Banner
    What Our Genes Tell Us About Race
    By Michael White | November 5th 2008 04:50 PM | 43 comments | Print | E-mail | Track Comments
    About Michael

    Welcome to Adaptive Complexity, where I write about genomics, systems biology, evolution, and the connection between science and literature,

    ...

    View Michael's Profile
    Which species is more diverse, humans or chimps? Most of us would be tempted to answer 'humans'. Unless you're a primatologist or you work at a zoo, you would likely have trouble telling one chimp apart from another, not to mention distinguishing between West African and Central African chimpanzees. By contrast, we can easily spot differences among humans - if asked to guess whether someone was from China, Pakistan, or Kenya, few of us would have any trouble getting the answer correct.

    By the measure of genes though, humans are amazingly uniform. Humans are genetically less diverse than chimps, and both chimps and humans are much less diverse than a common species of fruit fly. Given our species' long history of racial conflict, our genetic uniformity may come as a surprise. Not too long ago people in polite company would debate whether different human races really all belonged to one species. Our DNA tells us that our genetic differences don't even come close to matching the variety found within a single, apparently monotonous fruit fly species.


    Measuring Genetic Diversity

    How do we measure genetic diversity? The basic concept is simple. Pick a few dozen or a few hundred genes and tally the differences in those genes in a sample of a few dozen or a few hundred people. Thus at gene A, person 1 and person 2 may differ at 5 DNA bases, while person 1 and person 3 may differ at 7 bases. After making all of the possible comparisons for gene A, you compute the average number of differences - in the case of gene A let's say we find an average of 6 differences between any two people. If gene A is 6000 bases long, we say any two people would differ on average once in every 1000 DNA bases.

    But when we look at gene B, we may get a different answer - maybe only an average of one difference every 2000 DNA bases. To get a good measure of how genetically different humans tend to be, it's obviously best to get results from many different genes, or even entire genomes. Most recent studies have found that humans differ on average once in every 1000-1500 DNA bases. Given that we have 3 billion DNA bases in our genome (just counting one copy - either your maternal or paternal version), any two people are likely to differ at about 2-3 million positions in their genomes. That number applies to differences within yourself as well - there are about 2-3 million differences between the DNA you inherited from your mother and the DNA you inherited from your father.

    (There exist more sophisticated ways of measuring genetic differences, and we could include genetic variants that involve more than just a difference at a single DNA base pair - we could count segments of DNA that have been inserted or deleted, called copy number variants. However the basic approach to measuring genetic difference that I've described is all we need here.)

    How does this genetic variation stack up against variation in other animals? While we tend to differ at 1 position in every 1000 bases or so, chimps differ at about 1 in every 750 bases, and gorillas differ at roughly one in every 650 bases. The fly species D. melanogaster is much more genetically diverse, and tends to differ at one out of every few hundred bases. Keep in mind that these estimates are just averages - genetic variation is not spread out evenly across an entire genome. Some portions of our genomes are highly variable, while others are relatively uniform.

    The bottom line is this: although fruit flies and gorillas may look largely the same to us, they beat us hands down in genetic diversity.


    Genetic Differences Among Human Populations

    Of course the genetic differences that do exist among humans are enough to generate much of the biological diversity we see around us - differences in skin, hair, and eye color, our voices, our physical stature, and our personalities. Obviously environment plays a big role in many traits, but as the differences between Samoans and Japanese illustrate, genetics can account for a great deal even when there is a large environmental influence.

    At first guess, you might think that most of our genetic diversity would fall along racial lines. Race differences often seem to be the most obvious differences among different human groups, so it wouldn't be surprising if genetic differences fell along racial lines as well. Genetically, a white guy like me should be much more similar to my white neighbor than my black one, right?

    Before we get to the answer, let's fill in some background. In any discussion of genes and diversity, it is critical to keep in mind one obvious but nevertheless subtle point: only genes that exist in different versions are responsible for our biological diversity. There may be a gene that is critical for intelligence, but if that gene is absolutely 100% identical in all humans, then it does not account for differences in intelligence among humans. We may one day identify all of the genes involved in producing intelligence (defining intelligence may actually be a harder problem), but we won't understand the genetic basis of differences in intelligence until we discover the human genetic variation in the genes responsible. This narrows our species' genetic palette - not all of the genes responsible for a particular trait will be a source of diversity in that trait.

    With the availability of genome-scale tools, researchers have devoted a significant amount of effort in the last eight years to finding those genes that exist in different versions among humans. We haven't mapped all of this genetic diversity yet, but there is enough to start looking at genetic variation among humans in different parts of the world. With this new data, we are developing a much more detailed picture of how races differ genetically.

    'Race' though, is a horribly sloppy term. Geneticists prefer to speak about populations, not out of political correctness but because race is extremely imprecise. We've all filled out some form or another asking whether we are Black, White, Hispanic, or 'none of the above.' It's obvious that this is much less informative than knowing whether someone's ancestry is African, Australian Aborigine, European, or East Asian.

    Thus researchers like Richard Lewontin have argued that "As a biological rather than a social construct, 'race' has ceased to be seen as a fundamental reality characterizing the human species." Race may be too imprecise to be biologically meaningful, but there has to be some biological reality behind the obvious physical differences in different human populations, right?

    Yes, there are genetic differences between different human populations, but the big surprise is this: genetic differences between human populations are few compared to the differences within human populations.

    Here is what that means:

    If you compare my genome with that of a Chinese grad student down the hall from me, you'll find that only tiny fraction of the 2-3 million differences between us tells you much about our ancestry. Among Chinese, there may be a tendency to have a DNA base 'G' at position XYZ in gene ABC on chromosome 12, while among Europeans (where my ancestors came from), there is a tendency to have an 'A' at that same position. What we find though, in almost all cases, is that these tendencies are not absolute: 90% of Chinese may have base 'G', while the other 10% have base 'A'. And maybe 70% of Europeans have base 'A' at position XYZ, while 30% have base 'G'.

    So in other words, the fact that my Chinese friend has base 'G' at position XYZ in gene ABC does not tell you with certainty that he's Chinese. In fact both my Chinese friend and I may have the base 'G' at that same position, even though it is less likely in my case. If you look at any one gene, you don't get enough information to make an accurate call.

    In order to really see differences among human populations, you have to look at many genes (or any place in the genome where humans vary - it doesn't have to be a gene). In the Chinese population, base 'G' may be common at position XYZ on chromosome 12, base 'T' may be more common at position TUV on chromosome 6, etc., etc. So once you look at dozens or hundreds of informative positions, you can say with high confidence, 'this person is Chinese, and that one is European.' (And of course we could all be American or Canadian or British by birth - we're obviously talking about ancestral populations here.)

    By examining enough genes, we can reliably use just DNA to correctly assign people to ancestral geographical populations. Researchers can do it blindly - they can look at DNA sequence for 1,000 people whose identities are hidden, and use DNA information to assign those people to geographical populations. Once the assignments are made, the researchers take a peek at the true identities of their sample group, and it turns out that their assignments are extremely accurate.

    But here is where it gets paradoxical: while there are enough genetic differences among human populations to make accurate classifications, those genetic differences make up only 5-15% of the total amount of genetic variation. Most of the genetic variation among humans has nothing to do with differences in populations. The genetic differences between 'races' are minor compared to the differences between people in general.


    Medicine and Race

    So what does this mean about the biological differences we can observe directly? One key area where this has implications is medicine. This result suggests that race is not actually a very good predictor of medical outcomes (at least those based on genetics, not lifestyle). Let's take a classic case: There is a genetic variant in an angiotensinogen gene that substantially increases one's risk for developing high blood pressure. This genetic variant shows up fairly frequently among some African populations, in up to 90% of the population in some cases. But, like most medically important genetic variation, it shows up in other populations as well. 30% of Europeans also have this genetic variant.

    Suppose there is a blood pressure drug that is ineffective in people with this particular angiotensinogen genetic variant. If you simply make treatment decisions on the basis of race, you will frequently make the wrong call - some Africans who would benefit wouldn't get the drug, and a fair number of Europeans who don't benefit would be given a useless prescription. As University of Utah geneticist Lynne Jorde argues (PDF):

    [Genetic] variation tends to be shared widely among populations, so race will often be an inaccurate predictor of response to drugs or other medical treatments. It would be far preferable to test directly the responsible alleles [genetic variants] in affected individuals.


    When it comes to medically important traits, we need to avoid typological thinking - when we consider genetic variation as a whole, humans don't fall neatly into racial or population categories. Yes, a minor fraction of genetic variation enables scientists to accurately categorize people by ancestral population, but the vast majority of the variation that produces diversity in our susceptibility to disease, response to drugs, and even our behavior does not fall neatly along racial lines. All human populations overlap substantially in their genetic diversity.


    Intelligence and Race

    The debate over race and intelligence has a long and tarnished history, although that doesn't mean it's not a legitimate scientific question to address. However, the debate has taken place almost entirely outside modern genetics, falling instead within the realm of psychology (such as work done by Arthur Jensen). Some writers would have you believe that science is converging on a consensus that the 'IQ' gap between various races is genetic (and that liberal conspirators are trying to cover it up). That claim is false. Researchers have not identified a single genetic variant with an impact on intelligence that falls along population lines. In fact several studies have recently tested variants in genes that appear to be involved in controlling brain size. No correlation with intelligence was found. Yes, genetics does play a significant role in intelligence, and many other traits. But there is simply no genetic evidence (and I mean real genetics, not psychology) for genetic differences in intelligence between human populations.

    Why is this so? Other traits, like skin color, obviously fall along population lines. While skin color is obviously not a 100% reliable predictor, skin color is a major indicator of race. Irish, Kenyans, Pakistanis, and Chinese populations all have clearly different skin tones.

    It turns out, not surprisingly, that the genetic variation for some (but not all) skin color genes does in fact follow population divisions, in contrast with most other genetic variation. This is most likely because skin color differences end up being relatively simple - a single variant of a gene (causing lighter skin, for example) can easily become common in a population through natural selection. The result is that you have different human populations with dramatic differences in skin color.

    Other traits, however, are much more complex than skin color. Physical differences which are determined not by one, but many different genetic variants, are unlikely to split neatly by population. Intelligence is probably one of the most complex traits humans possess. It is almost certainly affected by variants in many different genes, and many of those genes have other important functions in the body. That means this: two different human populations could have easily developed differences in skin color between them, but differences in intelligence would have been extremely hard to develop, by chance or by natural selection.

    Racial conflict has long been a part of human societies. Along with that conflict has come frequent speculation (most famously, but not exclusively among whites with European ancestry) that one race is inferior to another. Some have been worried that modern genetics would substantiate that belief, but our best genetic evidence to date shows those worries unfounded. Genetics does play a large role in the diversity we find among human beings. That diversity, in spite of some dramatic but superficial exceptions like skin color, is shared in common among all races.

    Comments

    jgerke
    There are species of roundworm that differ 1 in every 20 bases.  That's essentially two orders of magnitude more diversity than in humans.   But, each roundworm looks essentially the same.  It's pretty astounding that there are organisms out there that have as many single differences in their DNA as humans and fish, but look exactly the same.
    adaptivecomplexity
    1 in every 20 bases? Wow, I thought flies were diverse. I would think that 1 in 20 has got to be close to the record for a eukaryotic organism. Maybe it's also dependent on the amount of transposable elements and other, largely non-functional stuff? I don't know the literature on this - I imagine salamanders and some plants, with high quantities of repetitive elements can accumulate a lot of nucleotide diversity within a species.
    Mike
    jgerke
    That's the synonymous site diversity, just to be clear.
    I'm really disappointed with this post. The technical explanations are mostly fine (although there are mistakes, see below), but it fails to engage with the actual issues and evidence and is highly misleading.

    "The bottom line is this: although fruit flies and gorillas may look largely the same to us, they beat us hands down in genetic diversity."

    Diversity in terms of polymorphisms without functional consequences, or in the form of different mixes of polymorphisms that result in extremely similar distributions of phenotypes across populations, is practically irrelevant. Neutral evolution is very important for genetic diversity, but selection is key for genotypic differences with interesting phenotypic effects.

    "If you look at any one gene, you don't get enough information to make an accurate call."
    This just wrong.

    "The gene comes in two versions, one of which is found in 99 percent of Europeans and the other in 93 to 100 percent of Africans, the researchers report in today's issue of Science."
    http://www.nytimes.com/2005/12/16/science/16gene.html

    Actually, there are hundreds of already identified variants that are found in almost all members of one of the HapMap groups but only very infrequently in others, or at 20%-80% frequency in one group but barely at all in others (this naturally changes for mixed populations). These were found using tests for selection, so they have significant phenotypic effects relevant to fitness, often apparently on the order of 5%. See, e.g.:

    http://www.plosgenetics.org/article/info:doi%2F10.1371%2Fjournal.pgen.00...
    http://biology.plosjournals.org/perlserv/?request=get-document&doi=10.13...
    http://www.pnas.org/content/103/1/135.abstract

    "But, like most medically important genetic variation, it shows up in other populations as well. "
    Certainly most, but the remainder is significant, since the differences between populations are disproportionately the result of selection in different environments.

    "Some writers would have you believe that science is converging on a consensus that the 'IQ' gap between various races is [this should say partially] genetic"

    So, to be clear, non-molecular behavioral genetics, psychometrics, neuroimaging, etc aren't part of science? The various lines of evidence in the review article I gave you a few weeks ago (http://www.udel.edu/educ/gottfredson/30years/Rushton-Jensen30years.pdf) are all non-scientific? The fact that I can predict 45% of variation in IQ between individuals using an MRI of the brain, and 16% using simple brain size and weight information, and that the things being measured show extremely high heritability is not scientific evidence? The fact that continental ancestry groups show systematic differences in brain size, such that if the relationship between brain size and IQ within continental ancestry groups held between them, it could explain 5 IQ points of the gap? That the IQ gaps are reduced when you match for brain size, and that the brain size gaps are eliminated when you control for IQ? What is the non-genetic explanation for the fact that the subtests where inbreeding depression in a Japanese population most lowered scores are also those for which group differences are largest? This is like saying that we can't have scientific evidence that heart disease has a genetic component until after all the GWAS results.

    A partly genetic explanation for group differences is in fact the leading position among anonymously surveyed behavioral geneticists and relevant psychologists (http://en.wikipedia.org/wiki/Snyderman_and_Rothman_(study)), regardless of their political attitudes on affirmative action or scores on a liberal-conservative scale. Attacks on the relevant evidence almost always work through sleight-of-hand and ignoring the literature in favor of straw men of various kinds. I thought that you were going to write something more careful and honest after the previous discussion, but this post is crafted in such a way as to mislead readers.

    "Along with that conflict has come frequent speculation (most famously, but not exclusively among whites with European ancestry) that one race is inferior to another. Some have been worried that modern genetics would substantiate that belief, but our best genetic evidence to date shows those worries unfounded."

    This talk of 'inferiority' is problematic in several ways. First, regardless of the extent to which genetic causes are responsible, there are real group differences in IQ, in brain anatomy and development, that matter for educational attainment, job performance, understanding of politics, etc, in the United States. It is true that Chinese-Americans collectively have a higher mean IQ than the American average, and this explains a substantial part of their success in America, while a lower mean IQ among African-Americans explains many disparities in outcomes for that population, but these facts don't make any group inferior. The well-being of a person, whatever their ancestry, with an IQ of 85 is not less valuable than that of a person with an IQ of 115, whatever their ancestry.

    Even if you thought that IQ was related to morally salient things (e.g. because higher-IQ people are more likely to support equal rights for gay people, reproductive freedom, and religious tolerance and secularism), group averages don't somehow reduce Barack Obama's very high IQ, or increase the IQs of the likely 99+% of white Americans with lower IQs.

    "some have been worried that modern genetics would substantiate that belief, but our best genetic evidence to date shows those worries unfounded."

    First, we don't have much genetic evidence because very little of the genetic influence on IQ has been pinned down at all. GWAS studies haven't yet found particular variants for most medical conditions known to be highly heritable from twin and family studies, does the fact that we haven't yet done high power studies, with sequencing to catch rare variants and CNVs, mean that the remaining genetic influence doesn't exist?

    Second, it's just not true that what has been found doesn't support that belief. One of the few genes to show a replicated effect on IQ (in this case performance IQ in particular) is CHRM2 (http://www.ncbi.nlm.nih.gov/pubmed/17160701). Likewise several of the other reported variants (in which I place little confidence until higher-powered GWAS are conducted) were distributed quite differently across the HapMap groups. HapMap data show that alleles associated with higher performance IQ in a European-American sample are more common among the East Asian HapMap sample, and it just so happens that East Asian populations also have higher performance IQs. The Ashkenazi intelligence paper presents evidence for effects of Gaucher's disease and torsion dystonia:, and the East Asian HapMap sample has been under massive recent selection at the Gaucher's locus:

    http://homepage.mac.com/harpend/.Public/AshkenaziIQ.jbiosocsci.pdf
    I quote:

    We do have strong but indirect evidence that one of these, Gaucher disease, does indeed
    increase IQ. Professor Ari Zimran, who heads the Gaucher Clinic at the Shaare Zedek
    Medical Centre in Jerusalem, furnished us a list of occupations of 302 Gaucher patients.
    Because of the Israeli medical care system, these are essentially all the Gaucher patients
    in the country. Of the 255 patients who are not retired and not students, 81 are in
    occupations that ordinarily average IQ’s greater than 120. There are 13 academics, 23
    engineers, 14 scientists, and 31 in other high IQ occupations like accountants, physicians,
    or lawyers. The government of Israel states that 1.35% of Israeli’s working age
    population are engineers or scientists, while in the Gaucher patient sample 37/255 or 15%
    are engineers or scientists. Since Ashkenazim make up 60% of the workforce in Israel, a
    conservative base rate for engineers and scientists among Ashkenazim is 2.25% assuming
    that all engineers and scientists are Ashkenazim. With this rate, we expect 6 in our
    sample and we observe 37. The probability of 37 or more scientists and engineers in our sample, given a base rate of 2.25%, is approximately 4 x10
    -19. There are 5 physicists in the sample, while there is an equal number, 5, of unskilled workers. In the United States
    the fraction of people with undergraduate or higher degrees in physics is about one in one
    thousand. If this fraction applies even approximately to Israel the expected number of
    physicists in our sample is 0.25 while we observe 5. Gaucher patients are clearly a very
    high IQ subsample of the general population.

    Ever since torsion dystonia among the Ashkenazim was first recognized, observers have
    commented on the unusual intelligence of patients. Flatau and Sterling (Eldridge, 1976)
    describe their first patient as showing “an intellectual development far exceeding his
    age”, and their second patient as showing “extraordinary mental development for his
    age.” At least ten other reports in the literature have made similar comments. Eldridge
    (1970, 1976) studied 14 Jewish torison dystonia patients: he found that their average IQ
    before the onset of symptoms was 121, compared to an averge score of 111 in a control
    group of 14 unrelated Jewish children matched for age, sex, and school district. Riklan
    and colleagues found that 15 Jewish patients with no family history of dystonia (typical
    of DYT1 dystonia) had an average verbal IQ of 117 (Eldridge, 1979; Riklan et al., 1976).
    http://www.ncbi.nlm.nih.gov/pubmed/17160701

    "(and that liberal conspirators are trying to cover it up). "
    It's neither liberal (the Bush administration has been making appointments for the last 8 years) nor a (secret) conspiracy. There are numerous published articles and statements explicitly calling for censorship, e.g. in this collection:
    http://www.genome.gov/13014159

    Proposals include making research showing a genetic role in group IQ differences a violation of human subjects rules ("stigmatization of a group"), banning any researcher receiving federal funds from publishing or citing any article suggesting genetic explanations, preventing ideologically suspicious researchers from accessing datasets, and more. The NIH just had an event on this, and highly placed people there are keen on moving in this direction. IRBs and university are already extremely hostile, as researchers in the field can attest to. Many journal editors reject scientifically rigorous papers on the subject, often admitting that they do so for political reasons, and certainly apply a double standard to them. Likewise academic publishers, e.g. the renowned Arthur Jensen published his magnum opus (which was very favorably received by many expert reviewers), The G-Factor, with a tiny obscure publishing house. Many researchers have been subjected to physical attacks, death threats, institutional harassment (as determined by independent arbitrators and scholarly organizations), etc.

    A while back a proposal was made for psychologists associated with more hereditarian and environmentalist models of the black-white gap in the U.S. to jointly design a study that would provide strong evidence to help settle the question of genetic causation, and then to jointly lobby to have it funded. The environmentalists refused, because they doubt that the data will support them, or know that it won't and just use their position to deflect hostile attention from their field or themselves. Nobelists Watson and Crick and Shockley, and most of the subject matter experts, didn't come to the taboo conclusion that genetic factors played a role in group differences in IQ because they were fools or racists, but because that's where the evidence points. If you could bet today on whether the gap will turn out substantially genetic when large-scale sequencing results tagged with phenotypic information are available (within 10 years), at what odds would you take the bet?

    "Other traits, however, are much more complex than skin color. Physical differences which are determined not by one, but many different genetic variants, are unlikely to split neatly by population. Intelligence is probably one of the most complex traits humans possess. It is almost certainly affected by variants in many different genes, and many of those genes have other important functions in the body. That means this: two different human populations could have easily developed differences in skin color between them, but differences in intelligence would have been extremely hard to develop, by chance or by natural selection."

    This is completely wrong. When many loci affect a trait, individuals with high and low traits will have different average frequencies for the lot, and the selective pressure will be divided up among those many loci. Empirically, the narrow heritability of adult IQ is 0.5-0.7. so if the mean IQ of parent pairs (weighted by number of children) is two points higher than the population average, you will get an increase of at least a point every generation and create a change larger than the differences observed between continental populations in a few hundred years. Cut heritability in half or to a quarter for increased environmental variation, and it's still easily feasible.

    jtwitten
    I'm putting this at the top in the hopes that it will be read.  Utilitarian, your arguments have consisted of a deluge "evidence" (just because it was published doesn't make it so) and interpretations based on several mutually exclusive and incorrect understandings of molecular and population genetics.  At no time, however, do you present a coherent set of hypotheses about the nature of the genetic variation that controls purported phenotypic distinctions between "races."  I challenge you to do so.  Below, I shall only address a few of your points as this is not my blog.  My failure to counter any specific point in no way implies a concession on my part of the veracity of that point.
    Empirically, the narrow heritability of adult IQ is 0.5-0.7.

    Narrow-sense heritability in its strictest sense does not represent the genetic contribution to a trait.  Historically, it was used as a measure for determining the breeding value of livestock.   The key to the assumption that the heritability directly related to the genetic determinacy of the trait was holding the environment constant for all subjects.  This can be done to a degree in animal breeding studies.  At its best, it can only be approximated via statistical methods in humans.

    Narrow sense heritability is, by definition, the ratio of additive variance for the trait to the total phenotypic variance.  High narrow-sense says nothing about the complexity of a trait.  Even if infinite loci, each with an independent, additive effect on the trait, were segregating in a population, the narrow-sense heritability could be 1.

    so if the mean IQ of parent pairs (weighted by number of children) is
    two points higher than the population average, you will get an increase
    of at least a point every generation and create a change larger than
    the differences observed between continental populations in a few
    hundred years

    No.  Response to selection diminishes over time as the alleles with the selected effect on the trait go to fixation in a population.  An individual with all the positive alleles for IQ would represent the maximum genetic IQ in this system.  An individual with all the negative alleles for IQ would represent the minimum genetic IQ in this system.  The difference between the two would represent the maximum difference.  The mutation rate in humans is not of a scale to support an assumption that additional variation is being added at an appreciable rate.  There has also not been a strong argument that IQ (a measurement of dubious value for assaying human intelligence) has been under continuing, strong, directional selection in modern human populations.

    First, we don't have much genetic evidence because very little of the
    genetic influence on IQ has been pinned down at all. GWAS studies
    haven't yet found particular variants for most medical conditions known
    to be highly heritable from twin and family studies, does the fact that
    we haven't yet done high power studies, with sequencing to catch rare
    variants and CNVs, mean that the remaining genetic influence doesn't
    exist?

    Finally, potential phenotypic differences between races are not equivalent to diseases.  Hypothetical racial differences will be widespread in that population.  Relative to another race, those alleles would have to be common and of strong affect.  Diseases on the other hand are infrequent.  That is why they are diseases and not the healthy condition.  GWASs are good at finding frequent alleles of strong effect.  They have failed to explain significant disease variation because the causative alleles are either rare or
    adaptivecomplexity
    I'm not disputing that intelligence is heritable. But none of the actual linkage studies you cite show population differences correlated with IQ. You claim that:

    HapMap data show that alleles associated with higher performance IQ in a European-American sample are more common among the East Asian HapMap sample, and it just so happens that East Asian populations also have higher performance IQs.


    And yet the authors of the CHRM2 study you cite say of the CHRM2 SNPs: "To our knowledge, this is the first gene yielding consistent evidence of association with IQ across multiple studies conducted by independent research groups."

    Even if we just take the CHRM2 SNPs that are significantly correlated only with fairly small differences in IQ measures - just because those SNPs show up with a somewhat higher frequency in an Asian population does not mean those SNPs are responsible for population differences, especially when the differences in population frequencies are not very large. Gene x gene and gene x environment interactions could swamp out any effects of CHRM2 SNPs. The bottom line is you have to do the test, rather than make improbable extrapolations. And so far, there is no molecular genetic evidence for genetically based IQ differences between populations.

    About disease mutations and GWAS vs. race-sepcific alleles: What is becoming clear from recent GWAS results is that many traits with high heritability are not explained by the common gene/common variant hypothesis. (Nature on the case of the missing heritability, subscription only) It's looking more and more likely that rare variants are responsible for many complex traits, including height and complex diseases. That's exactly the opposite of what you need for significant racial differences - alleles that are common in one population but not another.

    I'm also not claiming that psychology is not a science. I am however skeptical of any science today that claims to be doing genetics without any actual genotyping. Especially when it comes to human psychology/social science studies where researchers all too easily delude themselves into thinking they have managed to control all of the relevant variables, especially across populations with significant cultural differences. All of the regression in the world can't fix that.
    Mike
    "And yet the authors of the CHRM2 study you cite say of the CHRM2 SNPs: "To our knowledge, this is the first gene yielding consistent evidence of association with IQ across multiple studies conducted by independent research groups."'

    DTNBP1 fits that description although its evidence is weaker, and the first prong has been met by several SNPs found by Plomin et al in large British twin samples. As I said in my comment, this evidence is light and shaky, and I don't put much trust in it, I'm objecting to false claims that there is no molecular evidence at all, and especially to claims that there is contrary evidence in the form of IQ-affecting alleles that are evenly distributed.

    If we do random draws of IQ-affecting alleles and find that their distributions systematically correspond to observed phenotypes, that should lead us to become confident of differences in the underlying genotypic distribution long before we know all relevant alleles. Likewise, if the alleles that have been found were all evenly distributed that would support a non-genetic explanation of group differences.

    "It's looking more and more likely that rare variants are responsible for many complex traits, including height and complex diseases. "
    Yes.

    "That's exactly the opposite of what you need for significant racial differences - alleles that are common in one population but not another."

    This is wrong. Imagine a trait where all variation is due to mutation-selection balance for rare deleterious variants. At any given locus population A and population B may have frequencies of deleterious alleles less than 1% but it might be that population B has 0.013% and A 0.011%, since A is under stronger stabilizing selection. Summing across all the loci, you'll get large cumulative differences. Variants of larger effect can point out the direction of selection: if Ashkenazi have been under selection that has driven up costly pleiotropic IQ-boosters like TYD1 torsion dystonia, then they should also show reduced frequencies of IQ-decreasing rare deleterious alleles.

    Also remember that the best IQ GWAs have been conducted on European samples with minimal admixture, so alleles that have already swept in Europe (or Asia) but not Africa (of which there are quite a number, including genes known to have important neurological effects) haven't been tested, since at 99+% frequency they would no longer be responsible for within-group heritability. Bruce Lahn tested two alleles that had undergone a partial sweep (with negative results) and has dropped the work since because of the controversy, but no one has tested the full sweeps, and if group IQ differences developed early the large effect IQ increasers (e.g. brain size boosters) could all have swept early on after a change in selective pressure.

    Of course, they would show up easily if you did mapping by admixture linkage disequilibrium on people of mixed continental ancestry like African-Americans, but that is probably the most verboten study design there is (it got an entire issue of the American Psychologist and permanent website dedicated to the proposition that this must never be done, lest the results be something Scientists Were Not Meant to Know). When some rich person (or the Pioneer Fund) decides to put a million dollars into it, or if someone somehow gets IRB approval to do MALD for racial disparities in Alzheimer's, we'll know. Someone may do it this year, because of the fall in SNP-chip costs.

    "At no time, however, do you present a coherent set of hypotheses about the nature of the genetic variation that controls purported phenotypic distinctions between "races. I challenge you to do so. ""
    I suspect you're assuming that my mention of variants that are sweeping or have essentially swept one continental ancestry group but not another means that I think all group differences can be attributed to such variants. The main reason I mentioned that was because it was a scientific error in the post .

    To answer your challenge: I would guess, based on GWAS work such as Plomin's and Goldstein's, that the high heritability of psychometric intelligence, and in particular Gf or working memory (which correlates at ~0.9 with the general factor of psychometric IQ), is mostly a result of very many alleles of very small effect size. With thousands of genes directing brain development, the rate of deleterious mutations affecting Gf or working memory will be high, and mutation-selection balance will be very important. The equilibrium of that mutation-selection balance will vary depending on the strength of stabilizing selection.

    There will also be alleles with pleiotropic effects, like the Gaucher's disease and torsion dystonia alleles, which boost IQ substantially but come with disease risk. Several of the Ashkenazi alleles look like cases of heterozygote advantage. The brain's consumption of 20% of calories is also relevant for pleiotropy. Alleles in these classes may have larger and positive effects.

    Finally, some mutations will be in the process of, or have already completed selective sweeps. We'll know within 12 months whether the intense selection in the Asian HapMap sample in the neighborhood of the Gaucher's locus is related to cognition.

    What selective pressures have been at play? Working memory or psychometric Gf is correlated with success and perceived intelligence in hunter-gatherer tribes, accident rates, etc. However, the correlation between Gf and success increases as occupations increase in complexity. Artisans, scribes, traders, etc face higher cognitive demands. Cognitively more demanding trades resulted in more income, and for most of history more income meant more surviving children (historical evidence such as wills backs this up). The increased selective pressure on intelligence would shift the mutation-selection balance for rare deleterious alleles, alter equilibria for cases of heterozygote advantage, and change net fitness for some IQ-boosters with negative pleiotropic effects. The latter may have included variants affecting brain mass, and in particular the size of the brain regions most associated with IQ under MRI (previously held back by energy costs).

    Also, selection for anything other than IQ, e.g. disease resistance or adaptation to a heavily carbohydrate diet, would result in 'hitchhiking" of alleles with deleterious effects on IQ in the vicinity of lactase or malaria resistance alleles. This would temporarily shift the IQ distribution to the left, until the strength of selection on the latter relaxed. High disease prevalence in tropical areas may have meant more selective pressure was from disease resistance, and early development of carbohydrate-rich diets and urbanization may have provided time to 'get over' some of the 'hitchhiking effects.

    "I'm putting this at the top in the hopes that it will be read. Utilitarian, your arguments have consisted of a deluge "evidence" (just because it was published doesn't make it so) and interpretations based on several mutually exclusive and incorrect understandings of molecular and population genetics."
    http://biochemicalsoul.com/2008/10/john-derbyshire-claims-the-barack-oba...

    "IQ (a measurement of dubious value for assaying human intelligence)"
    It predicts accident rates, performance on almost every activity the military measures, educational success, job performance (correlations from 0.2 for simple jobs to 0.8 for the most cognitively demanding jobs), and viewers' impressions of intelligence when watching video of an individual talking. The American Psychological Association disagrees with you:
    http://www.indiana.edu/~intell/apa96.shtml
    I discussed this at length in this thread, and won't repeat the info and links there:
    http://biochemicalsoul.com/2008/10/john-derbyshire-claims-the-barack-oba...

    "Narrow sense heritability is, by definition, the ratio of additive variance for the trait to the total phenotypic variance. High narrow-sense says nothing about the complexity of a trait. Even if infinite loci, each with an independent, additive effect on the trait, were segregating in a population, the narrow-sense heritability could be 1."

    I didn't say that heritability indicates that IQ isn't polygenic (GWAS thus far strongly indicates many alleles of small effect), just the opposite: I said that selection can act powerfully on a polygenic trait like IQ, which was falsely denied in the post.

    "No. Response to selection diminishes over time as the alleles with the selected effect on the trait go to fixation in a population. An individual with all the positive alleles for IQ would represent the maximum genetic IQ in this system. An individual with all the negative alleles for IQ would represent the minimum genetic IQ in this system."

    *Human families have mean IQs substantially above and below the means of continental ancestry groups in good environments,* it isn't a theoretical question. I gave an example in which selective pressure was quite weak over a number of generations, but observed group differences could be replicated in a *single generation* by picking founders for a new population 2 SD from the American mean. Your argument is completely irrelevant.

    "The difference between the two would represent the maximum difference."

    That difference is empirically known to be much larger than what would be needed, from individual and family variation. Do you doubt that natural selection could shift the distribution of genotypes influencing height by 1 SD in either direction? That's what we're talking about here. Sure, it would be silly to think that selection could quickly produce 20 foot tall giants, but this is just moderately moving the mean of the distribution within the normal range.

    "The mutation rate in humans is not of a scale to support an assumption that additional variation is being added at an appreciable rate."

    http://www.anthro.utah.edu/PDFs/accel.pnas.smallpdf.pdf

    "Finally, potential phenotypic differences between races are not equivalent to diseases. Hypothetical racial differences will be widespread in that population. Relative to another race, those alleles would have to be common and of strong affect. Diseases on the other hand are infrequent. That is why they are diseases and not the healthy condition. GWASs are good at finding frequent alleles of strong effect. They have failed to explain significant disease variation because the causative alleles are either rare or"
    I'll respond when you finish this paragraph.
    http://www.sciencedaily.com/releases/2005/02/050212185123.htm

    jtwitten
    ...extremely small effect. 

    "Other traits, however, are much more complex than skin color. Physical differences which are determined not by one, but many different genetic variants, are unlikely to split neatly by population. Intelligence is probably one of the most complex traits humans possess. It is almost certainly affected by variants in many different genes, and many of those genes have other important functions in the body. That means this: two different human populations could have easily developed differences in skin color between them, but differences in intelligence would have been extremely hard to develop, by chance or by natural selection."

    This is completely wrong.
    Apparently, you no longer believe that Michael was completely wrong.  It appears that you are asserting that intelligence is a highly complex trait, more so than say skin color.  This is why adverbs are dangerous. 

    If, however, you are saying that intelligence is determined by many alleles of extremely small effect, then it is even less likely that selection has acted on those loci.  The efficiency of selection is related to the inverse of the effective population size.  As a result, loci with alleles of extremely small effect size behave as if they are neutral.
    "The mutation rate in humans is not of a scale to support an assumption that additional variation is being added at an appreciable rate."

    http://www.anthro.utah.edu/PDFs/accel.pnas.smallpdf.pdf
    This paper says nothing about the scale of the mutation rate or the distribution of effect sizes.  Selection is more efficient in large populations, but we are talking about the addition of new variance.  Their model of human population growth is not consistent with any I have seen previously (not evidence in of itself).  Furthermore, many of their conclusions (high substitution rate, LD blocks) would also be consistent with the hypothesis that the human population was bottlenecked.
    *Human families have mean IQs substantially above and below the means of continental ancestry groups in good environments,* it isn't a theoretical question. I gave an example in which selective pressure was quite weak over a number of generations, but observed group differences could be replicated in a *single generation* by picking founders for a new population 2 SD from the American mean. Your argument is completely irrelevant.
    You have changed your argument.  My argument was addressing your inaccurate assumption of a linear response to selection over multiple generations as a method to demonstrate that weak selection could over time could manifest itself in large, final population mean differences.  Those hand picked founders would effectively be stuck at that mean.  They would not be able to respond to selection of any strength without addition of new variance of a magnitude that could be acted on by selection.

    "Apparently, you no longer believe that Michael was completely wrong. It appears that you are asserting that intelligence is a highly complex trait, more so than say skin color. This is why adverbs are dangerous."

    Yes, if one were to take that phrase without considering the constraint on my meaning provided by the following discussion, it would be unclear that I was referring to a particular part of the quoted text. Since it was also needlessly rude, I hereby retract it.

    " If, however, you are saying that intelligence is determined by many alleles of extremely small effect, then it is even less likely that selection has acted on those loci. The efficiency of selection is related to the inverse of the effective population size. As a result, loci with alleles of extremely small effect size behave as if they are neutral."

    When I said very small effect size I was using the phrase to refer to the percentage of IQ variation explained by the variant (as in this paper: http://www.sciencedirect.com/science?_ob=ArticleURL&_udi=B6W4M-4J84SSW-1...) not the change in an IQ in an individual with the variant. The potential distributions of alleles are constrained by empirical results: the GWAS on the one hand and the empirical facts about IQ in families, including the variance in the IQs of twin and non-twin siblings, on the other.

    jtwitten
    The only situation in which I can imagine this,
    When I said very small effect size I was using the phrase to refer to the percentage of IQ variation explained by the variant (as in this paper: http://www.sciencedirect.com/science?_ob=ArticleURL&_udi=B6W4M-4J84SSW-1...) not the change in an IQ in an individual with the variant.
    happening would be if the variant was incompletely penetrant.  My point about alleles of small effect behaving as if neutral remains valid.  Incomplete penetrance reduces the average effect of the allele in any given individual.  It is the relative fitness effect of an allele relative to the other allele at the same locus that matters.
    Low frequency.

    jtwitten
    Please explain how low frequency (below the GWAS detection level) and larger effects would result in average population level differences.  Having slightly more extreme individuals in one group does not constitute a difference based on race (i.e., gives me no predictive power).  Those individuals could affect the mean, but would not affect the median.
    "You have changed your argument."

    I presented an example to draw your attention to the small amount of selection involved. Group IQ distributions greatly overlap, e.g. in the U.S. ~1/6th of African-Americans have higher IQs than the median WASP. Response to selection declines as more selection is applied, but it is largely linear over short intervals.

    "My argument was addressing your inaccurate assumption of a linear response to selection over multiple generations"

    Human populations have a mix of alleles with positive, negative, and more complex effects on IQ phenotypes. We could tabulate allele frequencies for each subset of the population with a particular measured IQ score, and we know empirical heritabilities for families with different IQ levels. Within the normal human range, those heritabilities remain high. So, if we apply selection that shifts the distribution of alleles to more closely mimic IQ 105 people than IQ 95 people, we are shifting towards an allele distribution for which we already know heritability is high. Thus, it's perfectly reasonable to project basically linear response over a modest range (we're considering differences of at most a SD or two).

    "as a method to demonstrate that weak selection could over time could manifest itself in large, final population mean differences. Those hand picked founders would effectively be stuck at that mean. They would not be able to respond to selection of any strength without addition of new variance of a magnitude that could be acted on by selection."
    Your claim implies erroneous empirical predictions about heritability and variance of IQ in high-IQ populations.

    Theoretically, consider a normally distributed trait where most of the variation within the population results from the small effects of positive and negative alleles at different loci. The trait value for an individual is equal to X (the sum of positive alleles) minus Y(the sum of negative alleles). Then, the more polygenic the trait, the greater the disproportion between (X minus Y) and the absolute value of (X plus Y).

    Are you related to Ed Witten?

    jtwitten
    Assumptions: Intelligence is a quantitative trait that varies within populations and that a portion of that variation is caused by genetic variation.

    H0: There are no significant differences in the distribution of intelligence between human populations.

    H1: There are significant differences in the distribution of intelligence between human populations defined by race.
     
    Disproving the null hypothesis (this is a true null as it assumes nothing other than random effects) would require the following:
    • Demonstration of the ability to genetically identify populations called "race" in a rigorous fashion
    • Measure intelligence in a manner to correct for environmental factors
      • Demonstrate that the environmental factors addressed represent the complete set of environmental factors contributing to the environmental variance of intelligence
    • Demonstrate that those genetically defined groups have distributions of intelligence (corrected for environmental factors)  that are significantly different from each other  
    The next step would be to then demonstrate that the distribution of genetic variation controlling intelligence is different between races.  Arguments about selection are tertiary.  The null hypothesis must be disproved before any hypotheses can be put forward about selection (in this case drift would be the null).  While it is perfectly plausible that selection could act on human intelligence, this would further require either that the selection pressure on different groups be different or the initial variation in those populations was significantly different.

    I am assuming that you are referring to the prominent string theorist.  I am related to Edward Witten inasmuch as we (as can all American Wittens of whom I am aware) trace patrilineal heritage back to the same settlers in Virginia.  Chronologically, it is quite possible that I am more closely related to Jason Witten the tight end of the Dallas Cowboys.  The relations, however, are so distant that it is quite probable that I am equally genetically identical to all the other commenters on this thread.
    "H0: There are no significant differences in the distribution of intelligence between human populations.

    H1: There are significant differences in the distribution of intelligence between human populations defined by race."

    Just use the psychometric terms rather than 'intelligence,' they're well defined, well-validated, and matter for the explanation of gaps in social outcomes. I'm talking about the general factor, g, that emerges from principal component analysis of IQ tests, powerfully predicts real-world educational and job outcomes, and can be predicted with an r^2 of 0.45 using an MRI scan, not a nebulous 'intelligence.'

    Also, your hypotheses are misstated to refer to phenotypes, and group phenotypic differences (in both IQ and neuroanatomy) are already the scientific consensus, it's the causation that's controversial. Given that we know that there are ubiquitous phenotypic group differences in countries around the world, differences which are greatest on IQ measures that are most heritable within continental ancestry groups, differences in extremely heritable brain mass and volume that are related to IQ in the same way within different groups, behavioral differences that manifest within minutes of birth, that different groups regress to different means, that family socioeconomic status and access to education don't explain the differences, it's not rational to assign equal Bayesian credence to H1 and H0 (if those were modified to refer to genotypes). H1 has much more support than H0, and arbitrarily privileging a null of your choosing doesn't change that.

    "Demonstration of the ability to genetically identify populations called "race" in a rigorous fashion"

    The continental ancestry groups picked out by Risch, et al. We criticize social disparities (educational, income, occupational, criminal) based on self-reported 'race' or the visual assessment of government employees such as census workers, and the continental ancestry groups are very closely related, such that if it is shown that differences between African and European continental ancestry groups have a genetic component of a certain magnitude, the finding would largely apply to disparities as measured by self-identified 'race' in the U.S.
    http://genomebiology.com/2002/3/7/comment/2007

    "The next step would be to then demonstrate that the distribution of genetic variation controlling intelligence is different between races. "

    If you map genetic variants affecting psychometric 'g' within different groups, and find that variants that predict 'g' equally well within the different groups are differently distributed, and systematically distributed in accord with the phenotypic distributions, and/or the MALD studies mentioned above show positive results, you've done it. You don't even need a complete inventory of variants affecting 'g' or working memory, because additional independent variants give you progressively more evidence about the overall distribution.

    "Measure intelligence in a manner to correct for environmental factors"

    Adoption studies, neurological assessments, admixture studies, etc. We do know that the gaps aren't substantially due to family SES, unequal access to education, or the like.

    "The null hypothesis must be disproved before any hypotheses can be put forward about selection (in this case drift would be the null)."

    Have you read the Natural History of Ashkenazi Intelligence paper linked to above? You can show selection in a number of functionally related mutations connected to increased intelligence (such clustering can't be explained by a bottleneck) directly. You could test it in weeks if you could get funding and approval to assess IQs in heterozygotes. Likewise MALD on African-Americans, as discussed above, which has already been done for hypertension.

    jtwitten
    There is no arbitrary weighting.  I have stated a formal null hypothesis in that the hypothesis is the expectation if groups only differ from each other by chance.  I have also laid out the things that must be done in order to test whether this null hypothesis can be rejected.  First and foremost, populations must be defined non-arbitrarily via their genetic markers, which has not been demonstrated.  This is a very different problem from demonstrating that alleles differe in frequency from population to population.
    " First and foremost, populations must be defined non-arbitrarily via their genetic markers, which has not been demonstrated. This is a very different problem from demonstrating that alleles differe in frequency from population to population."

    The U.S. Census collects self-reported 'race.' The social controversy is about possibly differing distributions of cognition-affecting alleles between the groups picked out by those categories, the categories we use to talk about health disparities, the 'test-score gap, 'to direct affirmative action and minority contracting, etc. The standard dogma is that disparities in IQ scores between members of those self-reported and government-recognized categories are not caused by genetic variants affecting brain development. If self-reported race maps to systematically different genotypic dispositions to IQ, that dogma is falsified. Continental ancestry, which can be easily and quantitatively assessed using microarrays, is very unevenly distributed across social 'races' so a showing of relevant genetic differences between continental ancestry groups would also show differences between the social and self-report categories that our society uses to condemn racial disparities in education and the like.

    "There is no arbitrary weighting. I have stated a formal null hypothesis in that the hypothesis is the expectation if groups only differ from each other by chance."
    Actually, you stated a hypothesis about phenotypes, which has already been established. We see large consistent phenotypic differences in highly heritable traits and neuroanatomical features, between populations that have been quite endogamous in different environments while undergoing strong selection for thousands of years. We might hypothesize that 80% of observed gaps are due to genetic factors, 60%, 40%. 20%, or 0% (with an extremely powerful environmental factor that no one has been able to isolate over decades of research causing differences in neuroanatomy and g). We should distribute Bayesian credence across these hypotheses, and then update based on further information.

    Your approach is to privilege 0% as the null, rather than a number derived from experience with other variation in heritable traits between populations, and declare that the null can only be rejected when various claims are known with complete certainty through limited subsets of evidence. In other words, to throw away large bodies of evidence that we should update on.

    jtwitten
    The evidence that you continue to bludgeon us with does not demonstrate that you can genetically define race and, as a result, you cannot state that there are significant differences between racial groups, let alone precisely put people into those groups.

    The discussion is about genes and race.  Any definition of race that relies on anything but genetics (e.g., geography, self-reporting, etc.) is irrelevant to this discussion.  General statements about characteristics about populations as a whole are completely inadequate, since we need to be able to put individuals in the correct group (since we are comparing distributions).

    My statement of a null is the expected result from chance.  Prior data has nothing to do with the formal statement of statistical hypotheses.   We are not evaluating two equal hypotheses and deciding which one has the better evidence.

    Again, the plausibility of selection is still irrelevant.
    Josh,

    Three things:

    1. Should people who want to perform the experiments I mentioned above (testing heterozygotes for the Ashkenazi disease alleles, MALD for IQ on admixed populations) be censored, harassed, and interfered with to prevent the research?

    2. Does the end of preventing such research justify deception and misrepresentation of scientific literature?

    3. If the differences in very heritable brain mass and fine-grained neuroanatomy between continental groups are unrelated to IQ (although differences in these things predict IQ variation within each group just as well, and equalizing IQ equalizes brain size), and some independent environmental factor X is responsible for score differences, doesn't that imply that, sans factor X, people of African continental ancestry have much more efficient brains than those of European or Asian continental ancestry? How is it that people in China are saddled with all of that metabolically expensive wasted brain mass? And if evolution could make African brains so much more efficient than others, why couldn't it have produced meaningful group differences?

    jtwitten
    1.  Censorship and harassment are never warranted.  Criticism is always warranted.  I do not know enough (and, honestly, do not have time to learn enough) about the specific case you refer to in order to make an educated judgement.

    2.  Misrepresentation of scientific literature is never permissible.   Intelligent minds, however, are permitted to interpret data  differently and argue about it.

    3.  First, the concept of genetically distinct groups is illusory.  Any dividing lines will be arbitrary.  You appear very devoted to a selectionist narrative.  Personally, I have found the argument of Michael Lynch that we should resist the urge to explain all results of evolution via selection compelling.  Therefore, I do not feel it is necessary to concoct selection narratives to explain these apparent dilemma.   I am also aware that the validity of the literature you have cited from JP Rushton has been hotly debated, but I do not have access to the body of relevant literature from this location in order to reach conclusions about this debate.

    Lost the bracketed text in an edit:
    Y(the sum of negative alleles) [+Z, with frequencies such that on average X=Y].

    ...good God by the time I got down to the bottom of this major verbiage assualt I forgot why I stopped in here! One if not ALL of you guys needs to get a date...somebody elbow Michael and wake him up...holy crap!

    jtwitten
    I'm married with children.  I have nothing better to do. :)
    adaptivecomplexity
    I prefer to keep an open comment policy around here, as long as things don't degenerate to just name-calling.

    But I don't feel the need to get deeply involved in every discussion once I've said what I have to say.
    Mike
    Great, so now we've found that we are all part of the same 'human race', can we please stop using the term "race" to classify ourselves. Since being 'black' means nothing and 'white' means nothing (genetically speaking), let's let go of the stereotypical terms that incite prejudice and division. The 'black' entertainment network should drop the term. The Congressional 'black' caucus.?. Come on. After all these years of affirmative action and it's divisive breed of reverse discrimination, Obama is President. Can we be done now? The forms we all fill out allowing ourselves to be boxed up into "races" of 'hispanic', 'black', 'white' and 'native' are propagating prejudice not equalizing it. I refuse to be placed in a faction. Until we refuse to be classified by how much melanin we have, we can never see true equality. We should let go of the prefixes before American...Korean-American, African-American, Hispanic-American. If we did that, we'd be a lot closer to ending "racism"

    So who am I? RACE: Human Ethnicity: American

    adaptivecomplexity
    You bring up a good point. As I noted in the original piece, race is really a sloppy term, and a better scientific one is ancestral population. When we discuss something like race and genetics, we really have to talk about populations, not race.

    Of course ancestral population often doesn't correlate with nationality, and a major point of my original post is that the human species is genetically very uniform. By the standards of the animal world, our differences within our species are minor.
    Mike
    Gerhard Adam
    As a latecomer to this debate, I'm still stuck on hearing what a coherent definition of "intelligence" is (let alone something as vague as Intelligence Quotient).

    Does someone have such a definition?
    Mundus vult decipi
    adaptivecomplexity
    Others can answer this better than I can, but there are several points to keep in mind:

    - For the purposes of measurement, psychologists have to pick something standardized and measurable, like IQ or g. That does not mean that IQ captures everything  that we really mean when we think of intelligence, but developing some kind of measurement is important for these studies.

    - Various people have offered multi-faceted definitions of intelligence (which aren't so easy to quantitatively measure). If you're interested in the subject, a popular book is Howard Gardner's book, but I don't know what his peers in the field think of his ideas.
    Mike
    Gerhard Adam
    I understand what you're saying, but my point is actually that despite decades of effort we don't have a good working definition of what we actually mean.

    The psychological interpretation, in the kindest terms, is hopelessly flawed.  It seems that any real definition of intelligence must be applicable to all species and measurable.  Anything less is automatically suspect.  If this isn't possible, then it suggests that we really don't know what we mean when we use the term.

    This becomes even more important when we consider education, experience, and cultural influences on our concept of intelligence, then the data becomes even more skewed and less useful.

    I guess from my perspective the word "intelligence" means nothing if it can't be as accurate when applied to my dogs as to other humans.  We wouldn't accept such vagueness in terms like "heat" or "energy", so why should it be acceptable when applied to humans.  In short, my conclusion is that IQ is a measure against a quantity that we can't define and consequently it's useless regardless of whatever apparent results it seems to show.
    Mundus vult decipi
    adaptivecomplexity
    I guess from my perspective the word "intelligence" means nothing if it can't be as accurate when applied to my dogs as to other humans.
    That's an interesting point, and my intuition suggests the same - that features of something called intelligence should apply to chimps and dolphins and dogs. But there perhaps a valid reason for human exceptionalism - I think a genuine measure of what we intuitively mean would also put the least intelligent humans (maybe with the exception of some extreme disease cases) well above the intelligence of the next runner-up species, like chimps.
    Mike
    Gerhard Adam
     "But there perhaps a valid reason for human exceptionalism - I think a
    genuine measure of what we intuitively mean would also put the least
    intelligent humans (maybe with the exception of some extreme disease
    cases) well above the intelligence of the next runner-up species, like
    chimps."

    That's where I have a problem, because when we can't define it, we can't know whether it really is exceptional or unique versus simply being biased.

    An excellent argument can be made that chimps have much greater levels of expertise than typical humans when it comes to survival in the jungles.  Does this render them more intelligent?  Clearly a test that measured results based on that capability would should markedly different results from tests that are biased towards human experience.

    We see the same issues with humans when an individual can be gifted in one area and and significantly less functional in others.  Does intelligence somehow average out? 

    I think that when we are forced to analyze intelligence in other species, we begin to see that what we call intelligence isn't necessarily so unique.  Perhaps a better measure might be "ability to abstract", or other criteria, but by labeling it intelligence we attempt to create uniqueness at the expense of everything else.

    Another point relates to the comment that most humans are well above the intelligence of the next runner up species ....

    I would dispute that statement by suggesting that a minimal level of intelligence would ensure that the application of brain power (over simple "instinct") would enhance the survivability of an animal.  If an individual cannot survive using their own brain powers, then it would be hard to argue that they are somehow more intelligent anyhow.
    Mundus vult decipi
    ...IQ stats seem to be important only to those who score well with it, and it's merits are dubious at best I suspect. I know that some with high number IQ's have a very hard time walking through the Mall and other such places where there is too much stimulous, so certain intellegence appear to be compartmentalized. And even more amazing than that is that today I agree with Gerhard...there is no clear understanding as to the probable accuracy of genuine intellengence. Josh...you are a brilliant dude, but I think it's time you and the wife get away for a spell... ;-)

    Gerhard Adam
    I agree.  I've always thought it would be a fascinating psychological experiment to give children an IQ test and then tell them that their score was 20-30 points higher than it actually was.   My point is whether expectations could ultimately result in actual scores being higher by setting the confidence level higher.
    Mundus vult decipi
    Michael....where is the box you can click for being notified??? Why is it gone? What are you up too?

    "Various people have offered multi-faceted definitions of intelligence (which aren't so easy to quantitatively measure). If you're interested in the subject, a popular book is Howard Gardner's book, but I don't know what his peers in the field think of his ideas."

    I think the problem is he hasn't got much empirical data to back his idea up. I think the Snyderman/Rothman survery of academics in the field showed that the tests measured ability to process information. IQ tests seem to be very effective in predicting academic performance in different areas. For instance, see Dan Seligman's book "A Question of Intelligence". He documents that East Asians tend to score well above average on the non-verbal component of tests and this is consistent with their higher than average performance on math/science subjects.

    Linda Gottfredson has an interesting paper discussing logical fallacies used against the tests:

    http://www.udel.edu/educ/gottfredson/reprints/2008logical-fallacies.pdf

    Nicholas Horton
    Josh Witten said:
    This is why adverbs are dangerous.
    I was once mugged by an adverb.  It was totally horrible!

    It's true that defining "intelligence" is notoriously difficult.  In fact, I'd say it is largely impossible beyond some heuristic know-it-when-I-see-it kind of sloppy definition.  We tend to mean that someone who is intelligent has a mental capacity that is greater than someone who is not intelligent.  But, musical ability originates in the brain.  So, is a musician more "intelligent" at music than a non-musician?  Athletic ability is (to a large degree) neurological.  Could we make an argument that athletes are more athletically-intelligent than non-athletes?  Maybe.  But, we're getting into fuzzy territory. 

    We are not all equal.  But, it's hard to tell us apart.
    Gerhard Adam
    Whatever definition you arrive at becomes more difficult when you attempt to apply it to animals.  When we remove the human element of academic achievement (specifically), the concept of an innate intelligence becomes more nebulous.
    Mundus vult decipi
    Michael, I stumbled on your site whilst searching for information on the "Lewontin fallacy" which is pushed hard by racists, who think they have a good grasp of genetics, all over the net. Thank you very much for your time and well written, accessible, and factual articles. They're much needed.

    Bravo, questa splendida frase appena inciso

    Todo puede ser