I was looking at a recent survey of current knowledge in psychological genetics. The gist is that common variants – which can’t have decreased fitness much in the average past, since they’re common – are the main story in the genetic architecture of intelligence. Genetic load doesn’t seem very important, except at the low end. Big-effect deleterious mutations can certainly leave you retarded, but moderate differences in the number of slightly-deleterious mutations don’t have any observable effect – except possibly in the extremely intelligent, but that’s uncertain at this point. Not what I expected, but that’s how things look right now. It would seem that brain development is robust to small tweaks, although there must be some limit. The results with older fathers apparently fit this pattern: they have more kids with something seriously wrong, but although there should be extra mild mutations in their kids as well as the occasional serious one, the kids without obvious serious problems don’t have depressed IQ.
Selection still acts on really nasty mutations affecting the brain, even today: people that are crazy or retarded have low fitness. We’re better at preventing early deaths from infectious disease than we used to be, but they were never super-related to IQ in the first place. Selection can change the frequency of those common variants, so differential reproduction with IQ could have happened and had consequences, probably bad ones. But how bad? I doubt if within-population selection decreased IQ by as much as a point a generation. That makes it hard to see how IQ could have declined by a standard deviation (15 points) since the Victorian era – there is no available genetic mechanism. Since it is also obvious (to anyone who has eyes) that no such change occurred, the lack of a feasible mechanism is not really a problem.
West Hunter 
Did you mean to link to this excerpt from this paper? or this other paper??
Since it is also obvious (to anyone who has eyes) that no such change occurred, the lack of a feasible mechanism is not really a problem.
The arts have declined, but we seem technically inventive enough. Heck, we’ve got more than enough technical skill in the arts, but our elites tend to believe a lot of retarded things, which tends not to inspire great art.
… And do not forget the cave paintings and what Picasso said about these paintings comparing them to today’s art:
“We have invented nothing new”, after seeing the cave paintings at Lascaux France.
“After Altamira, all is decadence”, when he exited the infamous caves in his native Spain.
Sure, those are disputed quotes (see http://timelineforhomosapiens.com/2013/12/11/picasso-quotes-and-primitive-art/),
but I would say the same thing….
Does not add up. If the variance in intelligence was caused by common variants, we would have found them.
Therefore, most of the variance is caused by rare variants of small effect
– which is to say, genetic load..
I have never thought that most of the variance in human intelligence is caused by hundreds to thousands of small effects and I can tell you why in very simple statistical terms. If that was the case then there would be an averaging out of the population and the standard deviation would be very small. What we have is a huge variation in human intelligence. A contributing physicist can hold in his brain a page full of calculations while joe average can’t add 4 single digits in his.
This huge variation in human intelligence is therefore not caused by either genetic load of very small effects or individual genes that cause improvements of a tenth of one percent.
So..what causes the huge variation? Is the whole world stumped? Individual genes have a minuscule effect and so do all but a few very rare mutations which never spread very far.
Lo and behold a study came out just before Christmas that gives an alternate answer. I defer to the blog Psychological Comments of one of our most esteemed commentators James Thompson for an explanation of what this study discovered. You need to scroll back in his blog to January 3rd to his blog post titled Intelligence Genetics: Motorways http://drjamesthompson.blogspot.com/
Read the whole explanation by James Thompson, please. A dumbed down version is this. An optimum grouping of genes, which these scientists isolated through what is called a gene coexpression network analysis, causes the world to be populated by 1 Greg Cochran, 100 Dave Chamberlins, and 10,000 Joe Averages.
Ma and Pa Kettle mix up 150 genes on this grouping of genes and out comes a little scamp that can’t add up four single digit numbers in his head. He didn’t win the lottery.
“I have never thought that most of the variance in human intelligence is caused by hundreds to thousands of small effects and I can tell you why in very simple statistical terms. If that was the case then there would be an averaging out of the population and the standard deviation would be very small.”
That’s just not true. The variance in a trait due to the sum of many effects is sum_i {p_i(1-p_i)/2 a_i^2), where a_i is the effect of the substituted allele at locus i (assumes no linkage; two alleles per locus; but a generalization can yield same general result). Obviously this sum can be as large as we want, regardless of the number of loci. In general, a polygenic trait can have any standard deviation. The extent of the variation of a trait provides no information on the number of loci. Its distribution does provide information: the closer to a normal (Gaussian) distribution, the more likely its due to many loci (central limit theorem). IQ distribution is quite Gaussian – hence there being a book called the Bell Curve.
It’s distribution does provide information and that large distribution cannot be caused by thousands of equally tiny effects or the distribution would not be so large. So what causes the large distribution which does exist, that is the million dollar question.
I’m telling you it can.
For “equally tiny effects”: Pick a number V as large as you’d like. Then set a_i = a for all i such that
V = a^2 sum_i {p_i(1-p_i)/2)}
(p_i is the allele frequency at locus i)
For any number of loci, it’s possible to pick a value “a” that satisfies the equation, regardless of the size of V.
QED
I think we are missing each other. I understand what you are saying. Would there not be a larger standard deviation in the bell shaped curve of human intelligence if this distribution was not made up of thousands of tiny effects but rather an optimum combination of genes. That is all I am saying. There isn’t a small variation in the working memory of a brilliant Physicist and and average human being, there is a huge one. Nobody makes these kinds of metrics, it would insure your upcoming unemployment if you did ;). Ok Joe Average we have measured your working memory and you are 18 inches tall and Greg over here is 10 foot 6 inches tall. Nope, wouldn’t fly.
okay, the central limit theorem takes a list of additive random variables of finite variance and produces a normal distribution of finite variance, the variance of a 50/50 is smaller than an 88/12
but why would they be additive?
working memory is just a story anyway, it sounds good and at some level it has to be true, but no one really knows how thinking works. i think non-accessible memory and whatever process gives me insights the next day is more important
Doesn’t the binomial distribution speak to this?
Np(1-p).
The larger N, the larger the variance?
Maybe I misunderstood what the argument was about.
Genetic load is rare mutations of small effect. If small effect, then likely intronic, therefore not excluded by this study. All this study tells us is that most coding mutations that are not subject to strong adverse selection do not adversely affect intelligence, which is not in itself evidence against the proposition that most variance IQ comes from genetic load.
You pretty much said what I wanted to say. Intronic and “intergenic” variants are typically ignored by these studies, even though most of the IQ hits have been in these regions.
Are you referring to Short Tandem Repeats (STR’s)?
This is concordant with the notion that intelligence isn’t just a matter of overall brainpower. The g-factor of IQ batteries is sometimes interpreted that way, but mistakenly so: EVERY human behavior is embodied in the brain, and all sorts of things we do that aren’t very “intellectual” involve tons and tons of brainpower. IQ tests don’t measure social intelligence or physical agility, and for good reason, but those things are still just information processing as much as visual reasoning or verbal fluency are. Whether or not a person is a high-IQ type seems to have a lot to do with which behaviors are most deeply engrained and what types of stimuli are treated as most relevant.
Physical agility is not just information processing. What about strength? Flexibility? Speed of nerves firing?
Does anyone still care about the Victorian IQ thing except Bruce Charlton? I thought we’d all moved on.
You’d be surprised.
I don’t know if you have looked at the very large amount of data on SAT scores in the USA, but they have been declining for perhaps 50 years (don’t have good data before that). The data is available at the College Board website….Despite having “re-normed”–e.g. added up to 100 points to push scores up–the tests around 1995, and slowly dumbed down the math portion, there is a very significant and steady decline, which converted to IQ, amounts to 1+ points per decade……College Board is very concerned about it, but no one else seems interested…
If they have data from a consistent test on a representative sample, it might be meaningful. So show me.
More test-takers?
The PISA study is also telling something. Germany was very annoyed when it was first ranked number 20 in 2000, behind most European countries. And that for the country of the Heisenbergs. I can testify that all my children had to go through brutal drills so that the PISA test would be passed better, which it does now. Germany now scores near France at least. But the question is: How would Germany have scored 30-50-70 years ago. Was that all a fluke at that time? Many would say YES, so this is a difficult argument. But again, this is just another puzzle stone to look at for a time wise view on the matter.
“I thought we’d all moved on.” Shame on you for using that horrible expression.
Translation–unpleasant subject dropped down the memory hole…
Can you possibly serious? IQ, its formation and heritability, and its effects on individuals and societies are among the fundamental principals of modern social and psychological science.
“Selection still acts on really nasty mutations affecting the brain, even today: people that are crazy or retarded have low fitness.”
Most likely true. They are not killed off. But they are unlikely to reproduce since their odd to find mate is lower than average people. Fortunately, lack of sexual partner is not on welfare list. Maybe sexual toys/substitutes and birth controls mechanism are contributing to eugenic process.
I think Greg disagrees with this, but imagine two types of genetic variation that reduces measured IQ:
a) random noise/mutational load/deleterious mutations– the stuff that makes you a “funny looking kid.”
b) an allele that reduces measured IQ but has some potential fitness payoff– quicker reactions, heuristic-based rather than abstract reasoning, more cerebellum less cerebrum, remembering where hundreds of different watering holes are found but not being able to learn algebra, etcetera.
You assume that within a homogenous population, the driver of variation will be mainly a. But if you include a bunch of heterogeneous populations, I could see the main source being b.
A great many genetic load mutations surely exist across a large assortment of types of functionality. Some very likely have little or no effect on physical attractiveness.
Consider, for example, genetic load mutants that reduce lysosome efficiency. They could have little impact during development (with lots of cells dividing and diluting accumulating junk) but much impact by middle age.
Mutations that trade off between types of mental function seem plausible. But at the same time, all the subtests of IQ tests positively correlate. So what fraction of the alleles that influence IQ cause trade-offs between types of cognitive function?
I think they cause trade offs with things that are not IQ; these could be totally unrelated to cognition (i.e., a bigger head might be harder to keep cool or to get born) or it might be related to cognition but not to abstract reasoning. Think Kahneman’s System 1 rather than System 2. If something makes you apply heuristics and biases more quickly and systematically, it won’t make you do better on an IQ test, but it might require some processing power to accomplish and it might boost fitness under some/many circumstances. Just because there are idiots all around doesn’t mean that all forms of idiocy are equally rewarded.
How much of IQ is genes for direct intelligence and how much is genes for brain development and repair?
(stem cells in milk, iodine – that kind of thing?)
I want to know whether some genetic variants that boost nutrient delivery or waste removal to the brain deliver IQ boosts.
Galton like to measure things the way others like to memorize baseball statistics. So his measurements of his neighbors and exhibition visitors have been preserved. Some moderns think that his measurements are somehow proof that IQ was higher then. This is a far fetched notion.
“Since it is also obvious (to anyone who has eyes) that no such change occurred, the lack of a feasible mechanism is not really a problem.”
Map of patents by country
If you compared that map with one from the Victorian era you’d expect Western Europe including Britain to have shrunk relative to most of the world but what about Britain specifically compared to Holland, France and Germany?
I’d expect a similar map from the Victorian era to have Britain as a similar relative size to France and Germany – if not a relative size similar to modern Japan.
It might not be anything to do with IQ of course but it seems odd to me.
Patents per country and other intellectual output has to do with more than just IQ.
Sure but is there a remaining question of a drop specifically in Britain, if not globally.
I think a map like that from say 1880 would make for an interesting comparison.
Well, if we follow this map, anyone from Canada and South America is lucky to find a way to put pants on in the morning.
Sure but if most patents come from a smart fraction then don’t you only need a small drop in the average IQ to get a big drop at the smart-enough end?
Even if the map does highlight something real the cause could be a lot of things like for example the disproportionate growth in the financial sector sucking brains into non-patent heavy areas but it’s an interesting map and it would be cool to see one from 1880 to compare.
As it’s total patents, Canada’s relatively small population is a consideration.
What is “common”? MAF > 0.01? Should we distinguish between rare < 0.01 and very rare < 0.001?
What is the line of reasoning?
GCTA shows a substantial amount of variance predicted by common variants. But is this causal? The only GCTA that Arslan and Penke cite is quite nervous about interpreting the result in terms of common variants, rather than LD with rarer variants. I’m inclined to interpret it as common variants, but I think it’s ridiculous that A&P don’t mention that their source disagrees with them.
Penke and Jokela say that GCTA shows that 28-51% of the variance, but I can’t seem to follow the citations. The 51% presumably comes from this GCTA, which used a narrow sample of people, specifically so that it would increase the LD. So the difference between their result and other GCTA is probably due to rarer variants.
A&P go on to say that h²/H² ~ 2/3, which is a lot less than I thought. So maybe the GCTA has captured all the additive effects. Are rare effects necessarily additive? Maybe that’s a way of ruling out further rare effects, but
Is this a reference to Marioni et al? Do you really expect counting mutations to do any good? That sounds like you are reifying mutational load. The mutations with a minor effect on IQ will be lost in the noise of the mutations that don’t.
Anyone who looked at the scatterplot from Woodley, te Nijenhuis, and Murphy, would know that their interpretation of the data is idiotic. There are, today, large samples that are pretty much as fast as the Victorians. If they believe their methodology, they should conclude that Australians are 1σ smarter than other moderns (Brits, I think). But they don’t, either because they’re too stupid to understand what they’re doing, or because they are frauds.
There are two steps in their method. One is to compare reaction times measured in different ways and the second is to interpret reaction time as IQ. Since we know that Australians are not 1σ smarter than Brits, we know that at least one of the two steps is faulty. Almost certainly it is the first step. But if they really believed the first step, they could choose a method and measure the reaction times of both populations with the same method. If Australians really were 1σ faster than Brits, that would be very interesting. But of course WNM don’t talk about it, let alone test it, because they don’t really believe their study.
“The results with older fathers apparently fit this pattern: they have more kids with something seriously wrong, but although there should be extra mild mutations in their kids as well as the occasional serious one, the kids without obvious serious problems don’t have depressed IQ.”
I wonder about statistical power. There’s a tendency to say “no effect” whenever p<0.05, when of course that’s an incorrect conclusion (better to say “the present data cannot conclusively demonstrate an effect,” or something). That’s a problem if their sample was small. Particularly because the effect we’re looking for is likely to be small. If there are indeeds thousands of loci contributing to IQ, then how many of these will be mutated in an old father? If it’s only a ~dozen, then the effect will be very small (12/1000 is a small number), and so the effect will be very hard to see, statistically. (Sorry – i’m too lazy to follow the link.)
Kind of like the neutrality debates. It’s hard to detect a selection coefficient of order 1/1000 because it is “small.” Unless your sample is very large, you’re probably not going to detect that difference, and you’ll conclude that the trait is neutral. But in a large enough population, s=1/1000 is not neutral.
“but moderate differences in the number of slightly-deleterious mutations don’t have any observable effect”
Moderate = how many?
Pingback: The Word From The Dark Side, January 18th, 2015 | SovietMen
The lack of a feasible mechanism … But, Sir, IQ is hereditary to certain degree, so differential reproduction must leave its stamp on a population’s average IQ. And in the last 150 years, the better classes are having less children. That is a feasible mechanism, or at least so it seems to me.
Sure. But not 15 points worth.
Yes. To reach 15 points worth in no time one needs mass immigration.
More than once ’15 points worth’ decline ‘since Victorian times’ was brought up on this blog. Why instead of questioning this, you are still looking for convoluted ways to support this?
We don’t support it.
The “Victorians had high IQ” idea comes from tests done by Galton in 1889…obviously no IQ tests existed back then, so he measured the reaction times of subjects using Gipp Chronographs.
Reaction time is somewhat correlated with IQ, but you can’t lean too heavily on it as a measure of intelligence. Women have slower reaction times than men, but similar IQs. You can do well at one but not the other.
Even worse, there’s no reason to think Galton took steps to make his data pool a balanced cross-section of the Victorian population. He basically set up a laboratory in a museum, charged people an admittance fee, and recorded their reaction times. His subjects would have been disproportionately from the upper classes.
As Greg says, 15 lost IQ points would have had consequences, none of which we’ve seen. The most intellectually productive period in human development came after Galton’s time, in the 20th century (remembering that in the 19th century there was still low hanging fruit to pick.)
Galton was generally aimed in the right direction, but if he was exclusively testing the top 20% of the population then his measurements are probably worthless. Statistics are annoying like that. Doing mostly good work isn’t enough – tiny mistakes can knock all your numbers into a cocked hat.
It’s like all those clickbait articles that say “Check out this college entrance exam from 1915! Boy, will you feel stupid!” ….completely forgetting that only the top 5% of the population went to college in 1915.
“remembering that in the 19th century there was still low hanging fruit to pick”: I’ve never seen that argument used other than in an entirely circular way.
Comparing UK 2016 to 1886, a drop of 8-10 IQ certainly seems plausible, if not 15. We seem about as far from Victorian England as the modern Balkans are from us, or slightly more.
Really? No.