John Derbyshire Misunderstands Race and Genetics

Michael White

There's been some comment recently about pundit John Derbyshire's belief that Obama will try to shut down biology because it has validated racism.

Needless to say, Derbyshire is full of it, and he has a poor grasp of what recent genetics has actually demonstrated regarding nature, nurture, and race.

This deserves a much longer discussion, which, fortuitously, is on its way, since I've been meaning to discuss a recent talk on genetics and race at our department given by geneticist Lynn Jorde.

The short version is this: Derbyshire claims that we can

"Name any universal characteristic of human nature, including cognitive and personality characteristics. Of all the observed variation in that characteristic, about half is caused by genetic differences. You may say that is only a half victory; but it is a complete shattering of the nurturist absolutism that ruled in the human sciences 40 years ago..."

What Derbyshire misses is that those genetic differences generally do not fall along racial or population lines. There is no current evidence to support the notion that genetic differences responsible for variation in complex traits like personality or intelligence are linked to race.

This deserves more discussion, but here is what the American Society of Human Genetics said about James Watson's comments on race and intelligence:

On October 14, 2007, The Sunday Times (London) quoted speculation by geneticist James Watson regarding alleged intellectual inferiority among Africans. ASHG find the comments to be tragically misguided and without scientific foundation. Watson later apologized "unreservedly" for his comments, stating that there is no scientific basis for such beliefs.

Derbyshire would have you think this is simply politics, but it's not - the science is clear.

(Hat tip to PZ Myers and Irradiatus; Irradiatus links to the original Derbyshire piece, if you're interested in reading it.)

Comments

Thanks for the link. I have updated the post linking back here as well.

Irradiatus (not verified) | 10/08/08 | 17:28 PM

I love PZ.

every biologist I know considers the Republican party to have been a disaster for American science

An oddly illogical statement for a guy criticizing someone else's logic. I rather doubt PZ spends his evenings with Republicans or pastors. The most famous example of that logic I know is Pauline Kael (The New Yorker film critic) apocryphally saying about Nixon's landslide election win, "How can that be? No one I know voted for Nixon!"

Like film critics, we can't take the opinions of biology teachers on politics too seriously.

Obviously Derbyshire is trying to mobilize the base for Republicans - much like PZ does for Democrats. While I am not one to dress up the pig real pretty like he is I don't think Obama will be a disaster for science regardless of his views.

His wife's and pastor's nonsensical tirades on race aside, Obama seems like a sharp enough guy and being president does make you a little more serious. Whatever personal views he may or may not hold in defiance of science, I don't think he will cripple it. But I don't think Bush crippled hESC research.

Derbyshire himself says it won't matter - other countries will still have science we don't - so neither McCain or Obama concerns me much.

However ...

Basically, his argument is: “Barack Obama is black. So Obama is anti-racist. Therefore, any science dealing with the nature of human variation will be outlawed by an Obama administration.”

... bits of rubbish like this from Irradiatus give science blogging a bad name. I get that he wants people to vote for Obama but inventing quotes and then going to war over them is for political bloggers. I felt my IQ drop reading it.

Hank Campbell | 10/08/08 | 17:50 PM

I didn't intend for this to be a political post, but to be fair to Irradiatus, Derbyshire has used this topic before as a political club to take a whack at liberals, suggesting that their devotion to multiculturalism has made them blind to the obvious results of biology. It's not just science bloggers who have made this political.

Mike

Michael White | 10/09/08 | 11:36 AM

You're first comment on the logic of PZ's statement is false. He did not say "every biologist". He said "every biologist I know", which anyone can reasonably take to mean "every biologist I have discussed the matter with."

The fact of the matter is that PZ knows ALOT of biologists, as do I, and I can easily make the same statement that PZ did.

As for my own quoted remark, it seems clear to me that my statement was a succinct summary of Derbyshire's implied thesis. Of course, Derbyshire never once actually explicitly stated what his thesis was, verbally skirting around his implied meanings. It's quite easy to see that “Barack Obama is black. So Obama is anti-racist. Therefore, any science dealing with the nature of human variation will be outlawed by an Obama administration.” is his implied meaning.

No he didn't actually use the words "black" or "race," but those are his clear implications.

My post had nothing to do with my own desires for Obama. It had to do with the idiocy of whispering of game-changing discoveries on race and genetics that will unsettle us all. It's just silly.

I stand by my post, even if it is admittedly less eloquent than PZ's and Michael's above.

Irradiatus (not verified) | 10/08/08 | 18:06 PM

"There is no current evidence to support the notion that genetic differences responsible for variation in complex traits like personality or intelligence are linked to race."

I don't think it's possible to honestly assert that there's no evidence with knowledge of the literature. Genome-wide association studies have turned up minimal results (although the most promising candidates that have held under repeated replication, e.g. CHRM2, show different frequencies in the HapMap groups), but there is a lot of genetic evidence using twins and adoptees, structural equation modeling, MRI, etc, even if you don't think it's conclusive.

If there is no evidence, why did a majority of 1,020 mostly left-leaning behavioral geneticists, sociologists, psychologists, and other relevant specialists attribute at least part of the black-white IQ gap in America to genetic factors, while only a quarter withheld judgment and a sixth said the gap was wholly environmental? Are they all stupid or racists? Why is everyone at NIH so terrified of the results of genomic testing of the hypothesis if they're confident that it's false, rather than just doing the experiments and proving the racists wrong?
http://en.wikipedia.org/wiki/Snyderman_and_Rothman_(study)

Here is an extended review of the literature and defense of Watson.
http://www.gnxp.com/blog/2007/10/james-watson-tells-inconvenient-truth_2...

Here is a comprehensive review article presenting much of that evidence:
http://psychology.uwo.ca/faculty/rushtonpdfs/PPPL1.pdf

One of the authors is the leading psychologist Arthur Jensen, of whom the egalitarian socialist, and discoverer of the Flynn Effect, James Flynn (http://en.wikipedia.org/wiki/James_R._Flynn) said:

"I never suspected Arthur Jensen of racial bias. Over the years, I have found him scrupulous in terms of professional ethics."
http://www.gnxp.com/blog/2007/12/10-questions-for-james-flynn.php

There are simple ways (like MALD) to rather conclusively test the hypothesis of genetic causation for group differences with one big grant for SNP-chips and the like. They will probably be deployed in the next several years, by the Pioneer Fund if no one else. At what odds would you take a bet on the results of those experiments?
http://www.apa.org/journals/releases/amp60160.pdf
http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=384950

U (not verified) | 10/08/08 | 23:49 PM

I don't think it's possible to honestly assert that there's no evidence with knowledge of the literature. Genome-wide association studies have turned up minimal results (although the most promising candidates that have held under repeated replication, e.g. CHRM2, show different frequencies in the HapMap groups), but there is a lot of genetic evidence using twins and adoptees, structural equation modeling, MRI, etc, even if you don't think it's conclusive.

The studies you mention have shown that genetic variation certainly plays a significant role in various traits - I'm not denying that. But there is more variation within races or populations than between them - the genetic differences between races are minor compared to the differences found within races.

Sorry for this drive-by post - I had been planning to go into this in more depth even before I saw Debyshire's piece. I'll try to get more up soon.

Mike

Michael White | 10/09/08 | 09:59 AM

I'm pretty sure now that you haven't read the relevant literature, and I think it would be a disservice to your readership to write that post before reading the Jensen article and perhaps also the defense of Watson. They're not that long, and the evidence is convincing (that there is at least some significant genetic component to racial differences) to a majority of researchers in related fields when they are anonymously surveyed, even if they're unwilling to face harassment for admitting it publicly. If you don't take it into account, your post will be shadow-boxing.

"The studies you mention have shown that genetic variation certainly plays a significant role in various traits - I'm not denying that. But there is more variation within races or populations than between them - the genetic differences between races are minor compared to the differences found within races."

Michael, that's certainly true for neutral variation and in general, but not for traits that have been under selection. Differences in skin color, maturation time, lactase, alcohol resistance, disease resistance, adjustment to carbohydrates, etc, exist between groups and can sometimes even be larger than average differences within races.

Also, we already have an instance of selection of personaity and resultant group differences in DRD4 alleles:

http://www.pnas.org/content/99/1/10.full
http://www.pnas.org/content/99/1/309.abstract

[You can read up on the following in the Jensen article and follow the citations there]

MRI brain scans can predict a big chunk of human IQ variation, with most of the predictive power from the size of the regions where size is most heritable (and that is very highly heritable). People of African, Caucasian, and East Asian descent have increasing brain size in that order (enough to explain a third of the gap alone), but if you match blacks and white for IQ, you find both have the same brain size. (This is what you would expect if it was one of several independently varying traits affecting cognition.)

The tests that show the greatest heritability when you look at twin pairs are also the ones where racial differences are largest.

Children of mixed racial ancestry show increases or decreases of adult IQ roughly approximating their degree of admixture in transracial adoption and other studies.

The IQ tests where inbreeding depression lowers results the most in Japan are also the ones with the largest racial differences.

"Sorry for this drive-by post - I had been planning to go into this in more depth even before I saw Debyshire's piece. I'll try to get more up soon."
No worries.

U (not verified) | 10/09/08 | 11:29 AM

Incidentally, insofar as IQ is a polygenic trait with many rare alleles of limited effect (which is what the genome-wide association studies indicate) in mutation-selection balance, then strong selection for other things will result in less relative pressure for IQ. Because we know that selective pressures varied a lot worldwide, we have reason to expect differences in cognitive ability.

Likewise, historical records in Britain and Europe show that wealth was highly correlated with number of surviving children for many centuries, plausibly creating selective pressure for things like future time orientation or IQ.

U (not verified) | 10/09/08 | 12:57 PM

A piece of Classic Derbyshire

I got to know John Derbyshire through his writings on mathematics. Then I came across this:

The Abolition of Sex Don’t worry — only in the law. Only!

What he was predicting is beginning to happen in practice, This side of the Pond!

Toilet signs 'too PC'"

Robert H. Olley
Physics Department
University of Reading
England

Robert H Olley | 10/09/08 | 09:10 AM

"atson later apologized "unreservedly" for his comments, stating that there is no scientific basis for such beliefs."

Not exactly. He said:

"To those who have drawn the inference from my words that Africa, as a continent, is somehow genetically inferior, I can only apologise unreservedly. That is not what I meant. More importantly from my point of view, there is no scientific basis for such a belief...

The overwhelming desire of society today is to assume that equal powers of reason are a universal heritage of humanity...."

To question this is not to give in to racism. This is not a discussion about superiority or inferiority, it is about seeking to understand differences, about why some of us are great musicians and others great engineers."

As noted on Gene Expression:

"Watson would only be retracting his intelligence claims if he considered those claims tantamount to claims of 'superiority' or 'inferiority', which he clearly emphasizes he doesn't. Watson is saying that questioning that all races are equal in intelligence is not racism, it is trying to figure out why the world looks the way it does, with the greatest engineers and the greatest musicians disproportionately coming, in a systematic way, from different racial backgrounds. In other words culturally separated people of African descent have been musical innovators across a diverse range of cultures (in the Middle East, Africa, Europe, North and South America, and the Caribbean), while culturally separated people of East Asian descent have excelled at math and science across a diverse range of cultures (in Asia, Europe, North and South America, and the Caribbean).

This is not a claim of racial 'superiority' or 'inferiority', either in terms of legal worth or even in terms of overall talent - since groups all have different strengths and weaknesses. It is simply the recognition that people of different genetic heritage, on average, reveal different talents wherever they are found in the world, and there is one explanation that best accounts for these observations: evolution.

In other words, Watson was thinking like a scientist. Which is exactly why he was punished.

The moral laws of our society dictate that we are not allowed to think scientifically about some issues. Especially not in public."

Ben Richards (not verified) | 12/04/08 | 22:06 PM

"But there is more variation within races or populations than between them - the genetic differences between races are minor compared to the differences found within races."

Isn't this the Lewontin fallacy (see AWF Edwards 2003 paper on it). Steve Hsu discusses this:

"Further technical comment: you may have read the misleading statistic, spread by the intellectually dishonest Lewontin, that 85% percent of all human genetic variation occurs within groups and only 15% between groups. This neglects the correlations in the genetic data that are revealed in a cluster analysis. See here for a simple example which shows that there can be dramatic group differences in phenotypes even if every version of every gene is found in two groups (i.e., 100% of the variation is found within each group) -- as long as the frequency or probability distributions are distinct. Sadly, understanding this point requires just enough mathematical ability that it has eluded all but a small number of experts.)"

http://infoproc.blogspot.com/2008/01/no-scientific-basis-for-race.html

Ben Richards (not verified) | 12/04/08 | 22:15 PM

Recent estimates for between-group variation range from 8%-15%, depending on which genes you're looking at. This recent paper looked at 650,000 SNPs in individuals from 51 populations and found that ~11% of the variation is between populations or regions when you look at autosomal SNPs.

Geneticist Lynn Jorde, who has been critical of Lewontin's claims about the non-existence of race, reviews similar numbers in a 2004 paper.

Jorde, when he spoke at our departmnet this fall, basically put it this way: 1. race is a sloppy and biologically confusing term, we should talk about populations instead; 2. there is enough population-specific genetic variation to enable scientists to assign most people to the correct ancestral population by just looking at DNA, but 3. population-specific genetic variation only makes up about 10-14% of human genetic variation.

This:

Further technical comment: you may have read the misleading statistic,
spread by the intellectually dishonest Lewontin, that 85% percent of
all human genetic variation occurs within groups and only 15% between
groups. This neglects the correlations in the genetic data that are
revealed in a cluster analysis. See here for a simple example which shows that there can be dramatic
group differences in phenotypes even if every version of every gene is
found in two groups (i.e., 100% of the variation is found within each
group) -- as long as the frequency or probability distributions are
distinct.

is a misunderstanding of what geneticists are claiming. That 85%-15% statistic does have to do with correlations among SNPs, and partitioning of genetic variance; it does not mean that 15% of all SNPs are only found in one population and not the other.

It's all about differences in allele frequencies, not presence or absence of alleles in different populations (although there are some alleles, not many, that are pretty much population-specific).

Jorde's paper which I linked to above has more details. This paper also discusses how geneticists calculate the partitioning of genetic variance.

Michael White | 12/04/08 | 22:52 PM

I think you misunderstand what Hsu is saying. The point is that genetic clustering *implies* frequency differences on certain alleles in different populations. This *might* imply group differences on certain genetically influences, measurable traits.

To claim that the 85-15 variance breakup rules out such differences *is* the Lewontin fallacy. There is a nice picture at the link below to clarify things.

http://infoproc.blogspot.com/2008/11/human-genetic-variation-fst-and.html

Anonymous (not verified) | 12/05/08 | 10:55 AM

To claim that the 85-15 variance breakup rules out such differences
*is* the Lewontin fallacy. There is a nice picture at the link below to
clarify things.

I'm not saying it rules out differences in any measurable traits. The most obvious example is known skin pigmentation alleles that appear to be largely responsible for differences in skin color between Europeans and Africans.

But I agree with Hu that:

There is no strong evidence yet for specific gene variants (alleles)
that lead to group differences (differences between clusters) in
behavior or intelligence...

Which is what Derbyshire gets wrong.

And I think that for a trait as complex as intelligence, it's extremely unlikely that there are significant population differences because of the genetic complexity of the trait and the likely pleiotropy of the genes, especially given the small amount of between-population genetic variation.

I do disagree with Lewontin and agree with Lynn Jorde who has criticized Lewontin - it's not right to say there is no genetic basis for 'race' (if you redefine the sloppy term race to mean populations).

Where Hsu is wrong is here:

you may have read the misleading statistic, spread by the intellectually dishonest Lewontin, that 85% percent of all human genetic variation occurs within groups and only 15% between groups. This neglects the correlations in the genetic data that are revealed in a cluster analysis.

That 85-15 statistic most certainly does not neglect correlations in the genetic data - that's what it's based on. So Hsu is right that correlations in the genetic data are important; he's wrong to say that the 85-15 statistic doesn't reflect that.

Michael White | 12/05/08 | 11:32 AM

This is splitting hairs, but what Hsu means there is probably that in using the 85-15 variance break up to claim that one can't identify separate clusters is to ignore *additional* information in the correlations.

The point is that even with an 85-15 breakup of within vs between group variance, there is still a lot information to be had. There are *many* directions of variation along which two groups can be almost disjoint. Along one of these principal components NO member of group 1 could be mistaken for a member of group 2. (Up to exponentially small number of exceptional cases -- see the figures here:
http://infoproc.blogspot.com/2008/11/human-genetic-variation-fst-and.html )

Now, all it would take is some correlation between one of these principal component directions and a phenotype like intelligence to produce a group difference. I am not saying this is the case, but it certainly can't be ruled out.

You seem to think 85-15 means that intergroup variation must be small. That is not the case at all. Imagine a simple model with N independent vectors of variation. Suppose that along a particular direction two groups either have (1) exactly the same distribution or (2) are entirely disjoint with a number of bp differences roughly equal to the SD of the distribution of type (1).

Then it is easy to see that we could have (roughly) .15N directions of type 2 and still satisfy the Lewontin 85-15 breakup. But .15N makes it quite plausible that phenotype distributions of *many* types will differ in the two groups.

Remember, we are not discussing anything as distinct as black-white skin pigmentation differences. We are talking about a shift in mean of a fraction of an SD or a 10% change in variance. These are small effects. It's worth pointing out that phenotype differences in, e.g., IQ are clearly observed. That is not controversial. What is controversial is to what extent the causes are genetic. Invoking a "small" 15% between group genetic variation hardly suggests that the causes can't be genetic.

Anonymous (not verified) | 12/05/08 | 12:54 PM

You seem to think 85-15 means that intergroup variation must be small.

85-15 means intergroup genetic variation is small.

Just given that fact, yes this genetic variation could produce a lot of population-specific phenotypic variation.

But given human population history, the genetic complexity of a trait like intelligence, and the likelihood that the vast majority of the SNPs in the human genome are phenotypically almost or completely neutral, I'm arguing this is unlikely.

Michael White | 12/05/08 | 14:47 PM

Oops

"This *might* imply group differences on certain genetically influences, measurable traits."

should be

"This *might* imply group differences on certain genetically influenced, measurable traits."

Anonymous (not verified) | 12/05/08 | 10:56 AM

BTW, "group differences" does not mean that the distribution of phenotypes is disjoint over the two groups (they could still overlap), but that the *distributions themselves* are different -- i.e., could have different mean or variance, or tail behavior. This is to be *expected* if the allele frequencies are different in the two groups.

If you have an argument for why current understanding of genetic clustering is able to rule out such group differences, I would like to hear it.

If you don't have such an argument then you have to admit that we *might* find that certain population groups (which correspond to folk characterizations of "races") might exhibit different distributions of certain characteristics like IQ, personality, disease resistance, height, etc. Note the last two are already demonstrated.

In the case of IQ or personality we don't know the answer yet, but I don't think any well-informed and careful scientist can claim they know the answer at high confidence at this point.

Anonymous (not verified) | 12/05/08 | 11:03 AM

I've written a longer piece on this subject, but just to recap:

I do not believe that any group differences are ruled out - clearly skin pigmentation (a trait for which we have found population-specific alleles) and height, for example are genetically influenced traits that show different distributions in different populations.

But given the small amount of genetic variation between populations, I'm skeptical that a trait as genetically complex as intelligence exhibits significant genetically-based differences between populations. You're right that the jury is still out on this (which is where Derbyshire, and Watson for that matter, go wrong - they both are claiming that we've largely settled the issue in favor of genetically-based population differences in intelligence). My prediction is that it is much easier, genetically, to produce population differences (and yes, I know we're talking about distributions) in skin color, stature, and disease resistance, than it is to produce intelligence differences.

Michael White | 12/05/08 | 11:52 AM

BTW, "group differences" does not mean that the distribution of
phenotypes is disjoint over the two groups (they could still overlap),
but that the *distributions themselves* are different.

I haven't said otherwise, but I want to emphasize that you are making a great point here. Our intuitions about race are often misleading because we frequently rely on what appear to be disjoint characteristics to make racial classifications, like hair, skin color, and facial features - those differences do partition quite well among populations. You don't really find much overlap in skin color between say, Northern Europeans and West Africans, unless you're talking about an individual of mixed ancestry.

Because those major differences are so easily visible, we tend to (incorrectly) think that many population differences are like that - sharply divided between racial groups.

Michael White | 12/05/08 | 12:00 PM

Comments

The Abolition of Sex Don’t worry — only in the law. Only!

Toilet signs 'too PC'"

Add a comment

Know Science And Want To Write?

What's Happening

Books By Writers Here

Not a Scientific Blogging member yet?

Register Now for a Free Scientificblogging.com Account

Already a Scientific Blogging member?

Sign In Now

Related Articles Here On Scientific Blogging

Comments

The Abolition of Sex Don’t worry — only in the law. Only!

Toilet signs 'too PC'"

Add a comment

Know Science And Want To Write?

What's Happening

Books By Writers Here