Around 2014, the Chinese government ramped up its campaign to “de-extremify” Muslim Uighurs in the Xinjiang province, a huge swath of land that makes up the northwestern edge of the country. China’s grasp on the region has tightened over the past few years, and the country has come under increased scrutiny as reporters have uncovered its extensive, constant surveillance of Uighurs living in Xinjiang, and its detention of more than 1 million Uighurs in communist “reeducation” camps, where people are forced to study Communist propaganda and even waterboarded.
And in 2016, Chinese authorities launched a program called “Physicals for All” that was ostensibly intended to provide health care. As part of that, though, the government also collected DNA samples, iris scans, and other biometric data from more than 36 million people in the Xinjiang region. In the years that followed, authorities shared that information as part of collaborations with scientists from around the world, who have published hundreds of papers including data from Uighurs and other vulnerable ethnic minority groups in China. Those studies are the foundation for the development of dangerous technologies like facial recognition that identifies ethnic groups, or DNA tests that can reconstruct a person’s face. This raises questions not only about the ethics of such technologies but also in the data that’s helping create them: Was that data collected voluntarily, and did participants provide informed consent for its use?
This week, two major academic publishers—Springer Nature, which owns the prestigious journal Nature, and Wiley, which publishes more than 1,600 journals—told the New York Times they’ll reevaluate previously published papers that include data from “vulnerable groups.” Springer Nature said it will “bolster” its guidelines, and editor-in-chief Philip Campbell said the company will contact editors to “request that they exercise an extra level of scrutiny and care in handling papers where there is a potential that consent was not informed or freely given.”
One 2018 paper, which compares Uighur, Tibetan, and Korean faces for “ethnic group facial recognition,” seems to have undergone this reevaluation process. I found a tweet from June with screenshots of the original paper. Now, it includes a line that reads, “The study was conducted with the approval of Dalian University and written, informed consent was obtained from each study participant who understood their photographs would be used for non-profit scientific research.”
Tightening up ethics standards and instating a reevaluation process sound like good ideas driven by good intentions. Journals refusing to publish data that was clearly collected unethically sends an important message to the scientific community that bad research hygiene will not be tolerated, and setting up a checkpoint to ensure that researchers have obtained the proper ethics board approvals and informed consent from participants is an easy first step for evaluating research ethics. But even if a study passes this basic checkpoint, that doesn’t mean it’s been done ethically, or that researchers have given careful thought to the implications of their work. How to determine whether studies are ethical is an open-ended question, and to answer it, researchers will need to go beyond checking things like universities’ approvals or records of informed consent.
Even enforcing that first guideline—verifying that researchers obtained the proper approvals and consent documents before beginning their work—would be quite complicated. Submitting a manuscript for publication in an academic journal is already a tedious, overcomplicated process I wouldn’t wish upon my worst enemies. You format your file to the exact esoteric requirements of the journal, upload it to a web portal that works with the speed and efficiency of dial-up AOL, and wait months for reviews from other overworked academics, who then eviscerate your manuscript.
I’ve submitted a handful of papers, all of which involved data from children, which ethics boards define as a “vulnerable population.” I’ve never been asked to submit any kind of evidence that I’d obtained the proper permissions from a research ethics review board, or for evidence that I’d gotten informed consent for each participant. I’m thankful for that, because it’d be a painstaking process to upload hundreds of pages of files, and someone on the other side of the editorial process—presumably a reviewer, working for free, or a journal editor, working for a paltry honorarium—would need to verify these documents. A much easier but clearly flawed process, which I have seen in journal submission materials, is to ask researchers to check a box saying they’ve obtained the proper permissions to collect participants’ data. While I’d like to believe most people are honest, it’d be hilariously easy for a researcher acting in bad faith to just check the box and pass this basic screening.
To find out more about the process, I emailed the lead author of the ethnic facial recognition paper as well as three of the journals’ editors, but none responded to my messages. I also emailed a Springer Nature communications official, but received no response there, either, as well as a Wiley communications official, to ask about their reevaluation process. (We’ll update this piece with any responses we receive.)
Let’s assume that journals have taken the most thorough route and resolved to verify that researchers have documents showing they’ve obtained the proper approvals and consent from participants before collecting their data. But signed forms do not necessarily indicate data has been collected ethically, says Yves Moreau, a professor of engineering at the University of Leuven in Belgium who recently wrote a Nature commentary advocating for a crackdown on “genomic surveillance”—researchers’ and companies’ use of DNA databases to track or profile individuals. Informed consent requires participants to freely consent—something that’s difficult for vulnerable groups like Uighurs, who face the threat of being sent to reeducation camps for refusing cooperation. Informed consent also requires not only acquiescence to data collection but also a real understanding of how that data might be used. “True informed consent would require including statements like ‘this technology could potentially be further developed to discriminate and persecute members of your ethnic group, your family, or yourself,’ which makes it clear that ‘informed consent’ is essentially an oxymoron in this context,” says Moreau.
Plus, if journals are relying on ethics committee approvals, that’s hardly a uniform standard. The process for obtaining research approval varies widely from country to country. “Even if consent forms and ethical approval is available, editors should not systematically defer to the ethics committee decision,” says Moreau. “I have [seen] multiple instances of papers on DNA profiling of Chinese populations where the ethical approval was provided by the ethics committee of the Institute of Forensic Science of the Ministry of Justice.” This is essentially a higher-stakes version of a boss approving his own request for time off, or a grade-grubbing seventh grader checking her own math homework.
Such a system may also end up continuing science’s legacy of Western paternalism. “Requiring a rule of assurance and submitting copies of the consent documents might be culturally elitist—it places the Western standard of autonomy above all other ways of thinking of people—placing the individual over the group,” says Craig Klugman, a professor of bioethics at DePaul University. Not every community privileges individual consent in the way Westerners do. Klugman points out that some people may seek permission from a parent, spouse, or village elder. Any good researcher should understand the norms and culture of any group they’re studying before they ask for permission to begin work, and when working with groups that might engage in, say, communal decision-making, researchers might approach community leaders, like a council or chief, before asking individuals to participate. In a paper on informed consent in international research, a group of medical researchers write that “developed countries’ informed consent procedures are so established and regulated that it is difficult for investigators to consider alternatives.” For instance, a communal consent on behalf of a community would not suffice by most ethics boards standards. In that case, researchers might still need to collect individual consent forms, a move that the researchers say could cause “concern and distrust,” especially if signatures are generally only used for serious decisions, like business agreements. Any rigid journal policies requiring such forms might disqualify great work with underrepresented communities, even if it was conducted ethically.
Enacting ethics review board approvals or consent forms as the standard might also “have the effect of dismissing science from China as not meeting ‘Western standards,’ ” says Klugman. As a Chinese American, I’ve heard my fair share of sweeping generalizations about Chinese scientists’ commitment to ethics, and anti-Chinese sentiments in general. There’s a careful balance to be struck here in seeking an ethical way forward, where Chinese researchers are held to high standards of ethics without Western scientists gatekeeping the field by strictly enforcing their norms and standards. It’s already the case that most prominent journals are run predominantly by editors in the EU and U.S.; whatever norms those journals enact will have implications for science at large.
As technologies and methods evolve, so too will scientists’ understanding of ethics. Moreau would like to see a system in which editors and publishers continuously collaborate to make decisions about what is ethical. “I am not advocating setting up a bureaucratic machinery to systematically check consent forms and second-guess the work of ethical committees,” he says. “What I am pushing for is for reviewers, editors, and publishers to fulfill their duty of vigilance.” Journals’ ethics policies can certainly head off major issues, if they catch problematic papers before they’re published, but the root of the problem remains: Researchers themselves must think carefully about the work they’re doing, how they’re doing it, and what their data might be used for, even if they themselves do not have bad intentions. (If you’ve seen Terminator, you know Miles Dyson didn’t mean to build Skynet.)
“I believe it’s time to do something,” says Moreau, even if it isn’t yet clear how to ensure published research has been completed ethically. “We need to bring these problems forward and force society to understand what’s at stake, and to make good choices about what to do next.”