I see that a story on AlterNet, by Anneli Rufus, entitled, “IQ Blackout: Why Did Studying Intelligence Become Taboo?” has been making the rounds. Popular Daily Beast blogger, Andrew Sullivan, picked up on the piece, prompting a back and forth exchange between him and Atlantic blogger, Ta-Nehisi Coates. The issue of race and intelligence, or more specifically IQ, is something that seems to reappear over and over despite being a remarkably unproductive topic. It is an issue of considerable cultural significance because of its near ubiquitous relevance – nearly everyone has some general opinion or thoughts on the topic – and therefore something I teach a unit on in my Race and Human Biological Variation course. Before getting into my own thoughts on the issue, it is worth looking at what the three writers above say, as it features many of the points common to discussions on the topic.
The AlterNet story begins:
Scholars used to avidly study human intelligence. They measured cranial capacity. They administered IQ tests. They sought to define what intelligence was and who had more or less of it and why.
These days, not so much.
The story might as well end there. It is not a bad thing “scholars” don’t spend a lot of time on these topics these days….there is not much to be said or gained from looking into them (see below). Most of the remainder of the article consists of quotes from Dennis Garlick, a UCLA postdoc and author of Intelligence and the Brain (Aesop 2010), largely focusing on the work of former Berkeley psychologist, Arthur Jensen.
In his pick up of the story, Andrew Sullivan comments:
The right response to unsettling data is to probe, experiment and attempt to disprove them – not to run away in racial panic.
This comment ignores a reality that, in fact, there was a huge response to Jensen’s work. The response was not to look deeper into the research begun by Jensen, but instead to reject the nature of the research and point out the flaws of it on first principles.
Ta-Nehisi Coates, responding to his former colleague (and boss, I believe), makes an astute observation:
With that said, Andrew’s ahistorical approach to race and intelligence has always amazed. The contention, for instance, that “research is not about helping people; it’s about finding out stuff,” may well be true in some limited sense. But it’s never been true, in any sense, of race and intelligence. In the 19th century helping out white people (however that is defined) was very much the point of intelligence research. Into the early 20th century, the rise of eugenics was equally linked the field to the advancement of “people.” Even the intelligence theorists whom Andrew, himself, has advanced over the years are motivated by a desire to presumably help people, if only in the form of deciding how a society should expend its limited resources.
This prompts a reply from Sullivan:
This is roughly the quality of the responses to The Bell Curve when it came out, but with added inaccuracy. No one is arguing that “that black people are dumber than white,” just that the distribution of IQ is slightly different among different racial populations, and these differences also hold true for all broad racial groups
Finally (or not…I’m sure this will go on in some form or another), Coates responds:
On the broad question–Should researchers be free to explore the nexus of race, IQ and intelligence?—Andrew and I are in harmony. Onward, indeed. Where we differ is the following: Andrew, like most conservatives who write about race, is more concerned with a vague p.c. egalitarianism than the forces that birthed such things. (Unlike “political correctness” those forces can actually be quantified, and their impact demonstrated.)
That his contention has long been linked to one of the ugliest strains of American thought, that it continues to be linked to actual white supremacists, is not particularly troublesome to Andrew. But that others might find it troublesome is deeply distressing. I don’t charge Andrew with defending slavery or sterilization. I charge him with bumbling through the ICU, tinkering with machinery, and wondering why everyone is so uptight and stuff.
People are biologically different from each other. Part of this biological difference is certainly both genetic and developmental differences that affect how our brain processes and utilizes information gathered from the world we live in. Additionally, these biological differences are almost certainly structured by evolutionary and demographic forces acting throughout the history of human evolution. However, as a plethora of work has shown, these differences are greater within (even appropriately categorized) groups than between them. But the larger point, and the flaw with Jensen’s work, is there is no really good way to measure this, or indeed, make anything meaningful out of it.
If you ask people to define intelligence, you almost inevitably get some assortment of statements about “how smart people are” or the “ability of people to do things well.” A criticism of IQ testing, of course, is that it is inherently context specific. The reality is probably that intelligence itself is context specific. I was thinking of this earlier today in writing about the use of fire in the archaeological record. One of our limitations in studying paleo-fire today is that we are somewhat limited by what we know about how fire works and how to control it. Just because we can step in the backyard and turn on a gas-powered grill doesn’t mean we have more intelligence with respect to fire. A couple years ago I sat amazed in a talk listening to an archaeologist describe how populations in Southern Africa more than 60,000 years ago developed techniques involving slow heating of lithic material under controlled temperatures (more than 36 hours!) in order to achieve proper lithic properties to allow for tool production. I would have very little idea how to do that. This might be an example of knowledge versus intelligence, but I think it is an example that reflects the broader issue of problem solving. We are almost certainly “smarter” in nearly every way that matters than our ancestors 60,000 years ago….but don’t expect to win in a head-to-head competition on their soil. Intelligence is inherently context specific.
And if your measure of intelligence is something more basal about the operations of the brain at the cellular level…people are doing that research. The rise of neurosciences is one of the biggest changes in the landscape of the biological sciences over the past 50 years. But even with those studies, intelligence as it matters to people is largely what we do in action not what can be measured in theory. And what we do in action is largely structured by issues above anything about our basic biology.
“these differences are greater within (even appropriately categorized) groups than between them. ”
Differences in strength and height are greater within male and female groups than between them, therefore men are not taller and stronger than women on average. Differences in athletic ability within male and female groups are greater than between them, therefore women are as athletic as men.
No…obviously men are larger and stronger than women on average. Globally, men are about 6-8% larger on average in total body mass, although the separation between male and female mean values varies somewhat between populations. But I think your example actually illustrates my argument very nicely. First, male and female are vastly more discrete biological categories than racial categories (even though categories of sex are probably not as discrete as you might think). So characterizing differences in size or strength are actually more of an acceptable pursuit because the basic categorizations have some validity. Racial groups simply don’t discriminate into categories nearly as nicely, for the obvious reason that ancestry is a more complex and long-term process than the segregation of male and female gametes during conception. Even if we discount the problem with the validity of the underlying categories, your example illustrates a few more nice points about race and IQ. Despite the fairly significant separation between mean male and female body mass values, correctly identifying any given individual as male or female based solely on body mass has a huge error associated with it. Male and female size distributions, despite their mean averages, overlap throughout nearly their entire distribution. As such, the measure (body mass) is not a particularly good way of describing male or females at an individual level. Like IQ, body mass is not a constant measure throughout life. Studies show that test taking “training” can lead to substantial changes in reported IQ scores for individuals, even within a very short span of time. Body mass, meanwhile, obviously changes throughout life as well (the recent Thanksgiving holiday probably increased most people a bit). So while adult body mass dimorphism is about 6-8% on average, if you took the same measure on 8-year olds you would probably find that males are smaller than females by about 2-3%. So now we have two problems…our categories (race) are ill-defined (because they, in fact, lack strong biological separation) and our measurement is not constant. This would not be a problem if we solely used IQ to look at individuals at a given place and time. Which, as it happens, is actually what Binet initially developed standardized intelligence testing for. The problem is that despite the lack of traction of races as a category, we are very prone to making assumptions about people based on perceived race (gender works in quite similar ways). So we are very prone to assuming that an average population/group difference (ignoring the problems in measurement and categorization) represents an accurate assessment of any given individual on the street. Which it doesn’t. The variation between you and your brother/sister at any given time is likely to be as big as the difference between you and someone belong to a different perceived racial group.
Race isn’t a discrete category, but human beings cluster into discernable races. You go from denying group differences in IQ, to worrying “assumptions” will be made about specific individuals. Such worries are not a valid reason to suppress the truths.
“test taking “training” can lead to substantial changes in reported IQ scores for individuals”
Training can’t close the racial gaps. If it could, this subject wouldn’t be controversial.
“characterizing differences in size or strength are actually more of an acceptable pursuit because”
Because the prevailing political ideology has not set up taboos around those characteristics.
“ancestry is a more complex and long-term process”
Which is why it’s absurd to pretend all populations evolved the exact same mental capacities.
“The variation between you and your brother/sister at any given time is likely to be as big as the difference between you and someone belong to a different perceived racial group.”
Variation in what?
“Race isn’t a discrete category, but human beings cluster into discernable races.”
Humans cluster in lots of ways, but those clusters don’t match up with notions of race that predominate. The reason is because of the complex pattern of biological/genetic ancestry that shapes patterns of human population variation. Using genetic data you can identify significant clustering of human genetic variation, but it occurs either at scales above the level we typically associate with race (i.e. continental variation) or below the level we associate with race (local population). But even these clusters are not good working units for analysis of this kind because they represent unique and differing historical and evolutionary processes spanning different time depths and encompassing more or less intra-cluster variation. The whole issue is then made more problematic by studies that show that our brain is predisposed to divide people into groups, but does so on the basis of very limited amounts of information in our learning environment. So we are basically set up to form “race-like” groups despite the failure of such groups to conform to a biological reality. It is these cultural perceptions of race (which vary tremendously across time and space) that then play a major role in shaping how we interact with the social world around us. Clarence Gravelee’s article from the American Journal of Physical Anthropology a year or so ago, “How race becomes biology,” is worth reading for a fuller exegesis of this point. We are different, but we race does a bad job of describing the evolutionary forces that have shaped patterns of biological difference. Race does a better job, taken in given cultural context, of describing how we move through a highly structured social world.
“Training can’t close the racial gaps. If it could, this subject wouldn’t be controversial.”
No, but training shows that “IQ” is not some innate entity. It is a measure of a given person in a given context. It is not a good way of assessing group characteristics, even if we ignore the problem of categorization of the groups.
““characterizing differences in size or strength are actually more of an acceptable pursuit…Because the prevailing political ideology has not set up taboos around those characteristics.”
No, it is more acceptable because both the groups are more salient (biological males and females) and the measure is less subjective.
“ancestry is a more complex and long-term process”…Which is why it’s absurd to pretend all populations evolved the exact same mental capacities.
No one is pretending that. I point out in the original post that “people are biologically different from one another.” But that, in and of itself, is not sufficient to extend that argument to characterize groups of people (again, ignoring the challenge in defining your groups). Human cognition is an extremely plastic phenotype, meaning it is highly impacted by environmental influences during development. Yes, people have innate tendencies towards what kind of information the brain processes and how it processes them, but those differences in innate ability are far weaker than the differences associated with environmental effects.
“those clusters don’t match up with notions of race that predominate.”
Actually they do. For example see Genetic structure of human populations, Rosenberg NA, Pritchard JK, Weber JL, Cann HM, Kidd KK, et al.(2002), or Genetic structure, self-identified race/ethnicity, and confounding in case-control association studies, Tang H, Quertermous T, Rodriguez B, Kardia SLR, Zhu XF, et al. (2005).
“race does a bad job of describing the evolutionary forces that have shaped patterns of biological difference”
Racial differences are the inevitable result of evolution. The usefulness of racial categories depends on the situation.
“those differences in innate ability are far weaker than the differences associated with environmental effects.”
That’s contradicted by multiple adoption studies.
In response to Django’s latest comment:
No…those clusters don’t match racial categorizations. Yes, it is possible to identify clusters of genetic variation (and you can identify as many as you set the algorithm to identify, really). But no, those clusters don’t represent the racial categories we actually use in our daily lives. The obvious test of this is to simply recognize that while any given individual has the same genetic variation wherever they are, they are very unlikely to be racially categorized in the same way where they are. Additionally, it is worth pointing out that the variation that allows us to establish those clusters is really a very miniscule portion of the entire whole (a point I will come back to). But using the Rosenberg et al. 2002 article as an example, the clusters they show are not racial groups. They are intermittently groups defined by ethnicity, language and geography which overlap to varying degrees with “race.”
If you don’t believe me, believe Noah Rosenberg. Here he is in a 2011 Human Biology paper:
“Most alleles are widely distributed, the fraction of alleles private to
individual regions is small, most populations contain most of the alleles present
in the human population, and the mean genetic difference for two individuals
from the same population is almost as large as that for two individuals chosen
from any two populations. We will see, however, that in the accumulation of
small amounts of allele frequency variation across many loci, it is possible to
make inferences about individual genetic ancestry from genetic markers.”
….
“From these results, we can observe that despite the genetic similarity among
populations suggested by the answers to questions #1-#4, the accumulation of
information across a large number of genetic markers can be used to subdivide
individuals into clusters that correspond largely to geographic regions.”
On a broader theoretical level, we know that one of the evolutionary forces that plays a major role in shaping the human genome is gene flow. Populations have been exchanging genes and admixing for several hundred thousand years at a minimum. Cognitive functioning is highly selected for in modern contexts and likely has been for the duration of recent human evolutionary time. The widespread nature of admixture means that major genes associated with cognitive functioning are likely widespread in the human species. There are likely many many rare variants that also impact cognitive functioning, but they are highly most likely distributed as rare variants in individuals within a given geographic-historic population, and not likely to characterize “racial” groups.
Getting back to Andrew Sullivan’s initial complaint…people actual do study this. They just don’t study it in racialized terms because that is the wrong way to study it!
“hose clusters don’t match racial categorizations…those clusters don’t represent the racial categories we actually use in our daily lives.”
That statement is contradicted by multiple peer reviewed studies.
“the variation that allows us to establish those clusters is really a very miniscule portion of the entire whole”
Down syndrome is even more miniscule.
“Populations have been exchanging genes and admixing for several hundred thousand years at a minimum.”
There’s been significant genetic exchange and admixture between Japanese and Africans? That’s news. You should publish something on it. Sub-Saharan Africans don’t have any Neanderthal genes.
“They just don’t study it in racialized terms because that is the wrong way to study it!”
They don’t study it in racial terms for political reasons.
I don’t see any point in additional comments.
I may well be wrong, but I believe that most (?) genetic studies seeking to divide up the human population are based on neutral markers (i.e. Hunley et al. 2009 -http://bit.ly/sLMru6), i.e. ones that are believed not to affect function and fitness. So even if genetic studies did vindicate traditional conceptions of race, they still wouldn’t really support linking the genetic ‘essence’ of a ‘race’ with IQ. The role of the (social) environment cannot be underplayed in discussions of ‘intelligence’ and other traits.
Thank you, Adam, for writing this and for the very informed responses in the comment section. It’s quite interesting that the latest line on race-and-IQ is all about innuendo and accusations of politically-correct suppressions rather than any new data.
Zachary Cofran is correct to note latest genetic research does not confirm the reasonings Django promotes here. The special issue of the American Journal of Physical Anthropology, Race Reconciled examines genetic, forensic, and other biological evidence to conclude “Race is not an accurate or productive way to describe human biological variation” (Edgar and Hunley 2009:2).
Nevertheless, many anthropologists could use more familiarity with this latest work in order to contest the kinds of studies that emerged especially 2000-2005. I write about this in a post called Anthropology on Race which is inspired by your work. Thank you again.
First, male and female are vastly more discrete biological categories than racial categories (even though categories of sex are probably not as discrete as you might think)
Actually , commenting on the Tang et al. study Django linked to above, Neil Risch noted that they found genetic clusters to correspond more reliably (99.9%) to racial/ethnic self-identification than markers on the X chromosome in the same sample corresponded to identification as male or female. This would suggest that in America races are more discrete biological categories than male and female.
While global genetic variation in humans may be clinal, genetic variation in the US really isn’t. America was settled within the last few centuries by people from different corners of the earth, and each group was highly genetically distinguishable from others in the beginning. America’s population is not a random sampling of humanity. There’s been some mixing between these gene pools, but the basic ancient divisions are still the major determinant of genetic structure in America, with the result that self-identified race in America is a robust biological category, as demonstrated by Tang, Risch et al.
IQ tests are highly reliable and valid measures of human capacity. Racial differences in IQ are an important factor influencing racial disparities in many outcomes in America. There’s nothing wrong with the “first principles” of Arthur Jensen’s research on race and IQ — it’s a legitimate and answerable scientific problem, and his more capable and honest opponents (e.g. James Flynn) have no problem admitting this. Research into this topic has been halted because its results challenge prevalent ideological presuppositions.
What Tang et al. show in their study is that they can reliably distinguish their study population into four groups based on 326 microsatellite loci. Of course they can. If you classified a group of people based on a huge set of variables you can classify them into however many discrete groups you would like to. That they chose four self-identified ethnic categories is of no significance. The issue of discrete categorization has less to do with your ability to create those categories than what those categories actually represent. Those 326 loci, of which the groups are discriminating on a fraction of, represent a miniscule portion of the genetic variation encompassed by those groups collectively and individually. It is an overly simplistic graphic, but the one here on the AAA’s RACE site based on Jeff Long’s work is helpful for visualizing this. Just because you can make groups doesn’t make them real, particularly real with respect to another variable of interest.
IQ tests are highly repeatable, I disagree on the notion that they are “valid measures of human capacity.” They can be reliable and repeatable measures of “societal performance,” on some sort of abstract scale, but that incorporates a much broader set of variables than is addressed in IQ testing.
If you classified a group of people based on a huge set of variables you can classify them into however many discrete groups you would like to. That they chose four self-identified ethnic categories is of no significance.
There’s nothing arbitrary about the close correlation between self-identified race/ethnicity and distinct genetic clusters. They both reflect differences in ancient geographic ancestry, and thus differences brought about by evolution.
Populations geographically separated in their evolution are distinct in the sense that the members of each population are genetically overall more similar to all members of their own population than to any member of the other populations. The sum total of heritable variation distinguishes populations, which we may want to call races, from each other. In the case of heritable complex traits influenced by hundreds of genes like general intelligence, allele frequency differences between races mean that the distributions of the trait are highly unlikely to be similar across races.
they can be reliable and repeatable measures of “societal performance,” on some sort of abstract scale
There’s nothing abstract about school success, job performance, health disparities, etc. that are all affected by IQ differences.
that incorporates a much broader set of variables than is addressed in IQ testing
Of course IQ does not measure all the things that are important, but it’s not meant to do so. To criticize IQ in that way is similar to criticizing weighing scales for not telling how tall you are. IQ is regarded as a more important variable than many others because its predictive validity is usually greater and because it can be measured more reliably than e.g. personality differences.
I should have been clearer in my use of “arbitrary”…the perils of blogging on Saturday morning between kids’ sports practices and PTA-mandated cookie-making.
Obviously human evolutionary history has shaped the pattern of genetic variation we see within and between populations. In the specific context of US history (and more broadly New World history), the primary demographic process is the coming together of three relatively distinguished population groups; Europeans, West Coastal Africans, and East Asians (with Native Americans included within this final group). This does mean it is possible, without plumbing too deeply into the depths of genetic variation, to find markers which reliably sort people whose ancestry comes predominately from one of these three regions. You can, this way, create a genetic definition of race by associating race with a certain combination of genetic markers. But this genetic definition is not the working understanding of race in any reasonable context. People identify race in themselves and others based on a whole host of culturally defined characteristics that are extremely malleable. And yes, it is of course possible to find a set of markers that map onto self-identified ethnicity/race, but this definition is essentially an instantaneous one for race, applicable to a given group of people and a given set of genetic markers in a specific time and place. I am sure it would be possible, given the motivation, to do genome-wide SNP screening of Cubs fans and White Sox fans and identify genetic markers that allow you to reliably separate the two self-identifying groups…but that is more of a statistical phenomenon than one with explanatory power. As you say, intelligence is likely to be influenced by hundreds of genes, which means that the underlying genes are likely to be widespread across different perceived racial groups. And yes, you would not necessarily expect the same distributions across groups, but neither would you expect dramatic differences associated with the relatively minor differences in allele frequencies. Alleles with a strong phenotypic effect on cognitive ability are likely to be subject to fairly strong selective coefficients, which means if they have been around for a long time are likely to be widespread, if at different frequencies, across populations. If they are of more recent origin, they are more likely to be present within localized geographic groups, not at the level that we use functionally as representative of race.
And of course there is nothing abstract about school success, but there are lots of different ways of defining success and lots of different paths to success.