Investigating human diversity in the twenty-first century

Julien Delhez

Anthropological Review • Vol. 83(3), 317–325 (2020)

Investigating human diversity in the twenty-first century

Julien Delhez

Seminar für Ägyptologie und Koptologie, Georg-August-Universität Göttingen, Germany

Abstract: In recent years, there has been renewed academic and public debate on the topic of race. The present essay compares two books dealing with this subject. Charles Murray’s Human Diversity states that the social sciences are permeated by a rigid orthodoxy that puts unnecessary strain on researchers working on sex differences, race differences, and individual differences. Far from being scary, Murray argues, these differences are interesting and can be lived with. Adam Rutherford’s How to Argue With a Racist examines various claims with respect to race. It includes four sections dedicated, respectively, to skin color, ancestry, sport, and intelligence. The essay assesses the authors’ factual claims as well as their tone, their general mindset, and their personal attitude toward science.

Key words: human differences, race, genetic variation, ancestral populations

Introduction gue With a Racist: History, Science, Race and Reality (Rutherford 2020). While both Research on human evolution and ge- books have already been reviewed sever- netics has greatly changed the way hu- al times by academics and by journalists, mans see themselves. While helping to I believe a comparative book review has understand the past and the present of several merits. Firstly, both authors are our species, research has also led spe- informed observers and have witnessed cialists and the broader public to revise the same scientific revolution, and yet their thoughts on some highly sensitive came out with very different conclusions; topics. it may be interesting to explore why. In this essay, I discuss two books ad- Secondly, an assessment of the authors’ dressing recent studies on human differ- viewpoints and writing may reveal many ences and what they mean for society: differences beyond factual disagreement. Charles Murray’s Human Diversity: The Thirdly and importantly, when one uses Biology of Gender, Race, and Class (Murray each book to cast light on the strengths 2020), and Adam Rutherford’s How to Ar- and weaknesses of the other book, it be-

Review Article Received: July 28, 2020; Revised: XXXX XX, XXXX; Accepted: August 25, 2020 DOI: 10.2478/anre-2020-0024 © 2020 Polish Anthropological Society 318 Julien Delhez comes easier to overcome the temptation tackles sex differences in what Murray of selecting the best arguments on one calls “cognitive repertoires” (Murray side while ignoring valuable points on 2020: 8), i.e. differences in personality, the other side. All in all, the main lesson mental abilities, and behavior. To enable I have drawn from comparing these two the reader to understand how large an books is that the topic of race requires a average difference needs to be in order great deal of intellectual humility. to matter for society, Murray provides Human Diversity is a thick volume sum- an interlude explaining what Cohen’s d marizing the results of many recent stud- consists of (Murray 2020: 23), and he ies on sex differences, race or population points out one of the most important differences (see below for the distinction facts about sex differences in cognitive between race and population), and individ- repertoires: even if the vast majority of ual differences in relation to class struc- effect sizes for such differences are tiny, ture. The book does not require advanced it does not mean than such differences training in genetics or psychometrics. are socially irrelevant, because “in the Nonetheless, given how complex the is- real world, it is taken for granted that sues dealt with are, one can expect that small differences add up” (Murray 2020: it will mainly find readers among those 28). A very convincing example of this who are willing to familiarize themselves is given by the comparison between fac- with technical notions. Despite its as- es of both sexes: for isolated traits, it is sumed goal of challenging “an orthodoxy often difficult to perceive any difference that is scared stiff of biology” (Murray between the female face and the male 2020: 2), Human Diversity is noticeable for face, but when one looks at a face in its its non-confrontational tone. entirety, most of the time, it isn’t difficult How to Argue With a Racist is a short to find out whether it is male or female book which focuses exclusively on the (see Murray 2020: 29). topic of race. Compared to Human Di- The next three chapters tackle sex versity, How to Argue With a Racist has a differences in personality, neurocogni- more combative approach, as Rutherford tive functioning and abilities, and voca- (2020: 3) explains: “This book is a weap- tional choices. Murray generally focuses on.” The book has neither footnotes nor on the issues that are least contentious endnotes, only a list of forty references among psychologists and neuroscien- (Rutherford 2020: 189–194). tists. The difficult question of whether Before comparing the arguments the sexes differ in g (the general factor of both books on the topic of race, it is of intelligence) is treated very briefly, worth presenting Murray’s claims on the with a footnote allowing curious read- two other sensitive issues he deals with: ers to find articles from both sides of the sex differences and class. debate (Murray 2020: 386–387, n. 61). For sex differences in vocational choices, Latest research on human sex Murray relies on the results of the Study of Mathematically Precocious Youth differences (SMPY). This study is especially help- ful for assessing sex differences in life A substantial part of Human Diversi- choices insofar as all participants, male ty (Murray 2020: 11–127 & 337–372) and female, had the ability to embrace Investigating human diversity in the twenty-first century 319 a career in engineering or the physical (2020) wrote: “Murray leans hard on the sciences. Even within the SMPY sample, work of Cambridge psychologist Simon substantial sex differences in life prefer- Baron-Cohen. But while Baron-Cohen ences have been found: among others, is well respected, some of his published on average, females were more likely to claims have not been replicated (that is, choose a part-time employment (either confirmed by subsequent studies). This temporarily or definitely), and males doesn’t mean they are wrong, but that were more likely to give weight to a “sal- Murray’s confidence is unwarranted.” ary that is well above the average per- In fact, Murray (2020: 35) clearly urg- son” (Murray 2020: 69). Females were es caution about one of Baron-Cohen’s also more likely to end up in the human- most publicized studies: “It is a single, ities. Importantly, both sexes viewed unreplicated study with a sample of 102, their lives as equally satisfying, which no proof to take to the bank, but its find- suggests “that there are multiple ways ing was in line with many other studies to construct a meaningful, productive, that have found personality sex differ- and satisfying life” (Lubinski et al. 2014 ences in infants.” cited in Murray 2020: 76). Another chapter deals with the differ- Class and individual differences ences that have been observed, on brain scans, between brains of both sexes. For Human Diversity’s section about class the time being, the most important thing mainly presents conclusions that are al- to remember is that this research field is ready accepted among behavioral geneti- still in its infancy (Murray 2020: 125). cists. I will summarize them very briefly. Appendix 2 deals with medical condi- The author accepts neither the pessimis- tions that may constitute exceptions to tic view that socio-economic class is pri- the sexual dimorphism of the human marily a function of economic or ethnic species. Appendix 3 tackles sex differ- privilege, nor the optimistic view that ences in brain size, whose implications “people can become anything they want are not yet well understood, and the phe- to be if they try hard enough” (Murray, nomenon known as “greater male vari- 2020: 203). Instead, Murray (2020: 204) ance” (see Murray 2020: 357). states: “Class is a function of the genet- Are there additional topics the au- ic lottery plus character, determination, thor should have covered? In the spirit hard work, and idiosyncratic circum- that a chain is as strong as its weakest stances.” To clarify the role of the genet- link, Murray has willfully refrained from ic lottery, Murray outlines his argument discussing the findings of evolution- in three steps. Firstly (chapter 11), in an ary psychology (see Murray 2020: 6–7) overview of solidly established findings and studies on rodents in relation to of behavioral genetics, he points out that biological sex differences in the brain all behavioral traits are heritable, and the (see Murray 2020: 103). Some readers g factor substantially so, while the shared will probably miss a discussion of such environment (parental environment) has studies. In the author’s defense, though, little impact on the children’s cognitive one cannot say that wariness about un- repertoires. Secondly (chapter 12), he fair critics was not warranted: in a re- provides evidence as to the crucial role view for the New Statesman, Philip Ball of g in determining educational perfor- 320 Julien Delhez mance, job performance, income, and Importantly, while the number of groups occupation. Thirdly and finally (chapter (K) is decided by the analyst, analyses 13), he deals with various claims about conducted with different numbers offer allegedly substantial effects of exter- results that are consistent with one an- nal interventions. The sections on the other: “different values of K do not pro- growth mindset (Murray 2020: 255–258) duce a radically different pattern of re- and epigenetics (Murray 2020: 260–268) sults. Instead, they augment the results, are very worth reading. giving a greater degree of definition to a previously identified pattern” (Murray Humanity’s genetic variation 2020: 155). In a second time, Murray (2020: 158) and recent evolution addresses the view “that humans left Africa so recently that they haven’t had The part of Human Diversity that deals time to differentiate themselves geneti- with race is shorter (Murray 2020:129– cally in ways that would affect cognitive 202) than those related to sex differences repertoires.” A key element of Murray’s and class. It is also by far the most cau- argument is that the so-called “neutral tious of all three parts. Murray (2020: theory of molecular evolution” – which 403, n. 1) explains that his attitude was states that while natural selection drives inspired by the reaction to the publica- most of phenotypic evolution, most of tion of Nicholas Wade’s A Troublesome In- molecular evolution is neutral and is ex- heritance (Wade 2014), as the second half plained by genetic drift – is being chal- of this book largely consisted of specula- lenged by numerous recent analyses, tions and attracted considerable criticism which emphasize the role of soft sweeps from the scientific community (Coop el in standing variation, i.e. changes in allele al. 2014). After a brief description of frequencies that affect a high number how social constructivism regarding race of alleles related to a specific polygenic came to be (Murray 2020: 129–132), the trait: “A change in the environment may author points out that a discussion of have only modest effects on the allele fre- race differences in cognitive repertoires quency at any one locus, but it has those would not make sense in his book, since modest effects on hundreds of the rele- such a discussion cannot establish that vant sites and thereby produces a cumu- the differences are genetic. Instead, he latively large effect” (Murray 2020: 171). chooses to focus exclusively on the ge- Crucially, principal component analysis netic evidence. After briefly presenting (PCA) shows than ancestral populations notions such as SNP, genotype, and genetic also differ in the frequency of alleles un- drift, Murray (2020: 143–148) offers an der recent selection pressure: “When ge- account of the most recent findings on neticists use noncoding genetic variation human evolution – an account he had to from multiple populations, those popu- revise many times, given how fast new lations are genetically distinctive in ways discoveries are made. Then, he intro- that broadly correspond to self-identified duces cluster analysis, which shows one race and ethnicity. When geneticists use can identify genetic groups within the genetic variation that is not only func- human species – groups that often corre- tional but has been under selection pres- spond to self-identified race or ethnicity. sure since the dispersal from Africa, the Investigating human diversity in the twenty-first century 321 same correspondence usually appears” may be the most serious problem of Hu- (Murray 2020: 181). man Diversity: the difference between race Then, Murray (2020: 182–202) deals and population is not explicitly addressed. with the sensitive question of whether This is important because the view that genetic differences in cognitive reper- “race is a social construct” historically toires are likely to be found between an- rested not on one, but on two pillars: (1) cestral populations. As the author notes, the idea that it is impossible to distin- current polygenic scores for IQ and be- guish genetic groups within the human havioral traits cannot tell us much about species; and (2) the idea that even if such between-population differences, since groups were to be found, they could not “the predictive validity of a polygenic be properly described as races, because score deteriorates as the genetic distance race refers to groups that were historical- between the test population and the ly perceived as sharply distinct from one comparison population increases” (Mur- another (see e.g. La Vie des Idées 2014). ray 2020: 184). However, due to strong Murray has successfully challenged the calls to collect genomes from non-Euro- first pillar, but did not address the second pean populations, it is expected that re- one. As a result, one may perfectly accept searchers will be able “to study genetic his data and still state: “race is a social differences in personality traits, abilities, construct.” For instance, evolutionary and social behavior across continental psychologist Nicole Barbaro (2020) populations” (Murray 2020: 185). Mur- wrote that “race is more accurately de- ray (2020: 186–196) also offers a person- scribed as a social construct rather than al analysis which shows that there are a biological reality” and also pointed out: continental population differences in al- “that human populations differ genetically leles related to cognitive repertoires. The is not synonymous with the claim that meaning of these differences is unclear, human races differ genetically” (empha- and the author warns: “I am not presenting sis in the original). proof that those differences cause phenotypic In my opinion, it is a mistake to frame differences” (Murray 2020: 186; emphasis the debate as a choice between two op- in the original). Nonetheless, such differ- tions – “race is a social construct” or ences may raise interesting questions in “race is not a social construct.” Both the future. statements are problematic, because race is not one thing. Historically, race has been Race or population? used to refer to a wide variety of con- cepts, either vertical (X son of Y, son of Around the end of the last chapter on Z) or horizontal (group living in a cer- race, Murray (2020: 196) wonders: “We tain area at a certain time), scientific or have known for years that biologically non-scientific, essentialist or non-essen- complex differences in continental pop- tialist, hierarchical or non-hierarchical, ulations have evolved since humans left and… social or biological. Instead of Africa. It is an unlikely assertion on its stating that “race is” anything, it would face – how can “race is a social construct” be more productive to ask: can there be a continue to be the received elite wisdom valid social concept of race? Or: can there if such differences are already known?” be a valid biological concept of race? The In my view, this question reflects what two views are not mutually exclusive. 322 Julien Delhez

Reflections on skin color, tioned on and off since its publication, ancestry, sport, and intelligence but remains broadly correct. The main challenge was formalised as ‘Lewon- Aside from its introduction, Adam tin’s Fallacy’ in 2003 by the mathema- Rutherford’s How to Argue With a Racist tician Anthony Edwards, which pointed includes four chapters on race-related out that if you aggregate multiple sites issues: skin color (Rutherford 2000: 27– of variation across a genome, you can in 65), ancestry and genealogy (Rutherford fact predict the population from which 2000: 67–107), sport (Rutherford 2000: a person comes accurately. Both results 109–136), and intelligence (Rutherford are true; it just depends on the detail and 2000: 137–175). The book is quite dif- the resolution.” This misrepresents the ficult to summarize given the author’s point made by Edwards, who criticized tendency to tackle numerous questions not the percentages but “the old statis- within the same chapter. tical fallacy of analysing data on the as- The first chapter starts with a discus- sumption that it contains no information sion of skin color and the genetics of pig- beyond that revealed on a locus-by-locus mentation, and goes on with a history of analysis” (Edwards 2003: 799; see also racial classifications, most of which were Dawkins and Wong 2016: 461–462). in great part based on skin color. Most if Rutherford then deals with the is- not all researchers will agree with the au- sue of ancestry testing and genealogy. thor’s statement that skin color is “a very He rightfully points out that there have superficial route to an understanding of been many migrations in human history, human variation, and a very bad way to and he explores the impact of the Viking classify people” (Rutherford 2020: 64). exploration of America, the slave trade, Several sections of this chapter are prob- and the Rwandan genocide. The end of lematic. On the issue of cluster analysis, the chapter discusses at length – and Rutherford (2020: 49–51) describes the derides – white nationalists “who claim results of Noah Rosenberg’s 2002 pa- racial purity and therefore racial superi- per (Rosenberg et al. 2002) but ignores ority” (Rutherford 2020: 99). Without a the more recent studies led by Li (Li et doubt, the author quotes statements that al. 2008), who used the same sample as are astounding and indefensible. But this Rosenberg but “was analyzing 642,690 begs the question: what should be done variants instead of 377” (Murray 2020: in the face of ignorance and extremism? 151), and by Xing (Xing et al. 2010), Rutherford (2020: 102) admits he has who included “data from 296 individuals little confidence in reasoning and - argu in 13 populations that had not been cov- ments: “As Jonathan Swift said in 1721: ered by previous studies” (Murray 2020: ‘Reasoning will never make a Man cor- 153). Also surprising is the treatment of rect an ill Opinion, which by Reasoning Lewontin’s 1972 paper (Lewontin, 1972) he never acquired.’” by Rutherford (2020: 49), who writes: As for the section on sport, the author “Lewontin found that the vast majority does not seem to believe that any safe (85 per cent) of genetic differences were conclusion can be reached. Rutherford within classical races, not between them. (2020: 119) even writes: “The point is Only 6 per cent of differences segregated this: elite sprinters in the Olympics are by race. This conclusion has been ques- not a dataset on which a statistician could Investigating human diversity in the twenty-first century 323 draw any satisfactory conclusion.” The unlikely due to the immense genetic di- chapter’s conclusion contains a puzzling versity that is now well established across statement: “As well as entertainment, that continent.” But a few pages later, he sport is a celebration of the extremes of states: “It would be perfectly possible human capabilities. To reduce it to mere for two populations with different sets unearned biology is racism, whether con- of genetic differences to get the same IQ scious or not” (Rutherford, 2020: 136). scores” (Rutherford 2020: 158–159). Fi- However, did any observer reduce it to a nally, while it is true that current poly- phenomenon determined 100% by biol- genic scores are “not particularly adept at ogy and 0% by training? At least not Jon dissecting the differences between popula- Entine, author of an influential mono- tions” (Rutherford 2020: 158; emphasis graph on the topic of black athletic suc- in the original), Rutherford (2020: 175) cess (Entine 2000) – a book which Ruth- lacks cautiousness when he concludes: erford does not refer to. Entine (2000: “People are born different, with different 18) explains: “the opposing and incom- innate capabilities and potential. How patible claims that black athletic success these abilities cluster within and be- can be explained by environmentalism or tween populations is not easily explained evolution are equally simplistic. Sports by fundamental biology, by genetics. In- success is a bio-social phenomenon.” stead, when digging into the data as best On intelligence, the author broadly as we can, the answers lie not in DNA, agrees with current intelligence research- but in culture.” That no genetic cause has ers on the most basic facts, e.g. the pre- been identified doesn’t mean that cultur- dictive validity of IQ tests for various life al explanations have been vindicated; it outcomes (Rutherford 2020: 146–149). only means that one should avoid hasty However, there are substantial flaws conclusions, and keep an open mind. in his discussion of group differences in intelligence. Rutherford (2020: 151) Discussion suggests: “It could therefore be sensibly argued that a big part of the alleged dis- Aside from factual claims, many dif- crepancy we see between some African ferences between the two books can be and European countries can be attribut- noticed. As explained above, Human Di- ed to the Flynn effect not having hap- versity is generally more restrained in pened universally, and significantly not tone than How to Argue With a Racist. in some African countries.” In fact, it is Language precision is another element: now admitted by researchers that the while Human Diversity is easily read, it Flynn effect and group differences in IQ is not always easy to understand what scores have different causes (see this re- the author of How to Argue With a Racist cent meta-analysis: te Nijenhuis and van means, for instance when he writes that der Flier 2014); James Flynn himself has IQ “makes a much better predictor of stated that these differences “cannot be many more things than a sprinting times explained by the Flynn effect” (Wilby does” (Rutherford 2020: 147), without 2016). With respect to the same topic, detailing what these “many more things” i.e. IQ scores in Africa, Rutherford (2020: are; or, when he writes, on the question 150) writes: “while it is not possible to of whether genetic variants are associat- fully exclude genetic factors, these seem ed with specific ancestral populations: 324 Julien Delhez

“The answer is yes. And no. And maybe” Race has now become an extremely (Rutherford 2020: 121). When it comes sensitive topic. There is urgent need for to the reference system, Human Diversi- more data, more curiosity, and more ci- ty has extensive endnotes and a full list vility. In this regard, Human Diversity is an of references at the end; in How to Argue important step in the right direction. The With a Racist, the author’s choice of list- same cannot be said about How to Argue ing forty articles or other publications With a Racist; having found A Brief History at the end (Rutherford 2020: 189–194) of Everyone Who Ever Lived enjoyable and makes it very difficult to find out which informative, I believe that Rutherford’s claim is backed by an article (in A Brief 2020 book could have been much better. History of Everyone Who Ever Lived, an- While we all make mistakes, underesti- other book written by Rutherford for a mating one’s opponents is a very dan- general audience, the references were at gerous one. Rutherford would be well least sorted by chapter; see Rutherford advised to pay attention to these wise 2017). Moreover, it is unclear which cri- words of Ernst Cassirer (1946: 296): “In teria led to the choice of these forty pub- order to fight an enemy you must know lications. The list of references includes him. That is one of the first principles of neither Lewontin’s 1972 paper nor Ed- a sound strategy. To know him means not wards’ 2003 rebuttal, although both are only to know his defects and weaknesses; discussed in the book. it means to know his strength.” Ultimately, however, I think the most striking difference between Murray and Conflict of interest Rutherford lies in their personal attitude: Murray spends most of the book analyz- The author declares that the present ing data and is generally charitable with work was prepared in absence of any the researchers he disagrees with, e.g. financial, professional or personal - rela Eric Turkheimer (see Murray 2020: 279– tionship with either author whose book 286). He also shows his excitement about is reviewed here. future discoveries. By contrast, Ruther- ford talks very little about the future and Corresponding author regularly mocks those he disagrees with, more than he helps the readers to under- Julien Delhez, Eichweg 19, 37077 Göt- stand why they think the way they do tingen, Germany (which does not mean the disagreement e-mail address: [email protected] is not justified). Among others, he- de scribes anthropologist Johann Friedrich References Blumenbach as “a particular sort of bib- lical creationist” (Rutherford 2020: 41), Ball P. 2020. The gene delusion. New States- writes that Thomas Huxley “used inde- man, 10 June. Available at: https://www. cipherably imprecise language” (Ruther- newstatesman.com/class-race-genet- ford 2020: 44), and calls James D. Wat- ics-science-human-diversity-charles-mur- ray-review. son “old and infirm” (Rutherford 2020: Barbaro N. 2020. Thoughts on Human Di- 143). Is it that difficult to reject an idea versity. Nicolebarbaro.com, 23 Febru- without utterly tearing down the human ary. Available at: https://nicolebarbaro. being who subscribes to this idea? Investigating human diversity in the twenty-first century 325

com/2020/02/23/thoughts-on-hu- Lubinski D., Benbow C.P., Kell H.J. 2014. Life man-diversity/ Paths and Accomplishments of Mathemat- Cassirer E. 1946. The Myth of the State. New ically Precocious Males and Females Four Haven – London: Yale University Press. Decades Later. Psychol Sci 25(12):2217– Coop G., Eisen M.B., Nielsen R., Przewor- 32. ski M., Rosenberg N. 2014. Letters: “A Murray C. 2020. Human Diversity: The Biolo- Troublesome Inheritance.” Stanford Uni- gy of Gender, Race, and Class. New York: versity, Center for Computational, Evo- Twelve. lutionary and Human Genomics. Avail- Rutherford A. 2017. A Brief History of Ev- able at: https://cehg.stanford.edu/let- eryone Who Ever Lived. New York: The ter-from-population-geneticists Experiment. Dawkins R., Wong Y. 2016. The Ancestor’s Rutherford A. 2020. How to Argue With a Tale: A Pilgrimage to the Dawn of Evolu- Racist: History, Science, Race and Reality. tion. Boston – New York: Mariner Books. London: Weinfeld & Nicholson. Edwards A.W.F. 2003. Human genetic di- te Nijenhuis J., van der Flier. 2014. Is the Fly- versity: Lewontin’s fallacy. BioEssays nn effect on g? A meta-analysis. Intelli- 25(8):798–801. gence 41: 802–7. Entine J. 2000. Taboo: Why Black Athletes Wade N. 2014. A Troublesome Inheritance. Dominate Sports and Why We’re Afraid New York: The Penguin Press. to Talk About It. New York: Public Affairs. Wilby P. 2016. Beyond the Flynn effect: new La Vie des Idées. 2014. Interpréter la diversité myths about race, family and IQ? The humaine. Entretien avec Bertrand Jordan. Guardian, 27 September. Available at: Available at: https://www.laviedesidees. https://www.theguardian.com/educa- fr/Interpreter-la-diversite-humaine.html tion/2016/sep/27/james-flynn-race-iq- Lewontin R.C. 1972. The Apportionment of myths-does-your-family-make-you-smart- Human Diversity. Evol Biol 6:381–98. er Li J.Z., Absher D.M., Tang H. et al. 2008. Xing J., Watkins W.S., Shlien A. et al. 2010. To- Worldwide Human Relationships Inferred ward a More Uniform Sampling of Human from Genome-Wide Patterns of Variation. Genetic Diversity: A Survey of Worldwide Science 319(5866):1100–4. Populations by High-Density Genotyping. Genomics 96(4):199–210.