“Race” is a Valid Scientific Category
Alexis Delanoir, one of the writers at the blog How To Paint Your Panda, uploaded an article in January talking about “Lewontin’s Fallacy” and race realism. She agrees with race realists who say that Lewontin committed a fallacy. Given the fact that there is more variation within human races than between them, it does not follow that racial classification is insignificant. But she still thinks that Lewontin’s conclusion was correct. Alexis denies race realism. A reader asked TRS to create a response to that article, and I was tasked with writing that response. After emailing back and forth with Alexis for while about what she said in her article, I have done so. So, here it is.What Race Is
Before getting into what Alexis wrote, I want to briefly outline the meaning of, and case for, race being a valid scientific concept. Categories, of course, are just labels man invents for sets of objects he decides to group together. Science is aimed at explaining and predicting the natural world. So a valid scientific category is a grouping of objects that aids in our ability to predict and explain nature. Racial classification groups people together based on ancestry. A person’s race is simply determined by where most of their ancestors lived in the distant past.
As it turns out, grouping people in this way gives us the ability to predict an awful lot about a person. As almost everyone knows, there are differences in observable physical features between the races concerning skin color, cranial morphology, and the like. Some psychological differences are less well known. For instance, some races have larger brains than others and some physically mature faster than others. There are also well established psychological and behavioral differences between the races. For instance, in Race, Evolution, and Behavior, the psychologist Phillipe Rushton documented racial differences in intelligence, criminality, parenting, and sexual behavior. Others have shown that the races differ with respect their to moral senses, financial behavior, and political attitudes. The list could go on. In fact, it is the rare exception, rather than the rule, to find that the races score equally on a behavioral or psychological test.
Race clearly has predictive power. But what about explanatory power? Broadly speaking, racial analysis offers us three mechanisms by which to explain group differences. First, there are genetic differences that exist between people as result of their ancestors living in different breeding populations in the distant past. For example, there are gene variants that are associated with individualism which are much more common in white populations than in non-white populations. Second, differences between races will sometimes exist, or be exaggerated, because of beliefs that each race has about what it means to be a member of their race. For instance, among black Americans being interested in school is considered “acting white” and is looked down upon. These beliefs likely increase the gaps between blacks and non blacks in intelligence and education. Thirdly, differences between the races will sometimes exist, or be exaggerated, because of how members of one race act towards members of another race. For instance, the high unemployment rate among blacks might be caused, in part, by the fact that liberals so persistently convince black people that they aren’t responsible for the outcomes of their own lives. Thus, race has a good deal of explanatory power as well as predictive power.
This explains why race is currently being used in a variety of scientific disciplines. In psychology differences between people are regularly organized and explained in racial terms. Studies of psychological anthropology journals in America suggest that about one in five articles published describing human variation utilize race to do so. In forensics, knowledge about the anatomy and genetics associated with each race has been used to help identify victims and criminals. And in medical science race is commonly used when determining how at risk a person is for a given disease.
On top of all this, there are two features of race that make it uniquely useful. First, it is easily discernible by sight and almost everyone knows which race they are a member of. This makes gathering and using information about race extremely easy. And second, if you look at the whole genome, pairs of people from the same race are almost always more genetically similar than pairs of people from different races. This is important in explaining social behavior because there is a good deal of evidence to suggest that people naturally get along better with people the more genetically similar they are to them.
And thus, grouping people by race can be useful in predicting and explaining data that scientists are interested in. It is already commonly used in many scientific fields and there are some facts which make it uniquely interesting. And so we see that race more than meets the standard for what makes something a valid scientific category.
At least since Darwin’s time, most race realists have claimed that races are not just valid scientific categories but also that the races are separate subspecies. Broadly speaking, a subspecies is a group of populations within a species which differ in some significant way from other populations within the species and which, for the most part, live in different places from the rest of the species. Unfortunately, there is no consensus within taxonomy on how to define exactly what “differ in some significant way” means. Human races clearly pass some standards for significant differentiation and clearly don’t pass others. I don’t see the debate about whether or not races are separate subspecies as being particularly important. Even if one chooses not to view races as subspecies they are still distinct enough to have predictive and explanatory value and should, therefore, be used in science.
A common theme in Alexis’s article is that racial classification is not biologically meaningful because it is a social construct. A lot of race realists misunderstand this claim. When someone says that race is a social construct they are not denying that there are obvious biological differences between Ethiopians, China men, and Swedes. Rather, race denialists are claiming that nature does not compel us to group all Africans south of Egypt together or people in Europe and Northern Africa together. Rather, it is up to humans to decide to group people together in this way as opposed to another way. The groups are, therefore, created by people and influenced by their attitudes and cultural norms. In other words, they are socially constructed.
It is true that race is a social construct: man can decide to group people together in what ever way he wants. And while there are some psychological predispositions that make all people categorize the human variation in a somewhat similar way, variation in racial classification between cultures clearly exists. For example, some Bushmen tribes in Africa think that they and east Asians should be grouped together (because both populations have similar epicanthic folds) and the rest of mankind should belong in a second category. Many anthropologists think that it follows from this that the concept race that modern Europeans came up with is not a valid biological category. In spite of its ubiquity, this is a rather simple mistake.
Whether or not a category was invented by man or nature has no impact on whether or not it can predict and explain scientific data. In fact, subspecies have long been recognized as socially constructed by taxonomists. For instance, consider this quote from the 1700’s by Carl Linnaeus, the founder of modern taxonomy, on the nature of varieties (the primary below-species taxa prior to the rise of the subspecies in the 1840’s). “Species and genera are regarded as always the works of Nature, but varieties are usually owning to culture.” Later writers were even more explicit in their views. Erynst Mayr, probably the most important taxonomist of the 20th century and the inventor of the most popular modern conception of subspecies, called subspecies “a purely subjective” category. He went on to say “the subspecies is merely a strict utilitarian classificatory device for the pigeonholing of population samples.” In fact, when responding to a complaint about the subjectivity of subspecies demarcation biologist J. Tilden wrote“We should, I feel, have a mental reservation that our systems exist more in our mind than in nature. However useful our system may be as a tool, we cannot assume that no other system could be devised to express the same concepts as well or even better. By this line of reasoning, the concept of subspecies should no more be under fire than any other level of classification, since all are equally the products of man’s ingenuity.” And so it is clear that the conviction of anthropologists, that race is not a valid biological concept because it is socially constructed, results from a misunderstanding about the nature of biological categories in general. They are all socially constructed.
Leftists will often claim not only that race was socially constructed, but that racial classification was invented for unscientific reasons related to racism. The division of the world’s population into races originated in the works of naturalists writing before the 20th century. Given that these scientists were only human, it is no doubt true that their political views had some influence on how they thought we should categorize human populations. And there is no doubt that many of them were racist. I think the left often over-estimates how great of an influence these political views had, but the truth is that it doesn’t matter. If a concept is scientifically useful then it is scientifically useful. The motives that people had for initially coming up with racial classification don’t have any impact on that.
What Race Is Not
In my opinion, the greatest flaw in Alexis’s article is that she gets her conception of what race is – not from race realists – but rather from statements denying the reality of race. In particular, she refers to statements by the Human Genome Project, the American Anthropological Association, and the American Association of Physical Anthropologists, to define and, in part, discredit race. I’ve covered some of these statements in more detail elsewhere and argued that they are not only misleading but intentionally dishonest. Here, I just want to look at how the arguments presented in these statements relate to the definitions of race that I and other race realists have given.
First, here’s a quick summary of how race realists other than me have defined race: before the 20th century almost all researchers (i.e Linnaeus, Blumenbach, and Darwin) defined races in terms of clusters of physical traits. E. Hooton, one of the most important physical anthropologists of the early 1900’s and probably the most influential race realist writer of the time, defined races as groups which, due to common descent from populations which were, in the past, geographically isolated, posses differing combinations and distributions of traits. Ernst Mayr, who was not only an eminent taxonomist but also a leader of the Modern Synthesis of biology, defined as a race a set of, mostly, geographically separated breeding populations that differed from one another in some respect that taxonomists thought were significant. Theodosius Dobzhansky, another architect of the Modern Synthesis, defined races as populations that differed in allele frequencies. Steve Sailer, a prominent modern race realist, has defined races as “extended families”, or groups of people who are more closely related to each-other than they are to members of other races. And Michael Levin, a contemporary race realist and philosopher of race, has simply defined races as groups of people who descended from the same part of the world.
None of the statements by any of these organizations address race as defined by any major race realist in modern history. Instead, the Human Genome Project and the Association of American Physical Anthropologists decided to issue statements against races where race is defined as a group of people who all share a gene that no one outside of the race posseses, or a group of people who are all genetically the same. The statement by the AAA did mention race as defined by pre 20th century anthropologists. Still, its arguments were mostly dishonest. And they didn’t mention any of the conceptions of race that have been prominent since man achieved a basic understanding of genetics. For this reason, I regard all three statements, and their definitions of race, to be totally irrelevant to a serious consideration of race realism.
The Size and Origin of Human VariationAlexis’s article contains a chart supposedly showing that humans posses less genetic diversity than other great apes do. I don’t actually think this is particularly relevant to race: small genetic differences can lead to large phenotypic differences and grouping organisms based on such differences is a perfectly valid thing to do. But it’s worth pointing out that the chart’s claims are not true. Without getting into the technical details, the measurement of genetic diversity used in the paper that that chart is from is called the Watterson Diversity Estimator. Any decent textbook book on the relevant subjects will tell you that this measure only accurately measures genetic diversity if the species is in a state called a mutation-selection equilibrium. There are literally no geneticists who think that humans are in such an equilibrium. Because of this, Watterson’s estimator cannot measure genetic diversity in humans (or, for that matter, other great apes). A measure of genetic diversity that can be applied to humans is heterozygosity. A species heterozygosity is a measure of the probability that two alleles picked at random from the population will be non-identical. Using this meaure of genetic diversity will show that humans posses roughly as much genetic diversity as Chimpanzees. And that amount of genetic diversity is enough to allow Chimpanzees to have somewhere between 3 and 5 recognized subspecies, depending on which zoologist you ask.
Alexis also states that the human races have been evolving separately for “a very short time.” Once again, I don’t think this is relevant to the question of whether or not racial classification is valid: if the races evolved/learned their current difference in an abnormally short amount of time then so be it. The differences still exist. There is nothing mathematically implausible about large genetic differences evolving in this period of time. 60,000 years is a couple thousand human generations. Given this time scale, a difference in the frequency of an allele between two populations wouldn’t need to increase more than, on average, about .01% per generation for there to be large difference between the races today. Further still, the populations that left Africa mated with various archaic human populations and, as a result, have varying levels of Neanderthal and Denosovian admixture present in their genomes. These archaic human populations spent hundreds of thousands of years evolving apart from homo sapien sapiens and so the differential mating with them likely increased the genetic distance between modern human populations.
The Pattern of Human Variation
It should be clear by now that there is a decent amount of genetic diversity present in humans and it had plenty of time to get there. That, on its own, doesn’t show that races are significant. After all, the differences could be distributed randomly across the globe. But that isn’t the case. The chances of someone having an identical gene as you at a given loci is about 5-15% higher if that person is of the same race as you. And, as I already mentioned, when looking at the whole genome, instead of a single loci, the chances of you being more genetically similar to someone of another race than to someone of your own race is basically zero.
In the early 2000’s a lot of new data about how human genetic variation is structured came out. These studies utilized new software and technology that allowed researchers to perform cluster analyses on human genetic data. When a cluster analysis is performed a computer with specialized software is given bunch of data about the genomes of study participants and told to group these people together into X number of clusters. The cluster software will then sort people into clusters which maximize the within cluster genetic similarity and minimize the genetic similarity of people from different clusters. Beginning in 2002 a bunch of studies came out showing that if you tell clustering software to group people into 3-6 clusters you you find that it groups people into categories that correspond almost perfectly to races. In the most widely cited of these studies, participants’ self identified race was correctly predicted by their genetic cluster assignment over 99% of the time. Race realists have utilized this information to argue against people who say that race is “only skin deep” or that it has nothing to do with genes. They’ve also used this data to show that, in certain contexts, human genetic variation is patterned racially.
Alexis argued that because the software has to be told how many groups to sort people into these studies support the idea that race is a social construct. This is true, the software does have to be told how many groups to form and so these categories are invented by man rather than given to us by nature. We can choose to divide people into as many races as we want. But as I outlined above, I don’t see that as problematic. The point that race realists try to make with this data remains intact. There are clear genetic differences between the races and there are certain ways of analyzing genetic data that shows a great degree of racial structure.
The most significant argument in Alexis’s article concerns whether or not genetic variation in humans is continuous or discrete. Alexis believes that human variation is continuous, in the sense that allele frequencies change very smoothly and gradually over geography, and that this implies that we can’t apply “discrete” categories like race to it. There are two principle problems with this argument.
The first is that human variation isn’t all that continuous. As I mentioned previously, in 2002 there was a famous study by Rosenberg et al showing that humans could be assigned to a few genetic clusters with a great deal of accuracy. Alexis links to a 2004 paper by Serre and Pääbo which tried to disprove that claim. The authors argued that Rosenberg had achieved misleading results by sampling populations that were not sufficiently geographically dispersed. Imagine that all human populations existed on a single 10,000 mile plane and that the frequency for some gene variant increased by .5% every 100 miles. If you preformed a cluster analysis and only sampled people from miles 1-100, 500-600, and 900-1000, you would find that all the participants could be assigned to clusters that corresponded to the area on the plane that they came from. You thus might walk away talking about “discrete clusters” even though, in reality, gene frequencies changed in a perfectly smooth and gradual way. This is basically what Serre and Paabo thought happened in the Rosenberg study. The samples came from areas that were too far apart and didn’t match the actual geographic distribution of humans. So they collected data on a more widely dispersed sample and found that humans could not be easily sorted into racial clusters. But Rosenberg responded in a 2005 study and showed that Serre and Paabo’s results were the result of using too few genetic markers. When a larger number of markers were used a discrete set of clusters emerged from the data even when a highly dispersed sample was used. Moreover, Rosenberg et al showed that if you take two populations of the same race and compare them to two populations from different races the populations from the same race will be more genetically similar even when both pairs of populations are, geographically speaking, equally far apart.
The second problem with this argument is that there is nothing wrong with using “discrete” categories to organize continuous variables. This is commonly done in science. For instance, “high blood pressure” is a discrete category applied to perfectly continuous variation. Even if it were true that human genetic variation was perfectly continuous, which it is not, it would not follow from this that racial categories are not useful in science.
A related argument against the existence of race that Alexis uses goes like this: we can group people together by race. But we can also group people together in other ways. For instance, if we grouped people together based on blood type we would group some-people together that are members of different races and some people apart who are members of the same race. Thus, the choice to categorize people by race is arbitrary and therefore invalid. This argument goes back a long way. It was initially used in taxonomy journals as an argument against the existence of subspecies in general. A lot could be said about it, but here I’ll just point out a few things.
First, it is true that we can categorize people in lots of ways. But so what? We do! Just because we categorize people racially does not mean that we can’t categorize them in other ways as well. We can talk about racial differences and differences between people who carry different blood types, and differences between the sexes, etc. The fact that other ways of categorizing people exists is no problem at all for racial categorization.
Secondly, the fact that alternative classifications exist does not make the use of racial classification arbitrary. As we’ve already seen, race is useful in many contexts. It helps in predicting and explaining a lot of data. When we think that dividing people racially will aid us in analyzing whatever data we are looking at, we should. When we don’t, we shouldn’t. Only using a tool when it is useful does not make its use arbitrary.
In conclusion, I don’t find any of the arguments presented in Alexis’s article to be compelling reasons for abandoning race realism. Her article’s largest flaw is that it rests on definitions of race supplied by race deniers rather than race realists. Her article also goes to great lengths to show that race is a social construct. This is true, but race is also a valid biological category. The two aren’t mutually exclusive. There were a few problems with data, such as how continuous human genetic variation is and how much genetic diversity there is present in humans. But, for the most part, I think the article suffers from conceptual, rather than empirical, problems.
Even if Alexis agrees with all of my critique, I don’t think she will be convinced of race realism. This is because, in her view, we should ultimately just defer to what scientists say about race. And scientists say that race doesn’t exist. In an email she stated “If most anthropologists say that race isn’t a meaningful or useful concept for modern humans, and geneticists agree for the most part, I can’t argue, since they have much more of the scientific data available to them than either of us do (sadly).”
In the first place, I don’t think we can trust scientists on race. I think that it is fairly well documented that scientists have been incredibly irresponsible when it comes to race. And the second part of her statement is just wrong. We have access to all the same data that scientists do if we are willing to exercise the agency to find it and examine it. So there is no need to trust them. We can directly evaluate their arguments. But even if I did think that we should just trust scientists on race, I don’t think it is true that there is anything like a consensus on race in science.
There are basically three methods for measuring the popularity of race in science: surveys directly asking researchers whether or not they believe in biological races, studies looking at what percentage of journal articles on human variation that utilize racial categories when explaining biological differences between people, and studies looking at how race is treated in textbooks. In America the most recent surveys and data on journals comes from the late 1990’s and early 2000’s. Survey data of scientists suggests that about one in three American anthropologists believe that biological races of humans exist. Surveys of articles published in physical anthropology journals suggest that about 20% of them utilize race when describing biological variation in humans. An analysis of high-school biology textbooks found that over 90% of them utilized race when describing medical differences between populations. And a recent study of 18 anatomy textbooks found that all of them that talked about human variation utilized race when doing so. Thus, in America, there is not a consensus one way or the other about whether or not race is biologically meaningful.
Alexis is aware of all this information. However, she argues that because this data is about 15 years old, and between 1985 and 2000 there was roughly a 20 point drop in how popular race was found to be in polls of anthropologists, we should infer that something like another 20 point drop occurred over the last 15 years. Given the resurgence of scientists defending race in the public that occurred in the early 2000’s following the cluster studies, I think this is highly unlikely. Moreover, there isn’t any good reason to think that the trend would continue as it has in the past. It’s just an assumption. And anthropology isn’t the only discipline relevant to race. At the very least, it is clear that we don’t know that there is a consensus about race in science. Since we don’t know that a consensus exists it doesn’t make much sense to defer to it.
It’s worth noting that race is more popular among researchers in other countries. Around the same time that the American Anthropological Associations issued their statements denouncing race there was a conference held in Moscow entitled Race: Myth or Reality. The conference drew in both geneticists and anthropologists from both Eastern and Southern Europe. At the end of the conference, three statements were endorsed by this group of experts. It read, in part:
- According to the old anthropological tradition big human morphological variations which are the result of polymorphism united by common origin in certain geographical areas had been given the name “races.”
- Reality of the racial subdivisions of Homo sapiens are supported by the totality of the scientific data investigated on the different levels of human organism: morphological, physiological and genetic. Racial classification created with regard for morphological criteria clearly enough reflect the phylogeny of the separate populations and groups of populations.
- Negativism to the race concept which became apparent during the last decades, in many respects might be explained by the psychological shock which all progressive humanity had felt in the epoch of Hitlerism.
Surveys given to Polish biological anthropologists find that roughly 75% of them believe in human races of some sort or another. And an analysis of 12 polish physical anthropology textbooks found that all 12 of them endorsed the existence of human races. Surveys have also found that most anthropologists in Central and Eastern Europe believe in human races. Perhaps the most surprising findings come from China. Studies have shown that virtually every article published on human variation in Chinese biological anthropology journals utilizes race as a biological concept. Thus, there seems to only be one place in the world that that has a consensus about whether or not races exist. That place is China and the consensus is that race is a biological reality.
As I’ve already stated, even if there was a consensus I don’t think it would make sense to assume that it was right. However, there is no consensus. Given this, we must evaluate the arguments that are presented the best we can and formulate our opinions based on such evaluations. And an honest analysis of the relevant arguments will cause to one to adopt race realism.