How does this correlate with identical twin studies which show that over time their IQ converges, even in the case of separate upbringings, adoption, etc?
Would you mind giving a source as well? I've been doing some googling and every paper that references this seems to describe correlations to IQ scores rather than variance of IQ scores. For example Bouchard & McGue (1981) reported a correlation of twin's IQ of 0.86 if living together and 0.71 if living apart. That particular paper doesn't fill me with confidence (it seems to have a sample size of 111 and it's a meta-study), but it appears to be the most cited paper I've seen.
I haven't seen anything that discusses a variance of IQ score, nor do I really understand what that means (what's the standard deviation of correlated multivariate distributions?) It would be interesting to see the analysis.
Sorry, why are you asking me? It's the nytimes that didn't explain their mental leap from "insignificant effect" to "probably many more genes involved".
They're looking through a huge number of genetic variants for genes that each have a small effect on expected intelligence scores. Two obvious problems that they run into:
1. When you test that many genetic variants, some correlations will appear purely by chance. To correct for this, they require very strong evidence for each variant that they declare significant. This means they end up having a lot of false negatives, and it's only through having an enormous sample size that they manage to get results this good.
2. There are a lot of genetic variants that they're not even looking at. (They try to pick a decent subset to measure.)
So they end up with some set of extensively-vetted genetic variants that, combined, predict a fairly small percent of variation. This is what you'd expect to see if there were a lot more that they weren't able to detect, which would be consistent with the much larger heritability numbers from things like twin studies.
That was my thought as well. Judging from the article, the model they have has statistical significance, but small predictive power.
"But in the past couple of years, larger studies relying on new statistical methods finally have produced compelling evidence that particular genes really are involved in shaping human intelligence."
and
"But all of these genes together account for just a small percentage of the variation in intelligence test scores, the researchers found; each variant raises or lowers I.Q. by only a small fraction of a point."
That doesn't sound like a compelling result, or an enormous success to me. I've taken a few IQ tests in my life and my own variation between tests is more than a dozen points. If they were showing genetics accounting for a meaningful amount of change in the IQ score, then yes, but in low predictive power models, I'd be more likely to look to a confounding variable with those genes that wasn't Intelligence. (e.g. certain genes express more when people eat a nutritious diet growing up - the study mentions that these genes in question are tied to regulating brain cell development).
I remembr a discussion here recently regarding "p-hacking". With so many thousands of genes, aren't a fair number going to be correlated or predictive with statistical significance by sheer probability?
The keywords (and ultimately, links to the published studies themselves) are "easily googled".
But understanding the true import of the studies -- and in particular, the various sampling and other statistical caveats in the foreground of them -- far less so.
Googling this is going to lead you into a morass. You probably want to be more careful than that. A better starting point would be a paper by a reputable scholar (this is a field where there are disreputable scholars, so be careful, but you can find people doing good work on pretty much all parts of the spectrum of hypotheses here --- stick to universities you've heard of) and then walking their references.
> In fact, all this suggest is that the genetic influence is likely non-zero.
It's been widely thought to be non-zero for a while. Now that they are identifying individual genes responsible we can start to better quantify exactly how non-zero it is.
It strikes me as bizarre that we are so reluctant to apply the same principles of genetics and heredity to humans that we have applied to canines, bovines, equines, and even vegetables, for centuries.
I agree and think that is the main point here.
Within our social norms, any type of generalized genetic comparison between groups of humans is very much frowned upon. I think that's because it bring back past traumas such as Slavery, the Jewish Holocaust and female oppression.
This is so strong that almost any grouping statement is quickly labeled as racism, sexism, profiling etc.
Imagine the popular reaction if it's discovered that these "intelligence genes" not only exist but are more prevalent in one of our go-to easily labeled human subgroups such as women, men, whites, blacks, rich, poor, brown-eyed. It'll be proof that that group is objectively genetically more intelligent. What would be the ramifications such a discovery? What should be the ramifications of such a discovery?
I think it's quite tricky to research these things objectively without it being explosive. That's why even suggesting that "the genetic influence on intelligence is likely non-zero" is a big thing.
It's not bizarre, we did exactly that (to varying degrees of scientific quality) until the events of the mid 20th century.
As humans we still suck at knowing what to do with the data, we look at headline figures and not distribution graphs. If we found out tomorrow that Icelanders on average had an IQ 10 points lower than most humans we would likely see a lot of discrimination against Icelanders, even the ones on the upper end of the distribution graph that are more intelligent than the average human.
I think that most of us find it unpleasant that humans can be "breeded" to become better.
Though we fight all the time about our differences, the sheer diversity around you is a wonder to behold. That it happened naturally and you too, though imperfect for sure, are unique and can become better as you grow.
We talk about advancing our human race but the reality is, most of us are only capable of seeing things from our personal viewpoint, and just a bit of the future. We care most about ourselves and our experiences, as well as those closest to us like family and friends - we don't actively do things in order to get better as a species. Only 'large entities' like governments and corps afford to see this bigger picture - and I think this is extremely cynical and one-dimensional.
The problem is that if you breed people like animals, you will end up treating people like animals, especially those that don't have the traits people want. With embryo selection people can decide which individuals deserve life and which don't. Hitler was a monster, but since embryos don't suffer it's OK to kill them (never mind the fact that you are deciding whether another person deserves to live or not), and everyone likes abortion nowadays. CRISPR won't change the fact that a person born with low IQ is doomed, which means that anyone living today will be obsolete as soon as people start to breed people.
I am not sure I remember the term but it was something along the lines of keeping the blood fresh. Supposedly the British did this between the gentry who had titles and self made men. A marriage of title and ability so that they two wouldn't fight. Kings did this between themselves as well.
My eyes grabbed on that fragment, also, and I pretty much stopped reading there. And yet, even though their influence is "minuscule", researchers want to use them to "determine which interventions would be most effective for children struggling to learn". What? Sounds potentially scary, actually.
It may be a compelling result, or maybe not. But from the way the NYT piece is articulated, you just can't tell.
I don't follow. How does that suggest anything? Perhaps intelligence is largely environmental.
In fact, all this suggests is that the genetic influence is likely non-zero.