“Yet the Problem Remains”: Why Genetic Determinism Still Haunts Biomedical Research

ABSTRACT

After the horrors of the Holocaust and its connections to eugenics were revealed to the world, many post‐war population geneticists sought to establish rhetorical distance from the Nazi's state‐led campaigns, without abandoning their belief that actively shaping the population's genetics would produce a prosperous society. While many advancements in equity have come, well‐intentioned geneticists continue to struggle to distance themselves from those weaponizing genetic data. Herein, we outline the practices which often prevent modern population genetics from ridding itself of the ghost of genetic determinism. These include the interlocking practices of: agnostically noting “genes matter” without detailing the degree or context of importance; using examples or explanations which can be ambiguously interpreted; using flawed data to assert that medical advances require genetic insights; homogenizing the environment in ways that ignore, rather than account for, deleterious exposures; insisting that populations be mathematically segregated prior to performing their analysis; then, when predictable harms arise, using lack of intent to harm as an excuse to refuse to change standard practices or to accept responsibility. Finally, discrete recommendations for improvements are presented. Unlike denouncements of overt eugenics, it is controversial to argue that those working on the genetics of diseases like asthma may unwittingly echo the eugenicists of the past. However, the consistency of associations between modern population genetics and historic eugenics demand mechanistic evaluation.

After the horrors of the Holocaust and the full scope and significance of eugenics campaigns were revealed to the world, many eugenicists experienced a crisis of reputation rather than a crisis of conscience. One prominent eugenicist, Robert Cook, who for many years was editor of the Journal of Heredity from 1922 to 1965, encapsulated the sentiment when he stated, “The adoption by Adolf Hitler of a programme of ‘race purification’ based on superficial and perverted derivatives from the naive concepts of the early eugenic enthusiasts understandably engendered opposition to any programme of eugenic reform. Yet the problem remains.” [1]. The “eugenic enthusiasts” of the post‐war period sought to establish rhetorical distance from the Nazi's state‐led campaigns, without abandoning their belief that actively shaping the population's genetics would produce an improved body politic where the ‘fit’ would be more numerous, and the ‘less fit’ nonexistent.

This commentary will outline some specific ways the biomedical research community recapitulates the genetic determinism which was instrumental to eugenics after WWII. We argue that the habits of detrimentally reducing historical, social, environmental, and biochemical complexities has remained an unintended contributor to the weaponization of genetics research against underprivileged groups. The practices outlined limit the ability of well‐intentioned geneticists to distance themselves from those weaponizing genetics, which harms both the public and those in the academic field. We conclude that the harms to both the public and academia we present should prompt discussion about the ethics of continuing any genetic assessments which can be racialized and exploited by extremist actors.

1. Bias Versus Weaponization

Before delving further, we acknowledge the deep roots of medical racism in the United States and point the reader to the works which demonstrate how the dynamics of race within the context of genetics are uniquely harmful [2, 3, 4]. These authors point out that, unlike other biologic research fields, the use of genetics to support racial determinism for disparities predates the molecular insights which are now ubiquitous (due to ancestry sites, lay reports on genetic studies, and popular media). However, while there are numerous examples of racial and ethnic biases in biomedical research [5], bias is different in kind from weaponization. Bias is often unconscious, even if still malicious. Weaponization has a specifically purposeful component which makes its harms less ambiguous when assessing whether the acting agent can be held responsible. When research causes unintentional harm through fostering implicit bias, mitigations may be achieved by education, communication, or updated guidelines. Examples include recent campaigns to remove racial biases in renal function calculations [6] or in pain management [7]. In contrast, when research is weaponized, researchers have a greater responsibility to articulate a favorable cost‐benefit analysis and to prevent future weaponization of their work. Thus, given the continued risk of weaponization, the greater accessibility of genetic research heightens the need to grapple with the historical perils of conjoining genetics to explain differential outcomes in behavior and health.

2. Genetic Agnosticism

The term “genetic agnosticism” could describe the contemporary practice among geneticists of relaying that “genes are important” without explaining precisely why, how, or to what degree “genes matter” within the context being discussed. Importantly, genetic agnosticism leaves such interpretations up to the audience to answer and actively invites speculation. The terms “complex” and “multifactorial” can be applied to both Alzheimer's disease and asthma even though asthma hospitalizations can be predicted by air quality monitors across genetically unrelated populations [8]. “Complex” may also be used to describe how possible genetic or environmental drivers of a condition may be so unclear that it is simply declared a “geneticist's nightmare”, as was done for diabetes mellitus [9]. Ascribing complexity to genetics remains a commonplace excuse for the lack of progress [10].

The ambiguity of genetic agnosticism is sometimes praised as a ‘middle ground’ between the perceived extremes of genetic claims. At one imagined extreme, critics of population genetic practices are falsely accused of “dishonoring diversity” [11]. In reality, critics are offering advice for avoiding misappropriation of research [12], highlighting evidence for industrialization as the driver of common diseases [13, 14], or noting the widening gap between the degree of genetic contributions imagined by twin studies and real world observations (often euphemized as “missing heritability”). At the other imagined extreme, genetic determinism is falsely defined as claiming genes are the lone and immutable determinant of health [15]. In actuality, the eugenicists claim that eugenics would noticeably improve society.

Too often, genetic agnosticism facilitates the erroneous practice of using the terms “genetic” and “biological” interchangeably when discussing health disparities [11, 16]. The more egregious conflations imply that race is a biologic category, rather than a social construct, and feed the misperception that ancestry dictates biological outcomes [17]. Overall, endorsing genetic contributions without specifying the extent leaves the door wide open for determinists to suggest that disparate outcomes between groups are substantially rooted in genetics.

3. Ambiguous Discussions

The “rabbit‐duck” is an early example of an ambiguous image [18]. The defining features of such images are the ability to be interpreted in multiple, but mutually exclusive, ways but only when the presentation is restricted to the specific frame of view. While the drawing can be seen as a rabbit or duck, the illusion would be ruined by only a slight shift in perspective or subtle addition of details. In genetics, this practice is exemplified by the analogy of genetic networks operating like a rowing team as presented in The Selfish Gene [19]. The analogy purports that an optimized team would be comprised entirely of English‐speaking crew members; should a German‐speaking rower be added to the crew, the language barrier would preclude optimal function (which, in this analogy, means reduced fitness and greater risk of death) [19]. While the analogy was put forward as a defense of cooperation, within the framework provided by Dawkins, such an analogy can also be viewed as an argument that mixing “genetic stock” will lead to “dysgenic pressures”. The illusion of this analogy erodes if the perspective were shifted to an environmental or behavioral framework, such as an analogy which contrasts combining professionally‐trained rowers who speak different languages vs. a mixture of English‐only amateurs. Comparable harm to the analogy ensues if the fact that genetics have no contribution to speaking a specific language is noted. The arguments for genetic purity that the rowing analogy supports, and genetic agnosticism permits, were common between Europeans, long before the extrapolations to miscegenation with non‐European populations [20, 21].

The so‐called “genetics of educational attainment” frequently provides modern examples. This field frequently claims that select European populations are born with a greater probability of genetically superior intellectual potential than other Europeans (often using agnostic language which obfuscates that the correlations are small, that their calculations ignore the influence of familial wealth, and mathematically erase the effects of racism, sexism, and generational change). Behavioral geneticists deflect concerns to present an image of merely outlining genetic contributions to education [15], but also deny accountability when the same data is viewed as an endorsement of racist ideologies resulting in murderous consequences [12]. The attempted denial of responsibility persists despite behavior genetics' long history of misappropriation [22, 23]. If the intended picture being drawn is meant to be a rabbit, but is frequently “misinterpreted” as a duck due to the limitations of the chosen perspective, at some point the artist can no longer blame the audience.

4. Mechanistic Prerequisites

Improved detection of monogenic disorders, advancements in cancer recognition, and expanded pharmacogenomic tools have come from the genomics era—even if the impacts are often overstated [24]. Yet the translational potential of genetic sequencing in many common diseases is less clear. Two related studies are frequently cited as demonstrating the essentialist nature of genetic data in drug development [25, 26]; the findings of which are often misrepresented as a claim that possessing a genetic target doubles the chance that a drug will achieve licensure approval. Since the drug trials analyzed were first selected on Phase I applications, the correct phrasing should be that drugs which successfully matriculated from Phase I to approval were twice as likely to have genetic targets. This phrasing does not negate the potential value of genetic data but makes it more obvious that genetic associations which failed to reproduce, failed to model, or otherwise failed to reach Phase I were excluded from the analysis (including decades of failed “candidate genes” in behavioral genetics [27]).

In addition to ignoring pre‐clinical failures, potential confounders in drug approval were not assessed. For example, the authors included applications submitted to treat “infection”, which definitionally include antimicrobials which lack any human target. Therefore, their methods mathematically misattributed the notoriously high rate of failure for antibiotics to their lack of human genetic targets rather than the risk of microbial resistance. Their methods also failed to account for differing resources of the supporting pharmaceutical companies, which would obviously impact drug approval given the cost of achieving pharmaceutical licensure [28]. Additional confounding may have occurred through heterogeneity in regulatory burdens, or the size of the targeted patient population; more common disorders offer easier recruitment for clinical trials and improved statistical power for alleging population‐wide genetic associations.

In contrast to the numerous public health achievements that did not require any genetic insights [29], publications highlighting the achievements attributable to Genome Wide Association Studies (GWAS) over the last 15 years can point only to retrospective associations, devoid of therapeutic translation and often justifying utility using the future tense [30, 31]. GWAS often does not lead to anything more than partial elucidation of genetic or regulatory architecture, and researchers continue to outline the gaps between results and clinical relevance [32]. A recent paper laments, “…leveraging genetic data to find disease genes, and ultimately drug targets, has proven to be challenging” with current post‐GWAS methods increasing in computational complexity only to offer a “promising approach” to identify targets, rather than to reproducibly identify the targets themselves [33]. Consequently, while molecular insights are one path to disease mitigation, the belief that deprioritizing genetic research funding will deter medical breakthroughs is not supported by the historical record.

5. Statistical Segregation

In the early to mid‐19th century, the then ubiquitous field of phrenology claimed that measurable physical features of the skull provided insights into behavioral and medical futures, despite lacking mechanistic insights. Phrenology also used differences in the frequency of these features to both define and assign hierarchical value to different populations. As scientific support for phrenology waned in the mid‐1800s, race scientists increasingly used genes (then defined merely as “a packet of heritability”) as the scientific patina under which they presented their bigoted views [20, 21]. However, rather than establish differing races, genetic sequencing identified that any two humans will be around 99.9% identical [17] and that there is greater genetic variation within racialized populations than between them (meaning that a randomly selected African and European could be more genetically similar than two randomly selected Africans) [34].

Despite the data on the relative homogeneity of genetics across human populations, genetics is the only “‐omic” assessment to continue to use racialized assays. As an example, proteomics, transcriptomics, metabolomics, as well as immunostaining and pH could each reliably distinguish a biopsy of a patient with active atopic dermatitis from a biopsy of healthy skin [35], while genetic sequencing could not [36]. Yet in genetics, there is a burgeoning claim that studies must use continental “ancestry specific” analysis to maximize validity [37], often using allelic frequency as justification [38] in ways that mirror 19th century classification schemes.

While common definitions of a person's ancestry invoke the individual's family tree (known as genealogical ancestry), genetic ancestry involves only the specific ancestors who contributed genetic material to the individual. While genealogic and genetic ancestors overlap, the genetic contribution from one's genealogical ancestors declines sharply after going back 4–5 generations [39]. The drop off is steep enough to mean that a person today is no more likely to share specific genetic material with an ancestor from 8 to 10 generations prior than they are to share those alleles with an unrelated stranger [39, 40].

Genetic ancestry groupings (such as “European ancestry” or “Irish ancestry”) are constructs which are most often defined by contrasting the single nucleotide polymorphisms (SNP; a difference in DNA sequence information which may, or may not, have functional meaning) to a database of individuals who have 4–5 generations of their genealogical ancestors tied to the same region (any further back and the groupings erode due to the aforementioned generational decay in genetic similarity).

Defenders of using genetic ancestry present it as another prerequisite for drug discovery [37, 41], while critics outline that genetic ancestry shares the flaws of all racial, continent‐based classifications and that the quantitative nature provides only the illusion of objectivity [42, 43]. The criticisms seem validated by data showing that when racialized classification schemes (genetic or otherwise) are used for medical decision making, they not only fail to improve health, they often worsen outcomes [20, 44]. Furthermore, the same frequency‐based classifications used to define ancestries between continents can just as easily segregate populations within continents [45, 46] or within individual nations [47, 48]. The disconnect leads generates the examples of people who think of themselves as “pure blooded English” unexpectedly learning that a sizable percentage of their DNA is ascribed to other nations [40].

Because of this ambiguity, genetics groups such as the American Society of Human Genetics, have long articulated that ancestry is a continuum rather than distinct categories [49]. Despite these statements, actions indicate that publications and grants reporting “ancestry specific” findings are occurring with escalating pace, as is the use of the historically problematic concept of “admixture” [43] (Figure 1A,B). Similar position statements written by editorial boards of major journals have also fallen short. For example, “How Nature contributed [emphasis added] to science's discriminatory legacy” only pointed to pre‐WWII examples [50] despite the murder of 10 people in Buffalo, New York being perpetrated by a man citing a Nature Genetics paper from 2016 [12].

Figure 1.

Rates of annual publications and NIH awarded grants invoking “ancestry specific” analysis. (a) Total PubMed citations for the indicated year using the indicated search terms. (b) Total US dollars ($) awarded for grants containing the indicated search terms as indicated by NIH RePORTER tracking (reporter.nih.gov).

Again looking to atopic dermatitis as an example, despite segregated statistics, “ancestry specific” analysis revealed that Black Americans are more likely to develop atopic dermatitis but are less likely to have “genetic predispositions” such as FLG null alleles [14, 36]. The authors correctly interpreted this finding as a sign that risk alleles are less potent than originally thought, and that environmental injustices should be investigated and mitigated [36]. However, others insisted that associations which were only significant if study populations were pre‐sorted by continental ancestry justified spending resources on the search for genetic explanations [37] or justified racializing the disease itself [51]. Tolerance for “ancestry‐specific” analysis may also facilitate invalid post hoc calculations when a negative finding in the total study population has a subgroup with positive findings [52, 53].

A claim that a specific chemical could be proinflammatory for one nationality but therapeutic for another would be readily dismissed in toxicology or environmental science. Similarly, protein and RNA researchers would never claim there was an “Asian insulin” or endorse ancestry‐specific transcriptomics. Even during the time when racialized calculations of renal function were seen as legitimate, there were never differing versions of urinary protein assays for Africans versus Europeans. However, statistical segregation allows population genetics to rebrand similarly failed reproducibility as a “flip‐flop phenomenon” [54] or “portability problem” [55]. The geneticists Jonathan Prichard and Molly Przeworski have raised significant cautions about the non‐portability of polygenic risk scores across socioeconomic status, age, and sex, as well as their extreme sensitivity to study design [56]. Excusing inconsistent genetic findings also encourages ableist hierarchies whenever genetic associations are used to biologize disabilities for which the relevant mechanistic insights implicate environmental exposures [57].

For a relatively young species like homo sapiens, evolutionary genetic adaptations which explain population differences are coupled to specific exposure with large impacts on fitness. Most human adaptation over the past 40,000 years has been decoupled from genomic sequence changes, with the notable exceptions of sunlight intensity, geographically restricted fatal infections, and historic advances in crop and dairy farming [58, 59]. Furthermore, pointing to genetic adaptation as a reason for racial disparities risks being interpreted as claiming that even multi‐generational Americans might be “better off back where they came from”. Similarly, while it may feel more scientifically accurate to report that a studied population has “an increased burden of deleterious variants”, such statements are easily conflated with notions that those populations have “bad genes” [60, 61] or the equally eugenic formulation of “mutational load” [62].

This realization counters the common refrain from geneticists that diversifying genomic databases is essential to improving health outcomes for marginalized groups [11, 37, 63]. There is indeed value in expanding the databases to elucidate allelic frequencies to better assess variants of uncertain significance, especially in those populations where rarer variants may be more common, and their phenotypic associations easier to understand; such practices may better catalog cancer markers or identify drug‐relevant alleles for pharmacogenomics. However, each of these examples help by contextualizing individual people within the full scope of human genetic diversity, not by acting as if the complexities of human biology can be satisfied by information gleaned through knowing the birthplace of one's great grandfather. Variance and penetrance indeed matter on an individual level, but extrapolating to populations (especially when devoid of environmental assessments) risks being seen as endorsing the false claim of biologic differences between racialized groups [64]. Thus, akin to the eugenicists of the past, contemporary geneticists hinge their hypotheses on “inborn defects” whenever they claim that the genetic assessments (which have yielded only highly confounded [65], non‐translatable correlations for European populations), are the key to improving the lives of marginalized groups [66].

6. Environmental Homogenization

As stated, genetic sequencing demonstrates that any two humans randomly selected from anywhere on the globe will be around 99.9% identical [17]. Despite such immense similarity, only identical twins (or derived mouse strains) would be referred to as having “the same genotype”. Yet, the observation that not all people within a 'similar environment’ will fall ill is frequently used to suggest that inborn differences are the catalyzing factor. To such claims we must ask “by what metrics would we objectively claim that two environments were the same?”, especially given that under appreciation of environmental confounding leads to increasingly biased genetic estimates [67, 68].

Just a single molecule of lead among either a million atmospheric particles or 100 million water molecules would distinguish the safety of two cities. Yet, most toxin detection systems have lower limits in the multiple parts per million (ppm) range. This means research cannot distinguish between pristine conditions and one in which even breath (calculated to contain 10e22 air molecules in 0.5 total liters) contained nearly 10 trillion particles of lead. Furthermore, the EPA deploys this low sensitivity tracking on only around 600 chemicals, despite nearly 20,000 having been newly added to the market over the past 50 years [69]. Thus, the environmental data available for assessment is meaningfully incomplete compared to the finite human genome.

Although geneticists have attempted to account for “more similar” environments in their approaches, these adjustments continue to lack specificity. Many approaches, such as for modern twin‐studies, attempt to derive “shared environment” without measuring a single environmental variable [70]. Other approaches correctly note that two people living in the same neighborhood would have a “more similar” environment than people living great distances apart [71], yet such approaches cannot inform specific exposures any more than knowing two people may have “more similar genomes” could account for any specific allele [65, 72]. For example, if immigration is associated with the outcome of interest—as is the case in allergic and other inflammatory diseases [14]—then assessing early life environment would be more apt than adjusting for current residence. While an improvement, the sibling‐GWAS approaches [67] still homogenize any differences between the types of in‐home exposures linked to inflammatory diseases like: antibiotics, tobacco smoke, industrial exposures, pets, home remodeling, ultra‐processed food ingredients, and more [14].

A variant of environmental homogenization comes in the practice of adjusting genetic assessments solely for socioeconomic status or zip code deprivation index. This homogenizes deprived environments by suggesting that the impacts of poverty would be the same across age, race, sex, and geography. While genetic ancestry is one way for a phenotype to “run in the family”, spending multiple generations near contaminated water or toxic air can present a form of “chemical ancestry”. This concern applies to both race and class as demonstrated by some interpretations of genetic differences between upper‐ and lower‐class neighborhoods in England as indicating that poor, White Britons lack the genetics needed for upward mobility [71] despite the researchers failure to assess for any of the numerous environmental variables known to impair cognition or the storied social and class barriers to social mobility [73, 74, 75, 76].

Population geneticists rightly defend incorporating genetic data into complex models to assure all contributory factors are assessed. However, we must ask what the likely response from those geneticists would be if their city council proposed building an oil refinery across from their child's school. No one would insist that the students undergo GWAS so that SNPs could be used to stratify risk. No one would argue that mitigation strategies should include giving children with “deleterious SNP” extra air filters or discounted asthma inhalers. Instead, the insistence on genetic modeling would wane if the at‐risk populations of marginalized communities forced to live near pollutant sources suddenly included the geneticist's own children. It is far easier to homogenize the environment and overlook such regional injustices when one does not reside within the plume.

7. Using Lack of Intent as Justification for Continued Harms

In stark contrast to the post‐WWII eugenicists, those generating genetic data in inflammatory disorders today aim to help the patient populations affected rather than to obliquely signal their support for eugenical policies. However, lack of intent to harm is only a viable excuse in the presence of ignorance and naivete. Typically, racists perpetuating genetic determinism in lay media (like blogs or chat boards) misappropriate the claims of scientists; one such example related to differing rates of lactose intolerance between ancestry groupings gained mainstream press attention [77], under captions like “Today's proponents of racial hierarchy are making their case by misinterpreting research on the human genome itself.” However, the racist, genetically motivated murders in Buffalo—which included direct citations to numerous genetic publications—left no doubt that perpetuating genetic determinism in science could have fatal consequences [78].

After the Buffalo murders and after Nature apologized to its readership for past racism, Kathryn Paige Harden, a psychologist whose work on the genetics of educational attainment was cited by the killer, was given a chance to comment in a Nature Reviews Genetics editorial [15]. She wrote that if genetic research were causally linked to racism, then researchers “have a particular responsibility to justify the benefits of their research as outweighing its social risks” [15]. She then presents the contrasting hypothetical in which “genetic essentialism does not actually cause social bias but is merely correlated with it … or is invoked as a post hoc justification for existing prejudices” [15]; in this scenario Harden concludes that “explanations of why research does not support essentialist beliefs, or curtailing the research altogether, might have limited or no benefits to reducing stereotyping and prejudice” [15].

The later claim requires one to believe that the odds of racialized murders are not impacted by exposure to racialized genetic data. This is a hypothesis that runs directly counter to the assailant's written statements, which leaned heavily into not only genetic studies, but the myth of the “Great Replacement”. Moreover, we assert that even if racialized genetic data were proven to be only a post hoc justifier of hateful actions rather than a direct cause, this would not negate the need for countervailing benefits. Once the potential for harm is predictable, subsequent harm would, at best, represent a failure to prioritize safety and, at worst, reflect a lack of concern for the harmed. Given that the replacement conspiracy is perhaps the most well‐known racist and antisemitic theory in social science, and the numerous works outlining the harmful misuse of genetic data [79], it strains credulity to assert the geneticists in question could not have predicted their work would be weaponized.

As importantly, it is a central tenant of medical and general ethics, that any benefits must significantly outweigh the harms, real or imagined, purposive or intended. One the basic principles of ethics since J.S. Mill and John Rawls (whom Harden is keen to cite) is that any harm must bring a significant good for a greater number. A year prior to her editorial, Harden and her co‐author acknowledged that “Even if we concede that, at a conceptual level, genes could cause average differences in human behavior, at a practical level, it is not readily apparent what we would do with this knowledge…” [80]. They shifted the claims that GWAS would one day improve educational outcomes to stating they would merely “provide us with basic knowledge we require for correcting schemes of environmental modification and intervention” [80]. However, claims that racialized genetic assessments intend to elucidate environmental factors do not seem credible given the environmental homogenization discussed previously.

While some may justify such research as “just asking questions”, discussions around how to act upon genetic deterministic claims often lead to examples such as Francis Crick's open perseveration on whether society could be enhanced by only giving “fit” individuals the antidote for a sterilizing agent placed into the food supply [81]. Post‐WWII, eugenics became increasingly future‐tense and speculative. Discussions of sociobiology, like eugenics before it, were often defended as part of the objective duty for scientists to debate everything, which melded decisively with anti‐communist rhetoric around “freedom of thought” and against the “regulation of inquiry” [82]. The implication of this stance towards data and inquiry as future oriented and speculative places the researchers in a state of malicious neutrality: they are not held to account when spending public funds on research which concludes that “it is not readily apparent what we would do with this knowledge…” [80], nor are they at fault when the white supremacist answers on their behalf.

8. Interactions and Implications

While a focus on genetic causation might be guided by altruism, eugenics has always been more invested in justifying discriminatory policies than uncovering targetable mechanisms of disease. As such, well‐intentioned researchers can distance themselves from eugenics while still struggling to eradicate genetic determinism from their methods and assumptions. At the core of this struggle is genetic agnosticism, adherence to which causes researchers to reflexively refer to traits or diseases as “polygenic” or “complex” even when: the genetic associations are weak, mechanistic connections are lacking, statistical significance requires biologically unfounded segregation, or when avoidance of experimentally validated exposures present a simpler solution. Then, ambiguous discussions causes research to be perceived as implying that marginalized populations lack the genetic material to protect themselves against diseases which is re‐packaged as applying to social phenomena like education [78].

These practices are justified by either assuming that deprioritizing genetics would fail to meet the molecular prerequisites needed to derive therapeutic interventions or that “missing heritability” assures that current shortfalls will be overcome at an unspecified point in the future when sufficient sample sizes or computational power have been obtained. Meanwhile, study designs which fail to enumerate a single environmental variable do not face accusations of ‘ignoring the environment’ due to the ubiquity of the environmental homogenization fallacy. Ultimately, when papers reporting the “genetics of racial disparities” find greater attention with hate groups than academics [78] or general misconceptions are entrenched [83], the lack of purposeful harm may be seen as sufficient justification for a refusal to alter approaches.

9. Repeating History

The realization that twin study‐derived heritability cannot establish the genetic contributions of traits or disorders has long been established [84]. While opinions may differ as to whether twin studies have become more sophisticated or just more convoluted, all such approaches continue to be based on the same equal environment assumption which concluded that wearing sunglasses at night is 130% genetic [85]. Yet, modern publications continue to present heritability as a proxy for genetic sequencing and spreading the flawed assumptions of genetics into fields like the microbiome [86]. For example, differing microbiome signatures of identical vs. fraternal twins have been presented as evidence for a “bidirectional relationship” between genes and commensals [86]; yet there is no mention of how assortative mating, population stratification, poverty [87], and environmental injustice may generate non‐causal disease correlations with both dysbiosis and genomics. While any potential future claim of an ancestry‐specific microbiome for cognition could undergo falsifiability testing and thus avoid feeding the racist ideologies linked to analogous genetic studies, the failure of population science to address the flawed assumptions of population genetics risks recapitulating opportunity costs through misdirecting resources away from truly causal factors.

10. Concluding Recommendations

If modern research is to rid itself of the ghost of genetic determinism, best practices should include explicitly outlining the degree of genetic involvement, rather than only the statistical significance of the correlation [88]. Genetic associations that are small need not be ignored but should be contextualized in ways that avoid leaving the reader with a sense of ambiguity. If planning to sort populations by ethnicity, control comparisons should be used to assess for environmental confounding or population stratification, jargon should be avoided, and pre‐registering ancestry‐limited hypotheses should be considered. Furthermore, research funders, journal editors, and reviewers [89] should demand that research models including genetic data should also incorporate environmental, microbiome, and social determinant assessments.

Conflicts of Interest

The authors declare no conflicts of interest.