Solving Race!

Unpacking Complexities on the Issue of Genetics and Race, with some Links to the Best Recent Philosophy-of-Biology Papers on the Subject

Let me begin this post by making something very important very clear: the topic of race is an extremely messy and complicated one. If you can’t articulate what a heritability score is (how it’s computed, using what kinds of studies, why legs have a heritability score of 0), if you have no clue what a “reaction norm” is, if you don’t understand the clinal variation of human groups, if you don’t know anything about palaeoanthropology and how human genetics compares to that of other species, if you don’t know about how taxonomical classification of subspecies works in other species,  if you don’t know about the Flynn Effect, if you haven’t read Guns, Germs and Steel (not to imply that it’s a flawless book ), and if you have no experience with philosophy of biology or analytic metaphysics (I think this is probably the most important), you really should shut the fuck up until you meet these criteria (yes, all of them). You cannot importantly contribute to a serious debate about race if you do not meet these criteria. You will be talking totally at cross-purposes to those who do, and without realising it (that is, you will be suffering from the very common ailment known as the ‘pretence of knowledge’).

Some people say that to say that races are unreal or that race is a social construction is muddle-headed on basically any construal. Two articles that position themselves rhetorically against all views that fall under the label of “social constructionism about race” , and argue that modern genetics shows that we can non-sophistically and non-essentialistically say that race is biological are the paper “The Social Destruction of a Biological Concept” by the philosopher Neven Sesardic (https://philpapers.org/rec/SESRAS) and this long blogpost from the very prominent (on the internet) genetics researcher Razib Khan (http://blogs.discovermagazine.com/gnxp/2013/05/why-race-as-a-biological-construct-matters/#.WR0CzWh942w). Sesardic is an anti-leftist polemicist who sees much of what philosophers and scientists say about race as politically rooted. His paper first appeared in the March 2010 issue of the journal Biology and Philosophy, and it has since been extensively discussed in the philosophical literature. While it is a flawed paper in several different respects (as we will see, when I discuss Hochman’s and Pigliucci’s (❤) replies (https://philpapers.org/rec/HOCATN; https://philpapers.org/rec/PIGWAW )), it is possibly the best recent philosophical defence of ‘racial naturalism’.

Now, I should immediately point out that Sesardic’s diction in the conclusion of the paper suggests that he does not want to be yoked to a strong label like racial naturalism (“My aim in this paper was not to prove the biological reality of race,” he pleads)  (I think that the rhetorical strategy behind this is to position himself as the sane, sober moderate), but he makes very clear throughout the paper that he doesn’t like any variation of the account of race as a “social construction” and strongly disagrees with academics who are perfectly happy to acknowledge the same genetic data and facts about skull morphology and IQ that he makes reference to (meaning that the difference lies in the metaphysics).  It is therefore hard to think of a better label (he clearly thinks there is a substantive disagreement at work; he is not merely saying that the rhetoric of his opponents (Naomi Zack, Dick Lewontin, Phillip Kitcher, Anthony Appiah, Sally Haslanger, Jared Diamond, etc) is misleading).

Khan’s article is not as strong as Sesardic’s; it would be erroneous to say that Khan defends a position that one would call ‘racial naturalism’. In fact, echoing the Kitcher quote that Hochman cites in his reply to Sesardic, Khan writes:

“From a scientific perspective in biology there are not ultimate and fundamental taxonomic facts. There are simply useful ideas and concepts to illustrate and explore the objective phenomena of the natural world. The Species Concepts debate shows us this reality well, as even species can be tendentious. But the debate often shakes out along disciplinary lines. Many more ecological scientists seem to be taken by the ecological species concept, while evolutionary geneticists are more keen on the biological species concept. That is because they are choosing the framework most useful for their ends. There is nothing “Post Modern” in this in that it denies reality. Rather, we are disputing the ideas which we use to capture the essence of real phenomena in compact semantic relations suitable for symbolic representation (whether with math or language).”

Another difference between his and Sesardic’s articles is that Khan is not positioning himself explicitly against fellow academic experts, but against the totally genetically uninformed and very weakly defended racial constructionism of the “literary intellectual” Ta-Nehisi Coates. The reason I grouped these articles together at all is simply because they both don’t really acknowledge the key point that Hochman makes, which goes as follows:

“The metaphysical problem [that “there is a nondenumerable infinity of possible accurate maps we could draw for our planet” [Kitcher, 2007: 209]] is best illustrated by its practical counterpart. Racial labels may be more or less useful proxies for biological variation in different scientific disciplines, and each discipline may support racial naturalism or social constructionism to a greater or lesser extent. It is crucial that these points be separated. For instance, forensic anthropologists are able to allocate skulls, with fairly high accuracy, to the racialized group with which the deceased would have been identified. Does this show that racial labels are useful proxies for biological variation in physical forensic anthropology? Yes. Does this support racial naturalism? No. Why? Because forensic anthropologists translate trait measurements to the “racial” taxonomies societies use, or would have used, to describe missing persons; they do not generally ask whether those taxonomies are scientifically valid. When they do it is in a research context, and their answer is firmly negative.[footnote 21, citing this: http://onlinelibrary.wiley.com/doi/10.1002/ajpa.21006/abstract] Forensic anthropologists are also able to allocate skulls to groups that are separated culturally, linguistically, politically, and historically, and at a finer grain than a racial taxonomy offers. “Race” latches on to human morphological variation, but in a crude way. Forensic anthropologists use racial classification, but this is because we, the public, classify our missing persons racially, not because it is a precise or privileged representation of human morphological diversity.

The debate between race naturalists and social constructionists is best framed not as a dichotomy, nor a continuum, but as a discussion worth having in various scientific, philosophical, and political contexts. This discussion may one day end in eliminativism about race. In the meantime, the social constructionist will expect correlations between racialized groups and various traits. Yet she will, in Gannett’s words, expect such correlations to be “statistical not universal, local not global, contingent not necessary, and accidental not lawful, and expect their corresponding cuts in nature to be interest-relative not mind-independent, dynamic not static, indeterminate not determinate, many not few, overlapping not nonoverlapping, and superficial not deep.”[footnote 22, citing the Gannett quote]”

As I made clear, what Hochman is saying here has parallels in Khan’s own words, but the key point made here that Khan fails to draw attention to and Sesardic totally omits is this: that if another species had the same group differences as our species does, we would not feel the need to come up with categories in any way paralleling our racial categories. As Hochman later argues, and as Pigliucci argues in his excellent reply, modern genetic evidence is firmly against the conclusion that human races rise to the level of subspecies, according to the criteria we apply to other species. Wright’s Fixation Index, or the F_ST, which compares the genetic variation within one or more sub-populations to the amount of variation within the total population using single-locus comparisons (this means (I think) comparing single-nucleotide polymorphisms, as it were, 'one by one', rather than looking for more complicated correlations between multiple loci), is the technique used for sub-species determination, with an F_ST of 0.25 the arbitrarily defined sufficient level for the demarcation of a subspecies. Lewontin’s famous 1972 study “The Apportionment of Human Diversity” showed that, according to this kind of standard, there’s no way you could say our species has a subspecies. Now, as Pigliucci and Hochman discuss (because Sesardic discusses this point, and because Khan possibly nods to it in his use of genetic-clustering data), it is often claimed that Lewontin committed a “fallacy” (https://en.wikipedia.org/wiki/Human_Genetic_Diversity:_Lewontin%27s_Fallacy) in inferring anything about the metaphysics of race from a single-locus study (the claim is that it makes sense to look at correlations between lots of loci than just compare this locus in individual A to that locus in individual B), when genetic-clustering studies have since shown that, e.g. ‘Africans’ stick out from the rest of us (although ‘Africans’, if you want to isolate them as a population group (arbitrarily, remember), have significantly more internal genetic diversity than ‘Europeans’ or ‘South East Asians’ (two more arbitrarily isolated population groups)).

Of course, calling this a fallacy is simply an abuse of language; it is no such thing. And here’s what Pigliucci says about the genetic-clustering studies themselves, and how they are abused by people like Sesardic:

“The Rosenberg et al. paper is a study of 52 human populations, whose genetic diversity was characterized using 377 autosomal (i.e., not sex-linked) microsatellite loci scored in 1056 individuals. It is by all means a large sample of genetic variation, and its empirical conclusions are robust (Rosenberg et al., 2005). The significance of the Rosenberg et al. study for Sesardic is that it “did allow an inference of group structure and that, furthermore, five clusters derived from that analysis of purely genetic similarities corresponded largely to major geographic regions” (Sesardic, 2010, p. 153). Yes, but this is an interestingly (and possibly revealing) exercise in selective quotation on Sesardic’s part.

First off, Rosenberg et al. actually found a variable number of major clusters (6, 5, 4 and even 3), depending at what level one stops the analysis. Why pick a particular one as the major finding of the paper, other than because five clusters happen to fit the author’s predilection for the true number of human races? At the very least this is blatant cherry picking of the relevant evidence. Second, and far more damning, Sesardic entirely ignores that Rosenberg and colleagues go on to say (even in the abstract of their paper!) that “we identified … subclusters that often correspond to individual populations.” Are each and all of these subclusters also races, in Sesardic’s opinion? One assumes not, but Sesardic has not given us any compelling reason to think that K = 5 is the racial level because his own basic meaning of ‘race’ (a genetically identifiable cluster of individuals) is compatible with multiple levels of human population substructure. Under a referentialist semantic framework this suggests that ‘race’ as intended by Sesardic has no referent, not that ‘race’ refers to K = 5 but not to K = 6, 4, 3 etc.

What of Tang et al.’s paper? Sesardic summarizes its import thus: “A group of researchers led by geneticist Neil Risch et al., (2002) analyzed genetic data for 326 microsatellite markers on a sample of 3636 subjects from the United States and Taiwan. The subjects identified themselves as belonging to one of the four racial groups (white, African American, East Asian and Hispanic). The genetic cluster analysis of the data produced four major clusters, whose correspondence with the four self reported races was near-perfect: the genetic cluster membership and self-identified race coincided in as many as 99.9% of the cases.” Besides the fact that there are specific methodological issues with the Tang et al.’s survey (if one looks closely at their sample, one realizes that all of their Hispanics are Mexican Americans from a single county in Texas, which makes the otherwise surprising “Hispanic” cluster a reflection of mere geographical proximity), again, one can identify legitimate genetic clusters of human populations at a variety of hierarchical levels, but Sesardic offers no principled reason for identifying one such clustering as more fundamentally indicative of races.

So, yes, there is structured genetic variation in human populations. But this is hardly a surprising or controversial notion among human population geneticists, and it does not at all imply any strong correspondence between the available genetic data and folk concepts of races.”

So what does this all mean? It means that we are ‘working backwards’ in a really important sense when we use genetic-clustering studies to bring out genetic differences between human groups that seem in some incredibly rough sense to correspond to our ‘folk’ ideas.

In conclusion, it’s not cool – and philosophically wrongheaded – to be a ‘race realist’. (Incidentally, Charles Murray is acknowledged at the end of Sesardic’s paper, which means this becomes yet another wonderful opportunity to slag off that total piece of shit Sam Harris, who, I am told, did not press Murray on anything when he invited him onto his podcast (doing his usual thing of arrogantly asserting completely false and asinine things on some subject on which he has zero expertise)).


**Postcript, 16th October 2017: just found this: http://bactra.org/weblog/520.html