The Razor Blade in the Apple - Part III
Coming to terms is risky business. And it always will be.
In
the New York Times column I mentioned in Part 2, and to a greater extent in the chapter of his book that inspired the column, David Reich attempted to subtly, even imperceptibly, square a circle that I contend is intractably round. He outlined the particulars of the 2002 Rosenberg-Feldman study, also detailed in Part 2, which seemed to excuse the tenacious persistence of categories like “Africans”; “East Asians”; “Australasians” (or maybe we’d call them “Oceanians”); “Native Americans”; and a broad group including Europeans, Middle Easterners, and South Asians that I guess we’d want to call “Caucasians.” (As mentioned previously, this was always a somewhat dubious conclusion. If the STRUCTURE algorithm was instructed to split human populations into six rather than five coherent categories, suddenly one group—the Kalash, a fairly isolated population who speak a language known as Kalasha, from the Indo-Aryan branch of the Indo-European family, and who live in the Hindu Kush mountains on the Afghan border with Pakistan—emerge as the sixth group. It is implausible to assert that this population of roughly 5000 people who are phenotypically indistinguishable from other inhabitants of Afghanistan, Pakistan, or Tajikistan are a “race.”) Reich chronicled how Rosenburg, Feldman, and their colleagues in later articles posited an explanation for the emergence of this set of human populations that seemed to so conveniently match our expectations. One of the principle concepts of genetics is the so-called “founder principle,” which dictates that when a migrant group breaks off from a larger parent population, the new population that gets formed by that splinter group differs genetically from the parent population because the founders of the new population were genetically different from most of the members of the parent population. The Rosenburg-Feldman team conjectured that the history of all human populations consisted of a continuous series of new founder populations leaving Africa, moving to new geographical areas, and then becoming genetically distinct: a “serial founder” effect.
In the self-congratulatory tone struck by many practitioners of paleogenetics, Reich argued that his field had (among so many accomplishments in so short a time) left the “serial founder” hypothesis unsustainable. The human population that exists at present, Reich clarified, is a mixture of once geographically distinctive populations that today we can only identify through the genetic markers they have left in the present-day population. “Most of today’s populations,” Reich wrote in his book, “are not exclusive descendants of the populations that lived in the same locations ten thousand years ago.” This conclusion afforded an expedient subtlety to Reich’s point. He may have some qualms about the “orthodoxy” being pushed by scholars like Evelyn Hammonds, but no one could accuse him of the same sort of duplicity Nicholas Wade was guilty of. “The findings that the nature of human population structure is not what we assumed should serve as a warning to those who think they know the true nature of human differences will correspond to racial stereotypes.”
The astute reader will surely have noticed something askew in the implications Reich is trying to draw out. Is it true that categories like (self-identified) “African American” or “European American” are “scientifically meaningful” because they accurately point to “where in the genome people harbor segments of DNA from ancestors who lived in West Africa or in Europe in the last twenty generations”? Is it also true that humans “are not exclusive [emphasis added] descendants of the populations that lived in the same locations ten thousand years ago”?
Well, harboring segments of genes from ancestors who lived in certain geographical zones in relatively recent generations basically means that those people are not exclusive descendants of the populations that lived in those locations ten thousand years ago. So—as long as we duly note the modifier exclusive—then that’s not a contradiction. But Reich wanted to say that the socially-constructed groups that we use to parse human variation—races—have some validity because they point to real-world differences. Reich wanted to have his cake and eat it, too: When it fit his purposes (namely, criticizing those who would extinguish the Promethean fire of true science because of its social inconvenience), then modern “groups” or “populations” with different features or proclivities are real; but when asserting that those populations are real suddenly smacks of racial essentialism, then no, Ancient DNA studies have shown that the modern human population doesn’t correspond to old racial categories. Well, which is it?
Reich was more nuanced in his claims about human categories than was Armand Leroi. He didn’t just ham-fistedly come out and say that, if looked at the right way, races do exist. But he did say the thing Leroi studiously avoided saying: that ever-improving genetic research will surely eventually show that some human populations exhibit statistical tendencies towards certain aptitudes—say, athletic prowess1 or, indeed, intelligence. And it is this claim that raised the ire of the signatories of the letter condemning Reich’s Times column.
I am sympathetic with the signatories. The suggestion that a group of humans that somehow co-aligns with a supposed category of race—like Africans, Eurasians, or East Asians, etc.—might one day be verified as more or less intelligent is nauseating, even if the genetic etiology of that phenomenon is more subtle and mixed up with social and environmental contributions than my visceral distaste causes me to imagine. For example, Reich cited a study by Augustine Kong and colleagues that found educational attainment in Iceland to have dropped by roughly 0.1 standard deviations over the last century because of selection. And the selective trait, it was conjectured, was having children at a younger age. That would mean that, because of societal pressure and social organization, more people who could have children at a young age did so, thereby passing along the trait to have children at a younger age to later generations (and therefore attain fewer years of education). We could imagine that cultures with other traditions and other manner of social organization might have less such pressure; and (just assuming genetics can have some kind of bearing on educational attainment) we could further imagine cultures having been separated for long enough in the past for selection to form more or less genetically distinct groups, some more likely to attain further years of education and some less likely to do so. In fact, by that model, the line of distinction between “biologically real” and “socially constructed” seems blurry.
Nonetheless, Reich’s conclusion snuggles up far too cozily to Nicholas Wade’s thesis for me. I don’t want the world to be that way. Reich would surely reply that only science armed with ample evidence can adjudicate on these matters, and my preferences can play no part. I suppose that’s right. It’s not wrong to imagine the world might be such that human groups with genetic proclivities distinguishing them exist.
But it would be wrong to insist that such groups exist—or look desperately to find such groups—when evidence to the contrary overwhelms. That’s what the signatories against Reich’s column contended:
Human beings are 99.5% genetically identical. Of course, because the human genome has 3 billion base pairs, that means any given individual may differ from another at 15 million loci (.5% of 3 billion). Given random variation, you could genotype all Red Sox fans and all Yankees fans and find that one group has a statistically significant higher frequency of a number of particular genetic variants than the other group—perhaps even the same sort of variation that Reich found for the prostate cancer–related genes he studied. This does not mean that Red Sox fans and Yankees fans are genetically distinct races (though many might try to tell you they are).
Baseball is an illuminating comparison. As I discussed in Book 2 Episode 10: “Ballpark Figures Part II,” well over a century ago, baseball pitchers began experimenting with how to grip balls and throw them so that they sped up, slowed down, bend, swooped, curved, dipped, or dropped off. And they eventually codified certain kinds of pitches: two- or four-seam fastballs, plus cutters and splitters; breaking balls like curveballs and sliders; and changeups. This cultural development afforded pitchers more or less success.
But the past decade has seen a revolution, as major league stadiums have been equipped with ever-more precise tracking systems that can observe all of the movements on the field—by arms, legs, bats, and balls—at a level of granularity impossible for the naked eye. Armed with expansive ball motion data, analysts have trained neural networks to distribute pitches in a field with multiple principle components (velocity, horizontal break, vertical break) and locate meaningful clusters. Sometimes there is overlap, but the clusters hold up across the set of all MLB pitchers.
It isn’t wrong to say those pitch classifications are real. Then again, it isn’t wrong to say they are socially constructed. The world is analog, and we want to (perhaps have to) break it up into digits. How we separate those digits and how we parse them depends on what we want to accomplish. Analysts could train neural networks to trace the movements of baseballs along other parameters, but it might not make pitching more effective in ways we are mentally equipped to consider, because we’re so attuned to thinking about pitches the way we do. And the same is true for genetic differences across groups. As the signatories against Reich wrote, “there is a difference between finding genetic differences between individuals and constructing genetic differences across groups by making conscious choices about which types of group matter for your purposes. These sorts of groups do not exist ‘in nature.’ They are made by human choice. This is not to say that such groups have no biological attributes in common. Rather, it is to say that the meaning and significance of the groups is produced through social interventions.”
The analogy to baseball pitches is once again enlightening: Those neural networks don’t only classify the set of all MLB pitches into categories; they also help each individual pitcher see what kinds of pitches they are throwing compared to their own self-reporting and account for changes in their repertoire diachronically across their career as well as in real-time during the game as they work on fixing the pitches they employ. It’s an incredibly useful tool. David Reich might point to this analogy and bemoan the shortsightedness of his critics: Their fear of reinforcing racial stereotypes is robbing medical patients and doctors of powerful diagnostic tools. That is exactly what AEI resident scholar and psychiatrist Sally Satel argued in a 2002 New York Times article, "I Am a Racially Profiling Doctor."
There are two problems: First, the baseball pitch analogy only goes so far. The critique Svante Pääbo and David Serre leveled against the Rosenburg-Feldman study supporting human population clustering was strong—there is genuinely significant overlap between those clusters. Among pitches, there is overlap between four-seam fastballs and cutters, cutters and sliders, and sliders and curveballs, but compared to human genome clusters, the pitch clusters are more clearly distinct. Second, the reliance on racial profiles in medical diagnosis has well-established disadvantages. Some disparities that by rights shouldn’t be present between racialized groups in the U.S. are almost certainly influenced by racially profiling medical practices. The risk of preeclampsia and death during childbirth, which is 50% more likely for Black women than white women in the United States, might be influenced by genetic mutations (“e.g., sickle cell disease in the mother, FVL or APOL1 mutations in the fetus”), but none of those factors sufficiently explains the disparity. The dilemma is seemingly exacerbated by elevated stress brought about by internalized racism, including racism either overtly or unintentionally exhibited by medical practitioners. Meanwhile, where polygenic risk scores (which investigate the total number of genetic variants a person has in order to assess their heritable risk of developing a particular disease) are in use, their utility has proven limited to European-heritage populations because of a systematic lack of diversity in the genomic data. There are indications that newer algorithms (e.g. EIGENSTRAT) that search for “different differences” in genomic data than the ones traditionally associated with race and ancestry might make for more salutary medical practices, but these new algorithms still have, by necessity, some reliance on geographically-based ancestry. And when scientists and journalists catch onto this fact “downstream,” and revert right back to associating these algorithms’ results with race.2
Obviously, pitch profiling doesn’t have that sort of deleterious effect on baseball. But it’s not unequivocally all for the better, either. The widespread use of machine-learning software and statistics has rendered baseball a game of strikeouts, home runs, and not much else. And the game is a drag to watch compared to when I was a kid in the 1980s.
So, assuming you wish to save lives and improve health advantages, there might be some benefits to using the loose and arbitrary categories of “race” we have inherited. Using no categories at all seems untenable. The multi-signatory letter condemning Reich said much the same thing: “This is not to say that geneticists such as Reich should never use categories in their research; indeed, their work would be largely impossible without them. However, they must be careful to understand the social and historical legacies that shape the formation of these categories, and constrain their utility.” And yet, Reich applied much the same caveat when he admitted that “the race vocabulary is too ill-defined and too loaded with historical baggage to be helpful.”
Are these people just talking past each other? It’s tempting to say yes, but it’s not satisfactory. All parties here are circumspect: Reich intended to provoke the defenders of an “orthodoxy,” but he knew that racism is an odious reality, and so he could not simply embrace race outright; the signatories of the letter condemning Reich knew that codifying some set of categories for discussing human populations is necessary, but they could not bring themselves to let Reich’s notion of populations with recent common ancestry pass without critique. These people aren’t talking past each other. They’re whistling past the graveyard.
I began Part 1 with a comparison, namely that the way science tackles uncertainty is comparable to the way individuals and societies tackle uncertainty. The contrast between everyday social discourse and scientific discourse doesn’t just break down upon closer inspection; it actually reconstitutes itself in the opposite direction. That is, if you invest enough time considering how values and biases wend their way through the processes of scientific inquiry, you begin to see parallels in everyday situations—science doesn’t seem more like normal human interactions; normal human interactions seem more like science. In every domain, we behave as though we have access to the most profound thing science promises us: what the philosopher Thomas Nagel calls the “view from nowhere.” We are aware that we have points of view, but we harbor the conviction that we nevertheless understand things as they really are. In the sphere of science, overcoming the limits of our perspectives might entail using workarounds, like arbitrary points of reference such as sea level or Greenwich Mean Time. But the view from nowhere has bolder ambitions. It claims it can overcome any belief or bias. Just think about it: the very belief “I have a point of view” is a belief that it is objectively true that I have a point of view. We believe that’s the way things really are. And we believe that if we try hard enough, we can be just as certain about everything.
We’ve documented many failures of that ambition. We have noted, for example, that well-enough-meaning scientists, working in good faith—like David Reich and the signatories of the letter decrying his work—can look at the same set of facts, can even draw many of the same conclusions, and nevertheless lay opprobrium in each others’ laps. But here’s the rub: It couldn’t be otherwise.
The German philosopher Hans-Georg Gadamer had a different reading of the view from nowhere. He maintained that we can neither remove ourselves from our personal situations and take on the reality of other people nor encompass an absolute knowledge of all reality within the confines of our individual minds. Each person, he insisted, is limited by the horizon of their personal background—their family history, culture, nationality, experience of victimhood or privilege, gender, education, etc. But he saw a way out of the blind alley. What we call “closed-mindedness” Gadamer referred to as having no horizon. If you do have a horizon, you have something to see past, and the encounter with other people and their own personal backgrounds can then be a Horizontverschmelzung—a melding of horizons.
That sounds right to me, but I would argue that we are forced to have our horizons meld. It’s less like gently emulsifying a sauce and more like pushing sausage through a meat grinder: We will insist that “race” is a social construct that has had inimical social effects for centuries, but we will continue to uncover genetic relationships that correlate with “race.”
One way or another, we will have concepts and categories and ways to collectively understand the world. Certainly, there are things that exist beyond the ken of our mental horizons. And I would even venture to say that things we have names for—tables, chairs, masks, baseball pitches, infection incidence rates, and races—are not completely divorced from the reality beyond our horizons. But in very important ways, those things are made by the encounter of our mutual fears, both rational and irrational. Some might stand the test of time, and we might choose to hold fast to them; others might eventually strike us as specious or even invidious, and we might choose to jettison them. The best we can do is take the process in stride. We risk everything if we can’t do that. That much is certain.
In fact, as evidence of this, he pointed out that either West Africans or people with recent ancestry predominantly from West Africa dominate in elite sprinting competitions.
Many thanks to Nathaniel Comfort for pointing these two studies out to me and, more generally, for helping me think through the nuances of this essay.