2. The scientific controversy

The mere fact that populations differ in IQ tells us little or nothing about the source of the difference. The difference could be due to genes, but it could also be due to some environmental factor(s) – education, nutrition, cultural bias in IQ tests, and so on. How can we disentangle the effects of genes and environment? Behavioral geneticists have two principal methods for probing the role of genes in phenotypic differences: (a) behavioral genetic studies on adoptees and twins, and (b) genome-wide association studies.

The heritability of a trait – the proportion of the variance in a population explained by genetic differences – can be determined by adoption and twin studies. If adopted children grow up to resemble their biological rather than their adoptive parents, this implicates genes. If (genetically identical) monozygotic twins raised together grow up to resemble each other more than (genetically related but nonidentical) dizygotic twins raised together, this, too, implicates genes (on certain testable assumptions, such as equal environments, or, in the case of adoption studies, random placement into adoptive environments). Adoption and twin studies show that, in the contemporary United States, the heritability of IQ in adulthood is around 60 to 80% (Plomin & Deary, 2015).

Even if IQ has high heritability within racial groups, this does not imply that race differences are genetic. We cannot infer between-group heritability from within-group heritability. Lewontin (1970, pp. 7–8) illustrates this point with the following example. Suppose we have a sack of genetically variable corn seeds, and we plant a handful on two plots of soil. Conditions within each plot are carefully controlled so that each plant receives the same nutrients, sunlight, and so on. Since there is no variation in the environmental factors that affect growth, differences in height within each plot will be due entirely to genes – height will be 100% heritable. But suppose one plot contains a low concentration of nitrates. The corn in the low-nitrates plot will grow shorter on average for purely environmental reasons. Within-group (i.e., within-plot) heritability will be 100%, while between-group heritability will be 0%.

Heritability studies cannot show definitively that race differences in intelligence have a genetic cause. It is always possible that there is some hidden environmental factor(s) – a so-called “X factor,” analogous to nitrates in the corn example – that explains differences between – but not within – races (an X factor that explains the Black–White IQ gap would have to affect all Blacks equally in order to preserve high heritability and shift the population mean downwards without changing the variance). Nevertheless, the high within-group heritability of IQ can be part of a package of evidence for between-group heritability. This point was made by Jensen (1969), who was widely misunderstood as naively inferring between- from within-group heritability (see Sesardic, 2005, pp. 128–138). Expounding on Jensen’s argument, Flynn (1980) writes that “the probability of a genetic hypothesis will be much enhanced if, in addition to evidencing high [heritability] estimates, we find we can falsify literally every plausible environmental hypothesis [i.e., X factor] one by one” (p. 40; quoted in Sesardic, 2005, p. 136). The obvious candidate X factor that could explain race differences is, of course, racism. If racism lowers IQ, this could explain why the mean is shifted downwards for victimized groups. However, as Flynn argues, attributing differences to racism

Now, after decades of intensive searching, the X factor remains elusive. The adult Black–White IQ gap has remained stubbornly constant at approximately one standard deviation (15 IQ points) among cohorts born since around 1970 (Murray, 2007). Dickens and Flynn (2006) report that “Blacks gained 4 to 7 IQ points on non-Hispanic Whites between 1972 and 2002” (p. 913), but these gains appear to be among Blacks born before the early seventies. Dickens and Flynn (2006, Figure 3) indicate that, in 2002, the Black–White IQ gap in among 20-year-olds was approximately one standard deviation, or 15 points. Nisbett (2017) writes that “Dickens and Flynn found [the Black–White gap in IQ to be] around 9.5 points,” but this is only the gap if we include children (as R. Nisbett confirmed in a personal communication, December 24, 2018). More recent evidence indicates that the gap has persisted or even widened. Frisby and Beaujean (2015, Table 8) find a Black–White IQ gap of 1.16 standard deviations among a population-representative sample of adults used to norm the Wechsler Adult Intelligence Scale IV in 2007. Intensive interventions can raise IQ substantially during childhood when the heritability of IQ is low. But despite some misleading claims about the success of early intervention programs, gains tend to dissolve by late adolescence or early adulthood (Baumeister & Bacharach, 2000; Lipsey, Farran, & Durkin, 2018; Protzko, 2015). Adoption by white families – one of the most extreme interventions possible – has virtually no effect on the IQ of black adoptees by adulthood. Black children adopted by middle- and upper-middle-class white families in Minnesota obtained IQ scores at age 17 that were roughly identical to the African American average. Adoptees with one black biological parent obtained IQ scores that were intermediate between the black and white means (Loehlin, 2000, Table 9.3).²

To reiterate, the high within-group heritability of IQ combined with the failure to find an environmental X factor to explain the IQ gap does not show decisively that race differences are genetic, because it is possible that an X factor will be discovered in the future. However, the environmentalist theory of race differences has not, by normal scientific standards, been an especially progressive research program (in the sense of Lakatos, 1970). Environmentalists never predicted that the Black–White IQ gap would, after reaching one standard deviation, remain impervious to early education, adoption, massive improvements in the socioeconomic status of Blacks, and the (apparent) waning of overt racism and discrimination. Commenting 45 years ago on environmentalist theories that appeal to an X factor, Urbach (1974) noted that “any data in the world can be made consistent with any theory by invoking nameless and untested factors” (p. 134). Nevertheless, we cannot technically say that the environmentalist explanation for the IQ gap has been falsified. The fact that the gap did narrow since the early twentieth century gives some credibility to the idea that environment is playing a role.

Let us turn to the second method for investigating the role of genes in development: genome-wide association studies (GWAS). Unlike heritability studies, GWAS can uncover specific genetic variants – or single-nucleotide polymorphisms (SNPs) – associated with IQ. In just the last couple years, GWAS has identified hundreds of such SNPs (Davies et al., 2018; Savage et al., 2018; Sniekers et al., 2017), which together explain around 11% of the variance in IQ (Allegrini et al., 2019).

If we find that the SNPs implicated in IQ are differentially distributed across racial groups, this would not necessarily imply that race differences in intelligence are genetic. SNPs might have different effects across races and environments due to gene–gene and gene–environment interactions. SNPs with no causal relation to intelligence can be genetically linked to SNPs that do have a causal relation in some populations but not others, so SNP–intelligence correlations may not always hold across races (Rosenberg, Edge, Pritchard, & Feldman, 2019). But if we find that many of the same SNPs predict intelligence in different racial groups, a risky prediction made by the hereditarian hypothesis will have passed a crucial test. Even then, however, GWAS will only establish a correlation between SNPs and IQ without revealing the causal chain linking SNP to phenotype. It would still be theoretically possible that these SNPs lead to differences in intelligence as a consequence of environmental factors (e.g., parenting effects) that can be manipulated so as to eliminate race differences. But if work on the genetics and neuroscience of intelligence becomes sufficiently advanced, it may soon become possible to give a convincing causal account of how specific SNPs affect brain structures that underlie intelligence (Haier, 2017). If we can give a biological account of how genes with different distributions lead to race differences, this would essentially constitute proof of hereditarianism. As of now, there is nothing that would indicate that it is particularly unlikely that race differences will turn out to have a substantial genetic component. If this possibility cannot be ruled out scientifically, we must face the ethical question of whether we ought to pursue the truth, whatever it may be.

3. Should science be absolutely committed to truth?

Truth was traditionally thought of – and is still often advertised – as the fundamental commitment of science (e.g., Dawkins, 2006). Some of the great heroes of science – most famously Galileo – are celebrated for resisting moral and religious authorities who feared that free inquiry would bring disaster. But the fact that the censors of previous eras were wrong does not mean that those of today cannot be right (Dummett, 1981, pp. 287–288). Sesardic (1992), who ultimately defends free inquiry, observes that since the days when “truth was … regarded as the categorical imperative for scientists,” we have “learned from some dramatic instances that there is no guarantee that … new knowledge [will] serve the good of mankind” (p. 129). Just as physics can produce an atom bomb with the potential to destroy us physically, perhaps social science can produce an equivalent – a discovery that would undermine the social order and destroy us just the same.

Many scientists claim to be absolutely committed to truth while at the same time advocating or tolerating varying degrees of suppression of controversial work. In his famous article “The Responsibility of Intellectuals,” Chomsky (1967) makes the bald statement that “it is the responsibility of intellectuals to speak the truth and to expose lies.” Elsewhere, however, he says that scientists should not pursue the truth about group differences in intelligence because this could have dangerous social consequences (Chomsky, 1976, pp. 294–295). He says that while people “surely … differ in their biologically determined qualities,” it is wrong to investigate an association between group membership and IQ because to do so is to indicate “that the answer to the question makes a difference; it does not, except to racists, sexists, and the like” (Chomsky, 1988, p. 164; quoted in Cofnas, 2016, p. 486). “To anyone not afflicted with these disorders, it is of zero interest whether the average value of IQ for some category of persons is such-and-such” (Chomsky, 1988, p. 164).

Cofnas (2016) documented several prominent scientists and philosophers opposing the study or dissemination of findings related to group differences in intelligence. Howard Gardner, for example, who is known for his theory of multiple intelligences (Gardner, 1983), writes: “I myself do not condone investigations of racial differences in intelligence, because I think that the results of these studies are likely to be incendiary” (Gardner, 2001, p. 8). He also says that we should reject the work of hereditarian IQ theorists such as Arthur Jensen (whom he calls “the bad guys”) because even if they

The leading intelligence researcher Robert Sternberg (2005) argues that good science is characterized by “taste in the selection of problems to solve” (p. 295), and that it is in bad taste to investigate the genetic basis of race differences. Sam Harris, a neuroscientist famous for aggressively promoting a “reason-and-evidence” based worldview, recently interviewed Bell Curve (Herrnstein & Murray, 1994) coauthor Charles Murray. Harris accepted all of Murray’s claims about hereditarianism, yet in the introduction to the interview he said:

Although he does not argue that the information should be suppressed, he does believe that facts about group differences conduce to “pain,” and we might be better off not thinking about them.

A degree of censorship is already in operation when it comes to findings supporting hereditarianism about group differences. Mainstream media coverage of the race-and-IQ controversy almost always falsely claims that there is a consensus among the relevant experts that hereditarianism has been refuted. In fact, anonymous surveys reveal that a substantial proportion of experts on intelligence believe that there is a genetic component to race differences (Rindermann, Becker, & Coyle, 2016, 2020, Figure 3; Snyderman & Rothman, 1987, 1988). Research supporting both hereditarian and environmentalist explanations of race differences is routinely published in major psychology journals, particularly in psychometrics journals like Intelligence and Personality and Individual Differences. However, work supporting hereditarianism can be much more difficult to publish and disseminate, and research testing the possible genetic basis of race differences is rarely funded. As James Flynn (after whom the “Flynn effect” is named) notes, “if universities have their way, the necessary research [on race and intelligence] will never be done. They fund the most mundane research projects, but never seem to have funds to test for genetic differences” (Flynn, 2012, p. 36). Flynn (2018) says that “scholars at one of America’s most distinguished universities … admitted [to him] that they had never approved a research grant that might clarify whether black[s] and white[s] had equivalent genes for IQ” (p. 128). When he suggested some ways to test if there is a genetic component to the Black–White IQ gap, “they evaded the issue” (J. Flynn, personal communication, January 7, 2019). And because the state of the science is often misrepresented to the public (hence the common belief that there is no evidence at all for hereditarianism), findings are subject to de facto censorship. In some academic fields such as sociology and history, virtually all mainstream scholars refuse to consider the implications of group differences in intelligence for the problems they address. The questions before us are: should research on group differences be conducted?; should findings related to differences be publicly disseminated?; and finally, what role should philosophers – who often act as intermediaries between science and the humanities, and between science and society at large – play in supporting or opposing work on the genetics of group differences?

Arguments in favor of restricting scientific research generally, or research on group differences specifically, are almost always utilitarian: they contend that free inquiry can, in some cases, be expected to have an overall negative effect on human well-being (e.g., Block & Dworkin, 1974; Chomsky, 1976; Dummett, 1981; Kourany, 2016). The present paper argues that utilitarian arguments in favor of restricting research on group differences in intelligence have generally given short shrift to the potential and actual serious harms that have resulted from the current practice of stigmatizing and dismissing controversial work, and they have also ignored the potential benefits that might follow from openly confronting evidence about group differences.

This paper suggests that there are also at least two compelling non-utilitarian reasons to allow free inquiry (compelling, that is, to those who are not utilitarians). First, truth is (to some degree) valuable independent of its felicific consequences. Proscribing the study of group differences undermines the integrity of huge swaths of work in biology, psychology, economics, sociology, history, and moral and political philosophy. If we accept that truth is intrinsically valuable, then this is a cause for concern. Second, if not all groups have identical distributions of potential, then it is unjust to assume that some people must be blamed for average differences in performance among groups.

4. Utilitarian reasons to favor free inquiry

In 2001, virologists at the State University of New York at Stony Brook created a synthetic version of the poliovirus using strands of DNA purchased by mail order. They obtained the RNA sequence of the virus from the Internet, where it is freely available for all to see. Other researchers have published what are essentially instructions for recreating a smorgasbord of viruses that could potentially kill tens of millions of people were they to be released in the modern world (Kourany, 2016, pp. 784–785). In response, prominent medical authorities have called for restrictions on creating deadly viruses or publishing information that could help terrorists to do so (National Research Council, 2004).

Kourany (2016) asks: “is there any reason new policy constraints like these for the life sciences should not be put into effect for the social sciences – new policy constraints that include research guidelines for weighing societal harms of research against societal benefits … ?” (p. 786). As to the suggestion that the potential harm from deadly viruses is greater than the harm from research on intelligence differences, she urges us not to minimize the latter: intelligence research

The reality of all these harms cannot, however, be taken for granted. As for the situation in the United States, it is a well-established finding that the average levels of self-esteem in ethnic groups has, contra common wisdom, a negative relationship with average IQ. At least since the 1980s, Blacks have scored significantly higher than Whites, who in turn have scored higher than Asians, on measures of self-esteem (Bachman, O’Malley, Freedman-Doan, Trzesniewski, & Donnellan, 2011; Twenge & Crocker, 2002). Is the success of members of some ethnic groups impaired by hearing about research on intelligence differences? That depends on the strength of the effect of stereotype threat. According to the theory of stereotype threat, telling people that their group has a low mean score on a cognitive test (i.e., priming them with a negative stereotype) impairs their performance (Steele & Aronson, 1995). In recent years, stereotype threat has run into trouble.

The “replication crisis” in psychology has revealed that many widely touted discoveries – particularly in social psychology and, within social psychology, specifically in regard to priming effects – may be completely false (Kahneman, 2012; Open Science Collaboration, 2015; Yong, 2012). Some psychologists have suggested that stereotype threat could end up being another casualty of the crisis. Ganley et al. (2013) failed to replicate the effect in three large studies of female mathematics performance. Using statistical techniques to test for publication bias, Flore and Wicherts (2015) found evidence that publication bias may “seriously distort the literature on the effects of stereotype threat” (p. 25) on math performance in girls. In a pre-registered study of more than two thousand Dutch high school students, Flore, Mulder, and Wicherts (2018) failed to find any evidence that gender stereotype threat affects math performance. There has not yet been an authoritative replication attempt of studies on the effect of priming with racial stereotypes. As of now, we certainly cannot say that stereotype threat “has already been shown to cause significant harm” (Kourany, 2016, p. 787). If stereotype threat is not real, then conducting research on group differences in IQ will not harm lower-scoring groups via a nonexistent effect. Even if it does turn out to be real, it is not clear that it is responsible for “significant harm,” since the Black–White IQ gap remains unchanged even when stereotype threat is eliminated. Stereotype threat is alleged only to increase the gap under controlled test-taking conditions (Sackett, Hardison, & Cullen, 2004; Steele & Aronson, 2004, p. 48). If stereotype threat turns out to be both real and to cause some devastating harm, that could raise ethical problems for intelligence research – but even its reality is far from being established.

There is also an important sense in which restrictions on publishing instructions to create deadly viruses and restrictions on intelligence research are not analogous. In the former case, the proposal is to conceal knowledge while being open about our motives: we should not (according to the proposals) publish instructions about how to make deadly viruses because we do not want terrorists to use them to make biological weapons. In the latter case, however, the proposal is to, if necessary, misrepresent the state of knowledge while being dishonest about our motives – to assert that there are no innate differences between groups while pretending that this is an established discovery of science. Given the disanalogy, we cannot so easily make an inference from the acceptability of limits on publishing work in virology to the acceptability of censoring or prohibiting work on intelligence, even if findings concerning intelligence differences could be shown to produce some sort of harm.

It is also easy to overlook the harms that have been caused by uncritical commitment to environmentalism. Kourany (2016) comments:

But the reason that these programs, which Kourany rightly says ought to exist, have never been created is not because of racism but because of the taboo on talking about genetic differences among policy makers. No mainstream politician can acknowledge that there are differences that might call for the creation of a program to “work with the strengths and work on the weaknesses of every [ethnic] group to help make them the very best they can be.” It is hereditarians who have advocated these programs and environmentalists who have resisted them. The abstract to Jensen’s (1969) paper – the first major modern defense of hereditarianism – says (written in the third person):

Contemporary hereditarians have also called for tailored training programs (e.g., Gottfredson, 2005a, 2005b, p. 318; Lubinski & Humphreys, 1997). Accumulating evidence suggests that education can indeed be effective in raising specific cognitive skills (Ritchie, Bates, & Deary, 2015).

Decades ago, Jensen (1969) was vilified for predicting that Head Start – a program based on the environmentalist assumption that early intervention can permanently raise intelligence and academic performance – would not have its intended effects. In The Bell Curve, hereditarians Herrnstein and Murray (1994) marshaled compelling evidence that Head Start was not working.³ In 2012, a congress-mandated report by the United States Department of Health and Human Services (HHS) stated that the effects of Head Start disappear by third grade (Puma et al., 2012, p. xxi) – that is, Head Start was not working. Of course, the report did not endorse hereditarianism or recommend that Head Start be discontinued. It suggested that the program may confer some as-yet-undetectable benefits that last many years after participation.⁴ Nevertheless, in the years immediately following HHS’s report, the government increased funding for Head Start by hundreds of millions of dollars (Enriquez, 2016). If we had followed Jensen’s recommendation in 1969 to devote money to programs that were tailored to the strengths of different groups rather than to Head Start, around two hundred billion dollars would have gone to improving lives instead of accomplishing nothing that can be detected. This is one of many examples of how basing social policies on questionable scientific premises can have enormous opportunity costs.

5. Non-utilitarian considerations in favor of free inquiry

6. Conclusion: The responsibility of philosophers

The strategy – advocated by some influential scholars – of stigmatizing, suppressing, or downplaying evidence in favor of hereditarianism about group differences has been tried and has not worked. Research on this topic has been done and the results are widely available. Major psychology journals continue to publish work that deals openly with group differences (though researchers still debate about the relative contribution of genes and environment, and the question has not been settled definitively). Any measures that would be effective in preventing further work, such as those advocated by Kourany (2016), would have to be so severe that they would only attract even more attention to the findings they aimed to suppress. Science will carry on, and these questions will be answered. We should prepare in advance for the possibility that the genes underlying intelligence differences will not be distributed identically among ethnic groups. Failure to do this will only create a vacuum for “cranks rather than scientists” to opine on the nature and consequences of group differences (Anomaly, 2017, p. 293). Reich (2018) warns that if scientists “willfully abstain from laying out a rational framework for discussing human differences, [they] will leave a vacuum that will be filled by pseudoscience” (p. 258).

This paper has argued that the usual utilitarian reasons given for restricting intelligence research are not convincing and, in fact, there are strong reasons, both utilitarian and non-utilitarian, to favor free inquiry. For philosophers specifically, there is an additional consideration. For decades, the contribution of philosophers to this debate has consisted mostly in providing alternative explanations for evidence seeming to support hereditarianism about race differences (see Sesardic, 2000, 2005), and advocating various kinds of restriction and censorship (see Cofnas, 2016). This may be because hereditarianism is controversial, and philosophers are strongly disincentivized from pursuing lines of argument that lead to truly controversial conclusions. Testifying to how serious this problem is, Jeff McMahan, Francesca Minerva, and Peter Singer recently founded the Journal of Controversial Ideas, which will allow scholars to publish pseudonymously. Singer (2017) commented that “it’s unfortunate that such a journal should ever be considered necessary to enable controversial ideas to be published, but perhaps we have got to the point where it is.” It is not clear what kind of controversial ideas Singer had in mind, and the journal has not yet released its first issue, but it is hard to find a more controversial idea than hereditarianism about race differences in intelligence.

There is a danger for the philosophical community in putting our credibility on the line over the claim that race differences are entirely environmental. If work on genetics and neuroscience within the next decade produces convincing evidence that differences in measured intelligence among groups have a significant genetic component, there will be no way to conceal this information. The hereditarian explanation will have to be accepted, and people will know that philosophers were on the wrong side of the issue both scientifically and morally: scientifically, because we are supposed to be careful, disinterested commentators on scientific controversies, not activists supporting only the politically popular side; morally, because we did not help lay the groundwork for responding in a moral way to these facts that we should have known might be coming.