Wikipedia has a fun list of effect sizes here: https://en.wikipedia.org/wiki/Genome-wide_complex_trait_anal...
Unsurprisingly, height is 0.40-0.60 heritable, but did you know that "nostril area" is 0.657 heritable, while cilantro tasting is only 0.087?
The NYT article seems to imply this is the first study of its kind to be done, which is absolutely not the case. The GCTA article linked above cites 18 GWASes on general intelligence, starting in 2011. The abstract on the Nature article focuses on the 15 new SNPs found that affect intelligence, in addition to the 336 already known. It's an incremental advance, not a stunning breakthrough.
HN user gwern has covered this field extensively. Shouldn't be surprised to see him show up in this thread and gently correct some of the confused thinking in the light gray comments.
There are some phenotypes (characteristics) which clearly can be associated with a small number of genes (e.g. autosomal recessive disorders) , however for many phenotypes such as intelligence, it takes a lot more effort to correlate them with underlying genotypes. The fact that intelligence is approximately normally distributed is an indicator that there are many individual effects at play.
Intelligence test scores are normally distributed by construction: the weighting of questions is adjusted to ensure normality on a sample population. That test scores are normally distributed in samples other than the normative one tells us nothing other than that the test designers succeeded.
Contrast this with height, the measure of which is not defined in terms of a population distribution.
In this case, it's not an additive, multi-factorial trait; it's epistatic. There's a 'black hair' gene that overrides whatever else is happening in your genome. This is often the case with color traits: a dark color overrides anything lighter. It's the classic example of dominance.
It's a little ambiguous, but that's not how I read the abstract. The relevant sentence:
> We identify 336 associated SNPs in 18 genomic loci, of which 15 are new.
That seems to be saying that 15 of the 18 genomic loci are new, not that 15 of the 336 SNPs are new. Given that they "identified" 336 SNPs, presumably they all are new?
'Classical' GWAS are a top-down approach, where we try to find strong associations betweens a genetic locus and a trait (in this case intelligence), and they rely on very stringent p-value corrections due to the number of tests performed. This is why although height is >0.6 heritable (meaning genetics explain more than 60% of the trait variance in a given population), only 20% of the variance can be attributed to GWAS finds , such as the 52 genes in this article. Bear in mind that the height results were obtained with the largest GWAS ever with more than 250,000 samples and 300 institutes .
So where is the missing heritability? This is the starting point of a bottom-up approach, where we try to predict a trait from the genotype. With some clever but simple calculation, we know that most heritability can be found in the data (even if it does not pass the stringent significant thresholds). This is what is called 'chip heritability', the heritability explained by all the common genetic variants present on commercial chips taken together. With height, the chip heritability reaches 62.5% . The bottom-up approach is especially looked at for psychiatric disorders, and already delivered some interesting results such as 0.8 genetic correlation between schizophrenia and bipolar disorder . But useful clinical diagnostics based on genetic data are still a long way off.
This should please the HN crowd to know that we still have one tough prediction problem to crack. Current results can be improved either with more sophisticated machine learning methods or by integrating various biological annotations to try to unearth the missing heritability that GWAS can't seem to find with the current sample sizes.
Here are a few more links for those interested  .
I don't see how you can assign numbers like this. As a simple example, say the amount of protein absorbed while growing up affects adult height. Then you can have a certain protein catabolyzing gene that is highly correlated with height when protein in the diet is low, but has little relationship when dietary protein is high (since the person gets enough absorbed protein either way).
So the variance explained (either by a given genetic locus or in total) is not going to be any kind of constant value, and can in fact vary wildly.
The inability to explain heritability as more than a "sum of the parts" is a problem which I'm not familiar with a solution for at the genome scale.
The more we look, the more we're likely to find confounding genes like this. Sure, it might be that those genes directly raise intelligence... but it might also be that those genes make the parents less likely to smoke, and the measured difference in intellect is entirely due to exposure to parental smoking.
If there's a gene for "less likely to drop babies on their heads", it will probably show up as being correlated with higher intellect -- even if it has no direct effect on intellect and instead acts solely though yielding improved parenting.
If your goal is to select for greater intelligence, then genes which prevent the carrier from smoking and dropping their offspring will increase intelligence, which is the point!
Some of these are not smart genes, if they still result in dimwitted people you'd hate to be trapped in a room with. They may be good genes that promote healthy survival, and statistically stongly correlated with smart people, but if they aren't the crucial genes that result in the sort of person that quickly grasps complex concepts, has nice handwriting and can keep up with high learning curves, then they are helpful health and survival genes, not intelligence genes.
Specifically, there is no such thing as an IQ of 100. Instead, there is a mean score on a test within a population. That score might be 80, or it might be 597, or it might 8. By definition, if your score on the test is identical to the mean, then you have an IQ of 100 relative to that population.
If a second population takes the same test, and the mean in that population is different, the score which represents and IQ of 100 is different than the score which represents an IQ of 100 for the first population.
So, it is possible for two people to get the same score on an identical test, and be assigned different IQs.
Because of this, IQ values do not constitute a metric space, and anyone who tries to use it as such, like these studies do, is building their conclusions upon false premises.
PCA doesn't perfectly control away population structure, especially if you have cryptic tiny subpopulations in your dataset that only few individuals are in. In that case you don't have enough signal to establish the subpopulations. It's also a function of the way you call your SNPs - if you don't have enough SNPs, or focus on a different method, PCA won't magically know about those unobserved populations.
I don't think they were being sarcastic though, because the rest of their replies are pretty serious.
Comments like these are why I come to HN
This sounds really interesting, how does that work exactly?
Researchers have conducted many studies to look for genes that influence intelligence. Many of these studies have focused on similarities and differences in IQ within families, particularly looking at adopted children and twins. These studies suggest that genetic factors underlie about 50 percent of the difference in intelligence among individuals.
So, going with the NIH, I'd say about half.
"Around 90% of experts believed that genes had at least some influence on cross-national differences in cognitive ability."
At these sample sizes, that is an interesting idea. Has anyone checked?
(also, do include some marker of sarcasm in your comments, it can get confusing)
Why do you believe that PCA successfully captures all the relevant population structure? Probablistically, it's a Gaussian linear model... no reason to expect allele frequencies to follow that. If that were the case, methods like coalescent theory would be unnecessary.
Why do you believe that high genetic correlations between cohorts is sufficient to ensure safe inference from GWAS meta analysis? These are tiny effect sizes, and the argument is essentially that the correlations with intelligence measurements are statistically significant. Unless you believe that your null model is an otherwise entirely accurate reflection of how the data is generated, of course you're going to see spuriously significant results in huge samples.
I'll respond to your other points when/if I catch up on the literature. I'd be grateful if you could link the papers about the enrichment sets and tissue expression results.
'It means there’s a long way to go, and there are going to be a lot of other genes that are going to be important,' Dr. Posthuma said."
Indeed. This result is fully predictable from the several previous GWAS studies on IQ that have been done. Any gene that influences IQ in the general population has at most a small effect, and the variance among human individuals in IQ is the result of interaction of hundreds of genes and the environment the individual lives in.
Here's a link to a lot of good writings by one of the leading researchers on human behavior genetics, for more details. I especially like one particular article I'll link just below the link to his collected writings.
Turkheimer, E. (2000). Three laws of behavior genetics and what they mean. Current Directions in Psychological Science, 9, 160-164.
I assume the authors (all three of them professors of some renown) did not write this headline, because the piece is far better, more careful, and more comprehensive than the title would suggest (I'm no fan of Murray but this is not a typical Vox hot-take).
Here's a good takedown of the Vox article: https://medium.com/@houstoneuler/the-cherry-picked-science-i...
Vox is one of my least favorite outlets; not because they're more inaccurate or dishonest than others, but because they intentionally (and somewhat successfully) foster a reputation of facts-first journalism and then use that as a fig leaf for a lot of the same low-quality crap you see everywhere else too.
* It doesn't refute the scientific claims of the authors, but rather takes them to task for disagreeing with points Murray didn't make --- that's fair rhetorically, but fails to address the central thesis of the Vox piece, which is that popular race/IQ ideas are largely junk science.
* It asserts the authority of Charles Murray (for instance, citing him as having written an article rebutting one of Turkheimer's findings), despite Murray not being an expert on the subject --- his scientist coauthor on The Bell Curve is deceased. At one point, in this very takedown, its author even cites Murray observing that he doesn't understand the science.
* It attempts to dismiss the authors by citing James Flynn's critique of black culture, which is ironic on two levels: first, that itself is an argument that Turkheimer and Nisbett didn't make, so the pot is complaining about the kettle, and second because Nisbett himself has gotten into some trouble for making similar critiques of black culture. Memetics isn't genetics.
* It makes seemingly well-grounded arguments about the statistical validity of interventions Turkheimer talked about which, on a second reading, dissipate into handwaving --- what we're left with is a Medium blogger's conjecture against a working scientist's published results.
Turkheimer and Nisbett are at pains to point out that their views aren't uniformly shared by everyone in the field. Rather, they take Murray (and, by implication, Sam Harris) to task for summarizing a contentious scientific debate as if it were settled, but for hysteria in leftist academia. The Vox piece is far more persuasive and credible in this argument than the "takedown" you cite is in its own argument.
2) Never said Murray was an authority. I wrote he rebutted Dickens and Flynn on the non-narrowing of the gap in the last 25 years, which is correct, regardless of whether he's an "authority." Also, I didn't say he "doesn't understand the science," I said he doesn't understand the details of the mathematical techniques in a paper regarding the nature of the Flynn effect. This seems unimportant because Wicherts's conclusion -- that the Flynn effect gains have a different structure than the B-W IQ gap -- is uncontroversial.
3) Not really. I was citing Flynn for making the same point as Wicherts's above re: structure of FE gains, a demonstration that it's not just Wicherts's theory. I quoted him saying un-PC things about black culture just to demonstrate that in order to rebut Murray's unPC claims many scientists come up with different unPC hypotheses.
4) Talk about "handwaving": it would be helpful for you to elaborate what arguments you think actually "dissipate" on a second reading but I doubt you will.
And I would care more about your ad-hominem attack were it not for the Rindermann survey. Which brings me to my final point: you have it completely backwards when you say "Turkheimer and Nisbett are at pains to point out that their views aren't uniformly shared by everyone in the field. Rather, they take Murray (and, by implication, Sam Harris) to task for summarizing a contentious scientific debate as if it were settled." The scientists at Vox rather explicitly say that no credible scientists attribute some group IQ differences to genetics, while the survey I cite in the Medium article is convincing that is an inaccurate assessment of the views of researchers in this field. Moreover it's pretty clear that Harris does not believe the science is settled, and Murray also thinks that we should wait for direct DNA evidence to say that it's settled.
Genes for skin pigment
If you've never heard of inequality before, I guess
Intelligence is of course huge. Trustworthiness is huge. Probably 20% of GDP is related to trust whether it's paying for security, police, the military, $100s of billions of excess health care spending from over-billing and over-testing.
I think engineered human super-intelligence will have inestimably more impact than AI...but it enables AI because in a world of a few billion Leonardos, everyone will have figured out ways to solve repetitive work.
The biggest issue with better ethics is that we can't define them, so we couldn't optimize for them. Nobody agrees on what a good person is because nobody agrees on the meaning of good. But meanwhile the real world outcomes are decidedly not good.
Why would it? Don't we have enough smart people? Smarts come at a cost. Its usually tied with anxiety disorders, neuroticism, poor ability to develop social skills, limited emotional expression, lack of empathy, autism, etc.
We are also discounting things like emotional intelligence and other types of intelligences we may not fully understand yet or even recognize!
I'm not sure if the world needs a bunch of cold and emotionally stunted people taking the place of more well-rounded people. I think its questionable to think evolution didn't account for the diversity of personality types needed for a successful human society. Claiming "This type one here, this is the only good one, so we'll make everyone like this," seems foolish if not disastrous. Personally, I'd gladly shave off more than a few IQ points to be more empathic and social.
Not to mention, society already tried this with eugenics fairly recently. Its not "smart" to repeat the mistakes of history.
>Not to mention, society already tried this with eugenics fairly recently. Its not "smart" to repeat the mistakes of history.
When did we "try this" with eugenics "fairly recently"?
I'm almost certain you're referring to Nazism. But was Hitler really killing all the Jews for some sort of eugenics experiment? No, I don't think so. Were his actions genuinely scientific at all? Nope. He believed that raw power conquered all and that anyone weak should be murdered. That's different than eugenics.
We need to get past this kneejerk reaction comparing doing anything beneficial within the human genome to Nazi eugenics. It's a completely absurd comparison, especially considering that the average intelligence of Jews is very high.
Hitler was evil because he was a mass murderer, it's as simple as that. I really shouldn't have to clarify this on YC of all places.
It seems "smart people" need to start reading some history books.
Its complicated to nation-build. Any that work now, do so because of competition and selection. Remove that with 'intelligent design' and all adaptation stops. Fitness plummets.
I'm thinking more along the lines of attenuating certain (hypothetical) genetic contributors to antisocial personality disorder, violent psychosis, etc.
Not engineering docile perfection, just less of the obviously really bad stuff.
Edit: I do share your concerns about stopping Darwinian selection, though. Choosing local optima for cooperation might come back to bite us in the coming TransNeptunian War of 2217, e.g., when psychopathic, ruthless strategy is all that saves the human race from the conflict of first alien contact.
If we don't achieve a technological singularity in the next couple decades, I doubt that an alien invasion will be a worry a couple centuries from now. I say this because of impending ecological collapse, which will end human civilization as we know it... if not humanity in total.
And I'll save for another time a rant about how expensive it is for anything to cross interstellar distances. At most we'll see a self-replicating probe that tries to subsume all matter in the solar system, the "invaders" won't physically travel here.
And even as software, I predict that new colonists would likely arrive by communications laser, well after the local infrastructure has been built. Why bother coming over during the boring building phase?
lol, that's a great line.
While you may be right that our current genetic state, attained through generations of ruthless competition and selective pressures, is optimal for some things, it is harmful for others. Notably, the health of biosphere and many millions/billions of other humans. What good does nation-building do if our actions destroy each other through war or create an ecosystem unable to sustain advanced life? I would rather engineer for increased prosocial behavior and take our chances down that route than where we are headed now.
- You can be bad (i.e., lack of consequences)
- You want to be bad (i.e., some sort of reinforcement, often $).
Now both of these are de facto there if you become a high level politician and that's IMO the reason for real world outcomes [...] decidedly not good
Some people are bad and don't want to be and it is a struggle.
Some people just do what they need to to survive, for whatever survival means to them at any given time.
Some people test the limits of their, and societal, boundaries.
This is too complex of a concept to limit it to two coinciding factors, IMO we are all all of these things in differing quantities.
Hopefully people learn that the consequences of bad actions are rarely beneficial long term and develop processes to mitigate.
So you say. I'm much more inclined to believe they believe they'll get away with it.
> Some people are bad and don't want to be and it is a struggle.
Sure, but that's almost exclusively due to mental illness, not "rationality".
> Some people just do what they need to to survive, for whatever survival means to them at any given time.
Survival is reinforcement.
> Some people test the limits of their, and societal, boundaries.
Mostly because closer to the limit means more reinforcement. Simple risk/reward.
I want to think that being good is non-religiously rewarded (ie. in this life), but I'm not entirely sure that's true. I think it takes effort not to take unethical shortcuts or make profitable but hurtful decisions.
> So you say. I'm much more inclined to believe they believe they'll get away with it.
Says Dylann Roof (convicted of mass shootings at a church in South Carolina), "I do not regret what I did. I am not sorry."
The idea that intelligence produces rationality is a myth.
But it'd be pretty strange that IQ would not correlate with real world adaptation efficiency. After all, intelligence is useless if you can't use it to better navigate the world, because it itself misleads you (then you are better off with just fight-or-flight instincts).
And probably tendency for rationality has a genetic component too.
I want to make the world better but it's a bad use of my time from an individual perspective.
Intelligence and trustworthiness are not what you're looking for. Obviously we would love to engineer humans to be healthier, smarter, etc, but these do little to enhance our long-term survivability in the face of human on human and human on environment domination that we see today.
What we are most in need of is to engineer a more compassionate human than exists now, by modifying key genes for species-typical behavior. All humans have a tremendous and largely unconscious ability to tune out highly-destructive socioecological acts. We might expect these problematic genes to occur with little variation across all human populations.
When safe, effective germline engineering for greater intelligence and health is possible, I predict that it will be embraced as enthusiastically by the majority of society as indoor plumbing was.
Yes, some of them were quite inappropriate, such as:
>His hope is that everyone is equal, but he counters that “people who have to deal with black employees find this not true."
However, after that he says:
>There is no firm reason to anticipate that the intellectual capacities of peoples geographically separated in their evolution should prove to have evolved identically. Our wanting to reserve equal powers of reason as some universal heritage of humanity will not be enough to make it so.
Now, this latter quote is in my opinion a perfectly reasonable presumption, considering that hundreds of millions of humans historically lived in a myriad of drastically different environments and circumstance. These genetic differences among populations (I'm not talking about the rigid contemporary interpretation of "race", I'm talking about geographical populations) do exist, and it would be absurd to presume that for some magical reason it wouldn't affect cognitive ability as well.
And yet people would still get emotionally triggered and shoot this hypothesis down, just because of the way they are socialized and because it strongly deviates from the mainstream idealization of equality.
Not related to this issue - reality generally is not simple, and I am sad that even contemporary developed societies nurture expectations that are so simplified, idealistic, fictitious instead of being more mature and grounded in scientific reality, regardless of the topic at hand. I just wish that more people were more educated and scientifically literate, and that this would translate into better, proper political positions, instead of the idiotic clusterfuck we have today.
Obviously, these concerns will be a thing of the past considering the future potential of genetic engineering to amplify intelligence, and presuming that eventually such services would be available to your average person.
>He says that he is “inherently gloomy about the prospect of Africa” because “all our social policies are based on the fact that their intelligence is the same as ours – whereas all the testing says not really”, and I know that this “hot potato” is going to be difficult to address. His hope is that everyone is equal, but he counters that “people who have to deal with black employees find this not true”
An IQ of 125 is well above average and puts him in the top 5% of the population. And on an individual basis, IQ tests have some degree of error and aren't perfect. But they are very useful statistically, especially on a population level where these errors average out.
So was phrenology. This search for the g is just cargo cult science.
Keep in mind that all they were likely doing was correlating the presence of certain genes with intelligence.
They couldn't begin to start determining combinations of those genes 50.
It will certainly be the case that the combination of the genes matters. What if gene A gives you a little boost by itself and gene B gives you a little boost by itself, but genes A & B together actually lower your IQ?
With just 50 genes (assuming no mutual exclusions), you'd have 1.1258999e+15 different combinations of them.
Then as the article mentions, they think that there may be many more intelligence genes to find.
Genes with linear interactions would still grant evolutionary advantages.
Seems like a good application for NNs?
Given the vast complexity of all this, really the only low hanging fruit could come from determining modal genes, and eliminating all the many mutants since the overwhelming majority will be deleterious.
If you could run such a spellcheck on a person's entire genome, the results could be very impressive. Although, I admit that if done at large scales you'd essentially be creating a monoculture species. Of course, people are unlikely to all voluntarily opt for having the same genes, so under a voluntary gene therapy system this is an unlikely outcome... stratification is more likely.
IQ test are also not completely accurate for those that are gifted and have a learning disability. Kids that are gifted + LD'd may not even be put in gifted programs because their IQ scores are too low, but too high to put in an LD program, so they are stuck in normal class rooms where it isn't a good fit for them. Imagine being just as intelligent, if not more so, as the students in the gifted classes but denied access. That's what IQ scores do for some of those kids.
IQ test tend to work for gifted people that answer questions quickly and those without lopsided talent. If you are gifted + LD, or gifted but have really slow processing speed, then IQ test will not identify you as really that gifted, despite being so.
IQ tests are (imperfectly) correlated to grades, and income, but these things are also correlated to, for example, how quickly you can tap your finger. You could probably use that as a measure as well.
There are some people who might have a joint problem, and they would do horribly on finger tapping.
The point I am making is that all we have are imperfect measures. There is some underlying matter of fact. Maybe we need a better understanding of what intelligence is.
Douglas Hofstadter describes Feynman as acting out a sort of deliberate "village idiot" in a lecture on intelligence and patterns, by giving deliberately simplistic answers. With no proof, I think this is what Feynman must have been doing on the IQ tests, being deliberately but defensibly pedantic. I know that scoring 125 on an IQ test requires no skill that Feynman wasn't near the limits of human capability at.
The passage I'm talking about, from Hofstadter's amazing "Metamagical Themas": https://books.google.ca/books?id=NSpMDQAAQBAJ&q=feynman+sat
If you go and read the literature the idea is that thete is a concept called g, general intelligence, and that IQ tests dont test that, but produce a result which correlates with g. In this framework it is acceptable that Feynman's g is different from his IQ score--they were only ever meant to be correlated.
Many things correlate with successful life outcomes. Height for instance correlates with income.
Unsurprisingly, a single metric does not encompass the entirety of human intelligence. Failure to identify a particular gifted individual does not in any way make it an invalid metric. It means that you don't understand what a population metric is and what it's useful for. To be fair, many people don't understand this and the layperson understanding of IQ isn't valid, but that has no bearing on the 'proper' IQ.
Generally, whenever considering methods addressing evolution or heritability, you cannot confuse individuals and populations.
It's a politically sensitive topic, so articles play up problems with IQ while playing down it's predictive success.
But even one of the articles you quote admits "The difference between a 79 and a 69 is highly noticeable, and the test can determine which is which and the reasons why."
Tests like ACT and SAT are highly correlated with IQ. As is the Armed Services Vocational Aptitude Battery (ASVAB) that the military uses. So when Google checks applicant's SAT scores, they're checking their IQ.
One important thing to note is that it's a better prediction of performance of a group than an individual.
Anyways Feynman's purported 125 IQ score translates to an expected SAT score (V+M) of 1210. Which won't even get you into any of the schools he excelled at.
If your issue is my comment about IQ tests in the 1930s, I can't actually back that up. I have no idea what test he took. But psychologists have done a lot of work on the tests since then.
"Feynman was tested at 125"
There are some points to consider. There are and have been various publishers of IQ tests over the years and various forms of test. We don't have any clear idea what IQ test Feynman took or what it emphasized.
The IQ test that Feynman took was probably scored by the quotient method, meaning that an attempt was made to estimate his 'mental age' and the result was then divided by his chronological age and, finally, multiplied by 100.
IQ tests aren't scored that way any more—instead, a distribution of test scores is formed by giving the test to a large sample of test takers. A result one standard deviation above the mean of the sample group gives an IQ of 115, and so on.""
"A typical human behavioral trait is associated with very many genetic variants, each of which accounts for a very small percentage of the behavioral variability."
This sounds silly, doesn't it? But you can't deny that this article is a firm step in that direction. It's not crazy to think that parents may make gene choices in the future.
This would be not just a new stage, but a new kind of evolution. (Not too scary in the form we're reading about, and close to everyday sexual selection of mates anyway.)
Why? Without support it looks like a "just world fallacy." He's good at physical activities so he must be bad at thinking.
Sidenote: I am a Messi fanboy.
NB I don't follow football but I've just finishing reading a book on the Panama Papers that features a section on him.
This is the same as to claim that a particular tech-process of a CPU and its particular instruction set defines what kind of porn would be watched on a laptop. While there are few statistical correlations, say, between the cost of a CPU and statistically significant preference to, say, MILFs, a direct causality is still rather difficult to establish, at least for people of some intelligence.
They would take roughly a thousand people who are >=3 standard deviations smarter than average and compare their full genomes against people who are of average intelligence. This way they expect to find out all the "easy" genetic determinants of intelligence.
People underestimate the value of small changes in some types of intelligence. They cascade. If you learn how to learn faster, or better, it cascades across all the new learning. Imagine a bookshelf vs a stack of books. Some people develop tricks early on that are the the shelves. Hit a golf ball a little crooked and see how crooked it ends up 100 yards away.
Different is better:
This social programming to search for equality amongst men ignores the value of evolution, competition, meritocracy, and the resultant emergent fitness and excellence that results. Different, sometimes better, sometimes worse, is a requirement for robustness and progress.
See higher resolution:
Measure mental ability more like physical ability. Look at the detail measured in the nfl combine: https://en.wikipedia.org/wiki/NFL_Scouting_Combine
When you create an IQ test, you use specific stratagies to avoid cultural bias, and measure useful traits, like memory, speed, visualization, rotating objects in ones mind, etc. Why not just declare the results by category. The test maker knew what he was measuring. There's no reason to mush it all together.
Obviously not ideal, but a step in the right direction imo.
> There are no “genes for” IQ in any but the very weakest sense.
I'm not super knowledgeable about this stuff. I'm not trying to kick a hornet's nest; just trying to point out how I found this news to be particularly relevant in light of what I'd been reading recently. I'd be interested in seeing the Vox article's authors would fit this news into their thesis.
That is the most ridiculous straw-man in the history of straw-men. No one thinks IQ is determined by a single gene. In reality it's probably caused by a large number of small-effect genes, leading to the characteristic "bell-curve" distribution. To oversimplify - within a population let's say there are a thousand genes or SNPs that each affect IQ by a small amount. If you have 500 "plus" (good) and 500 "minus" you end up with an IQ of 100 (average). If you have 450 "plus" and 550 "minus" you have 90 IQ. If you have 550 "plus" you end up with 110 IQ. If you have 800 you are Newton or von Neumann. The numbers are obviously simplified and made-up, but that's the general idea. Of course this is assuming the nutrition and environment to allow that IQ to develop.
Since the individual effect sizes are tiny, you need a large number of people in a study to find them. But as big data gets better, we're going to find them, and the anti-science left will have to figure out a way to deal with it.
And intelligence has noting to do with "right" vs "left". That sort of political framing language is an excellent smell test for bad scientific information.
So when many genes are expressed simultaneously, you can get complex interactions between these distinct characteristics. But not a straight linear additive effect. If 5 genes all encode the same characteristic, then by definition they are actually just 5 copies of the same gene.
Intelligence is not related to genetics through some simple scalar factor. It is far, far more complex than that.
It mostly is when you're talking about "why person X has a higher IQ than person Y". It's not when you're asking why or how humans have intelligence at all, obviously.
The evidence conclusively shows
1. Intelligence is largely genetically determined
2. There is some genetic difference between races
3. Races differ in mean intelligence
As a result, Harris and Murray are correct in saying that it would be a miracle if none of the racial difference in intelligence is attributable to genetics, but that's not a very substantial claim. While they were clear about how differences within racial groups are much larger than differences between, it would have been better if they had taken care to also point out that we simply do not know how much of the difference between races is attributable to genetics, except to say that it's extremely unlikely that the answer is "none" or "all".
By identifying genes relevant to intelligence, we can start to answer this question. That will require someone to perform such an analysis, however, and seeing how Murray has been treated, few people besides him and some actual white supremacists are likely to be interested in doing such an analysis.
The more important question is why would it matter? Who cares if there are differences within or across racial groups with regards to intelligence? What reasonable social benefit is there in researching this? Intelligent people should seek ways to enrich the lives of others through their intellect, not find ways to build gates and limit possible outcomes.
The fact that intelligence is highly genetic in origin is extremely obvious. But the implications this fact has on how people interact with each other are extremely disturbing.
That is my primary issue with this line of research. Clearly it can't be banned or made illegal, but I think it's worth acknowledging that racism and sexism are significant problems and to spend a little time considering how this line of research will make these problems even more intractable.
that's not even close to true. on the most basic level, you agree that certain traits are heritable in totally predictable ways, yes? eye color is an example of this.
Eye color inheritance isn't "totally predictable"; this is so obviously wrong to anyone who's ever met more than a couple humans that perhaps I'm misunderstanding the assertion you're making here.
Come to think of it, I think you might be saying that _conditioned on the alleles you receive_, there are traits that are fully predictable. In that case, I think you're just misunderstanding the comment you're responding to, since conditioning on the alleles received is begging the question.
Alternatively, you may not know what a random variable is: https://en.wikipedia.org/wiki/Random_variable. It doesn't mean that the trait is "random" in the colloquial sense of uniform distribution across all possible values.
other combinations of eye color at the grandparents lead to predictable distributions of the expression of that trait in the grandchild generation. this may be what was meant by "random variable", but it's really a population effect so modeling it as a random variable is peculiar. it's not random at the level of the individual, or at least it isn't after the recombination phase of fertilization in the embryo.
given a genetic sample of an individual it is totally predictable what their eye color will be based on examining their genes. I don't see how this is begging the question.
Right, this is what I suspected you may have been saying. I think we were talking past each other a bit.
I still don't think it makes much sense to challenge someone who says "X is a random variable" by saying "that's not remotely true, X is not a random variable if you condition on the probabilistic component of X". Well, duh.
There's more to the output than the blueprint.
The genes that all humans share (most of them) contribute a lot more to intelligence. That's a different subject altogether.
"The rate of increase in fitness of any organism at any time is equal to its genetic variance in fitness at that time."
(There is the Flynn effect, but that's too quick to be genetic. Whatever its cause is apparently more important than genetics?)
There's apparently some speculation (according to the Wikipedia article on the Flynn effect at least) that the Flynn effect is due to non-genetic factors, in which case the underlying genetic trends are being obscured. There's a claim that genotypic IQ could be decreasing but we're not seeing that reflected effect in phenotypes due to environment improvements (which probably won't continue forever).
Bickering over the exact mechanism and arguing that our current ignorance of those exact mechanisms still casts doubt on the very existence of the observed differences isn't logical.
The vox article is very politically charged and admits to an unscientific politically correct approach to the whole subject. If your concerns are emotional and political, I guess it would be persuasive. If your concerns are scientific, there isn't really anything there that is an argument against anything said by Murray or Harris.
I think you should read it again more carefully. I can understand the inclination to dismiss Vox articles. This isn't a typical Vox article, and it clearly isn't an appeal to emotion. Its authors have almost certainly studied these issues far more carefully than any of us.
Modern DNA science has found hundreds of genetic variants that each have a very, very tiny association with intelligence, but even if you add them all together they predict only a small fraction of someone’s IQ score.
That is true, but, it's evading the overall question of "how much does genetics affect intelligence". It's focused only on the question of "how many specific genes have we identified".
As far as I can tell, it shouldn't be controversial to say that genetics has a huge effect on intelligence. For example, the NIH states on their website:
These studies suggest that genetic factors underlie about 50 percent of the difference in intelligence among individuals.
Sounds like our best scientific knowledge is pretty simple - intelligence seems like it's half genetic. The Vox article is quite misleading because it dances around this simple statistic.
You were correct in your other comment that I skimmed the article the first time. It set off my BS detector and I didn't bother to painstakingly scrutinizing it to see if it had some other value.
Taking your challenge, I re-read it and now I'm going to go one further. This was a hit-piece article that was designed to go after Murray and satisfy the typical Vox liberal red meat audience. The science that was in the article was incidental and the best they could do given their overall motive of striking at Murray while generating hits from their typical readership.
The title of the article is "Charles Murray is once again peddling junk science about race and IQ"
I think that someone came up with the title of the article first to capitalize on the Middlebury College controversy and then sought to justify that title with as much science as they could muster. Vox knows that they'll get tons of hits just from the title and that the information in the article won't be judged too harshly by their typical readers. My guess is that after coming up with the title, Vox editors got some scientists in the field who are critics of Murray to write up a justification.
Despite the article's title, the scientists admit: "Some well-informed scientists hold views closer to Murray’s than to ours. And there are others who challenge views that we accept about the utility of the general concepts of intelligence and heritability."
So which is it? Is it junk science or do some well-informed scientists agree with Murray more than they agree with the authors of this piece while some other well-informed scientists consider the authors of this piece to have flaws in their basic understanding of concepts that they touch on? My guess is that the scientists wanted this disclaimer in there but that the Vox editor would have preferred to not even make that admission. Those two short sentences in the article completely destroy the title and show that this article was politically (or financially, depending upon how you look at it) motivated.
And it addresses your "twin studies, separated twin studies, adoption studies" dismissal, rather directly.
Well, it agrees with them and even admits: "These “DNA-based” heritability studies don’t tell you much more than the classical twin studies did, but they put to bed many of the lingering suspicions that twin studies were fundamentally flawed in some way."
I wrote more but while trimming it up realized that this nonsense political hit piece has already wasted too much of my time. My BS detector was functioning perfectly the first time and I should have trusted it and avoided going back to that link and contributing to Vox's ad revenue.
As I mentioned, I had a bunch more that I threw away while editing. To summarize, I mostly pointed out how their "science" was straw men and other misleading statements. As I was editing my responses, it occurred to me why the logic was so poor. It was obviously constructed after the headline was created and so they were working really hard to put something out there.
Other responders in this thread spent their time pointing out some of the logic flaws I also saw. Take it up with them if you think this is a good use of your time. I've now wasted even more time on this, but I normally respect your posts - especially when you're discussing security. I thought I would at least give you an explanation as to why I won't be spending further time on the Vox article.
Update: I just skimmed through the HN discussion on this. You have plenty of other people who have responded to the Vox article quite thoroughly. HoustonEuler in fact looks to have specific knowledge on this subject.
Why aren't you hashing it out with him and others?
Consider that the issues involved (and the flaws they see in Murray's arguments) are fairly subtle, so this "bickering" as you call it is actually pretty important to understanding what's going on.
It called Murray's conclusions "dangerous". What place does an accusation like that have in the assessment of his scientific findings? You can't have a credible scientific discussion about a subject like the racial correlations of intelligence without a high degree of detachment and willingness to go wherever the data leads.
On top of that, the article was disingenuous. I listened to the podcast a couple of weeks ago and I recall a very interesting statement by Murray to the effect that "the IQ differences between members of each specific race are greater than the IQ differences between the different different races." And yet the Vox article implied that Murray was making sweeping and damning generalizations about the races.
Because the statement that there are any differences between population groups still leads to sobering/disconcerting conclusions for the top (Phds, C-level executives) and the bottom (poverty/crime) in a meritocratic society. (And if people are marrying only inside their class (or education level) this is self-sorting over time)
That you can't conclude that your neighbor in the subway is a genius or dumb as a brick just from her ethnicity (and vice versa) doesn't change these extreme edge cases.
But talking about the disturbing ways misleading scientific claims get used is on-topic. They do talk about the science since that's essential to making their argument.
The first 2/3 of the piece is entirely about the science itself. Obviously, even a glance at the piece reveals that it begins with a hierarchy of assertions about the science of genetic bases for intelligence, ordered by their acceptance in the field. Its conclusion includes this graf:
> Asserting that the relatively poorer intellectual performance of racial groups is based on their genes is mistaken theoretically and unfounded empirically; and given the consequences of promulgating the policies that follow from such assertions, it is egregiously wrong morally.
> Moreover, a reflexive defense of free academic inquiry has prompted some to think it a mark of scientific objectivity to look at cognitive differences in the eye without blinking. To deny the possibility of a biological basis of group differences, they suggest, is to allow “moral panic,” as Harris puts it, to block objective scientific judgment. But passively allowing oneself to be led into unfounded genetic conclusions about race and IQ is hardly a mark of rational tough-mindedness. The fact is, there is no evidence for any such genetic hypothesis — about complex human behavior of any kind. Anyone who speaks as if there were is spouting junk science.
This is hardly a stipulation that the science behind this debate is settled and the issues are political and social, and it's the graf that opens the conclusion of the piece.
Earlier they state:
"That is not to say that socially defined race is meaningless or useless. (Modern genomics can do a good job of determining where in Central Europe or Western Africa your ancestors resided.)"
"In reality, the racial groups used in the US — white, black, Hispanic, Asian — are such a poor proxy for underlying genetic ancestry"
What they don't do is link these statements. That a persons genetic ancestry to the three major groups: African, Caucasian or East Asian is strongly established as they admit. And that difference in IQ test scores linked to this ancestry is based in fact. They use the words "socially defined races" to avoid this but later make a much more broader conclusions of Murrey doing "junk science"
There are good counter arguments to each of their bullet points. Specifically this one:
"Murray’s assertion that it is hard to raise the IQs of disadvantaged children leaves out the most important data point. Adoption from a poor family into a better-off one is associated with IQ gains of 12 to 18 points."
They don't mention that the studies of high SES black vs low as a counterpoint to how this increase is gained.
None of the arguments they made falsify Murray's hypothesis. Murrey does not state that environment doesn't matter. He uses evidence from Twin studies and Adoption studies to show that genes matter more. We also know how to reduce intelligence via environment far better than we know how to increase it.
Where that point gets political is if people believe that the "environment" commonly surrounding certain socio-economic populations is the natural baseline for that population, or instead an environment that is depressing their likely natural performance.
Like, if you move someone from their status quo, then is that putting them in an artificial environment that ultimately will not help? Or is that removing them from an artificial environment that is artificially depressing their performance, thereby giving them a permanent relative boost when you move them?
Where it gets even more political is when people believe that certain disadvantaged environments are "natural" specifically because the intelligence of that population leads them there.
"No genes for IQ" is better interpreted as a statement that it's not just a small handful of genes that determine IQ (or any behavioral trait), but very, very many (plus nurture, but that goes without saying). It cites http://europepmc.org/articles/pmc4635473, which mentions:
The SNP rs9320913 is estimated to account for only 0.02% of the overall variability in educational attainment, but biometrical studies show that the total percentage of variability owed to genetic differences is three orders of magnitude larger (Heath et al., 1985; Rietveld et al., 2013). Since the SNPs with the largest effects are the easiest to find, these results suggest that educational attainment is a phenotype affected by thousands of undiscovered genetic variants, each responsible for a minuscule fraction of individual differences.
...which is not to say that Murray's claims that intelligence and race correlate strongly have merit, since as the article you posted points out, his argument has some severe deficiencies.
Specifically (and this is coming from me, not the article), when a measured trait (say IQ) is a function of a large number of genes, then unless that trait has highly variable selective importance depending on habitat, we would not expect it to diverge significantly between races. Traits that do diverge significantly tend to be either subject to different selection pressures between environments, or involve a small number of genes (with a small number of genes, it's easy for random noise to freeze unusual variants into segregated populations, but as the count goes up the law of large numbers kicks in and we see similar bell curves in all populations). This is almost trivial to prove.
As I understand it (and I'm not that knowledgeable either), people who are more closely related (from a genetic point of view) are likely to have more similar intelligence. However, that's pretty much all you can say. How closely related you are doesn't map to anything else cleanly.
It seems like that's pretty much what you'd expect from a lot of genes that contribute a tiny effect and are nearly randomly passed on to descendants.
This may seem almost trivial, as most people probably have this opinion anyway. But it still has inconvenient ramifications for egalitarism. For example: is someone in Harvard because their filthy rich family bought them a place and had connections ... or are they maybe clever because they come from a clever family?
> How closely related you are doesn't map to anything else cleanly.
What do you mean?
And similarly for other traits you might notice.
The hot potatoe is of course "race" (what US-Americans equate for some reason with skin color), or more specifically shared-ancestry-ethnicity (northern european vs southern europeans vs slavs vs ashkenazi jews vs sephardi jews vs south eastern asians vs east asians … etc)
It turns out that one genetic variant has caused two different mental illnesses and they still don't know why. The treatment for them is the same though. Of course they are not twins, so there may be other genes in play that may account for the difference in manifestation.
For me the bottom line is that evolution is still hard at work changing the human brain to work the way we need it to in modern times.
(Anecdote: I found it no problem to get degrees in both Computer Science AND English with minors in French and Philosophy. Some people find it strange that someone could do both types of things, but I just like learning everything!)
"If you can get a degree in math, you prove yourself to be intelligent and hard working. Companies like those kind of people."
IQ is alt-right propaganda. Definitely no IQ. As such, we need to teach everyone to code. If only the evil white men would let everyone spend 60 hours a week coding.
And the reason some people go to top universities isn't because they're smart. It's their class. Their class allows them to handle abstraction & complexity dude! We need to level the playing field & equally distribute ability to learn via capital redistribution. Being able to buy better brands is def going help shepherd us into a nu-world of equality.
For someone who lives in a remote area and survives on hunting wild animals, being able to tell where game is situated, is more valuable than matching some geometrical figures or knowing capital city of a country he will never visit. Such tests that take into account language and cultural differences don't exist.
Not knowing something you are not interested in, doesn't make one dumb. It is just not important to that person. I also noticed that the older I grow, the higher my IQ due to voraciously reading technical books and articles.
If you have studied engineering or simply maths, and like learning maths, you already have some background on geometry, algebra, etc. (and specially geometry/spatial stuff). At least the tests I've seen have some stuff that seems quite obviously been taking from maths... So, IMO, you can obviously score higher than someone that is, perhaps, as intelligent as you are, yet doesn't know or care about maths, or sorting things in certain way, or whatever.
While I don't really know much about the topic, IMO these tests ignore the fact that you can learn to solve certain problems, and yet "experts" still say they are about general intelligence. And by that logic, if you weren't to measure intelligence by a set of arbitrary characteristics, some of which can be learned in some specific social contexts... How would you even measure general intelligence? As you say, intelligence is relative to context. You can "learn" intelligence measured by these tests, or you can be "dumb" according to these tests, simply because you are ignorant of some things.
This part sums up what I was trying to say.
Being able to tell where the game is situated is not intelligence. It's a domain-specific skill. Intelligence is a general ability that makes one better at learning any domain-specific skill that utilizes cognitive capabilities.
On the other hand, the idea that there is a single number that correlates highly with most other definitions or aspects of "intelligence" has strong experimental backing, like it or not.
While that is certainly the goal and ambition of the people creating the tests, there still seems to be much debate as to what extent they are succeeding.
I can guarantee you this is not the case. This comment is honestly so thoroughly full of misconceptions about the nature of IQ and IQ testing that it reads like a troll.
IQ is meant to measure g, or 'general intelligence'. it's roughly the ability to solve complex tasks, and tends to span across domains (ie, if you are above average in one area of problem-solving, you are likely to be above average in others): https://en.wikipedia.org/wiki/G_factor_(psychometrics)
>While at it, show me an IQ test that was compiled using my first language since you are dismissing facts I put on the table.
just for instance: https://en.wikipedia.org/wiki/Raven%27s_Progressive_Matrices
it looks very defensive to simply ignore my responses that directly answered your challenges and say 'oh yeah well i can google things'. you may also try not to rely on tabloids for science reporting.
There's an interesting debate about why this is the case: were Neanderthals sufficiently distant from humans that their genetic variants tend to be incompatible, or was it because the smaller Neanderthal population made it harder to purge bad mutations so more of their variants were simply bad in general, or something else entirely?