Ethics and Practice in Communities of Care

Introduction

Should Science and Technology Studies (STS) aim to do ethics, and if so, what form should that work take? I raise the question of ethics or “doing good” here in the context of the histories of STS. I raise this question in part because this award is named for John Desmond Bernal who argued that science should be useful by helping to solve pressing social problems, and because my research has reached a point where I want to suggest ways in which it can be helpful to problems of public health. This question is not new. In its early period, STS moved back and forth between positions for and against providing policy recommendations or making other interventions using STS knowledge. From the 1970s, many STS scholars argued against making prescriptions, and the resulting debates about the interventionist role STS could or should play raised significant issues of complexity, uncertainty, and reflexivity. Here I will briefly discuss and reflect on these historical debates, then present our specific research, and then use that research to return to these larger questions about ethics in and with STS. I will end with our current attempts to engage in ethics and prescription.

STS research in the 1970s produced critical discussions about complexity, uncertainty, and reflexivity to argue that we (qua science) “have never been modern.” That is, we have never achieved the Cartesian division, and scientists have always been implicated in scientific processes and outcomes. Some scholars asked, then, what can STS claim about their/our knowledge? Over three decades later, Guggenheim and Nowotny (2003, 231) argued it was time for STS to aim toward “a possible future of STS … in a multitude of forms of engagement in policy, in consultation and in various forms of education. Engagement here is taken as a sign of progress, a strategy for escaping repetition.” Some STSers responded within the frame of reflexivity. They questioned whether STS can provide evidence that is any more credible than those of the scientists we analyze and critique. Others, like Lynch (2009, 113), raised the potential for “unintended consequences.” He reminded us that “engagements in public organizational affairs can be full of surprises, or, worse, subject to unintended reversals,” partly because as scholars we are seldom in a position decisively to shape the terms of engagement. More broadly, Lynch argued against an all-too-easy acceptance that the future of STS depends on its capacity to engage in either the organizational worlds of business and government, or the activism promoted by “the academic left.” Lynch suggested that doing a proper job as an academic may require time out from utilitarian and emancipatory demands. Have we now arrived at an end to that time out?

Nevertheless, we should also be alert to the fact that many STS scholars have been working on ethics and policy-oriented research for some time. Early feminist science studies scholars beginning in at least the 1960s had already been doing research aimed at improving the situation of women and others in the world. They took their research and knowledge claims seriously and argued for change. Among them is Donna Haraway's (1991) approach to interweaving multiple partial perspectives of implicated actors in specific situations, articulating all knowledges as situated knowledges. Following this approach, my discussion here aims to suggest ways to do ethics and policy even while acknowledging complexity and uncertainty. At the end, I point to scientists’ own effort to promote and study different ways of knowing which is not dissimilar from work in STS, and I point to some changes in science policy decisions in the directions we propose.

My own early research used my training in the qualitative sociology of work and professions, the symbolic interaction tradition, and early STS to understand how decisions were made about what kind of research to conduct in cancer research (Fujimura 1996). I wanted to redirect cancer science toward what I believed to be more complex approaches to studying the disease, but at that time I did not feel equipped to say what exactly those new approaches should be. So, my work was descriptive and critical of the focus on molecules/oncogenes, but not prescriptive about how cancer research should be done. But I did have another agenda. I wanted to learn how to conduct critical research on science, so I could later study science that impacted on sex/gender and race, two areas of research in sociology that had been part of my graduate studies. My later projects tried to understand how sex gene research was conducted and how its results were interpreted through the lens of social assumptions about sex and gender (Fujimura 2006); and how race categories were used or developed in human genetics research. My aim was to gain insights that could be used to generate changes in science and politics.

Ethics, Race, and Genome-Wide Association Studies (GWAS)

For my studies of the use of race in genetics research, I built a team of researchers who could study the science, the sociopolitical infrastructure of science, and hopefully also develop policy agendas for changing the science. We focused on GWAS which at that time were receiving enormous push and funding from the U.S. National Institutes of Health (NIH) as the best way to use the results of the human genome project and find medically related disease risk genomics. Unfortunately, our team's studies have shown that these new large-scale data searches for genomic risk factors for common complex diseases have used U.S. sociocultural categories of race and ethnicity to collect, build, and categorize data, with attendant ethical and scientific concerns. Elegant algorithms have been used to find disease risk factors within the data, but the results of these searches cannot be easily applied to individuals in clinical contexts, in part because they use race categories to collect and build the genomic data. Race categories and genetic categories are not equivalent, our research showed.

The work of my research team raises critical concerns within large-scale biomedical research, a resource-intensive “vast, finely tuned machinery” (Fortun 2023, 182) that invests effort into identifying genetic underpinnings of disease, and can often risk conflating the social with the biological, as we have made clear in our articles (e.g., Rajagopalan and Fujimura 2018; Fujmura and Rajagopalan 2020). The current use of race categories in U.S. biomedical research—and especially biomedical genomic research—is a flawed practice because U.S. race categories are deeply social, cultural, historical, and political; these categories are not biological in any way. If there is any congruence between race categories and common complex health disparities, for example, we argue that they are based on the effect of racism on people's health and not on inherent biological or genomic causes (Rajagopalan, D’Antonio, and Fujimura 2024). Using sociopolitical categories to collect and organize genomic data is to misleadingly transmute these categories into biological essence, simultaneously ignoring much of human diversity, and the growing proportion of individuals of increasingly diverse (or “mixed”) ancestries (our new research), resulting in mis-treatment/malpractice for those already disenfranchised by medical, financial, and political systems. This assumes, of course, that people have not always already been “mixed.” Clinically, the use of race categories generates opportunities for confusion, ambiguity, and ultimately, inequities in care (Rabay et al. 2024).

When I presented our research and conclusions at an Artificial Intelligence (AI) Conference at the University of California, Berkeley, an AI researcher told me confidently that he could fix the “bias” by fixing the algorithms used. In the cases we study, what would “fixing” an algorithm entail? Statistical geneticists use a long-standing statistical tool called principal component analysis (PCA) to try to “control for” shared genomic differences within a group of individuals that is assumed to share population history. Statistical geneticists argue such shared genomic differences/similarities need to be accounted for, or else they will confuse the search for genomic associations with disease risk. And PCA is their tool of choice.

We argue that PCA in itself is not the problem. Instead, the critical problem in GWAS is the assumptions made by geneticists that then affect the way data are collected. Statistical geneticists argue that ancestry must be accounted for to avoid spurious findings based on ancestry differences rather than disease risk differences. That is, they argue that shared genomic differences and similarities need to be accounted for before conducting association studies between individual genomes and disease risk, or else they will confuse the search for genomic factors that truly impact disease. But how do they know that individuals from whom they want to collect data share ancestry? To account for genetic differences and similarities they use sociocultural concepts of race. Together, these assumptions organize their data collection to create data that cannot be fixed by an algorithm. The data are always already structured by their assumptions, so that GWAS ends up in a circular production of data and assumptions. Through their work processes and choices, GWAS researchers produced a particular version of “populations” that eerily mirrored continental race, ethnicity, and nationality descriptors used as social identifiers.

Our ethnographic research studied the choices made throughout the process of statistical genetics. We studied layers and layers of data production infrastructures to understand how this congruence between populations and continental races came to be (e.g., Fujimura and Rajagopalan 2020; Rajagopalan and Fujimura 2018). In short, we argue that one cannot fix the problem by fixing the algorithm, given that the many layers of data infrastructures contain biased data in many different forms. We argue that algorithms cannot just be applied to old data or even to new data. Researchers have also to examine the production of the data they employ, to understand what parameters are used and what problems are created by the data. Our hope is that our study will help to make statistical geneticists aware of how their social assumptions affect their research, so that they will work toward rethinking and changing their methods and aims.

Methodologically, our research is also multidisciplinary. I was trained as a qualitative sociologist of science, and Ramya Rajagopalan learned the same from me as well as bringing her expertise in genomics gained during graduate studies at Massachusetts Institute of Technology (MIT). Together, we ethnographically studied biospecimen collection practices and the production of research and databases. We examined the data architectures and the microarrays that were used to identify genetic variation among individuals in these various collections, including the choice of DNA loci for use in microarray analysis. To study the clustering algorithms based on PCA that are used to sort and categorize genomic differences into genetic ancestry groupings, we were lucky to be aided by my late husband and team member Kjell Doksum, who brought his expertise as mathematical statistician to help us understand what the geneticists were doing. He could understand when and how statistical tools could be used, and when and where they could not, and he was the one who said the problem is not PCA. For ethnographers of science, understanding the outcomes of biomedical genetic research means that we have to study data production, analysis, parameter setting, and other practices up close and in dialogue. Understanding algorithmic knowledge outcomes does not mean just attending to the algorithm.

We believe that STS studies can perhaps help genomicists become aware of ethically problematic data. It is often not obvious how, when, and where the specimens and data they use were collected, analyzed, and organized. This is where multiple perspectives are critical to include. We bring to our study a personal commitment to decolonize scientific practice and hope that our findings based on this commitment help to shed light on and raise awareness of these problems within the data and in the analytic tools used by genomicists. We ultimately hope that our studies can and will spur change.

But there also are economic and academic disincentives that work against change. For example, political scientist Yoshiko Herrera has studied the problem of inaccurate categorization of data on race and ethnicity held in databases used by U.S. political scientists. They demonstrate that ethnic categories were non-standard across databases, rendering meaningless the conclusions of cross-national quantitative studies on diverse views on social issues (Herrera and Marquardt 2015). With respect to academic researchers, and not even considering other actors involved in data collection, Herrera noted that despite their acknowledgment that these databases are flawed, political scientists continued to use them because building new databases with comparable data parameters would take too much time, given academic expectations for publication, recognition, and grants intrinsic to career advancement within current systems. Thus, to build new databases, the academic reward structure needs to change, even as it is generally becoming even more demanding on researchers. The problems created by disincentives for change are even worse in GWAS, given that researchers in this field assume they can use PCA to validate or invalidate ethnic or racial identifications.

A major consequence of the use of race categories in medical genomics is that it can reify race as biological. Is there an alternative to using race categories in future genomic data collection? We have written about other options that GWAS scientists cast off in favor of the current paradigm because they were deemed less convenient and feasible (Fujimura and Rajagopalan 2011). The use of race categories remains one of the enduring challenges of genomics, and there appear to be no easy solutions. Those who advocate for using some version of lineage descriptors to handle the population substructure problem—be those descriptors ancestry, ethnicity, or race—promote methods for ever-finer delineations of populations based on presumed shared history, shared demographics, or shared gene pools. But these risk reifying race even further. Those, including ourselves, who advocate against using race categories are faced with population geneticists’ adamant contentions that their work becomes intractable without some accounting for the complex, layered, history of genome recombination that gives rise to the contemporary diversity (and similarity) of human genomes. But their idea of “intractable” is defined in terms of their own production timelines and does not take into account the concerns we raised earlier—the impact on peoples of color who are blamed for their illnesses.

So, we have taken what we call an ethical (policy?) position. Given that the historical and contemporary research on race shows that race and ethnicity in the United States and elsewhere are slippery, geographical and temporally contextual, and social concepts, we recommend that biomedical researchers instead adapt their current approach to address critical public health needs by more robustly aligning genomic investigations with an accounting of the many social determinants of health, such as clean water, nourishing food, education, employment, parks for play and exercise, time, and access to medical care. These are far more proximal (closer) factors impacting health outcomes. Our version of ethics is therefore less about an analytic philosophical exercise in individual justice, autonomy, free will, and so on, and more about a pragmatic, community justice, interventionist praxis; about responsible conduct as academics and researchers, with all the power relations that entails. We want to do more than just criticize science, although we still appreciate and participate in criticism. Indeed, critical research is needed to realize where science needs to change. But we also want to propose some solutions.

Despite recent attempts to undermine science, we cannot abandon critique, because we use it to work on improving science and society through STS. We do not assume that things have gone or will go smoothly or as we wish. But to not engage at all, despite the pitfalls, means that we stand by and allow what we know to be incorrect findings to govern treatments for communities already fractured by structurally racist systems in the United States. Here I take from Parvin and Pollock's (2020) argument that “unintended consequences” often take a disciplinary/industry approach to developing new technologies where anticipated consequences are already known and ignored. In our case, we have enough knowledge to anticipate dire health consequences for some people to suffer should we not change biomedical practices. At the same time, critique needs to be accompanied by denunciations of its weaponization, and with attempts to change science for the better (Fujimura and Holmes 2019).

These questions about ethical conduct as STS scholars matter, and are matters of collective concern, because they ask the field to weigh in on the question of responsible conduct as academics and researchers. These are not just questions about whether STS should remain passively observant, or take a stand within ethical debates or policy-making. Rather, these questions recognize that a decision point in which STS scholars decide whether and how to use their knowledge and expertise, whether to intervene or take a stand in the “real” world, is itself an instance of reflexive, ethical grappling, of deciding what kind of being and action in the world is ethical as a field to engage in.

4S Honolulu and the Revival of Hawaiian Indigenous Knowledge

Today's STS scholars are different in many ways from their predecessors. As an example, the keynote and plenary speakers at 4S 2023 were invited specifically for their interventionist research. Our keynote speaker Kamanamaikalani Beamer and the special plenary speakers Manulani Aluli Meyer, Mehana Blaich Vaughan, and Malia Akutagawa spoke of how Hawaiian knowledge ways had sustained and protected the Hawaiian people, communities, land, and sea prior to U.S. colonialism. They are developing their research and knowledge together with their communities and histories of Native Hawaiian knowledges. Some of it is restoring knowledge that has been suppressed through colonization, some of it is creating new knowledge as communities work with current environments. Their aims are to restore health and wellness; malama ka aina (protecting the land, growing healthy food, and feeding people). Postcolonial science for these scholars is about engaging with nature, communities, and the future in ways that restore knowledge ways from the ancestors, the kupuna, to address current and future problems.

The second plenary speakers Bruce Ka'imi Watson, Aurora Kagawa-Viviani, and Kyle Kajihiro spoke about making Kānaka science reign here, and returning Hawai'i to the Kānaka Maoli. They spoke about the poisoning of the land and people by business and the military, and about getting the military off Hawai'i and indeed away from the entire Pacific Rim. Getting the military out of Hawai'i is a topic not without debate among local residents, but it is debate that needs to address the complexities of life, health, and the environment on the islands, seas, and skies today. Other problems addressed by the second plenary include past bombing and destruction of Kaho'olawe, continual spills of poisons in Pearl Harbor, and myriad other dangers presented by the military, latest among them the current PFAS (per- and polyfluoroalkyl substances) poisoning of Oahu waters. As the speakers noted, these issues are not new, and reminded me of how during my Kaua'i high school years the military tested Agent Orange on the flora around the Wailua riverbanks, and buried metal barrels of Agent Orange on land in Wailua, which eventually poisoned the groundwater. Young people on the island were developing unusual cancers, which we high-schoolers believed were related to Agent Orange.

The health of Kānaka Maoli cannot be discussed separately from colonization; from attacks on their culture, language, families and kinship networks, lifestyle, and economic systems; from the stealing of their lands and water; or from the subsequent and ongoing socioeconomic consequences. The “first” Hawaiian Renaissance, beginning around King Kalakaua's reign, began efforts to restore the cultural, familial, and socioeconomic systems in Hawai'i. The “second” Hawaiian Renaissance brought in new energy in the 1970s from activists, musicians, wayfarers, and scholars like Trask (1999) in conjunction with Third World movements in the continental U.S. and anti-colonial movements around the Pacific. Relatedly, the island of Kaua'i has long been known to fight against development from the days of Prince Kaumuali'i, through land-use battles in the 1970s and 80s, and current efforts to control development. I participated in these land-use battles during my graduate school days, shuttling between the classroom and the struggles on Kaua'i. The Hawaiian Renaissance has worked to restore the teaching of Hawaiian language, culture, arts, land use, and history, and some of my grandnephews and nieces have benefitted from learning Hawaiian in immersion schools and at Kamehameha Schools. Despite these changes, land use and political struggles continue, as evidenced in the State Legislature sessions, county land-use decisions, the development of school curricula, the battles over the building of the latest international Thirty Meter telescope on Mauna Kea, and continued support for the restoration of a sovereign Hawaiian nation (see Goodyear-Ka’opua, Hussey, and Wright 2014).

Nevertheless, collaborations to restore Native Hawaiian knowledges continue. As just one example, Hawai'i imports 85–90 percent of its food. In ancient times, Kānaka Maoli produced enough and more food to feed its people without depleting the land and waters. Mehana Vaughan (2018) writes about how a “community of care” on the north shore of Kaua'i is restoring fish and sustainable fishing in the old and new ways, so there will be fish in the future. She argues for growing communities of care all over to restore ancient farming on land and sea, but to do it in ways that also learn from new possibilities to meet the need to feed more people. Given the current population of Hawai'i, the ahupua'a will not be enough to feed everyone, but combining Kānaka science and Western science may ameliorate problems created by colonization, its attendant growth, and colonial ways of fishing that depleted the oceans. But as Vaughan makes clear, communities of care have to be sustained through their kuleana to the land and sea in order for them to create this kind of sustenance. Relatedly, Kamana Beamer (n.d.) is working to build a circular economy:

In ancestral Hawaiʻi, a similar “give, take, regenerate” circular system led to the development of balanced structures of resource management. One example of this can be seen in the ahupuaʻa, a unit of land division and an efficient socio-political management structure that enhanced ecosystems health. Ahupuaʻa, in partnership with a sophisticated governance structure, ensured a successful ancestral circular economy, where resources were managed effectively to promote abundance. Combining contemporary Circular Economy solutions with ancestral knowledge creates integrated approaches to sustainability that are both environmentally regenerative and socially just.

Many Kānaka Maoli have not been able to participate in these communities of care in Hawaiʻi because of the lack of affordable living; rising costs in this capitalist economy have forced many off island (Trask 1999). I cannot do justice to this discussion in this essay, but I want to give credit to those who are here and are working to restore ancient fishing and farming knowledges and practices, through growing communities of care and including non-Native Hawaiians through their mutual kuleana to the land.

This kind of combination and collaboration of old and new practices and knowledges is found elsewhere as well, for instance in Native American communities on the continent. The U.S. National Science Foundation has begun to acknowledge Indian knowledge systems by helping to build centers of “two-eyed seeing” or knowledge-making that integrate Indigenous knowledge and Western science, such as the Center for Braiding Indigenous Knowledges and Science (CBIKS) based at the University of Massachusetts Amherst, which is “designed to weave together traditional ecological knowledge (TEK) and Western science” (Mervis 2023). According to anthropologist and CBIKS director Sonya Atalay, “Indigenous knowledge is place-specific, whereas Western science tends to seek universal rules that apply everywhere. Indigenous knowledge is rooted in the relationship between humans and their environment rather than isolating study targets from their surroundings.” Combining the knowledge traditions is, Atalay states, “a new model for studying the natural world,” drawing on “Indigenous ways of thinking [which] are not about ownership but about who has responsibility for that knowledge and who cares for it.” This is the same for Kānaka science, where the concern is for kuleana or responsibility for the knowledge and its use. The fact that NSF is funding CBIKS and other related projects is a phenomenal change from the past. I also note that some Indigenous scholars are critical of this “braiding” approach, so my focus on Vaughan's and Beamer's approaches is specifically mine.

Conclusion

The NIH’s National Human Genome Research (NHGRI)’s has focused on funding biomedical investigations into the genetic and genomic underpinnings of disease. The argument of the Human Genome Project was that it would elucidate genetic disease factors which could then lead to treatments. There have been genetic successes that came out of the Human Genome Project, having to do primarily with relatively rare monogenic diseases, that would have not been possible or would have taken decades longer without the HGP. But more prevalent common complex diseases like Type II diabetes, heart disease, most cancers, etc. do not have primarily genetic causes and have proven more elusive. After the HGP, NHGRI supported large-scale research into genomic factors through prospective genome-wide searches for associations with common complex diseases. This approach to disease has yet to deliver significant preventive measures or treatments for complex diseases, primarily because causality for these conditions cannot be attributed to single genes. Rather, it is complexly derived from interactions among biological factors, and social, cultural, political, environmental, and historical factors, often collectively referred to as environmental factors.

Relatedly, quantitative disparities in health outcomes have been used by some to argue for essential biological differences between races, or to blame the victim despite contrary evidence provided by public health researchers like Merlin Chowkwanyun (2022), Arline Geronimus (2023), and Nancy Krieger (2024). Our research has shown that genomic differences created by GWAS have also been used by some to argue for essential biological differences by race (e.g., Fujimura et al. 2014).

To guard against misunderstandings of race and health, we advocate for collaborative, integrative work with qualitative and quantitative researchers on racial disparities in health, including STS scholars, genome scientists, and communities of care. Communities of care have scientific practices developed over centuries that can contribute to making better science on genetics and disease. We see truly collaborative work between biomedical scientists and communities of care in research design and governance as avenues for Kānaka Maoli to both prevent the geneticization of themselves, and to bring their concerns and knowledges to develop studies that can improve their health. This approach does not draw them in only to greenlight scientific “progress” in ways that objectify nature and humans (Claw, Anderson, Begay, Tsosie et al. 2018). Kānaka knowledges are complex and have worked with uncertainties through the centuries. Our research aims for biomedical science to be planned and executed by and with their communities of care, not by their becoming mere objects of study (as in having “their” genomes genotyped).

Many younger STS scholars are already working on developing partnerships across communities, policymakers, and funders, in addition to partnerships with scientists. They will make STS even more expansive and interdisciplinary—more heretical, more imaginative, more crafty—in its thinking and in the strategies of possibility that it innovates. So, STS has already changed for the better, I think. But I am also grateful for the work that came before, especially since I am also of that era. Thank you for allowing me to speak about our research, and I thank Ramya Rajagopalan and the late Kjell Doksum for their contributions to our research projects and this reflection. ¹