Policy Experiments, Informed Consent, and Democratic Authorization 

A. What’s the Point of Informed Consent?

A full theory of informed consent is well beyond the scope of this article.²⁵ But like other commentators, I maintain that the most basic function of consent is that it permits others to interact with us in ways that would otherwise be wrong. Though a full theory of consent is not possible here, it is still important to say more about the two values that informed consent promotes because these values correspond to two important ethical objectives in human subjects research: (a) protecting individuals’ autonomy rights and (b) protecting and promoting individuals’ welfare interests. As mentioned at the outset, both (a) and (b) are constitutive elements of a more fundamental principle in research ethics: respect for persons. In my framing, informed consent is an ethical mechanism for achieving (a) and (b). And as I argue further on, democratic authorization is another ethical mechanism that achieves (a) and (b).

B. Autonomy

Start with (a) the protection of individuals’ autonomy rights. Autonomy, as I understand it for present purposes, is comprised of two elements. In order of importance, these elements are: (i) personal sovereignty and (ii) self-authorship. Personal sovereignty is the idea that there is a sphere of control over which persons are rightful rulers, viz., individuals have “final say” over what happens within this sphere. Joel Feinberg’s analogy with a sovereign state is helpful: “The politically independent state is said to be sovereign over its territory. Personal autonomy similarly involves the idea of having a domain or territory in which the self is sovereign.”²⁶ Understood this way, our autonomy rights are violated when there is illegitimate interference within our “jurisdiction” or rightful sphere of control. However, our legitimate sphere of control not only limits the actions of others, but also grants us a sphere of permissible action. Within our sphere of permissible action is a range of choices over what kind of life to lead. This is the notion of autonomy understood as self-authorship. Joseph Raz captures the spirit of autonomy as self-authorship as follows: “The autonomous person is a (part) author of his life. The ideal of personal autonomy is the vision of people controlling, to some degree, their own destiny, fashioning it through successive decisions about their lives.”²⁷

Both personal sovereignty and self-authorship are reflected in how bioethicists understand autonomy. In their influential Principles of Biomedical Ethics, Tom Beauchamp and James Childress write: “The autonomous individual acts freely in accordance with a self-chosen plan, analogous to the way an independent government manages its territories and establishes its policies.”²⁸ We can then further say that a person’s autonomy rights are violated when they are not granted adequate control over the direction their life takes.²⁹

One of informed consent’s functions is that it (a) protects individuals’ autonomy rights. But we also saw above that autonomy is composed of (at least) two elements: (i) personal sovereignty and (ii) self-authorship. This suggests that there are two corresponding types of autonomy rights that it is important to distinguish between. As I understand them, the autonomy rights corresponding to personal sovereignty are claims to non-interference or negative rights. These negative rights are possessed by all humans, exist pre-institutionally, and serve as a constraint on the actions of both individuals and governments.³⁰ One of the most important negative rights is the right to bodily integrity. In clinical contexts, it is typically the right to bodily integrity that accounts for the “consent” aspect of informed consent requirements. Because persons are sovereign over their bodies, clinicians and researchers must obtain permission to introduce or collect a substance into or from someone’s body in a way that is reminiscent of how governments regulate what enters or leaves its territories.

Not all autonomy rights correspond to personal sovereignty. What I refer to as entitlements correspond to autonomy understood as self-authorship. By entitlements, what I have in mind are claims to assistance that allow persons to meaningfully pursue a self-authored life plan. Because they are claims to assistance, entitlements are positive rights. Entitlements provide the basis for the “informed” part of informed consent requirements. In clinical research contexts, potential research participants are entitled to the information they need to ensure enrollment is congruent with their values and life plan. Researchers, in turn, have a correlative obligation to disclose such information to avoid illegitimate control over someone’s decision-making.³¹

C. Welfare

Autonomy as self-authorship is not the sole basis for an entitlement to information. Informed consent also (b) protects and promotes individuals’ welfare interests, and it is important to stress that (a) and (b) are distinct objectives corresponding to distinct values. After all, an individual’s autonomous choice may involve consenting to an interaction they know makes them worse-off.³² However, these two objectives often coincide. There are circumstances in which an individual’s autonomous choices best promotes their welfare. When reasoning adequately and sufficiently informed, individuals are in the unique epistemic position of knowing what is likely to promote their welfare interests.³³ And this further means that individuals are often in the unique epistemic position of knowing what risks to accept (or refuse) to (b) protect and promote their welfare interests. Note that in making this point, I take no position on the issue of what the ultimate constituents of well-being are. I only adopt an evidentialist view of the relation between informed choice and welfare.³⁴ What this means is that, when one is acting on self-regarding preferences, reasoning adequately, and sufficiently informed, this provides strong evidence that one’s consensual choice protects and promotes their welfare.

D. Minimalism and the Understanding Requirement

Despite the centrality of informed consent in the prevailing ethical and regulatory framework for human subjects research, it is no secret that the consent mechanism can fail to promote (A) autonomy and (B) welfare if our benchmark is some combination of an “unencumbered self”³⁵ and a utility maximizer. We know from fifty years of research in behavioral economics that cognitive constraints (sometimes) lead humans to make autonomous choices that deviate from rational norms.³⁶ Bioethicists have similarly taken note of these constraints. There is now an abundance of empirical evidence documenting the various ways research participants fail to fully understand their participation in research.³⁷ Even if researchers go to great lengths to disclose all the relevant information that a decision maker ought to be provided with to make an informed decision, there is bound to be a discrepancy between what information is disclosed and what is actually understood by consenting research participants. Yet if we accept there is often misconception about what involvement in clinical research entails, and we accept a view of informed consent on which the contents of what is understood must be identical to the contents of what is disclosed (the “standard view”), we arrive at the startling (and implausible) conclusion that most clinical research should come to a halt.³⁸ At the same time, no one is calling for a moratorium on clinical research.

The tension highlighted above has prompted philosophers and bioethicists to think more carefully about the understanding requirement for valid consent. Careful analysis of the problem has led Joseph Millum and Danielle Bromwich to defend minimalism with respect to understanding.³⁹ On their account, a research participant must understand three things to give valid consent: (1) that she is giving consent; (2) how to exercise her right to give or refuse consent; and (3) to what she is being asked to consent. The exact details of Millum and Bromwich’s proposal and whether it is ultimately the correct minimalist account need not concern us here.⁴⁰ What is important to emphasize is that minimalism about informed consent can shed light on the ethics of real-world policy experiments. As we proceed, it will be important to not let an idealized understanding of informed consent set an unrealistic benchmark for democratic authorization. In the same way that perfect rationality is not required for valid consent to a clinical trial, I will suggest that a perfectly democratic society is not required for valid authorization to a policy experiment.

A. Healthcare Research Without Informed Consent

Lukas Gelinas, Alan Wertheimer, and Franklin Miller have developed a framework for assessing when and why research without consent is permissible.⁴¹ Though Gelinas et al.’s focus is the healthcare context, their approach extends to policy evaluation. Drawing on an analysis of informed consent like the one above, Gelinas et al. propose two conditions that are jointly sufficient for making research without consent permissible: (1) the research stands to infringe on no right of the participants and (2) it is impracticable to obtain consent.⁴² Before elaborating on (1) the individual rights condition, it is important to say something about (2) the impracticability condition. Gelinas et al. emphasize that it is still important to obtain consent in cases where it is practicable even if failure to do so violates no one’s rights. This is because doing so promotes public trust and transparency in the research enterprise. While this is an important consideration, what promotes public trust in the research enterprise is an empirical matter and cannot be settled by philosophical analysis. To avoid speculation, I set this consideration aside for now but return to it in my concluding remarks.

What is important to focus on is Gelinas et al.’s (1) individual rights condition. Take Gelinas et al.’s analysis of quality improvement (QI) studies undertaken at the institutional level. An institutional level QI study will evaluate the effectiveness of an intervention on healthcare delivery or patient outcomes in a healthcare facility or healthcare system. QI studies are regularly done without any informed consent.⁴³ Gelinas et al. argue that, because institutions such as hospitals and healthcare systems possess rights of control, QI studies without consent are permissible when they evaluate interventions within their legitimate sphere of control. Like Gelinas et al., I will not specify the exact boundaries of where a healthcare institution’s rights of control end and where a patient’s right to exercise their autonomy rights over their healthcare decisions begin. Denying that hospitals and healthcare systems have a legitimate sphere of control would be highly implausible, as evidenced by the example below, and which also illustrates Gelinas et al.’s approach to research without consent:

institutions have a right to require that clinicians wash their hands with a particular type of disinfectant soap, without consulting or obtaining consent from patients for this practice. Given that this is a practice within the decisional authority of the institution, so long as the institution ensures a standard level of care with regard to hand washing and sanitation, institutions should have the right to conduct research comparing two types of disinfectant soap for use by clinicians without patient consent. The rights of patients will not be infringed by the conduct of such QI research without consent.⁴⁴

With respect to research without consent, the main insight to extract from the passage above is the following principle:

Intervention without Consent: If X possesses the right to implement intervention I in population P in a non-research context, then X does not violate any constituent of P’s autonomy rights by implementing I without consent in a research context.

In the example above, the healthcare facility (presumably) has the right to decide which disinfectant soap a clinician uses in its hospitals. Therefore, the QI study on disinfectant soap is permissible from an autonomy-based perspective. The autonomy rights of clinicians and hospital patients are not violated since they lack a right to determine which disinfectant soap a hospital uses. The disinfectant soap a hospital uses is not within the legitimate sphere of control possessed by these individuals.

Contrast the QI study with a typical clinical trial. The kinds of interventions and data collection procedures involved in a clinical trial require bodily interventions. Clinical researchers cannot just force a promising experimental drug down the throats of research participants, nor can they perform a nonconsensual lumbar puncture while a research participant is unconscious. These kinds of bodily intrusions are forbidden because people have a right to bodily integrity. This entails researchers also have a correlative duty to respect such a right, hence why informed consent is so important in the context of clinical research.

B. Policy Experimentation Without Informed Consent

Douglas Mackay and Averi Chakrabarti have extended Gelinas et al.’s framework to government policy experiments.⁴⁵ The key insight to MacKay and Chakrabarti’s analysis of policy experimentation without consent is that governments are also institutions with rights of control. Specifically, government institutions have rights of control over policy decisions within a given territory. As these commentors put it: “The normative relation of governments to their residents is thus similar to the relation that Gelinas et al. understand to obtain between clinical institutions—e.g. hospitals—and their patients.”⁴⁶ Using this insight, MacKay and Chakrabarti propose two jointly sufficient conditions for permissible policy experimentation without consent: (1) the government institution conducting or authorizing the experiment possesses a right to rule over the spheres of policy targeted by the research; and (2) data collection does not involve the violation of participants’ autonomy rights.⁴⁷

The analysis of informed consent provided earlier helps fill in the details of the second condition in MacKay and Chakrabarti’s framework. Because persons possess a right to bodily integrity, any policy research involving bodily intrusions (such as blood draws) clearly requires consent regardless of institutional context. Other data collection procedures that interfere with an individual’s personal sovereignty (e.g., their privacy rights) are similarly forbidden.

Matters are less straightforward with respect to MacKay and Chakrabarti’s first condition (the right to rule condition). MacKay and Chakrabarti are certainly appealing to Intervention without Consent combined with the notion of a government’s “right to rule” to justify this first condition. However, MacKay and Chakrabarti never explain where a government institution’s right to rule over policy comes from. This is an issue fundamentally connected to whether a government has political legitimacy, i.e., whether it is morally justified in wielding political power.⁴⁸ MacKay and Chakrabarti do suggest that, unless one is a philosophical anarchist⁴⁹ or holds a theory of political legitimacy that no real-world governments currently meet,⁵⁰ one can maintain that there is a wide range of policy interventions that democratic governments (and their various institutions) have the right to implement. Though I do not disagree, a more complete account of the ethics of policy experiments requires further engagement with democratic theory for reasons I expand on below.

C. Three Difficulties with the Right to Rule Condition

There are three difficulties with MacKay and Chakrabarti’s right to rule condition (as initially presented) that are in tension with informed consent’s two objectives, and which can ultimately be remedied with insights from democratic theory. The first difficulty is that even though autonomy as personal sovereignty is accounted for in condition (2), something should be said about the background conditions (i.e., the institutional framework that governs a society) that ensure such rights violations do not occur in a frequent and systematic fashion. This not only applies to the data collection stage of a policy experiment but the intervention stage as well. The issue of favorable background conditions is often taken for granted in discussions of clinical research ethics, but it should certainly not be taken for granted when analyzing policy experiments—especially policy experiments in LMICs.

The second difficulty concerns autonomy understood as self-authorship. There is an unexplored tension between a government’s right to rule and its residents’ autonomy rights in condition (1) of MacKay and Chakrabarti’s framework. Governments (and other social institutions) have a profound impact in shaping the course of its citizens’ lives. An important way that governments affect their citizens’ lives is through its policy decisions. This suggests that policy experiments have the potential to undermine autonomy understood as self-authorship. Policy experiments involve manipulating some aspect of an individual’s environment for the sake of generating policy relevant knowledge. More specifically, they can involve manipulation of an individual’s entitlements to socioeconomic resources and opportunities. This connection between autonomy as self-authorship and access to resources and opportunities should not be downplayed. After all, one of John Rawls’ most influential insights is that these entitlements bear on an individual’s ability to pursue a meaningful life plan.⁵¹ Poverty is not only bad because of its effects on human welfare, but also because it deprives individuals of the ability to make meaningful choices about their lives.⁵² Similarly, one’s access to healthcare resources (such as ITNs) can affect whether one is healthy enough to partake in the social, political, and economic activities of their society.⁵³ And the duration and quality of one’s education is important not just because of the economic returns, but also because it enables one to develop the knowledge, skills, and critical thinking abilities necessary to make informed decisions in life.⁵⁴

The third difficulty concerns human welfare. Like with much research involving human subjects, policy experiments come with risks of causing (non-trivial) harm to individuals. In addition to protecting autonomy rights, informed consent also shows respect for persons by only exposing research participants to (substantial) risks to their well-being that they are willing to accept. After all, sacrificing the (substantive) welfare interests of an individual (or minority of individuals) to promote a socially valuable end (e.g., scientific research) is the paradigm example of ignoring the “separateness of persons”⁵⁵ and thereby failing to show respect. Since it is impracticable to obtain consent in some policy experiments, there is the troubling possibility of doing just this. And even in cases of policy experiments where it is possible to obtain consent for intervention and data collection (and their associated risks), there is still the troubling possibility of imposing risks on (non-consenting) bystanders. This consideration is especially salient in LMICs, where there is a tendency for socioeconomic interventions to have unintended effects.⁵⁶

D. The Need for Democratic Theory

The first two conditions of MacKay and Chakrabarti’s framework for policy experimentation without consent are worth maintaining. However, their framework requires further elaboration to address the three difficulties above. To address these worries, I argue that a clarificatory—but nevertheless important—third condition is needed: (3) a government institution possesses a right to rule over a sphere of policy if the institution is authorized to do so by a democratic procedure. With this addition, MacKay and Chakrabarti’s framework would comply with the requirement that human subjects research shows respect for persons. This is because, as I argue below, (P2) democratic authorization is like informed consent in that it is an ethical mechanism that (a) protects individuals’ autonomy rights and (b) protects and promotes individuals’ welfare interests.

Though I argue below that democratic authorization is the source of a government institution’s right to rule over a sphere of policy, it is important to make clear that there are policy interventions that not even democratically authorized actors can experiment with because they violate an individual’s autonomy rights. Consider, for example, a (completely hypothetical) policy experiment that evaluates the effects of limiting free speech on educational outcomes. Unlike the proposed SOAP or Tennessee STAR experiments discussed at the outset, no democratic government could permissibly conduct or authorize such a policy experiment. The intervention of interest conflicts with an individual’s right to freedom of expression, and as such, would not be within the legitimate sphere of control that even democratic governments possess. I maintain throughout that democratic governments cannot conduct or authorize non-consensual policy experiments that involve interventions (or data collection procedures) that conflict with individuals’ autonomy rights. However, for reasons I expand on below, this is an unlikely to happen due to the close connection between democracy and autonomy.

A. Democracy and Personal Sovereignty

Implicit in the common understanding of democracy is the idea that democracy is a system of governance that respects individual freedom. This idea can be cashed out in (at least) two ways (a second way is discussed and rejected in Section IV.B). The first way appeals to an empirical regularity between the protection of personal sovereignty and democratic institutions. This empirical regularity licenses itself to social scientific investigation. There is strong empirical evidence suggesting that democracies do a better job of protecting a special class of negative rights than non-democracies.⁵⁷ These negative rights are often referred to as personal integrity rights. Personal integrity rights discussed in this context include the right to not to be tortured, the right to not be arbitrarily imprisoned, and the right to not be murdered or disappeared by the state. These personal integrity rights presumably stem from a more fundamental right to bodily integrity, which I suggested provides part of the ethical basis for informed consent requirements in clinical research, and which also provides the basis for a right to not be medically experimented on without consent.

The link between democracy and the protection of personal sovereignty is an important first step in showing that democratic authorization is an ethical mechanism that (a) protects individuals’ autonomy rights. The fact that a policy experiment is conducted in a democratic society provides strong evidence that background conditions are such that they do not enable fundamental rights violations by public or private actors conducting human subjects research (including policy experiments). Specifically, the link between democracy and personal sovereignty kicks in when a society that meets the conditions for what Thomas Christiano calls minimally egalitarian democracy. The conditions are:

1. Persons have formally equal votes that are effective in the aggregate in determining who is in power, the normal result of which is a high level of participation of the populace in the electoral process.

2. Persons have equal opportunities to run for office, to determine the agenda of decision making, and to influence the processes of deliberation. Individuals are free to organize political parties and interest group associations without legal impediment or fear of serious violence, and they are free to abandon their previous political associations. They have freedom of expression at least regarding political matters. In such a society, there is normally robust competition among parties and a variety of political parties that have significant presence in the legislature.

3. Such a society also acts in accordance with the rule of law and supports an independent judiciary that acts as a check on executive power.⁵⁸

As Christiano makes clear, a minimally egalitarian democracy “need not by any means be fully just nor need it fully live up to the ideals of democracy.”⁵⁹ For example, a minimally egalitarian democracy is consistent with wealth inequalities limiting opportunities for substantive (rather than formal) equal influence over political decision making. A minimally egalitarian society may also not guarantee the full range of rights associated with the ideal of personal sovereignty if these rights are not connected to the democratic process. But to qualify as a minimally egalitarian democracy, it is crucial that racial and ethnic minorities should have “the protections of the rule of law, free association, and free expression, as well as equal opportunities for organizing politically effective groups.”⁶⁰

The conditions for minimally egalitarian democracy are worth highlighting because they provide a realistic standard for whether a government is justified in wielding political power. Going forward, I maintain that a real-world government is legitimate if the society it rules over meets the conditions for minimally egalitarian democracy. This further implies that a government institution possesses a right to rule over a sphere of policy if it is authorized to do so by a legitimate government, i.e., a government that rules over a minimally egalitarian democracy.

Despite the above, the link between minimally egalitarian democracy and personal sovereignty can only carry the argument so far. As specified earlier, autonomy is comprised of another element, i.e., self-authorship. My defense of P2’s first conjunct is incomplete without some discussion of democracy and autonomy as self-authorship.

B. Democracy and Self-Authorship

Though it is hard to deny that personal integrity rights are of the utmost importance, personal sovereignty cannot fully account for democracy’s importance. If it could, then then there would be nothing objectionable with a society ruled by a benevolent autocrat. After all, there is nothing incoherent or contradictory about an autocrat who goes to great lengths to protect the personal sovereignty of his subjects. But in such a society, the autocrat’s subjects would have no control over political decision making, which has a profound influence on the course of anyone’s life. Many are inclined to think that there is something regrettable about institutional arrangements that do not grant individuals some control over the influence of politics on their lives. But unfortunately, the intuition many share about democracy and autonomy as self-authorship is hard to make sense of. There are well known complications with the view that democracy promotes individuals’ autonomy by making them part of authors of law and policy, the most notable being the possibility of being on the losing side of a vote.⁶¹ One can still make the case for democratic arrangements facilitating joint authorship of law and policy and thereby promoting a form of collective autonomy distinct from individual autonomy.⁶² But in the case of minimally egalitarian democracies (i.e., real-world democracies), it is pretty clear that the decisions of both elected and unelected government officials do not typically reflect the will of the people, so citizens are often subject to laws and policies that they did not jointly author in any meaningful sense.⁶³

None of this should be taken to suggest that democratic authorization conflicts with autonomy as self-authorship. The value of autonomy as self-authorship should not be confused with complete control over the direction one’s life takes. Complete control is an untenable standard as there are aspects of any person’s life that they may deem important but are beyond their individual sphere of control.⁶⁴ For example, a bachelor’s life plan may involve marrying and having children with a certain kind of person, but this plan may be curtailed by the autonomy rights of eligible partners. By exercising their autonomy rights over who to marry, these eligible partners may incur obligations that prevent them from marrying the bachelor. Yet we would not say that the bachelor’s autonomy rights are violated by the choices of the partners he deems eligible.

An individual’s life plan may also be curtailed by a legitimate government exercising its rights and discharging its obligations. When it comes to policy experiments, one obligation is particularly salient. As a condition of possessing a right to rule, governments incur an obligation to, as Raz puts it, “create an environment providing individuals with an adequate range of options and the opportunities to choose them.”⁶⁵ This obligation stems from the basic interest all individuals have in formulating, pursuing, and revising a life plan of their own making. Policy experiments are important precisely because they help government institutions successfully discharge the obligation to create the conditions for self-authorship.⁶⁶ Among the most important ways to create the environment Raz envisions is by ensuring that policy effectively produces autonomy-related outcomes in a population. Yet at the same time, what class size best promotes educational outcomes and what rate of subsidy for health insurance best promotes health are policy questions that bear on entitlements but have no a priori answer. These are empirical questions that policy experiments can help answer in an evidence-based and cost-effective manner.

Combining the two points above, the main takeaway is that, despite potentially affecting the course of one’s life, non-consensual policy experiments do not ultimately conflict with any specific individual’s right to (partial) self-authorship if authorized by a legitimate government. In many cases, government institutions are simply acting on their obligation to promote autonomy as self-authorship in the population they rule over, including those targeted by an intervention as part of a policy experiment.

C. Taking Stock

My aim in this section has been to highlight how the connection between democracy and autonomy can be leveraged to (partially) justify non-consensual policy experiments. At this stage, I have only argued that democratic authorization achieves one of two ethical objectives associated with informed consent. In the next section, I turn my attention to the second, welfare-based ethical objective associated with informed consent. Those who worry that I have not said enough to rule out the possibility of a benevolent autocrat authorizing a non-consensual policy experiment will find refuge in learning that, though benevolent autocracies are perhaps a real possibility, these governments are unlikely to have the institutional machinery needed to permissibly authorize non-consensual policy experiments.

A. Acceptable Risk

When it comes to risk assessment, some may be inclined to think informed consent is important because it is never permissible for researchers to impose risks on others without their consent. This is a mistake. Setting autonomy considerations aside, the reason consent is important in clinical settings is due to the significant potential for harm associated with a bodily intrusion. A general right against risk-imposition is also a non-starter. Virtually every form of social organization—including the assignment of rights and obligations—creates unconsented-to risks for individuals.⁶⁷ Governments routinely make decisions or implement policies that not only impose risks, but also and knowingly benefit some citizens at other citizens’ expense (e.g., trade policy, monetary policy, etc.). Not only do governments sometimes impose risks and set back the welfare interests of some of its citizens, but governments may justifiably do so.⁶⁸ This is just a consequence of governments exercising their right to rule over policy. After all, it would be miraculous if the exercise of a government’s right to rule respected its citizens’ autonomy rights but never affected anyone’s welfare.

As a general matter, it is uncontroversial for a surrogate decision maker to make risky medical decisions on behalf of an incapacitated person. Surrogate decision makers may also permissibly enroll incapacitated persons in a clinical trial under certain conditions. The reason that family members are often deemed acceptable surrogates is that, due to their close relationship, family members are the most responsive to the incapacitated persons’ interests. I want to suggest something similar in the case of democratically authorized actors. As I argue below, the reason that democratic authorization makes research risks acceptable is that democratically authorized actors are responsive to those whose interests are potentially affected by a policy experiment.⁶⁹

B. Democracy and Government Responsiveness

Earlier I discussed the causal link between personal integrity rights protections and democratic institutions. This is not the only empirical basis for an instrumental justification of democracy. Democracy is also valuable because it promotes human welfare. There are different ways which democratic institutions promote human welfare.⁷⁰ A particularly illustrative example is the empirical relationship between democratic institutions and famine prevention in LMICs. This empirical relationship has most famously been studied by Amartya Sen, who stresses that the relationship between democratic governance and famine prevention in LMICs is no coincidence.⁷¹ The empirical regularity has an underlying causal basis: government responsiveness. Examining the causal basis for Sen’s findings helps address the issue of ultimate concern here: why democratically authorized actors are in the position to conduct non-consensual policy experiments. Per Sen, democratic institutions incentivize those in power to protect and promote the welfare interests of those who put them in power. In explaining the link between democracy and famine prevention, Sen emphasizes the political incentives that operate on governments and on the persons and groups that are in office. Sen writes that in democracies, “rulers have the incentive to listen to what people want if they have to face their criticism and seek their support in elections.”⁷² For Sen, these political incentives translate into government responsiveness.

William Easterly is another economist who emphasizes the link between democratic institutions and positive welfare outcomes, particularly in LMICs. He stresses that democratic governments are responsive to their citizens welfare because they are accountable to them:

accountability is a crucial mechanism in development to ensure that government does good and not ill to those affected by its actions. Under democracy, citizens can use many mechanisms—such as voting, popular protests, and spoken and written criticisms—to penalize governments that are harming individuals (even if it is only a minority of individuals). The same mechanisms reward political actors that do good by, for example, supplying public goods. When such mechanisms work, the government is accountable to its citizens. The opposite of accountability is impunity—the government can do whatever it wants to the citizens without consequences.⁷³

What Easterly’s helpful synopsis suggests is that democratic authorization is a causal mechanism in addition to an ethical mechanism. Accountability explains why democratically authorized actors are responsive to citizens’ welfare interests. Under democratic institutions, citizens enjoy the coordinating benefits of centralized political power while ensuring their welfare is not only protected from the exercise of that power, but also promoted.

With the link between democratic institutions and government responsiveness in place, we can now turn back to non-consensual policy experimentation. Democratic governments are in a sufficiently good position to conduct or authorize non-consensual policy experiments because of the political incentives democratically authorized actors face. Democratically authorized actors are already incentivized to design and evaluate policies that promote their citizen’s welfare interests. Further, democratically authorized actors are also disincentivized from implementing policy interventions that are ex ante harmful or that concentrate risk on some subpopulation. For this reason, democratically authorized actors are also unlikely to conduct or authorize policy experiments that are expected to harm individuals or unfairly impose risks on some subpopulations. This is because those targeted by an intervention have political representatives that must account for their welfare interests to remain in positions of power. And while it is impossible to guarantee no one is ever harmed over the course of a policy experiment (or any research, for that matter), bystanders and those targeted by an intervention have the option of seeking rectification through the ballot box or via their elected representatives if their interests are unfairly set back because of a policy experiment.

C. Taking Stock

I have argued in this section that the causal link between democratic institutions and government responsiveness shows that democratic authorization is an ethical mechanism that (b) protects and promotes individuals’ welfare interests. This completes my main argument. Non-consensual policy experiments show respect for persons when the government institution conducting the experiment is democratically authorized to rule over the spheres of policy targeted by the research. We can further say that, just in the same way that informed consent permits others to interact with us in ways that would otherwise be wrong, democratic authorization permits government institutions to interact with us in ways that would otherwise be wrong.

While critics may acknowledge that my argument is sound in the abstract, they will object that it is too idealized for real-world circumstances. Anyone who has the slightest interaction with the real world knows that my characterization of democratic authorization is simplistic. After all, unelected government actors are not directly accountable to citizens so they may not always be responsive to their interests. Further, due to the well documented cognitive constraints alluded to earlier, voters may elect representatives who fail to protect and promote their welfare.⁷⁴ Looming in the background is also the sordid history of research abuses sanctioned by the agencies of (purportedly) democratic governments. Black and brown persons have especially felt the brunt of these abuses.⁷⁵

My account of democratic authorization does still have bearing on the ethical permissibility of real-world policy experiments. First, it is worth noting that the prevailing ethical and regulatory framework for the protection of human subjects in the U.S exists today because of an increasingly democratic US government responding to its past inability to protect the interests of vulnerable populations, which includes racial and ethnic minorities.⁷⁶ With respect to the charge of oversimplification, consider an analogy with informed consent. Hardly anyone questions the ethical importance of informed consent in the abstract. Yet when we look under the hood of the consent mechanism, we find that it only has a strong tendency to promote positive welfare outcomes and does not necessarily guarantee them. This does not mean that it is impossible for researchers (or anyone, for that matter) to obtain valid consent, much less show respect for persons. Democratic authorization should be seen in a similar light. The accountability mechanism so crucial to justifying democracy’s ethical importance also has a strong tendency to promote positive welfare outcomes but does not guarantee them. Unless some other currently feasible ethical mechanism can more effectively reconcile a government institution’s obligation to implement evidence-based policies with the need to (b) protect and promote the welfare interests of those targeted by an intervention, we should accept that, for the purposes of conducting non-consensual policy experiments, democratic authorization shows respect for persons.⁷⁷