Social and Behavioral Genomics: On the Ethics of the Research and Its Downstream Applications

Abstract

Social and behavioral scientists increasingly work with geneticists or adapt the methods of genetic research to investigate genomic variation in a wide variety of behavioral and social phenotypes. Using genome-wide association studies, these social and behavioral genomics (SBG) researchers generate polygenic indexes (PGIs)—weighted sums of the estimated effects of each genetic variant on an individual’s phenotype. This review examines the ethical, conceptual, and social issues in SBG research and its downstream applications. In particular, it focuses on PGIs for ethically sensitive SBG phenotypes—those that (a) can be viewed as consequential to social status (e.g., obesity and substance-use disorders), (b) are contributing or have historically contributed to harmful stereotypes about minoritized groups and threaten to reify the biologization of social identities (e.g., financial prowess and athleticism), and/or (c) are central to a minoritized group’s identity (e.g., sexual orientation and sexual behavior).

1. INTRODUCTION

Before it was possible to use genome-wide association studies (GWASs) to create polygenic indexes (PGIs), also known as polygenic scores or polygenic risk scores (10) (for definitions of key terms used in this article, see Table 1), psychologists used twin, family, and adoption studies to investigate phenotypes as diverse as performance on IQ tests and schizophrenia. One of their key findings, enshrined in the so-called first law of behavioral genetics, was that all behavioral phenotypes are heritable—that is, within a population and a given environment, variability in a phenotype is associated, at least to some extent, with genetic differences (122). In the 2010s, with the advent of GWASs and PGIs and the availability of vast amounts of data, other social scientists, including economists, demographers, political scientists, anthropologists, and sociologists, started using these tools to study behavioral and social phenotypes such as income (61) and educational attainment (99). This more recent molecular genome-wide research is variously referred to as sociogenomics, social genomics, social science genomics, or genoeconomics (13, 26, 40, 56, 92). We use the term social and behavioral genomics (SBG).¹

Table 1.

Definitions of key terms used in this review

Term	Definition	Reference
Ancestry	A person’s origin or descent, lineage, roots, or heritage, including kinship.	97
Blanket consent	A form of consent that permits the use of identifiable private information or biospecimens in specific research without any restrictions on the kind of research that may be conducted. [Example: “Their research may be on nearly any topic” (2).]a	—
Broad consent	A form of consent that permits the use of identifiable private information or biospecimens in future research without needing new consent from a study participant each time. It is intended for the storage, maintenance, and secondary use of identifiable private information or identifiable biospecimens for future research that has not yet been specified. [Example: “I give permission...for long-term storage and use of this and other information about me, for health-related research purposes” (127).]	—
Ethnicity	A sociopolitically constructed system for classifying human beings according to claims of shared heritage, often based on perceived cultural similarities (e.g., language, religion, and beliefs); the system varies globally.	97
Ethically sensitive SBG phenotypes	Phenotypes that (a) can be viewed in a society (rightly or wrongly) as being very consequential for social status (e.g., obesity, substance-use disorders, intelligence-test scores, educational attainment, income, and criminalized behaviors), (b) are or have historically been part of harmful stereotypes about minoritized groups and threaten to reify the biologization of social identities (e.g., financial prowess, academic diligence, hysteria, hypersexuality, musical beat synchronization, and athleticism), and/or (c) are central to a minoritized group’s identity (e.g., sexual orientation, sexual behavior, and gender identity). Nearly all social or behavioral phenotypes have some potential to be sensitive.	87
Genetic ancestry	The paths through an individual’s family tree by which they have inherited DNA from specific ancestors. Genetic ancestry can be thought of in terms of lines extending upward in a family tree from an individual through their genetic ancestors. Shared genetic ancestry arises from having genetic ancestors in common (i.e., overlapping lines of ancestry). In practice, shared genetic ancestry is typically inferred based on some measure(s) of genetic similarity.	97
Genetic ancestry group	A set of individuals who share similar genetic ancestries. In practice, a genetic ancestry group is constituted based on some measure(s) of genetic similarity. Once a set is designated as a genetic ancestry group, its members are often assigned a geographic, ethnic, or other nongenetic label that is common among members.	97
Genetic similarity	A quantitative measure of individuals’ genetic resemblance, reflective of the extent of their shared genetic ancestry.	97
Genome-wide association study (GWAS)	A method for studying vast amounts of genomic data. GWASs scan the genome to identify common genetic variants associated with a trait of interest (from cardiovascular disease to educational attainment). GWAS is a theory-free analytic approach that is now commonly used in a wide range of genomic research.	18
Narrow or study-specific consent	A form of consent that permits the use of identifiable private information or biospecimens for a specific research study. Participants generally need to be recontacted and reconsented for any future research using their data that falls outside that scope. (Example: “We will use your data to study diabetes.”)	—
Polygenic index (PGI)	An index generated using a GWAS, comprising a weighted sum of the estimated effects of each genetic variant on an individual’s phenotype. The predictive power of a PGI is typically measured by R2, the proportion of variation in a phenotype among individuals that is explained by the PGI in a hold-out sample. PGIs are associational, not causal, and are not themselves causal mechanisms. They are also referred to as polygenic scores or polygenic risk scores.	19
Race	A sociopolitically constructed system created for classifying and ranking human beings according to subjective beliefs about shared ancestry based on perceived innate biological similarities; the system varies globally.	97
Social and behavioral genomics (SBG)	The study of whether and (if so) how genetic differences between individuals correspond to differences in human behaviors and social outcomes.	56

^aThe All of Us Research Program, which uses this language in its sample consent document (2), does have a policy against “potentially stigmatizing” research; however, this restriction is not imposed in the consent form itself, and a preamble in the consent states, “Researchers will use this data for lots of studies. By looking for patterns, researchers may learn more about what affects people’s health.” Therefore, we believe the All of Us consent is best characterized as blanket consent.

This review broadly focuses on SBG research but places particular emphasis on PGIs for ethically sensitive social and behavioral phenotypes. Following Meyer et al. (87), we define ethically sensitive phenotypes as those that (a) can be viewed in a society (rightly or wrongly) as being very consequential for social status (e.g., obesity, substance-use disorders, intelligence-test scores, educational attainment, income, and criminalized behaviors), (b) are or have historically been part of harmful stereotypes about minoritized groups and threaten to reify the biologization of social identities (e.g., financial prowess, academic diligence, hysteria, hypersexuality, musical beat synchronization, and athleticism), and/or (c) are central to a minoritized group’s identity (e.g., sexual orientation, sexual behavior, and gender identity).

Applying advances in molecular genetics, SBG researchers hope to demonstrate “the utility of genetic information toward understanding social dynamics” (27, p. 460), to “deliver richer, more precise answers to old questions in psychology, sociology, economics and related fields”(56, p.567), and to develop “a more complete understanding of social life” by taking into consideration “nature and nurture as well as their interplay” (26, p. 275). However, genetic assertions regarding human differences have long been used to perpetuate harms, including sustaining an essentialist view of genes as immutable and useful for dividing people into discrete groups and justifying existing status hierarchies (97). Further, ethically sensitive phenotypes are laden with values: Research on them will always be accompanied by the risk that people will use the results of such research to rank individuals based on those values. For instance, a society like the United States (or at least its elite institutions) treats people who are deemed to be more intelligent as more socially valuable than those who are considered less intelligent—assigning moral value based on one’s phenotype (110). When viewed in this context, SBG investigations into phenotypes such as criminalized behaviors and educational attainment raise ethical and social concerns, in terms of both the research itself and its downstream applications. This review explores those concerns.

2. BACKGROUND AND CONTEXT

2.1. Controversial Uses of Social and Behavior Genomics: Past and Present

In both the past and present, claims regarding genetic differences in human behaviors and social outcomes have been used to validate social and racial inequalities, legitimize social hierarchies, locate blame at the level of the individual, obscure structural inequalities, and absolve society of responsibility. More specifically, these claims have, among other examples, been wielded to justify slavery (36), outlaw interracial marriage (104), and restrict immigration (75). SBG evokes controversy in part because many social and behavioral traits can be used in attempts to place individuals high or low in social hierarchies. For instance, the US Supreme Court, in an 8–1 decision, justified the forced sterilization of “feebleminded” and “promiscuous” individuals (a widespread practice in the United States in the early to mid-1900s) with arguments that both traits were passed genetically from generation to generation and threatened the health and success of society (16, 108).

Harmful ideas about biological inheritance and human behavior date back to Francis Galton, who is widely considered the father of behavioral genetics, a precursor to SBG (100). Galton, who coined the phrase “nature versus nurture” (43), devoted himself to the study of inheritance—specifically, the inheritance of “genius” and the “mental peculiarities of different races” (43, p. v). His seminal 1869 work, Hereditary Genius (43), helped to usher in the eugenics movement, and in his 1883 book, Inquiries into Human Faculty and Its Development (42), he coined the word eugenics. As part of his eugenic vision, Galton sought to both discourage the “unfit” from reproducing and encourage procreation among the wealthiest members of British society.

Importantly, pronouncements in the Galtonian spirit persist today. For example, popular media outlets have published articles reporting on recent SBG studies using headlines such as “Britons Are Evolving to Be Poorer and Less Well-Educated” (68). Charles Murray, coauthor of the controversial 1996 book The Bell Curve, maintained in his 2020 book Human Diversity: The Biology of Gender, Race, and Class that recent genomic research provides evidence of the genetic underpinnings of educational disparities (96). And White nationalists have exploited academic journal articles and preprints that use GWASs and PGIs to support extremist ideologies (21) like “race realism”—the false ideas that biologically distinct racial groups exist and that a natural racial hierarchy is ordained (101). For instance, when genetics researchers published studies on lactase persistence (the ability to digest lactose), White nationalists selectively interpreted the results as evidence of their distinctness from, and biological superiority to, groups in which lactose intolerance is relatively high, notably Jewish and Black people. To celebrate what they took to be evidence of their superiority, they held so-called milk-chugging parties (57). More disturbingly, in May 2022, a White supremacist murdered 10 Black Americans in a supermarket in Buffalo, New York, and left behind a manifesto that cited a number of genetics studies, including social and behavioral GWASs. Using these studies, the shooter drew scientifically invalid inferences to claim that White people were biologically superior to Black people in their intelligence (137).²

At present, mischaracterizations of genomic studies circulate among alt-right White nationalist groups on social media platforms such as Discord, Stormfront, or 4Chan as images, figures, videos, or quotes called scientific memes (22). Researchers have noted a steady increase in the spread of these scientific memes since 2016 (22); their proliferation has given rise to calls for careful consideration of the data selected for analyses and to “overhaul how [geneticists] present their analyses visually” (22, p. 447).

In short, racists, classists, and misogynists have used and continue to use claims regarding biological and genetic differences to explain their hateful views and violent actions. As a result, past and present uses and interpretations of SBG research feature heavily in the ongoing debate over whether and (if so) how to conduct SBG research.

2.2. Genome-Wide Association Studies, Polygenic Indexes, and Their Limitations

In SBG, GWASs and PGIs are used to study a host of social and behavioral traits, including sexual orientation (44), educational attainment (99), and smoking behavior (37) (for a brief history of genetic research methods, see Supplemental Figure 1). However, GWASs and PGIs have notable limitations. First, the vast majority of genomic data to date have been collected in individuals whose ancestors likely lived on the European continent within the past few thousand years (83, 136). While researchers are working to recruit more diverse populations for biobanks (e.g., 126), develop reporting standards for PGIs (135), and calculate PGIs that work equally well for everyone (e.g., 91), at present PGIs are significantly more accurate for individuals who are more genetically similar to the GWAS training set from which the PGIs were generated; researchers call this issue the portability problem (32,95). Because of the portability problem, uses and applications of PGIs threaten to widen existing disparities, particularly in clinical contexts (1, 77).

Second, GWASs flag genetic variants that are associated or correlated with a phenotype; they do not themselves illuminate causal mechanisms or provide other explanations. When SBG researchers use GWASs to generate PGIs, they aim to create a tool that can predict a phenotype in an individual. PGIs cannot provide a definitive measure of an actual phenotype, and in many cases, SBG PGIs are only weakly predictive of individual phenotypes.

Third, GWASs and PGIs are plagued by environmental confounding due to genetic relatedness and passive gene–environment correlations (18). That is, “people who are more genetically similar (i.e., more closely related, even distantly) also tend to develop in more similar sociocultural, political, and physical environments, which influence most complex social traits” (18, p. 13). As a result, genotypes and environments may be correlated for noncausal reasons (18). Thus, what precisely a PGI captures is poorly understood, though PGIs “can and do reflect racism, sexism, or other prejudices, as well as more benign environmental factors” (124).

It is very difficult, and some argue potentially impossible, to distinguish genetic influences from environmental influences (18). Increasingly, researchers are turning to within-family GWASs to try to account for some confounding, but issues remain (130, 131). PGIs derived from within-family GWASs typically exhibit shrinkage in R² compared with their between-family (or population) GWAS counterparts, though the degree of shrinkage varies by phenotype (64).

3. CONCEPTUAL AND ETHICAL CONSIDERATIONS FOR CONDUCTING AND COMMUNICATING SOCIAL AND BEHAVIORAL GENOMICS RESEARCH

3.1. Engaging Communities and Publics in Social and Behavioral Genomics Research

Calls to engage members of the public, and more specifically to empower communities that historically have been exploited by scientific research, have grown in recent years. Although engagement can take many forms and can have different levels of impact (128), ethical guidelines and frameworks, as well as recent consensus reports and funding calls, now stress the value of engaging communities and publics in the scientific process (e.g., 97, 115). Specifically, engagement with communities and publics at the earliest stages of research and throughout the research life course is recognized as potentially important for building trust in and the trustworthiness of research and realizing the core research ethics principles of respect, beneficence, and justice.

The ethical imperative for community and public engagement extends beyond SBG research, but it takes on special significance in this field. Scholars argue that engagement in SBG is critical given the historical misuse and potential misapplication of this research against marginalized groups (79). This engagement is particularly important given gaps in existing ethical oversight mechanisms. For instance, the Common Rule—a US federal policy that governs research ethics—only addresses risks to direct research participants, not bystanders, groups, or society at large. This policy explicitly states that institutional review boards should not consider “possible long-range effects of applying knowledge gained in the research (e.g., the possible effects of the research on public policy)” as these are not “among those research risks that fall within the purview of [their] responsibility” (28). Yet the implications of SBG findings can affect entire groups or populations.

Engaging with communities and publics potentially recruited into and impacted by SBG research may (a) be a tangible way to demonstrate respect; (b) be a mechanism for acknowledging and addressing community and public interests, concerns, and expectations (97); and (c) help SBG researchers better identify and address who receives the benefits and bears the burden of the risks of the research (73). Furthermore, open and honest discussions about the risks and potential benefits of SBG research can lead to more informed decisions regarding SBG study design, conduct, and reporting and support researchers to better assess and intervene on potential risks.

Without engagement, SBG researchers may risk losing public trust and conducting research that conflicts with affected communities’ values or uses ethically and socially problematic analytic approaches (97). Importantly, engagement as a form of respect must go beyond mere consultation and requires actively listening to and incorporating the voices of those who are potentially most affected by the research. For SBG researchers to effectively address its ethical and social implications, they must recognize and, where relevant, include community and public perspectives.

3.2. Population Descriptors in Social and Behavioral Genomics

SBG researchers, like other genomic researchers, make inclusion, exclusion, and sampling decisions to ensure scientific rigor. Since SBG research examines genetic variation among people, these decisions often entail grouping people. These groups are often designated by population descriptors such as race, ethnicity, and ancestry and/or by genetic ancestry “populations,” as in the 1000 Genomes Project (see Table 1).

Imprecise descriptions and conflation of population descriptors are common problems in genetic research (24), with potentially significant consequences in SBG research. This is because notions of races and ethnicities as fixed biological categories of humans incorporate (implicitly or explicitly) social and behavioral traits as well as physiological ones. Such biological conceptualizations of race have historically been and continue to be misused to (a) divert attention from social determinants of health, (b) rationalize unequal social structures that privilege some racial and ethnics groups at the expense of others, and (c) justify racial or ethnic superiority or inferiority (97). SBG research could inadvertently reinforce the idea of innate racial or ethnic traits or behaviors by using racial or ethnic labels to describe a study population when genetic similarity (see Table 1) would be more appropriate (for more on genetic similarity, see 97).

While race and ethnicity should generally not be used in genomic studies (97), some SBG research on ethically sensitive phenotypes still attempts to compare groups defined by race and/or ethnicity, in which genetic ancestry could easily be misunderstood as race or ethnicity (87). Such group comparison research, historically and currently used to justify social hierarchies, is both scientifically invalid due to environmental differences and ethically problematic, as it suggests inherent biological differences between racial and ethnic groups while ignoring complex intersecting factors. Despite these issues, such group comparison studies continue to be conducted and published (for more, see 47).

To address the harms of inappropriate uses of population descriptors in genetic research, the scientific community has issued guidelines and consensus reports (67, 97, 139). These materials underscore the importance of careful selection of terminology and clear definitions and justifications for language choices. Given the interdisciplinary backgrounds of SBG researchers, the field is well-positioned to reconcile how genomic researchers have used concepts of race and ethnicity with how social scientists have historically used those concepts in their research. For example, understanding sociological concepts like racialization, or the sociocultural process by which people are assigned to a particular racial group, demonstrates why using race in genomic research is often inappropriate. Recognizing race as a dynamic sociopolitical concept rather than a fixed biological category is particularly important for research that investigates genetic variation across groups as it relates to social and behavioral traits and conditions.

Existing guidelines and consensus reports also advocate for partnerships between academic researchers and study participants and “the communities [they] are trying to serve” to “improve communication, study coordination, and long-term collaborations”(97, p.109). Such engagement becomes particularly crucial for SBG research given the historical and ongoing associations made between population descriptors, such as race, and social and behavioral phenotypes, such as intelligence. As mentioned in the previous section, engagement with potentially affected communities and publics is one step toward mitigating broad social harms that may stem from SBG research.

3.3. Phenotype Definition and Measurement

Social and behavioral traits can be challenging to define and/or measure; many exist on a continuum rather than as discrete categories. Moreover, in some cases, the measurement of a phenotype is fundamentally subjective and reflective of social norms and biases. For instance, aggression is a phenotype whose measurement could be influenced by the subjective evaluations of others: What behaviors count as aggression (e.g., physical violence, threats of physical violence, raising one’s voice, manipulation, humiliation, and dismissiveness)? In what context is aggression being defined or measured? In legal and business negotiations, for example, the use of forceful tactics to try to gain an advantage could be categorized as aggression but often is not. Moreover, when women exhibit assertive behaviors, like being self-assured or confident, they may be labeled aggressive, whereas men exhibiting the same behaviors are not. In short, there are socially embedded assumptions underlying how the traits studied by SBG researchers are defined and measured.

These challenges of defining and measuring phenotypes in SBG research can have ethical considerations. This is particularly the case when phenotype definitions inform genetic studies that may have far-reaching social implications, potentially reinforcing or creating harmful social categories or misconceptions about the genetic basis of complex human behaviors and traits. Furthermore, given variation within social contexts in terms of how an outcome such as aggression or subjective well-being is perceived, defined, and measured, it may be difficult to appropriately harmonize among datasets collected in different settings; this, in turn, can make it challenging to replicate study findings.

Some scholars consider genomic studies on some phenotypes to be scientifically fruitless regardless of whether the phenotype can be measured consistently across contexts. Although they grant that it is possible to discover associations between genetic differences and phenotypic differences, they think that, insofar as those associations will not lead to mechanistic understandings of disorders or ethically sensitive phenotypes (18), they are of no real benefit—but do pose the real risk of promoting erroneous ideas about how those phenotypes are determined by genes (12). For instance, while some LGBTQIA+ activists are supportive of genetic research on sexual orientation or same-sex sexual behavior, as long as the findings are conveyed accurately (53, 120), others have raised concerns that this research could contribute to the pathologization of minoritized sexualities and be used to buttress attempts to “correct” or change them (14).

3.4. Biobank Participants and Consent

Calculating PGIs requires access to quality-controlled genomic data from large databases like the UK Biobank, third-party organizations like 23andMe, or large longitudinal studies such as the National Longitudinal Study of Adolescent to Adult Health. These biobanks collect participant samples and other data (e.g., electronic health records and survey responses) using broad or blanket consent, which allows the use of information or biospecimens for future research without needing explicit consent from a study participant each time (72) (for examples of blanket, broad, and narrow consent language, see Table 1).

Many of the databases used to generate PGIs were established with the explicit aim of improving health, and with consent forms and processes that similarly emphasize health and health research (although some are somewhat broader in including “wellness”). As such, some commentators question whether SBG researchers should be allowed to use data from health-focused biobanks for their studies. That is, does using biobank data for SBG research fall outside the scope of the informed consent given by biobank participants?

One challenge in answering this question is that, as the concept of social determinants of health suggests, many social and behavioral phenotypes are critically important to health outcomes, and social and behavioral PGIs have in fact been used in research that is clearly health related (37,114). This may be why the data access committees and other governance bodies of the UK Biobank and many other large biobanks have allowed their data to be used in SBG research. At the same time, however, some researchers may use biobank data to conduct studies that go beyond the scope of the original research proposal without notifying or seeking additional approval from the appropriate entities (for more, see 132); if biobanks were aware of such uses, it is possible that they would not approve of them.

Furthermore, even if a social or behavioral phenotype like educational attainment is relevant to health, and therefore is in that sense legitimately covered by the consent language, this implication of the consent language may not be obvious to prospective biobank participants, who therefore cannot consider how studies of SBG phenotypes, including those that produce PGIs that could be (mis)used in various ways, might impact their willingness to participate. Empirical studies have shown that people can have complicity-based moral, religious, or cultural concerns about how their data are used, even when these studies are conducted with deidentified data (29, 50, 121).

3.5. Justifying Social and Behavioral Genomics Research

Scholars also debate what constitutes a legitimate justification for undertaking research that entails the sorts of ethical risks associated with SBG research. Critics of such research are skeptical that the benefits associated with investigating how much genetic differences are correlated with SBG phenotypes can outweigh the potential ethical risks associated with undertaking such research (18, 19). Moreover, these scholars are skeptical that SBG research will be able to provide novel information about the problems it claims it wants to help solve, like education disparities (12). The proponents of such research, however, argue that PGIs can heighten our understandings of social policies or interventions (56) or that using social and behavioral PGIs as control variables in social science studies can minimize potential confounding and increase rigor (138). For instance, a PGI for educational attainment could conceivably be used as a control variable in an educational intervention and would require a smaller sample to achieve adequate power, which would be more cost-effective and perhaps faster (87).

Some scholars—including both proponents of SBG research and those who have concerns about it—also worry that prohibiting certain research questions erodes academic freedom. Regardless of whether decisions by journals not to publish, funders not to fund, or biobanks not to permit SBG research meet formal definitions of censorship, some worry that combining science with what may be perceived (rightly or wrongly) as ideological considerations could, in an era marked by increasingly political polarization, have unintended negative consequences well beyond SBG research, such as undermining trust in science—which was, for instance, dangerously politicized during the COVID-19 pandemic (87, 141). In short, there is debate over the appropriate marginal value, in terms of explanatory power and identifiable interventions, that justifies conducting ethically sensitive SBG research and developing and using PGIs from it.

Although there are debates about whether such research should be undertaken, we suggest that if it is, researchers have an ethical obligation to meet a higher threshold of evidence (i.e., in terms of quantity of evidence and quality of evidence) than do researchers whose results are less ethically fraught.³ We recognize that operationalizing such a norm could be challenging and welcome further work exploring this. There is also agreement that once researchers have undertaken such research, they have a heightened obligation to take care in disseminating it (87).

Where there is disagreement between critics and proponents of SBG research, there is no easy resolution. As is the case with polarization in scientific research broadly (62), disagreements about generating scientific knowledge and evidence in SBG are shaped by value differences among individuals and research communities—regarding what counts as benefit, how to measure such benefit, and when that benefit promises to be large enough to warrant pursuing. Indeed, these disagreements go to deep value questions about the goals of science. Is the goal of science to reveal knowledge for the sake of knowledge? And/or is the goal more instrumental—to increase tangible benefit? As long as there are differences regarding such questions, it will be essential to continue the professional and public conversations about how to justify SBG research and, when such research is undertaken, how to conduct and disseminate it.

3.6. Opportunity Costs and Resource Allocation

Debate about the threshold of confidence and evidence required of SBG research raises additional questions about the opportunity costs of conducting it and whether existing resource constraints are such that conducting SBG research is an inappropriate allocation of resources. For instance, some scholars question the value of focusing on genetic contributions to social and behavioral phenotypes like educational attainment in a society where social and environmental determinants strongly impact educational outcomes. These critics warn that SBG research may detract from other kinds of research aiming to achieve the same outcome (e.g., to reduce social inequality) (12, 13)—what is sometimes called genetic distractionism (87).

3.7. Misinterpretation of Polygenic Indexes

The idea that genetic differences among people are associated with social and behavioral differences among them can challenge people’s intuitions about free will and agency (123). People can think of genes in determinist and essentialist ways—viewing genes as predetermining a person’s behaviors, health, and life outcomes and/or as dividing people into discrete, nonoverlapping groups (59). Researchers (18), laypeople (46), and journalists (94) have all been thought to, at times, overstate the power of genes. Such misunderstandings can be tied to the failure to understand that heritability estimates can and do change from one environment to another, to the erroneous idea that heritability estimates entail that traits are fixed and thus resistant to environmental intervention, or to the mistaken idea that PGIs offer causal genetic explanations for traits. However, there is also evidence that many laypeople do not misunderstand and that they deploy what determinist and essentialist views they may have in flexible and strategic ways rather than in a uniform manner; this flexible deployment is called strategic essentialism (25).

4. RESPONSIBLE CONDUCT AND COMMUNICATION OF SOCIAL AND BEHAVIORAL GENOMICS RESEARCH

SBG researchers have an ethical obligation to anticipate and, where possible, take reasonable measures to mitigate against risks that may arise from SBG research and its communication (87). The sidebar titled Ten Simple Rules for Socially Responsible Science lists some recommendations from Zivony et al. (142) for conducting socially responsible science; the sidebar titled Recommendations for the Responsible Conduct and Communication of Social and Behavioral Genomics Research, drawn from Meyer et al. (87), provides recommendations developed for responsibly conducting and communicating SBG research specifically. These recommendations span the entire scientific process, from defining the research question alongside relevant communities and publics to reporting conclusions.

Some of Meyer et al.’s (87) recommendations for facilitating the responsible conduct and communication of SBG research are equally applicable to genomic research broadly (and even to some nongenomic research). For instance, the harmful conflation of population descriptors in genomics requires both SBG and other genomic researchers to move away from language that conflates genetic ancestry, race, and ethnicity and to justify their use and definition of population descriptors. Moreover, the variable predictive accuracy of PGIs across populations is one reason to include diverse populations in research; this will entail diversifying biobanks and the genomics workforce and engaging a diverse set of peoples so as to create collaborative research environments that can facilitate benefit sharing (119).

However, other recommendations are more specific to SBG. For instance, given debate over phenotype definition and measurement for ethically sensitive phenotypes, it may be especially important for SBG researchers to be transparent about and justify their phenotype definitions and measurements. Additionally, given the heightened risks of SBG research on sensitive phenotypes, preempting misinterpretations, embedding caveats and context, and taking special care not to exaggerate effect sizes may prove particularly impactful. Furthermore, engaging communities, publics, and other interested parties⁴ can help to enhance the trustworthiness of SBG by producing research that reflects the preferences and values of those who are at greatest risk of being harmed, those who are the intended beneficiaries, and/or those who may affect the conduct (e.g., funders and academic journal editors) or translation (e.g., clinicians and members of industry) of SBG.

Scholars have gathered insights from a wide set of groups regarding genomic research on ethically sensitive phenotypes, including vulnerable groups such as members of the LGBTQIA+ community (53) and members of the public who represent communities that have been historically exploited by scientific research (e.g., Black/African American, Indigenous, and low-income people) (79); they have also solicited the perspectives of parents (113), educators (78), and clinicians (9). These insights were sought for both empirical and normative research questions and use a variety of methods that range from surveys and interviews to implicit association tests to forms of community and public engagement.

Given concerns about the misinterpretation of SBG research, it is vital that researchers clearly describe what their findings can and cannot conclude via accessible materials. One way that SBG researchers have sought to responsibly communicate is via the creation of FAQ documents to accompany their GWASs on ethically sensitive phenotypes.⁵A repository now exists to centralize these resources (80). Journalists have also been directed to these materials to assist them with their reporting (58). As additional examples of efforts at responsible research communication, SBG researchers have developed explainer videos (55), engaged with journalists (11), and written for public audiences in media outlets (137).

Finally, research suggests that genetics education that enhances genetic literacy can help to mitigate misinterpretations and reduce genetic essentialist and racist beliefs (33–35). For instance, a survey experiment on genetic predispositions for depression found that the negative effects of being told one had a genetic predisposition for depression went away after education about the nondeterministic nature of genes for depression (71). Finally, there are examples of revamped efforts to educate genetics trainees on the ethical and societal impacts of their work (105).

5. ETHICAL IMPLICATIONS OF DOWNSTREAM APPLICATIONS OF SOCIAL AND BEHAVIORAL GENOMICS RESEARCH

The increasing accessibility of genomic data is pushing genomic research beyond labs and clinics, leading to various applications that raise ethical concerns, particularly in SBG. This section covers current downstream applications of SBG [e.g., SBG PGIs in direct-to-consumer (DTC) genetic testing] as well as future possible applications (e.g., SBG PGIs in the legal system). We describe potential ethical concerns associated with these applications while simultaneously remaining aware of their methodological limitations (e.g., the portability problem and the limited ability of PGIs to predict individual outcomes), which may decrease the likelihood that a current use becomes more widespread or that a potential use even materializes.

5.1. Direct-To-Consumer Genetic Testing

DTC genetic testing entails returning individual genetic results to consumers without a healthcare or other professional intermediary. The DTC industry is increasingly incorporating findings from SBG GWASs and PGIs. Consumers can access genetic tests for various social and behavioral phenotypes, including empathy, leadership potential, attention span, intelligence, reading ability, and math ability (e.g., from companies such as GenePlaza or Xcode Life). Importantly, in the United States, most DTC genetic tests have not been validated or approved by the US Food and Drug Administration (FDA) or (in the case of many SBG DTC tests) are not subject to FDA oversight (103).

When DTC genetic testing first emerged in the mid-2000s, media outlets reported that some parents were genetically testing their children for social and behavioral traits like athletic ability to guide extracurricular choices (76). As testing technology has advanced, some parents outside the United States have begun conducting “genetic talent testing” on their children, hoping to identify potential educational advantages for their children (7). Professional societies like the American Society of Human Genetics have recommended against genetic testing for children and adolescents without a compelling medical justification (15). Still, parents can conduct DTC genetic testing on their children with little oversight—raising concerns about privacy (60), genetic discrimination (39), and a child’s right to an open future (93). Parents wanting to afford their child every opportunity for success might use these test results to seek special services from clinicians and schools (81).

DTC consumers understandably often fail to appropriately interpret their results. In one study of hypothetical DTC-styled PGI reports for social and behavioral, anthropometric, and clinical phenotypes,70% of participants misinterpreted the PGI percentile as representing the individual’s relative risk in the general population rather than among other test users (88). Moreover, 48% incorrectly confused actual for predicted phenotype (88).

Beyond concern about misunderstanding results, there is concern about the potential for negative psychosocial impacts of receiving genomic information about social and behavioral phenotypes. At present, very little is known about the psychosocial impacts of receiving genetic information about ethically sensitive SBG traits. One of the first studies examining this issue found that online survey respondents who were asked to imagine receiving a low PGI report for educational attainment scored lower on measures of educational potential, academic efficacy, self-esteem, and self-perceived confidence than those who were asked to imagine receiving a high PGI report (84). However, a similar study found that respondents did not consider PGI reports of low educational attainment, intelligence, or subjective well-being to be any more anxiety inducing than other standard or hypothetical scores (e.g., standardized tests) for the same traits, and estimated that any negative affect would significantly diminish within six months (88).

5.2. Polygenic Embryo Screening and Selection

Recently, some genetic testing companies (e.g., Orchid Health, Genomic Prediction, and MyOme), in partnership with fertility clinics, have begun offering prospective parents the option to learn the PGIs of embryos prior to transfer—a practice known as polygenic embryo screening, or preimplantation genetic testing for polygenic disease risk (PGT-P). PGT-P is typically offered to in vitro fertilization patients as an add-on, though it is possible that some highly motivated individuals might seek in vitro fertilization for the sole purpose of accessing PGT-P.

Currently, PGT-P companies provide PGIs for many phenotypes, including cardiovascular disease, schizophrenia, autism spectrum disorder, intellectual disability, and various cancers. The first baby born through this process was reported in 2020 and generated significant media attention (48, 107). Some parents have reportedly taken PGT-P a step further by uploading the raw embryo PGI data they receive from the PGT-P company to a third-party DTC company (i.e., a DTC company that does not handle biological specimens to conduct genotyping or sequencing) that offers PGIs for a wider array of social and behavioral phenotypes, and then making their embryo selection decision based in part on that additional information (49).

Researchers have demonstrated the limited predictive power of PGT-P, at least for the near future (66), and determined that the utility of PGT-P will depend on the selection strategy and the selected phenotypes (74). Nevertheless, the number of companies offering PGT-P is growing, along with debates over PGT-P’s ethical and practical implications (8, 20, 65, 70, 117). Some are concerned that people are making reproductive decisions based on unvalidated and unproven diagnostic information (38). Additionally, those concerned about equity and justice raise questions about how premature applications of PGT-P are distracting from research on PGIs “and their relationships with the environment in which we live” (38, p. 494). Further, PGT-P may only be accessible to individuals with social and financial capital, devalue certain traits, and ultimately exacerbate social inequalities (125). Over time, it could also alter population demographics if widely adopted (125).

Yet research indicates that a significant portion of the US public is interested in and supportive of PGT-P (9, 41, 90). For example, one study found that nearly 40% of Americans would use this service to increase their child’s chances of getting into a selective college if it were free (90). However, there may be divergence between how laypeople and clinicians view PGT-P, with clinicians generally being more cautious about the practice (9).

Finally, a technology like PGT-P may be a precursor to polygenic gene editing, in which specific genetic variants associated with a polygenic trait are altered (133). Polygenic gene editing may someday be possible as the predictive accuracy of PGIs increase and gene editing technologies continue to develop (133). This potential future application raises additional questions regarding the ethics of human enhancement.

5.3. Health Settings

SBG PGIs could be used in health settings. For instance, some SBG PGIs (e.g., for subjective well-being and neuroticism) lend themselves more naturally to the clinic because the extreme tail of their distribution predicts a phenotype that constitutes a psychiatric diagnosis (e.g., depression or anxiety). Indeed, the National Institute of Mental Health has invested in translational research examining the feasibility and acceptability of implementing PGIs for psychiatric conditions in the clinic (98; https://psychemerge.com). As another example, clinicians could decide to use a PGI for body mass index to identify adolescents at risk of developing obesity, possibly for a referral to a nutritionist. However, because obesity is stigmatized in most societies, so, too, might be a prediction of developing it. Perhaps less obviously, the educational attainment PGI more accurately predicts risk of several diseases than the disease PGI alone (99). It thus could be incorporated into routine preventive health screenings alongside other PGIs in a way that may improve the understanding and prediction of certain health conditions affected by social determinants.

Finally, some SBG PGIs could also be used to estimate the severity or phenotypic expression of a condition. A clinician could, for instance, use a PGI for educational attainment or intelligence to try to estimate the severity or phenotypic expression of an intellectual disability to help families prepare and plan. Such a PGI could also be used in the preimplantation genetic diagnosis context, raising concerns about eugenics.

5.4. Education Settings

Policymakers and educators are exhibiting growing interest in exploring how PGIs for phenotypes such as educational attainment, dyslexia, and attention deficit hyperactivity disorder (ADHD) could be applied to education settings. Some US K–12 education institutions have been approached by SBG researchers to genotype children (54). For example, in the New Haven Lexinome Project, researchers at Yale University partnered with New Haven Public Schools to assess the reading and cognitive abilities of first-graders to develop a genetic screener for dyslexia (52). Novel simulation methods have since demonstrated that PGIs could, in theory, be used alongside other screening tools to identify children with learning disabilities who may struggle academically (116).

The rapid inroads SBG is making into education research and settings is garnering the attention of policymakers (85), educators (78), and parents (7, 113). SBG researchers have testified before the UK House of Commons Education Committee about the role of genetics in educational outcomes (63). The Early Intervention Foundation, a UK nonprofit for youth, held a workshop series and published a 2021 report on the use of genetic data in early intervention and social policy (5). And a nationally representative survey of US parents found that many are interested in the use of genetic data to screen for learning disabilities (113).

Given the growing availability of SBG data, some researchers argue that SBG PGIs can enhance personalization in classrooms (6) or guide resource allocation (51). For instance, researchers in the United Kingdom proposed to identify low-income children with high PGIs for socially desirable education-related outcomes (e.g., educational attainment) and direct resources to them (51). However, using genomic data in education might deepen social inequalities, especially if only affluent schools or parents can access and act on PGI information (82). Furthermore, there is ample research showing that teachers’ expectations of their students can impact those students’ academic performance, including their likelihood of completing college (102); thus, there are also potential psychosocial harms associated with introducing PGIs into educational settings, because teachers might treat students differently based on their PGI results.

Current US laws, such as the Genetic Information Nondiscrimination Act of 2008, protect against genetic discrimination in many workplaces and in health insurance but not in small businesses, schools, or life insurance (89, 111). This raises the risk that students with stigmatized PGI results could face exclusion from certain jobs, schools, insurance, or other programs, while those with valued PGI results might receive preferential treatment based on genetics rather than actual work or academic performance or demonstrated insurance risk.

5.5. Legal System

To date, and to our knowledge, SBG PGIs have yet to be utilized in any significant way in any legal system. However, findings from behavioral genetics research (a precursor to SBG research) have, in limited instances, been presented as evidence in criminal trials (30, 31, 112). Despite concerns that behavioral genetics research would be used to indicate the risk of a defendant to others, research finds that it is most often introduced in court on the side of the defense to reduce perceptions of culpability (31). Further, a representative survey study of US adults found that behavioral genetics evidence did little to alter people’s judgments of culpability and punishment (3).

Still, SBG PGIs may one day be used in the legal system to provide proof of a defendant’s “condition.” The defense may try to argue that a defendant’s PGI suggests they are at heightened genetic risk for criminal or aggressive behaviors and should not be penalized for a condition that is outside of their control. Conversely, the prosecution could argue that a defendant with a high PGI for aggression is a danger to the public and should be treated accordingly.

It is also possible that PGIs could be used to screen for individuals at risk for entering into the criminal justice system. The US Centers for Disease Control and Prevention, for example, currently lists several risk factors (e.g., ADHD, antisocial behavior, and poor academic performance) associated with youth violence for which there are now related PGIs (23). Many youth violence prevention efforts currently use risk factors to identify whom to engage in mentoring programs or after-school programs. It is thus possible that PGIs could one day be used as an additional tool for determining who to include in violence prevention efforts.

Because SBG PGIs predict poorly at the individual level, using them in the legal system to impose penalties on individuals is ethically concerning. Even using PGIs to argue for leniency is problematic in the eyes of some, to the extent that such arguments can presuppose genetic determinism and undermine the idea of human agency.

5.6. Insurance and Other Consumer Goods and Services

Some profit-maximizing entities, such as financial or insurance firms, which already do some form of risk management, will likely be willing to pay for statistically noisy PGIs, even if the PGIs provide only a small increase in those entities’ ability to predict outcomes (89). Although these profit-maximizing entities might use PGIs to tailor their products in ways that benefit consumers, their use of PGIs also raises concerns about privacy and the potential for discrimination (89). The discussion about the ethical and legal implications of such entities using genetic information in general is now familiar, but the discussion regarding the use of PGIs by insurance companies or other consumer goods and services is just beginning. Attention should be given to considering how firms might access PGI information and the subsequent benefits and harms to both firms and individuals of utilizing such information, especially as PGIs grow in their predictive power (89).

6. RESPONDING TO DOWNSTREAM APPLICATIONS

Downstream applications of SBG raise a range of ethical concerns regarding, for example, the public’s understanding of what SBG PGIs do and do not mean and the potential for profit-maximizing entities to use SBG in ways that potentially harm consumers. As a result, a range of responses will be needed.

One set of responses involves government regulation of SBG PGIs that are intended for clinical use. For example, in 2010, the FDA called for DTC companies offering genetic tests for health-related conditions to subject their genetic tests to regulation as medical devices. The FDA expressed concerns that consumers were making health-related decisions based on information that had not been sufficiently validated or tests conducted without the methodological rigor required of clinical laboratory tests. When companies were slow to comply, the FDA undertook enforcement action against 23andMe, requiring it to stop selling its health-related genetic tests until they had gone through regulatory review (140). Currently, 23andMe is the only DTC company to offer FDA-authorized tests.

In 2024, the FDA amended its regulations to make explicit what it had long asserted—that it has oversight over laboratory-developed tests (LDTs) intended for clinical use as medical devices (86). More importantly, it announced plans to phase out its historical practice of exercising enforcement discretion over LDTs. Instead, the agency promised to improve the safety and effectiveness of LDTs by newly enforcing regulatory requirements, such as premarket review, quality-system requirements, adverse event reporting, establishment registration and device listing, labeling requirements, and investigational use requirements (129). This new policy stance will likely apply to companies that offer PGT-P and those that seek to market PGIs for clinical indications (which could include DTC companies that tout the health implications of their tests, as 23andMe did). The FDA and GENinCODE, a company that developed a PGI for cardiovascular disease risk, have already agreed to create a new medical device regulatory class for the submission of PGIs for premarket approval (45).

FDA regulation is not, however, a panacea, and there have already been legal challenges to the recent FDA ruling on LDTs (17). Unlike the UK’s Human Fertilisation and Embryology Authority—an independent regulator of embryo research and fertility treatments—the FDA’s oversight is limited to matters such as clinical and analytical validity as well as appropriate labeling and patient disclosures; it does not extend, for instance, to whether it is ethically appropriate to select embryos on the basis of sensitive social and behavioral phenotypes, no matter how accurate the PGI. Nor is it within the FDA’s purview to determine whether it is appropriate to risk-stratify patients according to those same phenotypes.

Another set of responses to downstream applications of SBG can come in the form of statements or policies from scientific communities. For instance, the Psychiatric Genomics Consortium, a research consortium devoted to understanding causes of autism, ADHD, schizophrenia, and bipolar disorder, explicitly states on its website that “investigators may not use these data to develop any type of risk or predictive test for an unborn individual” (see the Supplemental Appendix). When a PGT-P company began selling prospective parents PGI information for schizophrenia, the consortium went public with claims that the company had violated its restrictions against data use for commercial testing and embryo testing (4). As another example, after a DTC company used a GWAS on same-sex sexual behavior to market a “How Gay Are You?” genetic report, more than 1,400 members of the scientific community signed a petition calling for its removal (69). The petition, which successfully led to the removal of the app and now has more than 1,700 signatures (134), raised concerns about the possibility that results could be used to convict members of the LGBTQIA+ community in countries where same-sex sexual behavior is criminalized.

A third set of responses to ethical concerns about downstream applications of PGIs can come in the form of fostering dialogue and building strong relationships with communities that may be affected by those applications. Additional value may be found in engaging and seeking to understand the perspectives of other parties, such as media journalists, academic journal editors, DTC industry members, and policymakers. These groups may think differently about (a) the harms and benefits of downstream applications of SBG and (b) their roles in supporting the ethical translation of SBG. Dynamic, multidirectional dialogue and integrated engagement between these different groups may permit a fuller accounting of the actors who bear important moral responsibilities and the actions (e.g., harm mitigation and benefit promotion strategies) required to ethically translate scientific research like SBG.

7. CONCLUSION

SBG research has the potential to illuminate how ethically sensitive phenotypes emerge from staggeringly complex and ongoing interactions between genetic and environmental factors. It also carries substantial risks. Addressing the myriad ethical concerns surrounding SBG is essential, and promoting genetic literacy is crucial, in that misunderstanding the scope and limitations of PGIs can result in misinterpretations and harmful downstream applications. Recognizing that genes do not act in isolation and that the traits associated with ethically sensitive phenotypes, along with their associated PGIs, are inherently shaped by the environments in which they develop is crucial. Ongoing critical reflection on the ethical, conceptual, and social dimensions of SBG research is essential to responsibly addressing both its risks and its possibilities.

TEN SIMPLE RULES FOR SOCIALLY RESPONSIBLE SCIENCE.

Zivony et al. (142) proposed the following rules for researchers aiming to conduct socially responsible science:

1. Get diverse perspectives early on.

2. Understand the limits of your design with regard to your claims.

3. Incorporate underlying social theory and historical contexts.

4. Be transparent about your hypothesis and analyses.

5. Report your results and limitations accurately and transparently.

6. Choose your terminology carefully.

7. Seek a rigorous review and editorial processes.

8. Play an active role in ensuring correct interpretations of your results.

9. Address criticism from peers and the general public with respect.

10. When all else fails, consider submitting a correction or a self-retraction.

RECOMMENDATIONS FOR THE RESPONSIBLE CONDUCT AND COMMUNICATION OF SOCIAL AND BEHAVIORAL GENOMICS RESEARCH.

Meyer et al. (87) made the following recommendations for responsibly conducting and communicating SBG research:

Responsible Conduct

1. Diversify biobanks and research teams (a precondition for the whole ecosystem).

2. Engage with relevant actors, communities, and publics.

3. Justify the use and definition of population descriptors.

4. Justify phenotype definition and measurement.

5. Conduct studies with adequate power.

6. Conduct within-family analyses, if possible.

7. Replicate findings in hold-out samples.

8. Extend research benefits to diverse people.

Responsible Communication

1. Develop a key-points box that includes how results should(n’t) be (mis)interpreted or (mis)used.

2. Divert misinterpretations or misuse via FAQs, videos, and careful press releases.

3. Report effect sizes in the abstract and avoiding exaggerating them, including in graphs.

4. Embed caveats and context in graphs and tables.

5. Define and justify the use of the term populations.

6. Move away from population language that is easily conflated with race or ethnicity.