How to Diagnose Abhorrent Science

Abstract

Methodological developments in genetics have given rise to a new era of genomic race science. Like the notorious race science of previous decades, the new version seeks to identify genetic causes of racial differences in cognitive traits and outcomes, like IQ and educational attainment. Faced with accusations that this research program is just classic racism in updated methodological packaging, proponents often justify their work by acknowledging that it is “controversial,” while simultaneously defending the notion that controversy is a hallmark of dynamic scientific progress: something to be embraced rather than shunned. As this new genomic race science feeds a growing, and occasionally deadly, weaponization of genetics, it is time to ask: is this research “just controversial,” or is it something else, something worse? This manuscript proposes an evaluative framework for past, present, and future scientific research. Treated as a case study, the new genomic race science is more than just controversial. It is, by virtue of being both valueless and harmful, abhorrent science.

1. Race Science Déjá Vu:

From 2014 to 2017, University College London held a series of secretive conferences on intelligence¹. The 2016 program opened with a quote from prominent psychologist and eugenicist, Edward E. L. Thorndike:

“Selective breeding can alter man's capacity to learn, to keep sane, to cherish justice or to be happy. There is no more certain and economical a way to improve man's environment as to improve his nature.”^v

These clandestine eugenics conferences, which have since been investigated and suspended, were spearheaded by researchers who are currently leading efforts to reinvestigate the genetics of racial differences in cognitive ability with new genomic data: Emil Kirkegaard, Davide Piffer, and the late Richard Lynn, among others. Presentations covered a range of topics from dysgenic fertility in China to destabilization in Europe from non-Western immigrants, which “threatens the sustainability of European democracy, welfare, and civilisation” (Ibid, p. 6).

These conferences were a modern renewal of an uninterrupted history of contentious race science that dates back decades, from early-twentieth century eugenic ideas about different races with different genetic constitutions to Enlightenment-era ideas about different races with different origin stories ^2-7. Readers are probably most familiar with more recent episodes from the last fifty years. The uproar in the 1970’s caused by Arthur Jensen’s⁸ hypotheses about a genetic basis for differences among racial groups in IQ scores, and Chapter 13 of The Bell Curve by Richard Herrnstein and Charles Murray, was followed by somewhat of a hiatus. Extensive counterarguments had been made; Herrnstein, Jensen, and J. Philippe Rushton died. There were no new data. The IQ gap persisted, perhaps diminishing slightly but not dramatically⁹. Structural racism in American society, needless to say, persisted as well.

Among individual people as opposed to groups, IQ test scores remained heritable in the weak statistical sense on the basis of traditional twin and adoption studies⁷, as had been known for practically a century¹⁰. Biological breakthroughs that would reveal the basis of intelligence in the brain or genome were often promised but never realized¹¹. Most of the Academe outside the immediate fields of intelligence and behavioral genetics probably thought that the issue had been resolved for the better: race was a social construct, IQ tests were invalid, twin studies were methodologically flawed, The Bell Curve was thoroughly debunked¹². Scholars could be both scientifically informed and progressively minded. The liberal social science establishment could breathe easy¹³.

Not so fast: “Race science” is back. While the old guard still gets shouted down by protesters in public forums^vi, a new wave of race researchers has emerged, using questionable scientific techniques to investigate alarming hypotheses regarding the genomic basis of differences in cognitive abilities between racial and ethnic groups. There are many reasons for the return of race science. For one, it never really went away. Researchers with intellectual ties to the Jensen era continued to generate a body of research about relations between IQ scores, race, and national origin¹. Mainstream scientists of human intelligence and the genetics of behavior are probably more conservative, scientifically and politically, than the Academe at large, and many investigators never gave up on the idea that group differences in behavior might prove to have a genetic basis^14,15. The internet provided a platform for the publication of race studies that would have a difficult time surviving peer review¹⁶. Some of the best-known young investigators of hypotheses regarding race, for example Kirkegaard, Piffer, and John Fuerst, self-publish much of their work in openly racist journals (e.g., Mankind Quarterly), or in journals they edit themselves (e.g., OpenPsych and The Winnower), or on their blogs, and publicize it on social media².

The recent resurgence of race science is also attributable to the development of new kinds of evidence, technology, and scientific methodology^17,18. Behavior genetics has undergone a major transition, as twin and family studies have been supplanted by DNA-based methods, thanks to the scientific revolution in genomics that followed the Human Genome Project^19,20. Group differences in IQ notwithstanding, the revolution has had profound effects on the effort to understand the relationship between DNA and human behavior. During the Jensen era, the genetics of behavior was studied using the tools of traditional quantitative genetics: comparing the similarity of twins and other family members with varying degrees of genetic relationship. The major conclusion reached by these studies – differences in behavior are almost always partly heritable – was well-known before the discovery of DNA⁷.

Nowadays, molecular genetics makes it possible to search the entire genome for associations between DNA and variation in any human characteristic, with intelligence (often in the form of its proxies, like educational attainment) prominent among them³. When conducted as a means of understanding the behavior of individual people, the new DNA-based genomics of behavior represents a new stage in the ongoing social scientific enterprise of investigating the role that genetics plays in determining the probability that individual people will achieve a variety of psychological and social outcomes²¹.

Just as traditional twin-based behavior genetics was put to work in the service of racist hypotheses about the innate inferiority of groups defined by their race or ethnicity, the most recent generation of race researchers looks to the new genomic data for vindication¹⁶. Proponents of this research grant that it is “controversial,” while contending that controversial science is part and parcel of the normal and, indeed, healthy scientific process:

“…one of the most morally controversial areas of science is the study of group differences in psychology, particularly in intelligence”^vii

“Differences between human racial groups are perhaps the most controversial topic in all of the social sciences, with almost every conceivable fact being contested by two or more opposing factions”²³

“… 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…” (Cofnas 2020, p. 133).

Editors have also characterized the research as being controversial in nature. When publishing an article (Cofnas 2020) on the genetic bases of race differences, for example, the editors of Philosophical Psychology included an editorial that used the terms “controversy” or “controversial” three times but still concluded that the ideas “deserve to be disputed rather than disparaged” (Leeuwen and Herschbach 2020, p. 149).

The continued propagation of such research poses the problem of how to respond to it, a problem faced in many contemporary domains where the values of the public come up against a desire to find reasonable boundaries of scientific, scholarly, or political discourse. Framing the problem as such suggests one solution, as implicitly endorsed by the “just controversial” position: there are no boundaries, and the public, in our case the marketplace of scientific ideas, is best accepted as an unregulated free for all, whatever the cost in misinformation and denigration. We reject that position as more of an abdication than a solution, but at the same time fully recognize the importance of scholarly inquiry that is as unfettered as possible while remaining true to standards of accuracy.

Our goal in this paper is to propose a conceptual framework for thinking about these problems, which we call the ‘value-harm grid.’ The value-harm grid permits evaluation of scientific research on two axes: value and harm. We contend that most scientific research is both valuable and harmless, and consequently non-controversial. A smaller but significant portion of scientific research is controversial when it is either valuable-but-harmful, or valueless-but-harmless. It is the rare instance in which a research program is both valueless and harmful, at which point, we argue, it becomes objectionable to the point of abhorrence. Thus, the value-harm grid should help diagnose scientific research into three types: non-controversial, controversial, and abhorrent.

We do not wish to relitigate the long discussion of the scientific validity of race, although such issues cannot be completely abstracted from our main goals. So, for example, we will not consider in detail whether race is a biologically valid construct (Duster, 2005, 2006, 2015; Mallon 2004, 2006; Fullwiley, 2007, 2008, 2014; Fujimura, Duster, and Rajagopalan, 2008; Roberts, 2011, 2012; Morning, 2014; Kaplan 2010, 2015); the appropriateness of DNA ancestry testing ^29,30; the relationship between race and ancestry ^31,32; whether the statistical assumptions of twin studies invalidate their conclusions (Kaplan 2000; Block and Dworkin 1974a; 1974b), or whether contemporary epigenetics has obviated traditional concerns with the separate roles of nature and nurture of human individual differences ³⁶. We also set aside several ethical topics that are usually subsumed under the rubric of “human research participant” concerns, involving, for example, the privacy of genomic data or the harm that might be done to individuals in the course of giving them IQ tests ^37,38. Evaluative criteria and institutions for such concerns are already in place. We will also not emphasize the real-world steps that might follow a determination that a program of research is abhorrent, in terms of hiring, retention, tenure, publication, or funding, although we recognize any proposal that abandons the free-for-all model will inevitably lead to difficult decisions in those domains.

Finally, we acknowledge that our use of the term ‘abhorrent’ may invoke for some readers Leon Kass’s famous argument for the wisdom of repugnance ³⁹. Kass argued that some practices invoke moral disgust, which should be interpreted as evidence for the repugnant nature of those practices. Our argument works in the opposite direction. Our claim below is not that the new genomic race science invokes moral disgust, and so it should be understood as abhorrent. Our claim is that the new genomic race science is abhorrent because it is both harmful and valueless, and that abhorrent nature warrants moral disgust.

2. Non-Controversial Science, Controversial Science, and Abhorrent Science

Different scientific research programs elicit more or less critical scrutiny. The vast majority of scientific research is non-controversial. Scientists go about their business trying to find some new compound, or better understand how humans think about saving for retirement, or develop a new antibiotic. If successful, maybe that non-controversial research makes news reports, or maybe it doesn’t.

Scientific research that generates controversy comes in a variety of forms ⁴⁰. When scientists sequenced the genome of the influenza virus that caused the 1918 pandemic, critics raised concerns about genomic information falling into the wrong hands and being used for bioterrorism (Smith 2007). Gene-editing technologies, such as CRISPR-Cas9, have been proposed to create organisms that can breed with invasive species and drive them into extinction (a “gene drive”), but critics worry that it will be difficult and perhaps impossible to contain that technology and the organisms it produces ⁴². Research that is not inherently harmful but has no obvious value – on, say, crab sex research ⁴³ – is sometimes targeted as controversial because it is perceived as being a waste of resources. To understand the nature of controversial science we need to unpack these ideas of harm and value.

2.1. Harm.

Some scientific research poses a risk of harm. Harm comes in a variety of forms: physical, psychological, social, medical, or financial. Also, different entities can be affected by harm, such as individual people, communities, non-human animals, embryos, and ecosystems. The risk of harm associated with scientific research comes in degrees, and there are various measures of harm to consider. Harm may be more or less likely and more or less severe. It may impact many or few entities. Critics raised concerns about sequencing the influenza virus responsible for the 1918 pandemic because of the potential harm it posed. On the other hand, the likelihood of that harm transpiring was quite small.

2.2. Value.

Some scientific research is valuable. We mean “valuable” in a very broad sense. Research can be practically valuable, leading to medical interventions, technological applications, or policy guidance. It can also be theoretically valuable, facilitating conceptual advances in a science. As with harm, different entities stand to benefit from valuable scientific As with harm, value comes in degrees, with various measures of value to consider. The impact of the valuable outcome can be more or less significant and affect varying numbers of entities. A gene drive that eliminates invasive mammals from New Zealand would do a great deal to help restore that island ecosystem to its pre-colonial state. It’s potentially so valuable, in fact, that the New Zealand government considered including this gene drive as a part of their Predator Free 2050 plan. Conversely, a genome wide association study that identifies regions of the human genome associated with the ability to smell asparagus metabolites in human urine may satisfy some curiosity, but it has no obvious value beyond that, as the authors of such a study seem to grant with their tongue-in-cheek title, “Sniffing Out Significant ‘Pee Values’” (Markt et al. 2016).

One aspect of the value dimension is especially important for the matter at hand. Scientific conclusions can be valuable simply by being true, and scientific research can be valuable based only on its ability to elucidate that truth. The idea of truth-as-value is sometimes invoked in discussions of race and human behavior: Why should there be any criterion for evaluating a research program other than the truth of the conclusions it reaches? This argument seems epistemically virtuous, and in principle we agree with it. Nothing in this paper should be taken as an endorsement of suppressing findings that are true, at a high level of certainty made possible by good empirical science. The value of a research program, trivia like crab sex notwithstanding, can be established based on nothing more than its ability to reach true conclusions.

Saying this, however, is a reminder that definitive proof of scientific conclusions is a rarity in human behavioral research. Social scientific research into the ways personality influences job success, or how marital quality influences recovery from depression, may have value for all sorts of reasons, but rigorous elucidation of truth isn’t among them. Human behavioral science is, in the absence of randomized experiments, intentional breeding, dissection, or random assignment to living conditions, inherently speculative. Usually, the low certainty of social science doesn’t matter, because the research in question is equally low on the potential for harm dimension, allowing it to be appreciated for whatever soft-scientific value it may have, even in the face of manifest uncertainty.

2.2.1. Value-Harm Interdependence and Scientific Validity.

In most cases – perhaps all – assessment of the harm associated with scientific research will impact assessment of the value associated with that same research. That is, the reasons scientific research on crab sex is relatively harmless is in part attributable to the relatively limited value of the research. Conversely, potential applications of CRISPR-Cas9 gene-editing research for gene drives is harmful, in part, due to the inherent value of the research: if CRISPR-Cas9 were of limited value, it would likely be of limited potential harm.

Although a rigorous philosophical assessment of scientific validity and its relationship to value is beyond the scope of this project, a few examples point to the potential independence of the two concepts. The aforementioned genome wide association study of asparagus anosmia provides an example of a valid scientific research project that exhibits limited overall value. For all practical purposes, the research is scientifically valid – the authors developed careful research design, followed meticulous and robust methodological practices, systematic data collection, appropriate statistical analyses, published the research in reputable, peer-reviewed journals, etc. However, we contend that if one were to place genomic studies of asparagus anosmia on a spectrum from minimally valuable to maximally valuable, it would likely fall on the former end of the spectrum, in comparison to most scientific research endeavors.

On the contrary, the potential value of scientific research or theories of limited validity warrants consideration. It can be argued that studies yielding invalid results can offer valuable lessons by enabling scientists to glean insights from their errors, fostering critical discourse about the formulation of hypotheses and research methodologies, and serving as catalysts for the exploration of new avenues of investigation. Moreover, the indirect utility of certain invalid research merits examination. For instance, investigations into the efficacy of prayer, while not meeting the rigorous standards of empirical inquiry due to the inherent limitations in demonstrating causal connections with divine intervention, may still provide meaningful contributions. Despite the scientific impossibility of discerning the direct influence of divine forces, such inquiries can shed light on the intricacies of the placebo phenomenon and the psychological and emotional implications associated with prayer practices.

2.3. The Value-Harm Grid.

Combining considerations of harm and value allows for diagnosing what makes some science controversial (Figure 1). Some science is non-controversial because it is both unlikely to produce harm and likely to produce something valuable (Figure 1, upper-left quadrant). Most well-known research programs, like the antibiotic nature of penicillin, fall into this category, but most non-famous scientific research falls into this category as well. If you haven’t heard of the research, then it’s probably non-controversial, and it’s non-controversial because it has at least the potential to produce some value without doing considerable harm.

Figure 1:

The Value-Harm Grid.

Research can be controversial in one of two ways. First, it can be controversial because it has the potential to produce harm but is likely to generate something valuable (Figure 1, bottom-left quadrant). The sequencing of the influenza virus responsible for the 1918 pandemic falls into this category. The potential for harm was described above, but the value was clear too. Understanding more about what made an influenza virus turn pandemic could be extremely valuable for assessing new influenza viruses as they come along, so as to judge their health risk by way of their similarity/difference to the past virus ⁴⁵.

Research that is not likely to produce harm but is unlikely to produce something valuable is also controversial, but in a different sense (Figure 1, upper-right quadrant). The controversy in these cases is about whether the time, money, and effort invested are worth it when no appreciable value can be identified or expected. There are, for example, investigations of the effectiveness of prayer as a means of intervening in the world ⁴⁶; there are also Joseph Rhine's (1934) infamous investigations of extrasensory perception. The University of Virginia Division of Perceptual Studies (DOPS) conducts research on mystical experiences ⁴⁸, out-of-body experiences ⁴⁹, telepathy ⁵⁰, parapsychology ⁵¹, children’s past life memories ⁵², and mediumship ⁵³. Reasonable people can disagree about exactly how valuable any of these programs are; although their value may be questionable, so is their potential for harm, aside from the resources they waste. Controversy surrounds research programs of this nature, as they lack a distinct demonstrable value, yet are deemed benign.

3. The New Genomic Race Science: a case study in abhorrent research.

There is one more cell in the value-harm grid: scientific research that has little value and poses significant harm. We call such science “abhorrent” (Figure 1, bottom-right quadrant). It is abhorrent because it serves no end other than to cause harm. Precisely when scientific research crosses over from valuable to valueless or from harmless to harmful (and, thus, from controversial to abhorrent) is open to debate, hence the blurry boundaries between the quadrants of the value-harm grid; our purpose here is not to precisely quantify the axes, but rather to establish that there exists scientific research that is so clearly harmful and devoid of value that it occupies a unique location on the value-harm grid.

Our case study is the genomic race science that we described at the outset, the newest incarnation of a long history of race research that is as old as human biology and anthropology ⁵⁴. Contemporary genomic race science dates back to a program of late 19^th century human genetics devoted to confirming the racist claim that Black people were both biologically different from and inferior to white people.

Eugenicists of the early-20^th century had a fixation on race. They believed races were clearly defined or even separately evolved, and the prevalence of desirable traits (like large brains) and undesirable traits (like criminality) could be tracked in those different races starting in the womb ⁵⁵. Biological claims were then used to justify policies that targeted Black Americans. There were anti-miscegenation laws in more than half the states of America, which aimed to limit “race mixing” by prohibiting interracial marriages (Lombardo 1987; Pascoe 2009). Involuntary sterilization was also legalized across the US and in many regions, particularly in the South, Black Americans were disproportionately targeted on eugenic grounds ^58,59.

Even though race science became stigmatized in the mid-20^th century, many researchers with an interest in the biological underpinnings of race differences continued looking for genetic causes (Tabery 2020, 2023). They just stopped calling themselves “eugenicists” and “race scientists.” Arthur Jensen’s approach to race science relied on heritability estimates derived from twins and adoptees. IQ, in his estimation, had a high heritability – as high as 80%. This meant that the group difference for IQ between Black Americans and white Americans was unlikely to be entirely due to environmental differences between the groups.

3.1. What is the New Genomic Race Science?

In examining the case study of 'the new genomic race science,' it is essential to recognize its evolution and distinguish its contemporary features from the rich history of race science described above. Rather than delving into intricate technicalities of modern genomics, here we will introduce a few pivotal concepts. The current iteration of race science predominantly relies on the methodologies and evidence arising from contemporary genomics. Unlike the era of Jensen's ‘quantitative genetics,’ which leaned heavily on twin and family studies, researchers now turn to the outcomes of genome-wide association studies (GWAS). GWAS pinpoint statistical associations between minute variations in the human genome, known as 'single nucleotide polymorphisms' (SNPs), and behavioral traits or outcomes like height (or IQ).

A critical outcome of GWAS is the formulation of polygenic scores (PGS). At its core, a PGS serves as a genomic predictor, assigning an individual's risk for a specific trait or outcome on a percentile scale, ranging from very low to very high, based on a sample of DNA. The 'contribution' of individual SNPs to a particular trait or outcome is quantified by their 'effect size.' PGS, then, is simply the cumulative sum of all effect sizes associated with SNPs linked to a given trait or outcome. For instance, a PGS for IQ encapsulates the aggregation of myriad minuscule effect sizes derived from SNPs statistically linked to IQ in the results of a GWAS.

While delving into the nuances of these cutting-edge genomic methodologies can be extensive, it is crucial to keep a key perspective for the ensuing case study. What distinguishes the 'new' in genomic race research is the incorporation and reliance on the concepts and techniques of contemporary genomics, such as GWAS, SNPs, and PGS. Understanding this shift provides a foundation for comprehending the advancements and implications within the context of our case study.

3.2. The New Genomic Race Science Is Harmful.

The new genomic race science is harmful in a number of ways. Most obviously and dangerously, it fuels the weaponization of genetics and genomics. When consumed by alt-right extremists and white nationalists, it has fed white supremacist ideologies including the 2022 mass murder of ten Black people in Buffalo ⁶¹. A recent qualitative study of online ethnographic domains revealed that contemporary genomics – including papers published by new genomic race researchers – are “mobilized” by white nationalists in various online forums, archives, and blogs (Panofsky, Dasgupta, and Iturriaga 2021). The human biodiversity movement, leveraging empirical evidence to advance hereditarian explanations for racial disparities in social and biological outcomes, serves as a primary repository for papers by new genomic race researchers and constitutes a foundational resource for white nationalists; many of these researchers' articles are published in openly biased academic journals, recognized for accommodating 'controversial' race science, such as OpenPsych and Mankind Quarterly.

The harm, however, needn’t be overt violence. There are other, more insidious ways that the new genomic race science does harm when it contributes to a long and disturbing history of research that has been used to promote or justify discriminatory ideologies and policies. Early research on race and IQ was used to justify demonstrably harmful eugenic policies, including anti-miscegenation and forced-sterilization laws in the US. Although such policies are no longer in place, it should not be difficult to see the degree to which research presented as “science” purporting to demonstrate that some races are inferior to others has, is, and will continue to give rise to social, psychological, and even physical harm. In and of itself, racism is a major source of human harm. The new genomic race science is harmful because it advances and promotes racist, discriminatory, and prejudicial attitudes, beliefs, ideologies, and volitions.

The harms caused by racism are extremely well-documented in three areas: medical, psychological, and social ^62-64. Glaring racial disparities in health outcomes are characteristic of medical harms caused by racism ^64,65. Racial disparities in health outcomes indicate that diseases that vary equally across races exhibit variable mortality rates across races ⁶⁶. The association between racial disparities in health outcomes and racism is also extremely well-documented in studies on racial disparities in diagnosis and treatment of disease. Although varying in disease phenotype and treatment, racial disparities in diagnosis and treatment are demonstrative of the following pattern: white people are significantly more likely to be diagnosed and treated for a disease than Black people ^67-69.

Psychological harms of racism are also well-documented (Schmitt et al. 2014; Pascoe and Richman 2009). Racist beliefs, ideologies, and volitions are sources of discrimination, prejudicial attitudes, and stigma, which in turn result in maltreatment and a greater number of traumatic or stressful life events (e.g., acts of violence) for members of groups subject to racism. Broadly construed, the combined effects of such detrimental events result in an overall negative impact on psychological well-being, which may comprise measures such as self-esteem, depression, anxiety, or other forms of psychological distress ^63,72.

Racism is also a major source of social harm ⁷³. Social harms are best represented by drastic racial disparities in social goods, such as voting rights, education, healthcare, employment, housing, incarceration, and income/wealth inequality. Black Americans are poorer than Non-Hispanic, white Americans and live in substantially less valuable homes ⁷⁴. Historically, Black Americans have had egregiously unequal voting rights compared to other racial groups. Despite decades of political efforts to rectify racial income inequality, white Americans still outperform Black Americans in hourly wages for the same jobs in both private and public sectors ^74,75. Racial disparities in incarceration rates were extremely pronounced from 1880 to 1950 ⁷⁶ and still persist today where the incarceration rate of Black youth is 4.6 times higher than that of white youth ⁷⁷. Importantly, such disparities are not the result of greater involvement in drug use or trade, but rather, result from “discretionary decision making by law enforcement agencies as well as enactment of harsh sentencing policies by both state and federal lawmakers” ⁷⁸.

As mentioned in our introduction, many of the researchers who comprise the new genomic race research program have either helped organize or participated in the series of clandestine, invitation-only (and now defunct) London Conference on Intelligence, which hosted presentations on a range of controversial topics from the genetics of racial differences in intelligence to eugenics. Further, many of the researchers or the journals in which they publish are funded by agencies and organizations with racist and white supremacist ties, such as the Pioneer Fund and the Ulster Institute for Social Research. Therefore, although the new wave of genomic race researchers are more careful than their predecessors with respect to making explicitly racist claims in print, their social, organizational, and institutional venues of public uptake are all intertwined with racist, discriminatory, prejudicial, white nationalist, and eugenic ideologies.

3.3. The New Genomic Race Science is Valueless.

There was a time when race scientists believed in the practical benefits and policy implications of their work. The original eugenic justifications of race science have been abandoned, or at the very least are rarely spoken out loud. Arthur Jensen, in 1969, asked “How Much Can We Boost IQ and Scholastic Achievement?” He was shining an accusatory spotlight on compensatory education programs like Head Start that attempted to eliminate the IQ gap between Black and white Americans by spending federal dollars on the cognitive-educational environment of Black youth. Taxpayer dollars should not be wasted on futile attempts to eliminate the IQ gap by improving the cognitive-educational environment of Black youth; the gap was at least in part genetic. Instead, Jensen suggested different educational tracks for Black and white youths, each uniquely geared towards their innate biological capacities ⁸. Once again, such explicit calls for race-based social policies are rarely heard nowadays, even at the fringes of scientific discourse. Sympathetic discussions of racial hypotheses now include the caveat that people (like job applicants) deserve to be considered as individuals, based perhaps on their IQs, but not on their race.^viii

So why conduct race research if it has no practical value? The implicit answer is that the question of genetic origins of racial differences in intelligence is now a matter of basic science. If the genetic basis of racial differences is true, and if there are research paradigms that can demonstrate that it’s true, why not do the science and let the chips fall where they may? As we stated in the explication of the value dimension in the grid, we agree in principle. If there were research programs that could provide reasonably definitive answers to these questions in a non-biased way, it would be difficult to justify not conducting them, and scientific genies are difficult to keep in the bottle in any event. But there are no genies. To be clear, we recognize that it is possible to collect data that are less than completely irrelevant to the question of group differences in intelligence, in the social scientific sense we have described. One could hand out questionnaires to people asking whether it seemed to them that Group X is more intelligent than Group Y, but doing so wouldn’t be remotely conclusive, so it wouldn’t have substantial scientific value, and any limited value it might have would not outweigh the obvious harm it would cause. But the race scientists aren’t just handing out questionnaires, so we will now consider what they are actually doing.

3.3.1. Variance and Causes Revisited.

In its present form, the new race scientists appeal to genomic data to build evidence for the claim that differences in cognitive ability between racial groups are caused, in part, by genetic differences between those groups. The hypothesis sounds like it makes sense, but what exactly does it mean for a group difference to be partially genetic? Coming to grips with the specific content of the genetic hypothesis is at the heart of the question of the value of the research programs it motivates. The conceptual ambiguity of genetic hypotheses about group differences means that any currently imaginable empirical result will fail to get the race researcher to the intended outcome of genetically based group differences in behavior. You can’t get there from here.

We will assert a straightforward meaning of the claim that a group difference in a behavior has a genetic basis: it means that some portion of the difference that is currently observed would remain even if all environmental differences between the groups could be eliminated or somehow controlled. Think about what this means for differences in intelligence between Black and white Americans. Over the last 500 years, Black people have been kidnapped in Africa, shipped to the Americas under unspeakable conditions, enslaved, disenfranchised, segregated, and discriminated against. For the last fifty of those 500 years, very partial legal remediations have been in place. Does anyone doubt the “environmental effects” of that history on contemporary populations? There is no way to equalize those historical conditions experimentally or control for them statistically. The very idea is ridiculous. The problem is only made worse by the fact that causal environmental effects on intelligence, while not exactly easy to study in humans, have been quite well established using a variety of quasi-experimental designs ⁸⁰.

The insurmountable inferential obstacle imposed by irreducible environmental effects render any imaginable investigation of genetic explanations of group differences inconclusive. Garden variety heritability of IQ differences among individuals is a demonstration that degree of genotypic similarity (in the form of either familial structure or molecular data arrayed on SNP chips) is correlated with degree of similarity in IQ. Although the relation between group membership and IQ cannot be reduced to a percentage in quite the same way as an individual-level heritability, an IQ difference between groups that differ genetically is itself a variety of correlation between genotype and IQ. Ultimately, this is the basis of the race scientists’ intuition that group differences must be at least partly genetic. Intelligence is heritable in individuals, and groups differ in both genotype and IQ, so why wouldn’t the group difference be genetic? As the science podcaster Sam Harris said to The Bell Curve author, Charles Murray, “it’s just straight biology” ⁸¹.

But it isn’t straight biology, for the usual reason that correlation doesn’t imply causation. Ultimately, the straight biology argument is defeated by a reductio: all human behavioral differences are heritable, so why aren’t all group differences in behavior, from the consequential to the trivial, “partially genetic?” If everything is partially genetic, nothing is. The argument from environmental confounds of correlational observations also offers the hereditarians a way out. If, outside the context of race, hereditarians could identify a gene with a generalizable and quantifiable (and especially a positive) effect on intelligence, the hereditarians would be halfway there. Much as we might doubt the likelihood of success, they would hear no objections to their effort from us. Gene hunting might be controversial, but it isn’t abhorrent. If such a gene were ever documented, it would be a short step to testing whether it occurred more frequently in some groups than others.

3.3.2. Comparing Polygenic Scores.

Aside from problems with causal interpretations of genome-wide association study results ^82,83, the new genomic race research program conducts analyses well-known to be problematic in the broader genomics community: extrapolating genetic results to populations of ancestry different from the original GWAS sample. To date, a vast majority of the SNPs identified by GWAS have been derived in genetically homogenous populations comprised of individuals of European ancestry ⁸⁴. Yet GWAS results are subject to a problem of portability: results estimated in one population do not generalize to others ⁸⁵. Polygenic scores derived from one single-ancestry population are next-to-meaningless when applied to populations of ancestry different from the original GWAS sample. For example, PGS for height derived from populations of European ancestry inaccurately predict Africans to be shorter than Europeans and slightly taller than East Asians ⁸⁶.

Although this problem of portability is well-documented and more or less universally accepted in population genetics ⁸⁷, these methodological cautions are routinely flouted by the new genomic race researchers who make inferences about populations of non-European ancestry from genomic data derived from populations of European ancestry. Consider Piffer (2015), who uses the SNPs that had been shown to be correlated (almost exclusively in European samples) with educational attainment to compute a polygenic score which he then compares to the average IQ scores of 37 countries. The mean polygenic score and the mean IQ are highly correlated.

There is a lot one could say about this study, none of it good. For present purposes, however, our point is not, or not only, that the conclusions Piffer draws are incorrect, in the sense of contending that if he had conducted or interpreted the study correctly, he would have reached environmental rather than genetic conclusions. Instead, we say this: once one is finished comparing invalidly estimated polygenic scores with invalidly estimated mean IQ scores, the results are not so much incorrect as they are irrelevant. It doesn’t matter what the correlation is between those two ill-gotten columns of data. The study simply has no scientific value.

3.3.3. Admixture Studies.

A somewhat more plausible form of genomic race analysis is called an admixture study. Admixture refers to the mixing of genetic ancestries that exist to one degree or another in everyone. In some biological or medical contexts, admixture can be used as a legitimate and valuable scientific tool for identifying genes responsible for medical conditions, as David Reich famously did with prostate cancer ^89,90. The new genomic race researchers use the same tools only in a simpler and less methodologically scrupulous way (e.g., Kirkegaard et al. 2019). In samples of African Americans, manifesting an admixture of African and European ancestry, they estimate percentage of European ancestry and correlate it with IQ. Finding those correlations to be small but positive and statistically significant, they conclude that their results are “consistent with” a genetic explanation of differences between African and European populations; the more European the DNA, the higher the IQ scores.

But wait a second. How do African American populations get “admixed” in the first place? The answer to that question is certainly very complex, but one thing for sure is that populations aren’t admixed at random. Presumably, African-European admixture has occurred over the generations because of consensual and non-consensual interbreeding, and it is perfectly plausible that certain Europeans and Africans were variously selected for interbreeding, and that the interbreeding granted the offspring of those pairings some of the unquestionable environmental racial advantages of their European parent. In any genuinely scientific analysis of this problem, the absence of environmental effects would be an assumption without which the whole enterprise doesn’t make any sense; in these admixture studies environmental effects are the alternative hypothesis.

The authors of the study are aware of the problem. A longer form of their conclusion is telling: “While the results we found are consistent with an evolutionary model, there are some potential alternative explanations, namely: phenotypic discrimination, confounding due to immigration status, confounding due to geographic location, and intergenerational environmental transmission” (Ibid). They then proceed to apply statistical controls for the competing environmental hypotheses, make post hoc theoretical arguments, compare their results with the previous literature, call for future research with larger and more representative samples.

To be fair, the feckless attempt to clean up the inconclusiveness of one’s main analysis in the discussion section is a hallmark of normal social science, and if the truth be known the authors do a reasonable job of it by the usual standards. If it didn’t matter so much, if this were another of the thousands of inconclusive non-experimental studies of human behavior that are published every year, it could be chalked up as not very valuable in its lack of rigor but mostly harmless in its consequences: perhaps controversial, but not worse. Unfortunately for all concerned, however, the conclusions of this study do matter. These questions are nothing to speculate about. The study is of little value because it is incapable of discriminating the alternative hypotheses, predicting cognitive differences given genetic ancestry, or providing meaningful explanations of racial disparities in IQ.

3.4. Additional Putative Examples of Abhorrent Science.

This analysis focuses on genomic race science as a case study of abhorrent research, utilizing a value-harm grid that can be applied broadly. Examples of similarly abhorrent research may include anti-vaccine studies, ineffective gay conversion therapy research, and historical instances like the Tuskegee Syphilis Study. Each case exhibits limited or questionable value, with anti-vaccine research fostering misinformation, gay conversion therapy being widely recognized as ineffective, and Tuskegee experiments continuing despite the availability of penicillin. The harms are clear: anti-vaccine research contributes to vaccine hesitancy, gay conversion therapy causes severe psychological harm, and untreated syphilis leads to unquestionable harm.

4. Conclusion

We repeat something we stated at the outset but bears repeating. Our analysis makes recommendations for the terms of the discussion about race research and is not intended to determine its outcome in any instance beyond the several examples we have presented. Consideration of particular research programs will continue to be controversial, as they should be. People can disagree in good faith about the value and potential harms of scientific endeavors. Scientific truth will out, and we will say again that nothing in our analysis should be taken as endorsement of suppressing the truth. But for better or for worse, in the domain of human behavioral science, scientific truth – as opposed to data collection – is a difficult goal to achieve. The truth values of social scientific conclusions are ambiguous, and as a result, the desirability of research programs in which anything is at stake cannot escape the precarious balance of value and harm.

Researchers behind the new genomic race science defend their work from critics by claiming that it’s “just controversial.” Indeed, they wear controversy as something of a badge of honor – a sign that they are willing to bravely follow the data no matter where it goes, even if the destination is socially and morally tumultuous. The conceptual framework we have introduced here – the value-harm grid – however, suggests otherwise. Whether a program of research is just controversial (or something worse) hinges on a balance between its practical or theoretical value and the magnitude and probability of the harms it may produce. On this analysis, the new genomic race science is not just controversial science, but rather, abhorrent science. It is abhorrent because it is both valueless and harmful.