Brains feeling stretched…

Cosma has to go and show off with a magisterial demonstration of why he is the smartest man on the internet: he's written an exceptionally thorough description of heritability and IQ. It's not a light read (statistics and genetics!), but it's probably the most informative thing I've read in a month or more.

I'm sure I'm going to have to read it a few more times before I've absorbed it all.

More like this

A marginally less cranky physics post than the previous: the big story in my area of physics this week is probably the Harvard experiment involving the storage and transport of light pulses. Like the ILC announcement, this has been written up in the Times, and you can also read the Harvard press…
Over a year and half ago (~1 eon in internet time) I wrote this blog entry in which I turned around the title of Dobzhansky's famous essay "Nothing in Biology Makes Sense Except in the Light of Evolution". I didn't think I was being all that clever when I came up with the following: NOTHING IN…
PZ, Bora, Orac, John, and others have all put up posts about a list of the 50 most significant Science Fiction and Fantasy works of the last fifty years. As the reigning Geek-Lord of ScienceBlogs, I figured that I had to weigh in as well. Here's the list: the one's that I've read are bold-faced…
It's Saturday and therefore time for some lazy non-science blogging, especially since after I finish this post I'm going to bury myself in grant writing. Multiple grant deadlines are approaching, and, given that most of my grant support expires towards the middle of next year, I have to go full…

I've got to admit that, at this point, I'm just going for the 500,000th comment. But I'm going to click that link and read.

This is like an IOU comment.

Shame on you! Posting just to reach the... oh, oh, not quite there yet.

comment number:499,297?

Thanks PZ -- it was a good paper.

It basically says that heritability of IQ is systematically overestimated by confounding variables and simplistic assumptions and study designs. Wow -- like anybody with half a brain wouldn't have known that. It's nice to see it laid out so neatly though. (Oh, and the nice plug about the joys of Bayesian statistics is appreciated too!)

Ah -- just because we can measure something doesn't mean we understand it.

By spudbeach (not verified) on 30 Sep 2007 #permalink

My thoughts are the same as PZ's for now.

From the essay:
"The best estimate I can find puts the narrow heritability of IQ at around 0.34 and the broad heritability at 0.48."

Does this mean that IQ is 34-48% heritable?

By Brandon P. (not verified) on 30 Sep 2007 #permalink

Some of Cosma's criticisms have been around for years-- even before The Bell Curve, which came out in the 1980's. (And let me add, I've met Murray. My impression was, he's, shall we say, racial-tolerance-challenged.)

It's been obvious, for instance, that maternal effects must have an effect on IQ, and that adoptive parents are deliberately chosen to come from a particular social background.

These criticisms never seem to have the slightest effect on people who want to argue that intelligence is hereditary. I've even heard people defend Burt, who was shown, by the biographer hired by his own family, to be guilty of fraud. (The conclusion, that his later studies were definitely faked and the earlier ones may have been as well, has somehow "evolved' into, 'the later studies MAY HAVE been faked, but the earlier ones were all right.')

So congratulations and good luck to Cosma for having laid out a really telling rebuttal to the Bell Curve. But don't expect it to settle the issue. Maybe somebody should make a movie: "The Bell Curve: The Thing that Wouldn't Die."

By hoary puccoon (not verified) on 30 Sep 2007 #permalink

comment number:499,297?

It has been saying "499,296 comments!" for about 12 hours now. And before that, it was at "499,244 comments!" for another half-day. The permalink to your comment is numbered 585,282...

By David Marjanović (not verified) on 01 Oct 2007 #permalink

"These criticisms never seem to have the slightest effect on people who want to argue that intelligence is hereditary."

The problem is that at least a substantial percentage of those people don't care about the heredity of variation in intelligence by itself. What they have on their minds is either eugenics or establishing and enforcing the supremacy of certain groups. Insisting on mental variation's heredity is just one step towards that goal. They are more interested in a political agenda (another source of dogma) rather than science.

Even if differences in intelligence between individuals were hereditary to a large degree, it does not follow that the heredity of intelligence (or g, IQ, whatever) differences between GROUPS is similar, since variation between individuals of a group is usually stronger than intergroup variation. In fact, why SHOULD there exist genetic differences in intelligence between people of African and European descent (the groups most commonly compared)? Surviving the predator-infested savannahs, disease-ridden jungles, and blazing deserts of Africa was probably ever bit as intellectually demanding to the ancestors of modern Africans as anything prehistoric Europeans experienced on the tundra.

PS. If African people really were so unintelligent, maybe all those UFO wackos speculating about an extraterrestial origin for the pyramids were right.

By Brandon P. (not verified) on 01 Oct 2007 #permalink

The permalink to your comment is numbered 585,282...

Which doesn't tell us anything if we don't know how much spam there has been. Thanks again to Coturnix.

By David Marjanović, OM (not verified) on 01 Oct 2007 #permalink

Perhaps next you could post a link to a technical explanation of how IQ is utterly malleable and an analysis of the success of various programs to increase IQ through education.

Oh, wait...

By Caledonian (not verified) on 01 Oct 2007 #permalink

How painful to read for a statistician. Heritability is defined, analyzed, modelized, etc. What do I have to say against this? Nothing, of course. It's work beautifully done.

But just because you can study heritability in lots of ways, doens't mean that there's really "something" to be inherited. We statisticians are very good at working with such concepts and building models around them. But it says nothing about the actual, concrete processes at work within our brains and genes.

IQ is really a terrible tool. But Gould explained all this much better than I do. And reification was one of his pet peeves.

By Christophe Thill (not verified) on 01 Oct 2007 #permalink

I'm sorry, let me rephrase that: any lasting effects on IQ that are not only statistically but practically significant.

The best available evidence shows that, as long as the environment isn't profoundly crippling, children eventually leave it behind. Even damaging environments can be overcome by children if they leave it before critical periods have passed.

Now, you might argue that this is because there's no special tendency for people raised in intelligence-increasing environments to remain in them once they're on their own, so the true effects of environment aren't perceived... but then you need to explain why decades of attempts to identify and utilize beneficial environments has failed to do so.

In short, you're so attached to a hypothesis that appeals to your political intentions that you're rejecting massive amounts of contrary evidence.

By Caledonian (not verified) on 01 Oct 2007 #permalink

I'm sorry, let me rephrase that: any lasting effects on IQ that are not only statistically but practically significant.

First, demonstrate that IQ measurments themselves are practically significant.

Caledonian wrote:

I'm sorry, let me rephrase that: any lasting effects on IQ that are not only statistically but practically significant.

If you read the article, the author does discuss some such studies and some of their limitations. In particular there is a paragraph devoted to the effect of staying in school on IQ measurements. Your comments indicate that you didn't bother to read the article before rushing to post criticism.

By Jeff Alexander (not verified) on 01 Oct 2007 #permalink

Caledonian says:
"In short, you're so attached to a hypothesis that appeals to your political intentions that you're rejecting massive amounts of contrary evidence."

The point of the article, Caledonian, is that there isn't any really solid contrary evidence. That doesn't mean genetic factors don't affect IQ. But it does mean that the particular correlations of "genes" (measured various ways, but not actually by genes) vs. "environment" (another slithery variable) with IQ are all plagued by methodological problems.

And Cosma doesn't even open the whole can of worms about whether you can apply a study of one population (English) to a completely different population (African Americans.)

Furthermore, the proponents of the genetic model of racial differences are using an ID-style argument from ignorance. "I can't, off the top of my head, think of any factors in the environment that I've left out, so all the rest of the variation must be genetic." This is ALWAYS a sloppy research design, no matter who uses it for what political reason.

Finally, my dear Caledonian, if my opposition to non-gene-based "genetic" studies is political, it's the political line that science should be left alone to do its job-- not burdened by pressure from either side to come up with "politically correct" results.

By hoary puccoon (not verified) on 01 Oct 2007 #permalink

Bookmarked, and I'll give it a read next time I have an hour with nothing to do. Looks very interesting, even if most of it is above my head.

First, demonstrate that IQ measurments themselves are practically significant.

They're used to identify learning disabilities.

(Thank you, thank you, I'm here all week. Try the vegan veal.)

By Caledonian (not verified) on 01 Oct 2007 #permalink

Actually, the really relevant work on this subject was done by Benchley, in 1936. His conclusion?

If we hadn't anything to hang our own superiority on, we should be sunk. We should be just like the Egyptians, or the Eskimos, or Grandpa.

All else is mere commentary.

By Ktesibios (not verified) on 01 Oct 2007 #permalink

They're used to identify learning disabilities.

Except that in that very context, one of the telltale signs of a learning disability is the discrepancy between the Verbal and Performance Wechsler scores, rendering the meaning of the full scale score quite iffy. Which doesn't exactly lend support to your impulse to reify IQ.

By Steve LaBonne (not verified) on 01 Oct 2007 #permalink

Shalizi wrote: I realize I'm inviting the suspicion that I'm protesting too much.

My complaint is that most of the criticism is concentrated on how we calculate and interpret heritability. Shouldn't the concept be scrapped in biology if it does not produce useful results in the animal that is perhaps best studied of all organisms next to laboratory models?

"I can't, off the top of my head, think of any factors in the environment that I've left out, so all the rest of the variation must be genetic." This is ALWAYS a sloppy research design, no matter who uses it for what political reason.

Like the non-gene-based genetics of Mendel?

(I'm being facetious, but let's not toss a century of pre-sequencing genetics out the window along with the bathwater, please?)

Sahotra Sarkar has written about this question too in his Genetics and Reductionism. Have any of the genetics-trained people read that? Any thoughts? It seems quite reasonable to me, but I'm only relying on my cursory self-study of a basic textbook on the subject.

Except that in that very context, one of the telltale signs of a learning disability is the discrepancy between the Verbal and Performance Wechsler scores, rendering the meaning of the full scale score quite iffy.

One of the ways learning disorders are identified is when there's poor scholastic performance in a subject, poor subtest results in that subject in IQ testing, but normal or high results on the others. If there's high subtest IQ scoring, then there's not a lack of specific ability or difficulty learning the subject involved.

Which doesn't exactly lend support to your impulse to reify IQ.

Since 'reification' is defined as a bad thing, I'd say you're well-poisoning. It also demonstrates that you don't understand the argument being made. IQ isn't a great measure of intelligence, but it's the only reliable measure of some aspects of cognitive efficiency we have. For a very narrow subset of cognition, it's a fantastic measure.

By Caledonian (not verified) on 01 Oct 2007 #permalink

One of the ways learning disorders are identified is when there's poor scholastic performance in a subject, poor subtest results in that subject in IQ testing, but normal or high results on the others.

That's NOT what I was talking about. Do you actually know anything about the Wechsler scale and the difference between the verbal and performance sections (and the discrepancy between scores on the two which is characteristic of some kinds of LD students), to which I was referring? Apparently not. But of course that never stops you from putting your $0.02 in.

The phenomenon to which I'm referring is precisely one that casts a significant amount of doubt on the idea that talk of some kind of "general" intelligence actually makes any sense- and it's particularly marked in the case of the Wechsler scale and LD students, because their performance renders the significance of the full-scale score- "the IQ"- highly suspect. For that matter, the different kind of discrepancy that you brought up is also a problem for the hypothesis that IQ is some kind of global measure of "intelligence". (Personally I think there is little if any reason to suppose that the idea of "general intelligence" makes any real sense, and even you seem to be edging toward a partial acknowledgement that it doesn't.) You need to come to grips with the fact that there is an entire, destructive mystique and ideology built up around IQ which is unsupported by scientific evidence.

By the way, IQ tests are NOT the only widely used measure of whatever limited aspects of "cognitive effiency" IQ tests may be said to measure. Several others- for example, the CogAT- are popular. So IQ tests could go away tomorrow without leaving any practical lacuna, and (at least under that ideologically charged name) they probably should.

By Steve LaBonne (not verified) on 02 Oct 2007 #permalink

(Personally I think there is little if any reason to suppose that the idea of "general intelligence" makes any real sense, and even you seem to be edging toward a partial acknowledgement that it doesn't.)

Edging towards? I take it that you're completely unaware of my past discussion of IQ tests on these forums.

How convenient! Arrogance and ignorance in one package!

Aren't you the LaBonne who's always suggesting that a) I'm a libertarian and b) libertarians are vile, evil filth whose destructive beliefs are disproven by True Science?

By Caledonian (not verified) on 02 Oct 2007 #permalink

If you were sincere, you'd stop the kneejerk reactions along the lines of #11.

As usual, your vituperative skills, though primitive, exceeed your reasoning skills.

By Steve LaBonne (not verified) on 02 Oct 2007 #permalink

If you were sincere, you'd stop the kneejerk reactions along the lines of #11.

Except that there are quite a few people who actually hold such ludicrous but ideologically-appealing positions on this site. Poking such nonsense is the point.

No one here is particularly good at retaining their rationality in the face of emotionally-charged positions, but you're particularly bad at it. May I suggest you sit this one out?

By Caledonian (not verified) on 02 Oct 2007 #permalink

We'd be happy to listen to your explanation of WHY you appear to have such an emotional attachment to defending the value of IQ tests. Go for it.

By Steve LaBonne (not verified) on 02 Oct 2007 #permalink

Ooh, beautiful well-poisoning there.

Are you sure you don't want to call me a libertarian a few times? C'mon, I have a bingo card that needs to be completed.

By Caledonian (not verified) on 02 Oct 2007 #permalink

No answer that you're willing to own up to, eh? I'm not surprised.

By Steve LaBonne (not verified) on 02 Oct 2007 #permalink
We'd be happy to listen to your explanation of WHY you appear to have such an emotional attachment to defending the value of IQ tests. Go for it.

Ooh, beautiful well-poisoning there.

I have to agree. Isn't "emotional attachment" what that Robert Wood guy always accuses us "Darwinists" of?

You don't think defending good science and defending a concept as shaky as IQ are two rather different things?
I do.

By Steve LaBonne (not verified) on 02 Oct 2007 #permalink

You don't think defending good science and defending a concept as shaky as IQ are two rather different things?

Don't you mean "Sound Science"?

Next you'll be telling me that there are more polar bears alive now than there were when Columbus arrived.

By Caledonian (not verified) on 02 Oct 2007 #permalink

Cognitive psychology has a great deal to say on what IQ is, what it isn't, and what it is (and isn't) useful for.

Mr. LaBonne seems to be arguing against a pop-psychology understand of IQ and its implications, which wouldn't be so bad, except that he does not seem to understand that the problem isn't IQ, it's pop-psychology.

Actually, it's pretty much pop-anything that's the problem, but that's another gripe and will be griped another day.

By Caledonian (not verified) on 02 Oct 2007 #permalink

For the last year since I discovered HER, I have wanted to marry Cosma, but I knew that I was unworthy. Who could be?
SHE was part wop, too.
So, I have worshiped HER adoringly from afar.

As a result of this post, I learned that She is a he, a male (a man). I made a mistake. Stupido!

Now, I know what I suspected. No lady, no female, no way could be this smaht, from Boston, and part wop too. Too perfect. Too smaht. Impossible.

Now, I have to re-frame.
More unrequited love

By gerald spezio (not verified) on 02 Oct 2007 #permalink

Drat. That should be *understanding. -ing!

curse these feeble fingers and the pathetic brain that runs them

By Caledonian (not verified) on 02 Oct 2007 #permalink

Now who's well-poisoning, eh?

IQ, whatever it is, and whatever its real value if any may be 1) very clearly does not signify what its name purports, 2) is surrounded by a cloud of very bad scholarship (eg. reification of "g" as some kind of genuine neurobiological property, confusion of heritability with susceptibility to amelioration), and 3) is a very popular ideological club wielded in some pretty dubious circles (it's actually the right-wing Charles Murray types who practice the same kind of "sound science" as the climate deniers, and some of them are even the same people).

Given the sordid history of bogus claims and political misuse of "IQ", in my well-considered opinion the onus is on anyone defending the continued use of the whole concept and name of "IQ" to demonstrate why doing so does more good than harm to both clear thinking and and good public policy. If you want to call that position well-poisoning, feel free; I've long been aware that you and Humpty Dumpty have a lot in common.

By Steve LaBonne (not verified) on 02 Oct 2007 #permalink

You don't think defending good science and defending a concept as shaky as IQ are two rather different things?

I think the "emotional attachment" argument is an annoying way of defending either of those.

Now who's well-poisoning, eh?

Oh, I'm not saying your arguments are wrong because you're an idiot. I'm just pointing out that you're an idiot.

By Caledonian (not verified) on 02 Oct 2007 #permalink

Sticks and stones. I have explained my position in #40, and like it or not I will continue to be very suspicious of the motives of those- including you- who seem unduly attached to the notion of "IQ" despite its manifold intellectual drawbacks and unsavory political associations. Bouts of namecalling merely serve to strengthen such suspicions.

By Steve LaBonne (not verified) on 02 Oct 2007 #permalink

Yes, yes, you maintain that healthy sense of paranoia you have. It'll come in useful when the libertarians and vaccinationists come for you in the night - you'll be able to escape their grasp and catch a boat to the glorious socialist paradise of Cuba.

By Caledonian (not verified) on 02 Oct 2007 #permalink

Wow! That was a long and informative post from Dr. Shalizi! I am graduate student of behavioral genetics, and it was very nice to read a post that goes into such detail about the methods we often use in our research. I've yet only skimmed the essay, and I don't think I have the knowledge or ability to really dispute the parts of the essay that I disagree with, except to say that a lot of his criticisms seem to be based on the weaknesses of comparing MZ twins raised together to MZ twins raised apart, whereas most twin research which derives heritability estimates from the study of twins, because of the incredible scarcity of MZ twins raised apart, uses instead a comparison of MZ to DZ twins (with both types being raised together). Of course, this method also has serious drawbacks and potentially invalid assumptions, but a lot of his criticisms are directed towards a target which for all practical purposes does not exist.

windy--
To get back to your response of a couple of days ago-- If Mendel had planted his peas willy-nilly, with more of the wrinkled peas going in the rocky part of the garden, and then described ALL of the differences he found as genetic, on the grounds of "I can't, off the top of my head, think of any environmental effects I've left out," yeah, that baby should be thrown out along with the bath water.

In fact, as far as I know (based on hearing presentations of ag. experiments at Big 10 universitries) Mendelian-type experiments are carefully set up to keep soil type, hours of sunlight, etc., equivalent. If you read The Bell Curve, on the other hand, you'll notice that there isn't any concerted effort to account for environmental differences like the womb environment, which we know are significant. That is sloppy research, no matter which side of the political spectrum it falls on.

By hoary puccoon (not verified) on 02 Oct 2007 #permalink

comment number:499,297?

It has been saying "499,296 comments!" for about 12 hours now. And before that, it was at "499,244 comments!" for another half-day. The permalink to your comment is numbered 585,282...

By David Marjanović (not verified) on 01 Oct 2007 #permalink

The permalink to your comment is numbered 585,282...

Which doesn't tell us anything if we don't know how much spam there has been. Thanks again to Coturnix.

By David Marjanović, OM (not verified) on 01 Oct 2007 #permalink