effect on height in, say, premodern Polynesia but a much smaller effect in the modern West).
And yet, height, obviously is strongly heritable (source: c’mon). So snark that leads us to haughtily dismiss it should perhaps be reevaluated.
IIRC height is also very much subject to the 'missing heritability' problem. Twin/pedigree/adoption studies generally find something gargantuan like 0.8 or higher, and molecular studies AFAIK have topped out at something like 0.5. So whatever point we are arguing about—I'm not actually sure anymore—I don't think pointing to height gets you very far.
To clarify, my position is neither "children are not similar to parents" nor "children are similar to parents, but for reasons that have nothing to do with genetic similarity". I agree those are non-defensible positions; though maybe, on the basis of the work reviewed in Scott's post, they are becoming rapidly more defensible!
Regarding haughtiness and snark, I will just say that all internet arguments could benefit from a serious consideration by all parties of Matthew 7:3.
Adoption agencies generally do a decent job of filtering out the very worst potential parents. Consider as an example, Steve Jobs' adoptive parents: they were quite blue collar, but were excellent people. It was not Jobs custom to be uncritical of other people, but he was highly appreciative of his adoptive parents.
Yes, and this is a relevant fact about adoption in particular, and not about the whole of the human experience or genetics or any given complex phenotype!
Adoption studies are one tiny but not insignificant element in understanding the human experience, so knowing a little about trends in adoption is useful for evaluating adoption studies.
I never understood the finding that parenting and family environment have basically no effect on traits. Maybe it's believable for IQ... but on character and behaviour? Surely they must, right? It feels a stretch to argue that something that's so cultural, like for instance good manners, is hereditary as opposed to learned. Perhaps a level of amiability and disinclination to rebel are hereditary, which makes some children easier to raise to be polite, but you'd still need a parent teaching a child good manners for that child to adopt them. So how does that not show up in the studies?
I don't think anyone has ever done a twin study on good manners!
If someone did, I would expect some shared environmental effect, but also some genetic effect.
(before we even get to either of those, there would be a massive dominating culture effect - no genes will make a barbarian who never encounters civilization perfectly imitate the manners of Victorian Britain. But these genetic studies are generally thought of as partitioning variance within a single culture).
Within a single culture, manners partly depend on whether your parents teach you good manners. But they also partly depend on whether you listen when you're taught, whether you're conscientious, whether you care a lot about offending other people, etc. If you do, you'll probably figure out some way to pick up manners from someone other than your parents. If you don't, you'll probably forget the manners your parents taught you immediately after leaving home.
(one common pattern in a lot of traits, including IQ, is that shared environment matters most during childhood, and then genetics gradually takes over as you get older and more distant from your parents' teachings)
I'm using manners just as an example - what about work ethic, kindness, sense of duty, clarity of thought, appreciation of art, etc? Some of those definitely have a genetic component (I can see that my kids have different inclinations in all of these by nature), but parents do shape their kids to some extent, right? And these must be predictors of happiness and material success?
Back in the 1990s, I reviewed Judith Rich Harris's influential "The Nurture Assumption" about the limits of parenting for "National Review" mostly positively, but I noted some limitations to her focus:
"To show that peers outweigh parents, she repeatedly cites Darwinian linguist Pinker’s work on how young immigrant kids automatically develop the accents of their playmates, not their parents. True, but there’s more to life than language. Not until p. 191 does she admit — in a footnote — that immigrant parents do pass down home-based aspects of their culture like cuisine, since kids don’t learn to cook from their friends. (How about attitudes toward housekeeping, charity, courtesy, wife-beating, and child-rearing itself?) Not until p. 330 does she recall something else where peers don’t much matter: religion! Worse, she never notices what Thomas Sowell has voluminously documented in his accounts of ethnic economic specialization. It’s parents and relatives who pass on both specific occupations (e.g., Italians and marble-cutting or Cambodians and donut-making) and general attitudes toward hard work, thrift, and entrepreneurship.
“Nor can peers account for social change among young children, such as the current switch from football to soccer, since preteen peer groups are intensely conservative. (Some playground games have been passed down since Roman times). Even more so, the trend toward having little girls play soccer and other cootie-infested boys sports did not, rest assured, originate among peer groups of little girls. That was primarily their dads’ idea, especially sports-crazed dads without sons.”
I have twins. The one with ADHD was significantly smaller than the other at birth. I suspect epigenetic expression based on the finite amount of nutrients available in the womb, some of which is modulated by each fetus. In other words, I think twin studies might be a good gauge of heritability, but there's plenty more work to be done.
I agree with all of this except the epigenetic angle - why can't it just be that a nutrient deficiency caused their brain to develop less well? I don't think deficiency causes ADHD any more than it causes other things like low height or low IQ, and I can't think of any reason for careful epigenetic regulation of any of these.
You mention rare variants and structural variants as potentially not being well captured by SNPs panels behind most of these tests (and iirc some types of structural variation can be problematic even for whole genome sequencing). But a lot of the later arguments (around evolutionary pressure, etc.) seem to focus more or less entirely on the 'rare' part there. Is structural variation (e.g. copy number variations etc.) particularly rare in the human genome? Is there some reason to think that structural variation might not be a significant source of heritable variability? Some of that will be correlated with nearby SNPs but now we're talking a correlation to correlation and depending on assumption the analysis power of that can fall off quickly. Has anyone ascertained how much variation might be due to these sorts of things?
Have there been any studies of the connectomes of the brains of twins? I would guess the answer is no, because the technology does not exist yet. AFAIK, we are up to the connectome of a fruit fly. What I am wondering is the extent to which differences in intelligence are a result of different connectomes and how much the connectome is determined by genetics. Perhaps there is a large degree of randomness and the way the connectome develops during pregnancy cannot fully be predicted by genetics. Perhaps, as the brain develops, axons grow and connect based on factors independent of specific alleles, such as local concentrations of certain proteins or nutrients at the growing tip of an axon, environmental variables, cosmic rays, etc. Then there could simply be a random (i.e., untractable) variability in intelligence, which would have a high evolutionary advantage. Just as genetic evolution depends on the availability of variation in alleles that results from mostly random effects (e.g., point mutations, gene doubling events, etc.), so intelligence and, more broadly, behavior plasticity, in a population would then depend on such random variation in the development of the nervous system.
The idea that genes are having linearly additive effects always felt like a simplification for the sake of the model. Changing the way a complex system is made in many ways and expecting them not to meaningfully interact feels like video game logic. I hear that the interactions have not shown up when looked for, but might that be related to the point about computational tractability?
"Since the family unit is a perfect natural experiment that isolates the variable of interest (genes) while holding everything else (culture and parenting) constant"
As a parent of two children I find this assertion to be deeply flawed.
There's the possible environmental effect of nine months shared environment.
They should look at children adopted as embryos.
Most biobank participants are middle-aged, and many are older (UK Biobank median age is 58 according to a quick search). Could the somatic mutations that accumulate during aging lead to errors in the sequenced genomes, which then add noise to the genomic studies and decrease their ability to detect relevant rare variants?
Hi Scott, I think it all lies in conditions in the mother/womb.
Your data in section 4 on creatinine levels and such.
" Identical twins don’t have more similar kidney function environments than fraternal twins." - But perhaps they do.
For two identical twins the ideal concentration of zinc in the womb for optmial kidney development may be 0.000001 percent. But two fraternal twins may have slightly different requirements.
The womb will supply certain conditions that may be optimal for one twin but not the other. With identical twins their requirements are matched as well as the supply.
In fact I would expect the fraternal twins to show similar heritability to regular siblings, perhaps a little more similar since they share womb conditions, whereas with regular siblings the changes to the mother's diet/age and other factors will come into play.
- Apologies for all the extra posts, I tried to delete them. Something strange happened when I tried to fix a typo.
Here's a review that covers some of the same territory as Alexander. https://www.sciencedirect.com/science/article/pii/S1090513824000722 Author Zietsch argues "Recently, converging evidence has emerged that much of the remaining gap is due to imperfect correlation between the SNPs used in GWAS and the ungenotyped causal variants. ... To the extent that GWAS SNPs and causal variants are imperfectly correlated, SNP-based heritability based on unrelated individuals will underestimate total heritability. There is evidence that SNP-based heritability is an underestimate for this very reason." (He gives citations.) Zietsch considers evolutionary implications of the apparent fact that genes of large effect account for very little of phenotypic variance: selection is mostly purifying, not balancing. In other words, within most populations, it's not like there was a niche for small folks and a niche for tall folks (maybe at different times) and selection favored a mixture. Instead, either folks with more tall alleles than average, or folks with more small alleles, or both, were dealt a bad hand in the genetic lottery and consistently had lower fitness. Genetic variation within populations is mostly genetic load resulting from mutation and drift.
But this doesn't apply to the immune system, where selection actively favors genetic variety.
Are there any polygenic traits that are more fully mapped/understood that could serve as a control for other types of studies?
My recollection is that a given person's score on IQ tests can vary substantially time to time (like pretty much any test score, really). Do these studies get everyone to take multiple tests at different times to get a more solid number? If not, how do they control for that variability?
In other words, if identically intelligent twins take an IQ test, how much variance do we expect in their results, and how will that error propagate to the rest of these measurements?
"more home truths are to be learnt from listening to a noisy debate in an alehouse than from attending a formal one in the House of Commons. An elderly country gentlewoman will often know more of character, and be able to illustrate it by more amusing anecdotes taken from the history of what has been said, done, and gossiped in a country town for the last fifty years, than the best bluestocking of the age will be able to glean from that sort of learning which consists in an acquaintance with all the novels and satirical poems published in the same period. People in towns, indeed, are woefully deficient in a knowledge of character, which they see only in the bust, not as a whole-length. People in the country not only know all that has happened to a man, but trace his virtues or vices, as they do his features, in their descent through several generations, and solve some contradiction in his behaviour by a cross in the breed half a century ago."
people today would never be able to grasp and see the hereditary argument play out the way even a shrewd village person would who never left their town all their life did. by knowing personally each family of the village, and their qualities and characteristics, and seeing their kids develop and grow, and their kid's kids, and how the triats are passed down from each parent and renewed in their children in the subequent generation, the weight of the hereditary argument in palpable.
instead we have people arguing about heritability who have never observed anything firsthand or took those observations to heart, from pure statistics and data, and contriving all sort of confounding theories.
when you go back farther to the 19th century and before, it was almost taken as a given that hereditary effects were very large, by anyone who seriously thought about it, reasoning from experience and their powerful intuitions.
The problem is that those seriously thinking people with thier powerful intuitions thought the Irish were good for nothing layabouts, the Chinese were too stupid and drug addicted to be used for anything other than building railroads, and the Japanese were too dumb and squinty eyed to build and fly modern airplanes.
The intuitive hereditarian take has been wrong for several hundred years now and I'm not hopefull that future events will improve thier track record.
Generally their opinions about the Irish, Chinese, and Japanese were not based on prolonged first-hand observation.
the chinese and japanese were actually considered really graceful and intelligent species. the japenese ink art was appreciated. the chinese boy servants were appreciated. i actually havent read too many negative things about asians from 19th century britons. the british society was way more advanced, but when they talked about the innate characteristics of the asians they were very glowing - the opposite to how they talked about blacks , indians, natives, for example.
the indians they understood contained much variety of ability and that underlying the castes was race, and that race would be approximated in a variety of ways like by taking measurements of the nose. an imprecise method ,yes, but the best they could do in their time. at least they were still acting like proper scientists, and dilineating and classifying and doing all that good stuff that scientists do to elucidate the underlying relationships. darwin knew evolution to be true from intuition, but he had to document a thousand small differences to give it scientific validity.
Risley "found a direct relation between the proportion of Aryan blood and the nasal index, along a gradient from the highest castes to the lowest. This assimilation of caste to race ... proved very influential."
.
Have any sources? Both the British and US passed laws to keep undesirable Chinese and Japanese out. The mongoliod race of Japanese and Chinese became synonymous with down syndrome in the western world. I'm not sure where your getting these ideas that they were widely respected from?
We have English statements from the 1100's about the Irish proclaiming "They use their fields mostly for pasture. Little is cultivated and even less is sown. The problem here is not the quality of the soil but rather the lack of industry on the part of those who should cultivate it. This laziness means that the different types of minerals with which hidden veins of the earth are full are neither mined nor exploited in any way. They do not devote themselves to the manufacture of flax or wool, nor to the practice of any mechanical or mercantile act. Dedicated only to leisure and laziness, this is a truly barbarous people. They depend on animals for their livelihood and they live like animals."
all that stuff about the irish sounds accurate to me . it is incredible the amount of eminent people scotland produced (robert burns, thomas carlyle, sir walter scott, john ruskin, to name a few) and eminent people with scottish heritage compared to irish, when they have similarly sized populations, and are similarly rural. likewise the scottish peasantry are talked about in a very different way to the irish , as being strong natured, hard workers, and so forth.
the reason underlying for this is that the irsh represent an archaic native population , while the scottish got a lot more recent influx of geneflow from western and nothern europe from notable groups. (this is simply a speculation, but this is what happened to england with the arrival of the french normans and other groups, who many of the most famous english are direct descendants from, like Byron).
i cant give you sources, but the britons that lived in the far east loved the qualities of the asians and wrote glowing things about them.
https://www.online-literature.com/chesterton/2604/ "the elf of japan" by g.k. chesterton. no briton every wrote such a nice thing about other races like indians.
This is interesting because modern DNA studies have found that castes have been reproductively isolated for at least 3000 years and that India is not a big mixed hodgepodge of 1,5 billion people, but a huge set of very isolated subgroups of 0,5-1 million each, separated by various caste characteristics.
China, on the other hand, seems to be fairly mixed.
isn't this one of thise things that the modern bias that believes race can't explain anything is surprised by but which people who just lived a few centuries back could have easily guessed? its amazing what can be intuited from simple observation - like the inference of the atomic structure of matter from dust in the sunlight by that greek.
there was a lot of praise for chinese . they said the asian boy-servants were better than any girl, because all of their delicate manners of a girl but having the physical abilities of a boy. the british women would even undress in front of them without worrying of the impropriety because they were so amazingly well behaved lol.
my impression was sociology only went off track in the 20th century. everyone at that time was trying to solve all societies problems thru nurture. eugenics became taboo, blacks were trying to be integrated. they was a bouyant hope in nurture to fix everything because that was the only hope left. but it required turning over and ignoring all the old wisdom.
only the blinding lights of things like twin studies got people slightly back on track.
to try to discount these now is going backwards again, but im not surprised. the quality and soundess of the research being done in sociology has been very sketch every since great scientists like the founders of the field like Francis Galton disappeared.
one of harvards top-paid professor whose research was entirely discredited just had her tenure revoked.
the Irish were recognised by the English fromn their earliest encounters as brilliant, brave, witty, word masters and yes prone to drunkenness, laziness, and boasting.
They were recognized as some of the worst savages in Europe.
The Spectator 1882:
The Tragedy at Maamtrasna, investigated this week in Dublin, almost unique as it is in the annals of the United Kingdom, brings out in strong relief two facts which Englishmen are too apt to forget. One is the existence in particular districts of Ireland of a class of peasants who are scarcely civilised beings, and approach far nearer to savages than any other white men; and the other is their extraordinary and exceptional gloominess of temper. In remote places of Ireland, especially in Connaught, on a few of the islands, and in one or two mountain districts, dwell cultivators who are in knowledge, in habits, and in the discipline of life no higher than Maories or other Polynesians.
Theodore Roosevelt wrote in his diary that:
There are some twenty five Irish Democrats in the house. ... They are a stupid, sodden and vicious lot, most of them being equally deficient in brains and virtue. Three or four however ... seem to be pretty good men, and among the best members of the house are two Republican farmers named O'Neil and Sheehy, the grandsons of Irish immigrants. But the average catholic Irishman of first-generation as represented in this Assembly, is a low, venal, corrupt and unintelligent brute.
RDR is unbiased in theory. But in practice it's underpowered and highly sensitive to noise.
It gets worse because in small samples, the heritability is BIASED downwards. Because the % of IBD between sibs varies so little, any measurement error leads to attenuation and the signal is small relative to noise in this design. So great in simulations, bad in practice.
You should consider the simple hypothesis that RDR is wrong and that Gusev is dishonest
To clarify, do you believe that highly-powered SibReg and RDR will converge to twin studies?
I confess that I have only skimmed Scott's article as of now. (I'm in the middle of a terrible project involving navigating 3 bureacracies at once.) So it may be that the point I'm making is not relevant to the question of why it's so hard to find genes that account for the heritability of IQ. But here it is, in case it is.
What about the impact of physical attractiveness on the developing mind? I have always wondered whether the impact of attractiveness is not taken seriously enough as something affecting life course, including ability to perform in intellectually demanding tasks. Last I knew, people, including teachers, rated physically attractive children higher on various other high-valued traits, including intelligence. Physically attractive kids certainly see more smiling faces than unattractive ones, and probably are given more opportunities, and probably have greater self-confidence and higher self-esteem.
To put another way, physical attractiveness affects the environment in which kids grow up. Imagine identical twins where one has some mildly disfiguring facial scars from an accident in early life and one does not. They have the same parents, toys, etc., attend the same schools, and let's say they take the exact same courses and play on the same school sports teams. If a researcher is thinking that these kids grew up in very similar environments, they're wrong.
I agree physical attractiveness is an important factor in life outcomes, but at least in my observation, lots of intellectual high-accomplishers are not necessarily good lookers. Think of many famous scientists, or even Bezos/Gates/Musk. Among the extremely intelligent, there doesn't seem to be a higher % of good looking people ( though of course, being good looking within this group could still be an advantage, especially today with ubiquitous marketing/media) , although in other things such as acting, politics, etc., looks do undoubtedly provide a larger advantage.
I was with 100 percent certainty going to blame measuring IQ being kinda bottle-shaky water-dowsey witch-doctory; ei It reliably works better than placebo but the mechanism of determination has nothing to do with the test itself.
I still believe this a bit, but everything having good prediction but missing predictors has me back in the pea soup fog level epistemic region with a solid carve out for "Personal Spiritual Experience that IQ tests suck and are deeply suspect".
Maybe I don't know enough about psychiatric practice, but why would you resist temptation urge to throw twin studies in the adoptive parents' faces? If parents are going through something terrible like that, telling them that it wasn't their fault and that they did the best that they could seems like a really kind and worthwhile thing to do.
Re the anecdotal evidence of kids you worked with: isn't it possible that the incorrigibility of these kids is due to being victims of horrific abuse before adoption by their new parents? Relatedly... isn't there a lot of stuff that shows just bad outcomes for adopted kids in general?
Excellent review.
What are the studies/evidence that suggests interactions are small? My math/physics intuition would be when you do a linear approximation (assume no interactions) and then change the whole genome your baseline assumption should be that the linear approximation breaks down and has large errors.
You say, "Turkheimer is either misstating the relationship between polygenic scores and narrow-sense heritability," but in the passage you quote I am perfectly clear that I am talking about the DGE heritability, that is Column E from Supplemental Table 3. The value of .048 is the median of 17 (arbitrarily classified by me) "behavioral" DGE heritabilities taken from that column.
There is so much about these studies that I don't like, producing large errors in both directions.
First, there is something very wrong with twin studies. The assumption that separating twins at birth randomizes their environments is flatly false. When twins are given up for adoption in, say, Wisconsin, one may end up in Wisconsin, the other in Illinois. Very rarely does one end up Burundi. But Wisconsin and Illinois are extremely similar places, compared to Burundi!
I hate Educational Attainment as an target outcome. It is deeply immeshed in social practices, many of which are hugely impacted that things that are obviously not genetic. Consider affirmative action practices, as one example among hundreds. People sometimes claim to try to control for these, but it's not a matter of small number of confounders; it's hundreds of confounders all the way down.
I think genetics matter a very great deal, but in ways that thread through social practices in complex ways that make the conventional hereditarian vs. anti-hereditarian discourse meaningless.
The simplest explanation would seem to be that Sib-R and RDR just don't work very well.
Getting out of the weeds, we have one method that we have every reason to believe is producing quality results, and two other methods that maybe should work if we're right about a lot of extremely complex stuff. That the two latter methods don't reproduce the results of the first would seem to indicate that they're just wrong. This just doesn't seem very mysterious to me.
Okay Scott, now I am completely confused. In a somewhat tongue-in-cheek way, I do not know whether to be racist now or not. Can someone how is neither hateful not overly PC but just reasonable, give me some very very rough estimate whether that the say five million Syrian refugees we have here in Central Europe will have kids who will become say civil engineers and pay our social security / pension or not?
This is really what matters in practice.
Please note that I am really not trying to be prejudicial, I bought this very computer from a young Afghan man who built it, but then he sold it to me because he needed to buy a laptop for that civil engineering university and he looked all smart to me and he built this computer well. I am really open to all views.
OTOH it seems we in CE are so far not good at putting the refugees actually to work and making them tax suppliers, not tax consumers.
Can this calc be ran on the back of the envelope?
Is it ever possible to have a really dumbed down answer? Like muh millions of brown refugees, are they gonna cost us or will they contribute money to the state?
This is what the average Euro really wants to know.
https://www.economist.com/cdn-cgi/image/width=1424,quality=80,format=auto/sites/default/files/images/print-edition/20211218_EUC232.png
No, don't count on them becoming engineers. Intelligence is highly heritable. Don't be racist but it was a mistake to let five million low-IQ people into your society.
Are Arabs low IQ? History suggests they've done quite well at times, only falling behind in recent centuries.
The national IQ of Syria is in the 80's. While national IQ's aren't super-accurate I think they're directionally informative. The bigger issue is that these aren't normal economic immigrants (which are a better-than-average sample of the source population) but refugees who aren't selected for either ability or drive. Selection effects determine almost every social outcome. If you let in a bunch of low-human-capital people then you're going to wind up with a worse society in the long run.
the papers "Brain drain in Syria's ancient capital" and "Sex differences on the Progressive Matrices: Some data from Syria" have pretty good samples for getting an upper bound of national iq in Syria , and found 82.
The national IQ estimates are rarely inaccurate, the more valid complaints are whether that low IQ is due to their genes or their environment and are they just a few decades from going up 20 IQ points with the flynn effect or whether IQ test results don't mean quite the same thing that they do in western countries. thinking that these things will solve everything is cope they're better than trying to deny the national IQ samples entirely which is what leftoids normally do (not you)
Really? I'm nothing like an expert but I feel like one of the issues is getting a representative sample. Syria is pretty ethnically diverse, are the national IQ figures representative or did the testing systematically focus on a low-iq cohort? It's sort of like India: its national IQ is high-80s but that glosses over significant caste variation. For very poor countries I'm also somewhat skeptical that you can easily test the IQ of people who have never taken a written test before. Not that it's meaningless, I just mentally assign 5-10 point error bars and assume that reported figures are near the bottom end of that range.
Thanks for the references, I'll check them out.
they tested schools in damascus. capital cities usually have higher average IQs than the rest of the country due to brain drain. If you want to get a nationally representative sample you're right it would be harder but if you just want an upper bound , it's not so difficult.
They tested school children so it's not like they've never taken a written test before.
and yeah an IQ score of 82 in syria might mean something a bit different from an IQ score of 82 in a more developed country , but that still doesn't mean that the average national IQ scores is inaccruate. if you tested everyone in syria the average IQ really would be close to 82.
Arabs conquered the Classical Civilization, which gave them a lot of boost, but their intellectual trajectory since the destruction of Baghdad by the Mongols until the European colonial expansion was one of obvious descent.
Neither Egypt nor Iraq look like heirs to the most developed civilizations of Antiquity, and don't seem to be in recent upwind either. Italy and China, on the other hand, were able to reconstitute themselves into highly developed places, and India is probably on the road there as well.
Contrast also the Arabs with the Iranians, who are still more obviously intellectually developed as a culture, not even the Islamic Republic could extirpate the local intellectual life.
Even at Western universities, you will meet more Iranian than Arab scientists.
Muslim countries have become very inbred due to hundreds of years of repeated cousin marriage which lowers IQ
https://en.m.wikipedia.org/wiki/File:Global_prevalence_of_consanguinity.svg
https://www.discovermagazine.com/mind/cousin-marriage-can-reduce-iq-a-lot
Gregory Clark in "for whom the bell curve tolls" also theorises that due to the cultural prevalence of cousin marriage in muslim countries there would be less assortative mating and so the selection pressure for IQ or general social competence at acquiring socioeconomic status would not act as intensely.
This would explain why the middle east cities of north africa were at the forefront of civilization before Islam and not long after Islam spread but the more centuries they were muslim , the more they decayed.
People think of 1683 as when islamic civilization started falling behind but in fact that's only when they stated being beaten back in territory. When you look at global discoveries in science and mathematics it's clear that they had started falling behind centuries earlier.
No, you shouldn't. The children of immigrants from the Middle east and Africa are still overall large net tax drains
https://fm.dk/media/0qmmvey5/indvandreres-nettobidrag-til-de-offentlige-finanser-i-2018-a.pdf see figure 2.8
each descendant of immigrants from the middle east or north africa cost denmark on average 50000 Danish krona (6700 Euros) in just 2018 alone, even when standardising for the age distribution.
and in terms of crime they're even more criminal than their parents.
https://bra.se/download/18.45e4b8e192705389a34b52/1729514247826/2021_9_Misstankta_for_brott_bland_personer_med_inrikes_respektive_utrikes_bakgrund.pdf see figure 3
Specifically, 17% of middle eastern male immigrants in Sweden are crime suspects, but 24% of the male children of middle eastern immigrants born in Sweden are crime suspects.
And for North Africans it's even higher. 19% for the male immigrants and 31% for the male children of immigrants.
you would have to believe that simply being born and living in Europe all their lives doesn't change their human capital, but one or more further generations born in Europe will. The idea that they'll assimilate and become culturally indistinguishable from Europeans will happen a little bit but seems very unlikely to happen totally or even mostly, especially in the future as the % of indigenous children in european countries like Germany, France, UK, Austria, etc. plummets.
Why would the grandchildren of middle eastern and african immigrants in 20 years time assimilate to indigenous European culture when indigenous Europeans children will only make up 40% of the children growing up around them (much less in practice given that ethnic communities are separated to a large degree and spend much less time socialising than would be predicted by their overall percentages of the youth population)
So hoping that liberal, laissez-fair cultural assimilation will happen like a USA melting pot and that will raise the human capital of these immigrants and their descendants is very unlikely.
Once upon a time, there was a journalistic attempt to frame the German police as racist because they used a specific abbreviation "Nafri" (Nordafrikaner) for North Africans.
This attempt fizzled out immediately after the police showed just how much Nafris were overrepresented in all sorts of crimes. IIRC several hundred of them were responsible for half (?) of assaults in Saxony, with 4 million inhabitants, etc.
I hope the study corrects a changing baseline of educational attainment over the years. At some point it becomes more worthwhile to ask which college your parents went to than whether your parents went to college.
I can’t help but wonder what would happen if we analyzed the quality of novels by the usage of particular words or phrases. There is probably something there, as better novels, or better selling novels, or whatever, probably use or avoid certain words/phrases relative to crappy novels. But not much. Probably just enough to make it seem worthwhile to keep looking harder.
Of course we know that what makes a novel great is how all the words work together. That it isn’t so much which words, but how all the words chosen work together to create a higher level of “novel goodness.” Plot and readability and character insight are emergent qualities. Yes, the quality is in part connected back to the words, but only in a very holistic way (that also depends upon the cultural context the book exists in).
Not sure if any of this even helps… but a boy has to wonder.
This is a very useful line of thinking
"Typical estimates for adult IQ found it was about 60% genetic, 40% unpredictable, and barely related at all to parenting or family environment." A similar statement can be made regarding height where we find an even stronger genetic link, close to "80%".
But, this type of statement can be misleading and is only true in a limited scope of conditions.
To further clarify, here is a concrete example:
Average female height in Poland 1896-1914: 152.62 cm, std 5.74 cm
Average female height in Poland, children born 2000-2001: 167.85, std 6.91 cm
The average woman born in Poland 2000-2001 would have been over 2 standard deviations above the average in 1914, the average woman today would have been in the top 2.5% back then! Another way of stating the same fact, the 2 standard deviations increase in height is 0% genetic and 100% environmental.
The same trends can be observed in all developed countries all around the world. Less developed countries, for example in Africa, has experienced a much smaller increase in height. A clear separation can be seen been North and South Korea after the division of the countries with South Koreans gaining much more height than their North Korean relatives.
We understand this process fairly well, nutrition, pollution and disease greatly affect the growth and development of a child.
So, this type of analysis is fine, we should look for the genes that shape our bodies and lives, but we should remember that there are limitations to these types of studies.
> The average woman born in Poland 2000-2001 would have been over 2 standard deviations above the average in 1914, the average woman today would have been in the top 2.5% back then! Another way of stating the same fact, the 2 standard deviations increase in height is 0% genetic and 100% environmental.
3-5 generations passed between 1914 and 2000. Why would you assume no genetic change? There definitely was genetic change.
You are thinking of individual genetic changes.
I am talking about population genetics where 3-5 generations for a population of millions is a very short time. You would have to have an extreme evolutionary pressure to select for such a drastic change in such a short time. The collective genome of the Polish population would have seen an extremely small change over such a small amount of time, a few changes due to genetic drift, maybe a few changes due to migration. The greatest source of change would have been ww2 when approximately 20% of the Polish population was killed, but even an event this big and tragical would be considered small in terms of population genetics.
In the lab we often select for new traits using heavy evolutionary pressure and even then selection often takes 100's of generations, and this is with genomes that are less complex than our own. For a polygenic trait like height that is decided by approximately 12,000 genes according to one recent publication, you would need an extreme evolutionary pressure and many generations.
No, I'm thinking that you stated explicitly that 0% of the increase in height is genetic. There is no reason to believe that. Your "rebuttal" makes the argument that more than 0% is environmental, which is unrelated to what you said earlier.
Apologies if this has already been mentioned - I've read the comments but couldn't see anything about it, but I could easily have missed it. GWAS use common SNPs as markers for putative causative variants (which may be anything from another SNP to large structural variants), but the thing is, that means that all those variants have to be ancient, both the causative variants and the GWAS markers. GWAS won't detect links to more recently-derived causative variants. Studies tend to focus either on common SNPs (as per GWAS) or ultra-rare ones (single family pedigrees and Mendelian segregation), but I wonder how much variation is down to semi-common variants which are too recently-derived to show up using a GWAS but too common and/or variably penetrant/mild in effect to be identified using traditional Mendelian segregation methods.
As an example, I would offer a 12bp in-frame deletion in the gene RIPOR2, which is associated with non-syndromic hearing loss and is very common in the Netherlands (and also exhibits variable penetrance - not every carrier develops hearing impairment). It's a bit too recent and rare to be a perfect example, but it's not far off.
https://pmc.ncbi.nlm.nih.gov/articles/PMC8120656/
https://pubmed.ncbi.nlm.nih.gov/37164627/
Height is just a number printed on a particular stick.
Banned for this comment.
Scott! It’s all about peanut allergies! Ignore my title if you like but read it! “GxE is small” is just obviously wrong; those studies are looking at GxE far too narrowly. https://lymanstone.substack.com/p/why-twin-studies-are-garbage
That post is subscribers-only and I cannot read it.
This is one of those moments where it's very annoying that Substack doesn't allow you to gift-link specific articles. I understand why you wouldn't want to subscribe to a million Substacks of people who leave their articles in the comments. But I figure you also understand why I would keep interesting and original work paywalled. I would comp you a subscription but that requires an email or that you're already a free sub. Or you could do the trial-and-cancel thing. Alas, no simple workaround!
> But I figure you also understand why I would keep interesting and original work paywalled.
Depends. Are you in the entertainment business, or the knowledge production business?
Also, your footnote on Akbari’s intelligence selection paper seems mistaken. Yes the IQ PGS score changed, but not by NEARLY as much as many other complex traits. Clearly there’s been less selection on IQ PGS than many other traits where we do find heritability still!
I have twin adopted daughters and another natural daughter. Ask me anything.
In computer science, there is something called a hash function. Or similarly a Pseudo Random Number Generator. This is an algorithm that, given the same input, produces the same output. But given any small change in the input, the output is completely different.
If these traits were partly determined by a hash function over the genome, then any 2 people who aren't identical twins would have independent rolls of the genetic dice.
So this would show up as twins being the same, and except for twins there would be no detectable pattern between genes and results.
Another option is a gene-environment interaction effects.
For example, suppose some kidney genes work well with a high salt diet, and others work better with less salt. Then there is a strong correlation between twins, which share both environmental salt levels and genes. And a weaker correlation between siblings who often have different genes.
Yet looking across all environments, there is no best genome for kidney function.
> This is an algorithm that, given the same input, produces the same output. But given any small change in the input, the output is completely different.
No, that's a cryptographic hash function. A hash function has much laxer requirements.
Wouldn’t that result in Great Danes giving birth to chihuahuas?
I'm not conjecturing that everything is a hash function. Only that there is some component that is a hash function. (Well any sufficiently complicated function that science hasn't spotted the pattern yet really)
Oh, got it.
Behavioral twin studies first appeared in the 1920s, not the 1970s. Your defense of twin studies is weak, relying on supposed validation from other flawed methods and studies (e.g., adoption studies). P-hacked studies based on false assumptions cannot be used to cross-validate similarly flawed studies. It’s time to abandon behavioral and psychiatric research using twin method MZ-DZ comparisons after a disastrous 100-year run, and to rewrite the social and behavioral science textbooks accordingly.
Tangential to all of this, but something I recall from Nevin (one of the proponents of lead-crime hypothesis) on Rationally Speaking a few years ago:
At first glance, lead is the biodeterminist's favorite example, but also seems to prove too much. Many lines of evidence suggest that it's a massive effect, yet shared environment doesn't seem to be a big deal.
But IIRC, Nevin went through many original adoption studies, and found that adoptions were often much later than we imagine - over one year old, rather than just a few months. This might mean that either maternal/womb or early-childhood environment is actually much larger than we give it credit for.
http://rationallyspeakingpodcast.org/wp-content/uploads/2020/11/rs224transcript.pdf