In my personal opinion it is still important to figure out *why* there is a heritability gap:
-- If it is GxE (or AxC in ACE model language) inflating the twin estimates -- that's a lot of GxE! -- and we could imagine re-running all of the studies in Turkheimer & Waldron 2000 using polygenic score interactions to try to understand the interactive effect.
-- If it is gross EEA violations (I know this is an unpopular view) then we may need to re-evaluate some of the null findings from classical BG studies in twins, as we've effectively had a group of defiers present in the analysis this whole time. It also tells us something interesting about how parents (and society) treat kids when they look identical.
-- If it is a massive contribution of rare variation that GWAS is missing, that has major implications for population genetics and the way we think about selection on common traits.
For me, too, the most interesting question is why there's a heritability gap:
a.) Why are people so much more alike the more related they are?
b.) If almost everything is environment, why are the responsible variables so hard to pin down? And why are twin studies so much against it, since they are statistically much more robust than the usual social science studies?
c.) If genetics plays such a small role, does that mean that Mendelian inheritance and Darwinian selection are no longer important for humans?
d.) What is genetic variance within populations good for or responsible for?
Possible answers that I think are plausible:
a.) As it now seems, for whatever reason, almost every psychological and behavioural trait is rarely different from 0, at most up to 0.18, explainable with genes, which makes all hereditarians look stupid and decades of research a misallocation of effort.
b.) It's not the trait x SNP correlation that matters, it's the specific genetic package and combination passed on to the children that is responsible for the higher phenotypic heritability. It may be a combination of many SNPs that are very different at the population level, reducing the explained variance to almost zero, but within the family they make the offspring more similar to the parents in their more similar environment, leading to more similar outcomes that are not easily explained by parenting, schooling, books at home, and so on.
c.) Everything is relative: the more similar the environment, the more heritable a trait is. This will always be true, and I think this is somehow underestimated in the discussion. So twin studies give reasonable estimates of phenotypic heritability, but not genetic heritability, and they can't make valid assumptions about genetic heritability. In other words, genetics is the seed that gives a starting bonus and sets limits (still not everyone is going to be a gold medallist or a Nobel laureate), but the soil, the sun and the rain can nourish it to a very high degree and explain almost all the variance.
d.) We don't have a theoretical model or measurement tool that will make sense of the heritability gap in ten years' time or more, i.e. we are still at the beginning of a revolution in understanding genetics and heritability, and what seems clear now (very low heritability estimates) will be different in the future.
e.) Statistical intricacies such as overlapping confidence intervals, measurement error, power, correction for repeated measures, and things beyond my social science understanding of statistics.
Would it be fair to say that the gap between classic heritability estimates and direct genetic effects reflects the messy, complicated, and looping interactions between genes (one's own genes and family's) and environment? Cause it doesn't look like we can attribute all of it to "environment" alone in a manner that makes no reference to genes at all. I am not quite sure what's next given these revelations of direct genetic effects, especially since I am outsider to the field. But it's not clear to me that a direct h2 of 2% means that we can simply ignore "genetics." Maybe we can ignore the person's DNA for that purpose, but our scientific understanding of what's going on would still be referring to all the indirect genetic effects at the population level. No?
The problem here is one of adjusting the premise without adjusting the conclusion. If you propose that it’s “the messy, complicated, and looping interactions between genes (one's own genes and family's),” you have a mathematically untenable premise for high heritability. The truth is that the conclusion itself, that behavioral traits are significantly genetically influenced, has been called into question, but few in the field are willing to say it out loud. This isn’t a small problem, because it leaves the general public with the impression that the traits are genetically determined and this affects social policy decisions and arguably justifies vilifying the “other” in terms of wars, capital punishment, lengthy prison sentences, etc. The fact that it leaves researchers and scientists uneasy is no excuse to continue to perpetuate it and searching for some minuscule heritability doesn’t justify the broad-sweeping consequences of continuing to promote an idea that was based on refuted twin studies and personal biases.
Heritability estimates from twin and adoption studies come from studies 1) based on false assumptions, 2) using unreliable/invalid diagnoses, 3) that depend on p-hacked conclusions, 4) any combination of 1-3 above.
Also, feeling that way about the dangers of people believing that traits are genetically determined, can you possibly claim to be objective in your analysis of whether traits are in fact, genetically determined?? Isn't that going to affect your thinking and motivate you to resist the conclusion that traits are genetically determined?? Free up your mind.
And one other thing, the idea that these traits are somehow not heritable leads to the conclusion that in a perfectly nondiscriminatory world we would have perfectly equal outcomes. This arguably justifies vilifying the "other" in terms of calling them oppressors, justifying racial discrimination and segregation. It can even lead to the conclusion that to be fair, you must punish (for example in a classroom setting) the oppressors far more severely for an infraction in order to have the same average level of punishment for all. The Asian kid gets suspended for dropping his eraser and the Black kid goes unpunished for punching another kid, because you have to have perfectly equal levels of punishment. Then the level of violence in the classroom rises, everyone is afraid and there is no learning. Yes, it is important to get this right.
But not in basketball, of course. In basketball it is perfectly reasonable to have unequal outcomes. That is because the god who was supervising the evolution of physical traits dozed off for a while, but the god who was in charge of supervising the evolution of mental traits was fully attentive.
OMG. I wrote a post about this at https://comment78.substack.com/p/bound-to-fail?r=3c6ol1 . Isn't it obvious that for a very complex trait like neuroticism or intelligence that the crude GWAS we have now only reflects a small portion of the total genetic information? Isn't it obvious that with siblings, the rest of the genetic information is much more likely to be the same or similar than for persons with an unknown degree of relatedness?? Isn't it obvious that persons with similar GWAS scores are more likely to be related than persons with differing GWAS scores? So, the similarity of phenotype evaluations between siblings is more likely to be similar, even when the GWAS scores differ, not because of the environment, but because the vast bulk of genetic information that does not affect the GWAS score is more likely to be similar. Hey, if I am wrong, please tell me why. We do not have the computing power to figure out a really good GWAS for intelligence. Perhaps deep machine learning, if properly applied (no easy task), can get us a little closer to an answer.
Just one further thought. So, far the supposed heritability bound, is much less restrictive for very simple traits, determined by just a few identified SNPS, than for complex traits. Why's that, if I am wrong???
Hmmm. Are you saying that efforts to show rigid biological human uniformity keep failing, but still you hold out hope that somehow, some day it will be shown?? Yeah, the theory of only cultural evolution over the last 70000 years sure bit the dust with the advent of DNA retrieval from ancient skeletons. But, hey, maybe some day your ilk will find some theory with some legitimacy. Until then it is at best indeterminate. So why, exactly, do we keep using racial discrimination to try to force equal outcomes?? (AKA DEI). And you persecute anyone who does not go along. That is the most disturbing part.
What you are saying is “Why can’t we assume that genetics is the cause of disparities even though we don’t have evidence for it”? The main answer is that it is harmful, and not just potentially harmful. It has caused great harm and misery. A secondary answer is that the idea that genetic variants will make someone smarter is asinine. For some reason, despite the current flawed thinking being identical to phrenology, and simply driven by racist and classist assumptions, you aren’t seeing that and you think it’s because you have some special discretionary skills.
That is not what I am saying, at all. What I am saying is that there is not a shred of evidence that differing populations are rigidly uniform in every material aspect. In 70,000 years of separation, differing population sizes, different environments, it is more likely than not that some significant differences developed. Hence, it is wrong to just assume rigid uniformity. The idea that differing genetics can make some smarter is a certainty. It is different genetics that makes people smarter than other animals. We think with our brains and that is a physical part of our bodies, grown from a genetic blueprint. The racial discrimination motivated by the idea of rigid uniformity is causing pain and misery right now. The belief in rigid uniformity has destroyed entire cities. Cairo, Ill is barely even there, anymore. The human suffering when cities die is out of this world. We must have the courage to challenge the taboo, to build a world that is more logical and fair.
I don’t like eugenics, so I’m too biased is a solid take. The only person who can formulate an opinion is someone who has no beliefs at all. Interesting idea. Presumably, you have no beliefs or opinions. Which straight facts lead to your conclusions.
Leads you to that conclusion. It’s actually your assumption. Sadly the genes aren’t there. Conclude what you want, but the studies refuted high heritability.
I don't think anyone could possibly believe such low heritability results, as they are plainly impossible compared with the family studies. I imagine most of these supposed upper-bound estimates will be beaten by actual polygenic scores in the next few years. Based on family studies, we have a very strong prior that something is off about the method for these results. Because of this, I am not really worried. In 10 years time, we will have large WGS datasets (500k in UKBB) with millions of people and we will have a better estimate directly from genomics. Haplotype modeling may be one way to avoid the extreme multiple-testing issues with WGS data (100s of millions of variants, most with almost no variation).
I published a paper in 2014 which asked what it would take for the null hypothesis of the Human Genome Project to be accepted. Turkheimer, understandably, suggested I was jumping the gun in suggesting the time had come for us have a rethink about what we teach undergrads. The paper also asserted that largely absent Direct Genetic Effect of all HGP studies demanded a reanalysis of the findings of twin studies. In particular, that where higher correlations MZ vs DZ are found, it might suggest that in the case of those traits (such as IQ), there is greater uniformity of parental nurture. The paper is at https://thebowlbycentre.org.uk/wp-content/uploads/2014/10/Not-in-Your-Genes%E2%80%94Time-to-Accept-the-Null-Hypothesis-of-the-Human-Genome-Project.pdf
That's fair, and I do think the most recent within-family heritability results do call for a re-evaluation of the meaning of heritability in human behavior. But, those twin values haven't gone away, and I don't think it would be right to dismiss them as just EEA or whatever. MZ twins really are similar in their behavior, and if you are going to take tiny SNP heritabilities seriously, I think you have to take that seriously as well. The same is more or less true of first degree relatives. After than it drops off pretty fast. I think the conclusion is going to be that quantitative genetic relatedness gives us a way to amplify genetic signals in a way that does not apply at the individual level. To be clear, that would be a major concession, an admission for example that personal genomics is basically useless for understanding behavioral traits.
Interesting. It is greatly to be applauded that you are genuinely trying to keep an open mind about all this. Is it not time for you to set this out in a journal as well as in your (much anticipated) book?
The variance explained by SNP heritabilities is usually so small that if it were found in a study of, say, the impact of maternal depression on longitudinally measured outcomes for offspring, it would be regarded as of little or no import? Indeed, given the billions that have been spent on the HGP, in any other field of science would not the research bodies have stopped backing proposals to do further studies of the HGP hypothesis, or at the least, greatly reduced expenditure on it? As for the explanation for differences in similarity MS vs DZ, I have always argued that there surely are biologically based factors other than genes which could be partly explaining them (pure autism is surely inborn?) - eg we already know foetal exposure to maternal stress in the last trimester is a significant factor in outcomes like ADHD and autism, might there not be purely biological mechanisms during the third trimester or before that, which influence outcomes more in MZ vs DZ? Keep up the good work!
I doubt that Thomas Bouchard showed a slide with the MISTRA DZ-apart IQ correlations, since those were omitted from the famous 1990 MISTRA science article, and have been hidden ever since.
Are there no other explanations other than the field of genetics has been totally wrong about one of its central claims? For me, I don’t know if this passes the “sniff test”. Is this study shaking the foundations of the field?
In my personal opinion it is still important to figure out *why* there is a heritability gap:
-- If it is GxE (or AxC in ACE model language) inflating the twin estimates -- that's a lot of GxE! -- and we could imagine re-running all of the studies in Turkheimer & Waldron 2000 using polygenic score interactions to try to understand the interactive effect.
-- If it is gross EEA violations (I know this is an unpopular view) then we may need to re-evaluate some of the null findings from classical BG studies in twins, as we've effectively had a group of defiers present in the analysis this whole time. It also tells us something interesting about how parents (and society) treat kids when they look identical.
-- If it is a massive contribution of rare variation that GWAS is missing, that has major implications for population genetics and the way we think about selection on common traits.
For me, too, the most interesting question is why there's a heritability gap:
a.) Why are people so much more alike the more related they are?
b.) If almost everything is environment, why are the responsible variables so hard to pin down? And why are twin studies so much against it, since they are statistically much more robust than the usual social science studies?
c.) If genetics plays such a small role, does that mean that Mendelian inheritance and Darwinian selection are no longer important for humans?
d.) What is genetic variance within populations good for or responsible for?
Possible answers that I think are plausible:
a.) As it now seems, for whatever reason, almost every psychological and behavioural trait is rarely different from 0, at most up to 0.18, explainable with genes, which makes all hereditarians look stupid and decades of research a misallocation of effort.
b.) It's not the trait x SNP correlation that matters, it's the specific genetic package and combination passed on to the children that is responsible for the higher phenotypic heritability. It may be a combination of many SNPs that are very different at the population level, reducing the explained variance to almost zero, but within the family they make the offspring more similar to the parents in their more similar environment, leading to more similar outcomes that are not easily explained by parenting, schooling, books at home, and so on.
c.) Everything is relative: the more similar the environment, the more heritable a trait is. This will always be true, and I think this is somehow underestimated in the discussion. So twin studies give reasonable estimates of phenotypic heritability, but not genetic heritability, and they can't make valid assumptions about genetic heritability. In other words, genetics is the seed that gives a starting bonus and sets limits (still not everyone is going to be a gold medallist or a Nobel laureate), but the soil, the sun and the rain can nourish it to a very high degree and explain almost all the variance.
d.) We don't have a theoretical model or measurement tool that will make sense of the heritability gap in ten years' time or more, i.e. we are still at the beginning of a revolution in understanding genetics and heritability, and what seems clear now (very low heritability estimates) will be different in the future.
e.) Statistical intricacies such as overlapping confidence intervals, measurement error, power, correction for repeated measures, and things beyond my social science understanding of statistics.
f.) Chaos theory.
What rules out significant heritability in your mind? It’s like ruling out that there is a treasure buried on an island.
Would it be fair to say that the gap between classic heritability estimates and direct genetic effects reflects the messy, complicated, and looping interactions between genes (one's own genes and family's) and environment? Cause it doesn't look like we can attribute all of it to "environment" alone in a manner that makes no reference to genes at all. I am not quite sure what's next given these revelations of direct genetic effects, especially since I am outsider to the field. But it's not clear to me that a direct h2 of 2% means that we can simply ignore "genetics." Maybe we can ignore the person's DNA for that purpose, but our scientific understanding of what's going on would still be referring to all the indirect genetic effects at the population level. No?
The problem here is one of adjusting the premise without adjusting the conclusion. If you propose that it’s “the messy, complicated, and looping interactions between genes (one's own genes and family's),” you have a mathematically untenable premise for high heritability. The truth is that the conclusion itself, that behavioral traits are significantly genetically influenced, has been called into question, but few in the field are willing to say it out loud. This isn’t a small problem, because it leaves the general public with the impression that the traits are genetically determined and this affects social policy decisions and arguably justifies vilifying the “other” in terms of wars, capital punishment, lengthy prison sentences, etc. The fact that it leaves researchers and scientists uneasy is no excuse to continue to perpetuate it and searching for some minuscule heritability doesn’t justify the broad-sweeping consequences of continuing to promote an idea that was based on refuted twin studies and personal biases.
Heritability estimates from twin and adoption studies come from studies 1) based on false assumptions, 2) using unreliable/invalid diagnoses, 3) that depend on p-hacked conclusions, 4) any combination of 1-3 above.
Also, feeling that way about the dangers of people believing that traits are genetically determined, can you possibly claim to be objective in your analysis of whether traits are in fact, genetically determined?? Isn't that going to affect your thinking and motivate you to resist the conclusion that traits are genetically determined?? Free up your mind.
And one other thing, the idea that these traits are somehow not heritable leads to the conclusion that in a perfectly nondiscriminatory world we would have perfectly equal outcomes. This arguably justifies vilifying the "other" in terms of calling them oppressors, justifying racial discrimination and segregation. It can even lead to the conclusion that to be fair, you must punish (for example in a classroom setting) the oppressors far more severely for an infraction in order to have the same average level of punishment for all. The Asian kid gets suspended for dropping his eraser and the Black kid goes unpunished for punching another kid, because you have to have perfectly equal levels of punishment. Then the level of violence in the classroom rises, everyone is afraid and there is no learning. Yes, it is important to get this right.
But not in basketball, of course. In basketball it is perfectly reasonable to have unequal outcomes. That is because the god who was supervising the evolution of physical traits dozed off for a while, but the god who was in charge of supervising the evolution of mental traits was fully attentive.
OMG. I wrote a post about this at https://comment78.substack.com/p/bound-to-fail?r=3c6ol1 . Isn't it obvious that for a very complex trait like neuroticism or intelligence that the crude GWAS we have now only reflects a small portion of the total genetic information? Isn't it obvious that with siblings, the rest of the genetic information is much more likely to be the same or similar than for persons with an unknown degree of relatedness?? Isn't it obvious that persons with similar GWAS scores are more likely to be related than persons with differing GWAS scores? So, the similarity of phenotype evaluations between siblings is more likely to be similar, even when the GWAS scores differ, not because of the environment, but because the vast bulk of genetic information that does not affect the GWAS score is more likely to be similar. Hey, if I am wrong, please tell me why. We do not have the computing power to figure out a really good GWAS for intelligence. Perhaps deep machine learning, if properly applied (no easy task), can get us a little closer to an answer.
Just one further thought. So, far the supposed heritability bound, is much less restrictive for very simple traits, determined by just a few identified SNPS, than for complex traits. Why's that, if I am wrong???
It’s season 12 of The Curse of Oak Island. Think this is the year they will find the treasure?
Hmmm. Are you saying that efforts to show rigid biological human uniformity keep failing, but still you hold out hope that somehow, some day it will be shown?? Yeah, the theory of only cultural evolution over the last 70000 years sure bit the dust with the advent of DNA retrieval from ancient skeletons. But, hey, maybe some day your ilk will find some theory with some legitimacy. Until then it is at best indeterminate. So why, exactly, do we keep using racial discrimination to try to force equal outcomes?? (AKA DEI). And you persecute anyone who does not go along. That is the most disturbing part.
What you are saying is “Why can’t we assume that genetics is the cause of disparities even though we don’t have evidence for it”? The main answer is that it is harmful, and not just potentially harmful. It has caused great harm and misery. A secondary answer is that the idea that genetic variants will make someone smarter is asinine. For some reason, despite the current flawed thinking being identical to phrenology, and simply driven by racist and classist assumptions, you aren’t seeing that and you think it’s because you have some special discretionary skills.
That is not what I am saying, at all. What I am saying is that there is not a shred of evidence that differing populations are rigidly uniform in every material aspect. In 70,000 years of separation, differing population sizes, different environments, it is more likely than not that some significant differences developed. Hence, it is wrong to just assume rigid uniformity. The idea that differing genetics can make some smarter is a certainty. It is different genetics that makes people smarter than other animals. We think with our brains and that is a physical part of our bodies, grown from a genetic blueprint. The racial discrimination motivated by the idea of rigid uniformity is causing pain and misery right now. The belief in rigid uniformity has destroyed entire cities. Cairo, Ill is barely even there, anymore. The human suffering when cities die is out of this world. We must have the courage to challenge the taboo, to build a world that is more logical and fair.
I don’t like eugenics, so I’m too biased is a solid take. The only person who can formulate an opinion is someone who has no beliefs at all. Interesting idea. Presumably, you have no beliefs or opinions. Which straight facts lead to your conclusions.
Leads you to that conclusion. It’s actually your assumption. Sadly the genes aren’t there. Conclude what you want, but the studies refuted high heritability.
Yes, the flawed studies refuted high heritability. Not a very specific response, there.
I don't think anyone could possibly believe such low heritability results, as they are plainly impossible compared with the family studies. I imagine most of these supposed upper-bound estimates will be beaten by actual polygenic scores in the next few years. Based on family studies, we have a very strong prior that something is off about the method for these results. Because of this, I am not really worried. In 10 years time, we will have large WGS datasets (500k in UKBB) with millions of people and we will have a better estimate directly from genomics. Haplotype modeling may be one way to avoid the extreme multiple-testing issues with WGS data (100s of millions of variants, most with almost no variation).
n =~ 12,000 for depression. 3 billion base pairs in the human genome. Looks like way too low power to to rule out h^2 = 1.
I published a paper in 2014 which asked what it would take for the null hypothesis of the Human Genome Project to be accepted. Turkheimer, understandably, suggested I was jumping the gun in suggesting the time had come for us have a rethink about what we teach undergrads. The paper also asserted that largely absent Direct Genetic Effect of all HGP studies demanded a reanalysis of the findings of twin studies. In particular, that where higher correlations MZ vs DZ are found, it might suggest that in the case of those traits (such as IQ), there is greater uniformity of parental nurture. The paper is at https://thebowlbycentre.org.uk/wp-content/uploads/2014/10/Not-in-Your-Genes%E2%80%94Time-to-Accept-the-Null-Hypothesis-of-the-Human-Genome-Project.pdf
That's fair, and I do think the most recent within-family heritability results do call for a re-evaluation of the meaning of heritability in human behavior. But, those twin values haven't gone away, and I don't think it would be right to dismiss them as just EEA or whatever. MZ twins really are similar in their behavior, and if you are going to take tiny SNP heritabilities seriously, I think you have to take that seriously as well. The same is more or less true of first degree relatives. After than it drops off pretty fast. I think the conclusion is going to be that quantitative genetic relatedness gives us a way to amplify genetic signals in a way that does not apply at the individual level. To be clear, that would be a major concession, an admission for example that personal genomics is basically useless for understanding behavioral traits.
Interesting. It is greatly to be applauded that you are genuinely trying to keep an open mind about all this. Is it not time for you to set this out in a journal as well as in your (much anticipated) book?
The variance explained by SNP heritabilities is usually so small that if it were found in a study of, say, the impact of maternal depression on longitudinally measured outcomes for offspring, it would be regarded as of little or no import? Indeed, given the billions that have been spent on the HGP, in any other field of science would not the research bodies have stopped backing proposals to do further studies of the HGP hypothesis, or at the least, greatly reduced expenditure on it? As for the explanation for differences in similarity MS vs DZ, I have always argued that there surely are biologically based factors other than genes which could be partly explaining them (pure autism is surely inborn?) - eg we already know foetal exposure to maternal stress in the last trimester is a significant factor in outcomes like ADHD and autism, might there not be purely biological mechanisms during the third trimester or before that, which influence outcomes more in MZ vs DZ? Keep up the good work!
Have you seen this paper? I go pretty far in taking tiny SNP results seriously, and have taken a lot of heat from the BG establishment for it. https://muse.jhu.edu/pub/1/article/923702/pdf?casa_token=yJpFVqHSVMEAAAAA:G6o-BOneropzMluq2_TT_YPDQkhFb4LPg-NYIaJipawvsYupQa98CnTiJVKENJU2N7lDdkkLsg
The paper looks v interesting. Do you have a link to the full paper - i dont have access to a uni library
Try this link: https://drive.google.com/file/d/1-ms6LmZ1pCxYSP-BgyJVelQ741Gey5kK/view?usp=sharing
I doubt that Thomas Bouchard showed a slide with the MISTRA DZ-apart IQ correlations, since those were omitted from the famous 1990 MISTRA science article, and have been hidden ever since.
Are there no other explanations other than the field of genetics has been totally wrong about one of its central claims? For me, I don’t know if this passes the “sniff test”. Is this study shaking the foundations of the field?
I guess we can henceforth just go ahead and disregard family history in diagnostic evaluations. Who needs heritability nonsense!!!
I discuss my ideas on the heritability gap here. James Flynn will be quite relevant on this issue I think. https://open.substack.com/pub/eclecticinquiries/p/twin-studies-exaggerate-iq-heritability?r=4952v2&utm_medium=ios