Political Legitimacy as Grounded in the Wills of Citizens: A Reply to Peter

Abstract

Fabienne Peter (2020) recently proposed a taxonomy of accounts of the meta-normative grounds of political legitimacy. In this article, I argue that there is an important distinction left out of that taxonomy that complicates the picture. This is the distinction between attitude-independent and attitude-dependent conceptions of normative truth. Through an examination of these conceptions of normative truth (and correlate interpretations of what counts as a normative reason) I argue that what Peter calls a fact-based conception of legitimacy may collapse into a will-based conception. Further, the distinction has important implications for what Peter calls the belief-based conception. Finally, I defend the will-based conception against Peter's arbitrariness objection through an examination of ideally coherent eccentrics.

Introduction

Theories of political legitimacy in contemporary political philosophy tend to focus on first-order normative considerations, such as whether procedural or substantive values justify the content of legitimate political decisions. However, as Fabienne Peter (2020) has importantly pointed out, there is an often overlooked distinction among conceptions of political legitimacy originating at the metanormative level.

Peter proposes a taxonomy of accounts of the grounds of political legitimacy, including will-based, belief-based, and fact-based conceptions. In this article, I argue that at least one further distinction is needed in order to articulate the space of possible views—importantly, within the fact-based conception we must distinguish between attitude-dependent and attitude-independent conceptions of normative facts. I argue that the fact-based view will collapse into a type of will-based view in cases where normative truth turns out to be attitude-dependent. I argue that this distinction has important implications for the belief-based account. Finally, I defend the will-based account against Peter's objections.

1. Peter's Taxonomy of the Grounds of Political Legitimacy

Peter's taxonomy is an innovative contribution for theorizing about conceptions of legitimacy at the metanormative level, so this is my starting point although my remarks will complicate her picture. In this first section, I give an overview of the three main types of conceptions offered by Peter and discuss the objections against these conceptions. Following Peter, I take a minimalist interpretation of the concept of political legitimacy, viewing it as a normative property of some political decisions. While analyses of legitimacy can differ in scope—sometimes applying to a state as a whole, individual or sets of political actors, or particular political institutions—Peter takes political decisions as a starting point because this need not presuppose a particular account of institutions. The minimalist interpretation ‘is compatible with an account that centers on the problem of justifying the state, as even those accounts will ultimately distinguish between legitimate political decisions—those made by a state that satisfies certain conditions—and illegitimate political decisions’ (Peter 2020: 372). Political decisions have the property of legitimacy when the political decision-maker is entitled to make the decisions and citizens have correlate duties to obey.

The notion of political legitimacy has at least two senses—a descriptive sense and a normative sense. The descriptive sense, or de facto legitimacy, refers to the case when an institution is believed by those under its jurisdiction to have authority. De facto legitimacy is important and often thought necessary for an institutional regime to govern successfully. The descriptive notion of legitimacy is often the focus of discussions about legitimation in political science and sociology. Of the descriptive sense, Max Weber writes, ‘the basis of every system of authority, and correspondingly of every kind of willingness to obey, is a belief, a belief by virtue of which persons exercising authority are lent prestige’ (Weber 1964: 382). But my focus (and Peter's) is the normative sense of legitimacy, which is concerned with the normative (or moral) justification of coercive political power. Henceforth, all uses of ‘legitimacy’ refer to the normative sense rather than to de facto legitimacy.

Full legitimacy requires more than just de facto rule, but instead requires normative justification for political decisions and the creation of obligations for the citizens under the authority's rule. Some accounts hold that legitimacy merely entails a government or political actor's right to rule with no corresponding duty to obey for individual citizens (see Applbaum 2010). Peter does not attempt to settle the issue of whether legitimacy entails duties or obligations for citizens though she notes that the view that legitimacy does imply duties to obey is the ‘mainstream view’ (2020: 373). I will not attempt to settle the issue, either, but I will proceed with the mainstream assumption that legitimacy involves the creation of political obligations or duties to obey for the subjects of a political authority.

In virtue of what can the exercise of political power obligate subjects? Peter proposes a taxonomy of (at least) three theories regarding the grounding of political legitimacy: (1) the fact-based view, which holds that legitimate decisions are grounded by normative facts about what ought to be done; (2) the belief-based view, which holds that legitimate decisions are grounded by expert beliefs about the normative fact of what ought to be done; and (3) the will-based view, which holds that the wills of citizens ground legitimacy and that legitimate decisions are those that correctly adjudicate the wills of citizens. Peter ultimately defends a hybrid disjunctive account, according to which priority is given to the sufficiently justified beliefs about normative facts held by an expert ‘normative authority’. On this disjunctive account, legitimacy is grounded in the wills of citizens in cases when we cannot identify a normative authority.

1.1 The Fact-Based Conception

In Peter's taxonomy, the fact-based conception of political legitimacy is such that legitimacy is grounded in attitude-independent facts about what should be done. The idea behind attitude-independence is that there are normative truths that stand apart from any actual or hypothetical perspective, and those truths govern what it is that we have most reason to do. On this view, when political decision-makers make the decisions that are warranted by such facts, those decisions are legitimate. Peter groups this conception alongside views in metaethics and epistemology that see our actions as justified when they are warranted by normative facts, often construed as facts that relate to our normative practical reasons. Following Peter, we can refer to such views as ‘factualism’.

One objection that Peter levels against the factualist conception of legitimacy is that it fails to satisfy an important constraint on legitimacy as a normative property, which she calls the access constraint. The access constraint ‘implies that at least some citizens, minimally one, must have access to the grounds of legitimacy’ (Peter 2020: 377). Peter illustrates the need for such a constraint with the following example:

Imagine a scenario in which Parliament haphazardly made a series of decisions that concluded in the United Kingdom's revoking Article 50 and committing the country to staying in the European Union. Suppose this was met with widespread protest from citizens who in 2016 voted to leave. But suppose also that this decision would have been one warranted by the normative facts. The decision would then have been legitimate, on a fact-based conception, even though nobody was in a position at the time to authoritatively judge that it was, in fact, the right decision and even though there was widespread uncertainty and confusion about the legitimacy of this decision in the government and Parliament and among the citizens. (2020: 375)

According to factualist views of the attitude-independent sort, it is entirely possible that there are normative facts about what should be done—such as a normative fact that says the UK should remain in the EU—that are wholly inaccessible to any citizens. We can easily go wrong and fail to grasp the normative truth, and we can perhaps act according to what is warranted by the facts only accidentally, as in Peter's example. What seems to go wrong in this case is that a purportedly legitimate decision is made despite the fact that it is impossible for the relevant set of agents to settle whether the decision is politically legitimate. Yet, when we set ourselves to form legitimate political decisions, we see our task as one of settling what to do.

It is possible that citizens and decision-makers are collectively ignorant of important empirical facts that bear on an attitude-independent normative truth about what should be done—for example, they may be confused about what the ramifications of a political decision like Brexit will truly be. They may not be capable of arriving at the correct judgment that corresponds to the attitude-independent normative facts. Nonetheless, it seems that if no one is capable of authoritatively arriving at some judgment, then that judgment is not the kind we are in search of when we are seeking the justification for a political decision. Political decisions should not be arrived at haphazardly. More plausibly, they should be arrived at through the careful examination of argument and evidence or through trustworthy procedures. This is because legitimacy is a property that serves an important function in political life that Peter refers to as the ‘settling function’.

Peter argues—correctly, in my view—that attitude-independent normative facts ‘are the wrong ground of political legitimacy because the property of political legitimacy must be such that it can settle political deliberation . . . and, in order to fulfill the settling function, the grounds of political legitimacy must involve the attitudes’ (2020: 376). If the legitimacy of political decisions is grounded in attitude-independent facts that are inaccessible to citizens, then it is hard to see how legitimacy can serve to settle political deliberation and decision-making. According to the access constraint, we—or at least one of us—must be able to learn something about the grounds of a legitimate political decision. Otherwise, those grounds could not be the kind of thing that can settle political deliberation. It is conceptually possible that there are some (attitude-independent) normative facts about which nobody could learn, and thus those normative facts could not serve as the grounds for the property of political legitimacy.

Peter concludes that this means that the grounds of political legitimacy must involve at least some citizens’ attitudes because these are accessible. Normative facts may still play an important role in settling political decision-making, but they will do so through the beliefs that citizens have about them. Peter fills out the rest of her taxonomy with two conceptions that can meet the accessibility constraint because they are grounded in the attitudes of agents (or at least one agent). These are the belief-based and will-based conceptions.

1.2 The Belief-Based Conception

According to belief-based conceptions of political legitimacy, sufficiently justified beliefs about what should be done are the ground of political legitimacy. In Peter's taxonomy, both factualist and belief-based conceptions hold that there are attitude-independent normative facts about what should be done. The conceptions differ with respect to the role they assign to normative facts. For the factualists, the normative facts do the grounding work themselves. But in a belief-based conception, it is sufficiently justified beliefs about those facts that do the grounding work—political decisions are legitimate in virtue of responding to our best beliefs about what should be done. Belief-based conceptions have the advantage that they can satisfy the access constraint. While we may not have access to the normative facts themselves, we have access to our beliefs and the justifications for our beliefs about those facts.

Peter cites Joseph Raz's (1986, 2006) service conception of legitimacy as an example of a belief-based conception. According to Raz's conception, the decisions of political authorities are legitimate when they serve the reasons that governed individuals already have independent of the authority. Raz calls this condition on legitimate authority the normal justification condition (or sometimes the normal justification thesis; Raz 1986, 2006). The normal justification condition requires that political decisions are such that the governed are enabled to act according to their normative reasons better than they would if they had simply relied on their own judgments. Raz's conception construes legitimacy as a property of political decisions that provides us with an important service, helping us to act in the ways that we should but without needing to do all the work of sorting out how to achieve all those actions ourselves. The normative facts about our normative reasons are crucially important on this account, but as Peter interprets Raz, the facts themselves do not do the grounding work. Rather, the grounds are the sufficiently justified beliefs of political authorities regarding which political decisions will satisfy the normal justification condition.

The idea that it is beliefs that ground the legitimacy of political decisions raises the question of who among us has beliefs that can serve the grounding role, especially in cases where there is disagreement. Peter posits the notion of ‘normative authority’—’the entitlement to make binding decisions that is afforded by sufficiently justified beliefs about what should be done’ (2020: 382). The normal justification condition on legitimacy is satisfied when a decision-making body either has normative authority or the decision-making body is prepared to defer to it. Legitimate decisions thus are those that are guided by sufficiently justified beliefs about what should be done. But the notion of a normative authority (and thus belief-based conceptions of legitimacy that rely on such authority) is vulnerable to what Peter calls the epistemic underdetermination objection. This objection is that it is often not possible to establish who has normative authority because it is difficult to form sufficiently justified beliefs about what should be done and to identify who holds those beliefs.

Peter accepts belief-based conceptions of political legitimacy as correct but grants that the epistemic underdetermination objection is a serious one—’The epistemic circumstances of politics are such that for most political decisions, there will not be a decisive normative authority’ (2020: 385). She writes that legitimacy cannot rest on the grounds of belief in cases where there is no decisive normative authority (2020: 385). Peter thus puts forth a hybrid disjunctive conception of political legitimacy, according to which sufficiently justified beliefs are the ground of legitimacy when there is a decisive normative authority; when we have no decisive normative authority, then political decisions are legitimate in virtue of how well they adjudicate the (sometimes conflicting) wills of citizens. Peter assigns priority to the belief-based conception, but supplements that conception with what she calls a will-based conception, which is the third major type of conception in her taxonomy.

1.3 The Will-Based Conception

According to will-based conceptions, political legitimacy is grounded in the will, or the actual or hypothetical consent, of citizens. Political decisions are legitimate in virtue of how well they adjudicate between the sometimes conflicting wills of citizens. Peter cites Rawls's (1993) liberal principle of legitimacy and Rawls-inspired public-reason conceptions of legitimacy as examples of will-based conceptions. Public-reason conceptions of legitimacy are such that public reasons—procedural or substantive reasons that are accessible to and endorsable by citizens—justify and serve as the ground of legitimacy. Other conceptions that count as will-based in Peter's taxonomy include unanimity conceptions that hold that political decisions are legitimate if they have unanimous (actual or hypothetical) consent. Another example is participation conceptions, according to which a political decision is legitimate if the process of arriving at the decision allows for equal participation and the ability of all citizens to express their will. What many of these conceptions have in common is that they accept something like Rawls's (1993) conception of persons as the self-originating sources of valid claims. Will-based conceptions include those that accept popular sovereignty, seeing the will of governed individuals as the source of legitimate political power.

Peter rejects the idea that persons are the self-originating source of valid claims, at least for a broad range of claims, and argues that will-based conceptions thus fall prey to what she calls the arbitrariness objection. The arbitrariness objection against will-based conceptions of political legitimacy states that these conceptions ‘support undue arbitrariness in political decision making’ (Peter 2020: 380). According to this objection, the claims we make from the first-personal perspective can go against morally weighty beliefs that are sufficiently justified from a third-personal perspective. If such (third-personally) sufficiently justified beliefs are held by a normative authority, Peter argues they must not be ignored in the political decision-making process.

For example, a normative authority may form a sufficiently justified belief that children should not be separated from their parents at the United States border. (In all likelihood, this is a highly intuitive belief with which readers agree.) It is a belief about the normative landscape that we find sufficiently justified even if it is not ratified by some citizens’ first-personal perspectives. Peter argues that we must defer to such beliefs in order to get political results that are legitimate, but the will-based conception apparently cannot explain this. The will-based conception has it that all persons are the source of valid claims, even those whose beliefs are contrary to the sufficiently justified beliefs held by a normative authority. But, surely, no matter the arbitrary beliefs that are held by some citizens, separating parents from their children at the border is not a legitimate political decision.

The will-based conception seems to have a serious defect, at least with respect to contexts wherein a normative authority can form a decisive normative belief. Thus the will-based conception cannot tell us the whole story of legitimacy—at least sometimes sufficiently justified beliefs must serve as the ground of legitimacy rather than the arbitrary wills of citizens. So Peter's argument goes. Owing to the arbitrariness objection Peter assigns priority to belief-based conceptions that allegedly do not suffer the same problem (although we have seen that belief-based conceptions suffer from a separate troubling objection). This leads to Peter's hybrid disjunctive account. In cases where an identifiable normative authority forms a decisive sufficiently justified belief, then that belief grounds political legitimacy. In cases where there is no identifiable normative authority that forms a decisive sufficiently justified belief, then political decision-makers must defer to the wills of citizens.

I promise to return to the arbitrariness objection and offer a response to it. First, I want to explain why the metanormative picture is even more complicated than Peter's taxonomy of conceptions of the grounds of political legitimacy shows.

2. Attitude-dependent vs. Attitude-independent Normative Truth

Peter's notion of a fact-based conception of the grounds of legitimacy assumes that normative facts are attitude-independent in nature. Her notion of a belief-based conception, too, assumes that normative beliefs about what should be done are aimed at attitude-independent facts. In this section, I argue that we must distinguish these conceptions from similar ones that assume attitude-dependence about normative facts. In order to help clarify how these frameworks regarding the nature of normative facts (or normative truth) differ, I compare how each deals with puzzles raised by so-called ideally coherent eccentrics. My aim is to show that the nature of normative truth has immense importance for any framework that seeks to describe the metanormative ground of a property such as legitimacy, a point that complicates the taxonomy offered by Peter, which presents the fact-based, belief-based, and will-based conceptions as distinct. The nature of normative truth has important implications for the aim of the belief of a normative authority in the belief-based conception. Further, if the nature of normative truth is attitude-dependent, then a factualist conception will turn out to collapse into a will-based conception. These distinctions have important implications for how we think about normative disagreement in politics.

While the question of the metanormative grounds of legitimacy has rarely been discussed in contemporary political philosophy, there are long-standing disputes about the nature of normative truth, normative reasons, and the grounds of normativity in the metaethics literature. One prominent view about the nature of normative truth holds that these truths are attitude-independent. As Shafer-Landau describes this view: ‘There are moral truths that obtain independently of any preferred perspective, in the sense that the moral standards that fix the moral facts are not made true by virtue of their ratification from within any actual or hypothetical perspective’ (Shafer-Landau 2003: 15, emphasis in original; also see Parfit 2011 and Enoch 2011). This view is typically associated with robust realism in metaethics or the view that there are objective and irreducibly normative truths that stand apart from the perspectives of (real or hypothetical) agents. Robust realism captures an intuitive thought about morality—that moral truths and moral reasons have authority over us whether we like it or not. When we utter moral judgments, such as ‘murder is wrong’ or ‘it is wrong to separate children from their parents at the border’, we often mean to say that these things are wrong no matter how it seems from the subjective standpoint of any given actual or hypothetical agent. The truth is objective, and the reasons not to murder or not to separate a child from her parents at the border apply to us whether or not we agree. This is the attitude-independent conception of normative truth that operates in the fact-based and belief-based conceptions of legitimacy as they have been articulated by Peter.

This view stands in contrast to the view that normative facts are attitude-dependent. Street explains this view as follows:

There are no facts about how an agent has most normative reason to live that hold independently of that agent's evaluative attitudes and what follows from within the standpoint constituted by them; instead, an agent's normative reasons are always ultimately a function of that agent's own evaluative attitudes and what is logically or instrumentally entailed by those attitudes in combination with the non-normative facts. (Street 2009: 274; also see Korsgaard 1996 and Smith 2015)

Attitude-dependence is accepted in constructivist metaethical frameworks, which hold that there are no normative facts that stand apart from the perspective of (real or hypothetical) rational agents or facts derived from rational agency itself. Normativity is not something that exists apart from rational agents that we can discover; rather, normativity is something constructed by agents as the self-originating sources of valid claims. The constructivist framework sets out to vindicate the seemingly objective nature of morality and reconcile it with the intuitive thought that morality is binding over us in a way that is not independent of our practical reason or our rational nature. For the constructivist, normative facts like ‘murder is wrong’ do not exist independently of us; rather, they spring directly from our nature as rational agents in a way that we cannot escape or simply choose not to care about. For the constructivist, features of our nature as rational agents are the source of normativity.

These two perspectives on the nature of normative truths correlate with two competing views about the nature of normative reasons. Bernard Williams pointed out that there are two ways that we may interpret the notion of a reason, coining the terms ‘internal reason’ and ‘external reason’ (Williams 1979). Those who accept that normative truths are attitude-dependent tend to accept the internal reasons thesis, which holds that A has a reason to φ if and only if A has some desire that will be served by φ-ing. On this view, all reasons are relative to an agent's subjective motivational set, which includes the agent's goals, projects, dispositions, and commitments. On this view, reasons are not subjective in a totally simplistic way—we can be wrong about the reasons that follow from our subjective motivational set, perhaps owing to a rational failure or false belief. Further, we may have internal reasons to φ that we do not yet grasp—reasons that we only become aware of through critical reflection or rational deliberation. But according to the internal reasons thesis, we have no reason to φ if there is no sound deliberative route from our subjective motivational sets to the conclusion that we ought to φ. In contrast, those who accept that normative truths are attitude-independent tend to accept the existence of ‘external reasons’, which are moral reasons that apply to agents even if those reasons do not serve any of the agents’ desires. According to this view, A can have a reason to φ even if A has no desire that will be served by φ-ing (nor a sound deliberative route from their subjective motivational set that could lead to the formation of such a desire).

It is instructive to think through how these different perspectives about the status of normative truth and correlative theories of reasons apply in hard cases because these cases help clarify how the perspectives differ. I have in mind the hard cases of so-called ideally coherent eccentrics (Street 2009), and here I focus on the case of the ‘ideally coherent Caligula’ who ‘aims solely to maximize the suffering of others’ (Gibbard 1999: 145). We do not know much about the actual moral psychology of Caligula, the third Roman emperor, but he appears in the literature as a hypothetical case of someone who is seemingly amoral and systematically engages in sadistic, cruel, and tortuous actions. We imagine for the sake of argument that such an agent aims at maximizing the suffering of others without any rational failure or desires to the contrary. We imagine that Caligula's subjective motivational set is constituted such that there is no sound deliberative route from his desires, values, and commitments to the conclusion that he ought to refrain from torturing innocents.

The robust realist, accepting attitude-independence about normativity, will tell us that even the ideally coherent Caligula has a reason to refrain from maximizing the suffering of others. Although Caligula has no internal reason to refrain, there is an objective external reason that applies to him (and to all of us), and we can appeal to that external reason to explain why Caligula's cruel actions are wrong. It does not matter that rational deliberation could never convince Caligula to believe in the reason or be motivated to act on the reason.

According to the internal reasons thesis, we cannot say that an ideally coherent Caligula has a reason to refrain from maximizing the suffering of others because we have stipulated that nothing in Caligula's subjective motivational set could lead Caligula to rationally conclude that he ought to refrain. Among constructivists, there are at least two competing views that lead to different analyses of the notion of an ideally coherent Caligula. So-called Humean constructivists, like Street, will tell us that it is conceptually possible that there exists a rational agent like Caligula who desires solely to maximize the suffering of others, and if it is true that there is no deliberative route from his subjective motivational set to the conclusion that he ought to refrain from torturing innocents, then normatively speaking he has no reason to refrain. The Humean constructivist might still think that as a matter of fact there are no (or not many) agents like Caligula in our society—most of us do have desires, commitments, and values that are rationally inconsistent with Caligula's actions—but the existence of such agents remains a conceptual possibility (Street 2009).

In contrast, many other constructivist theories, such as the Kantian constructivism defended by Korsgaard (1996, 2009), hold that there are constitutive features of rational human agency that will rule out the conceptual possibility of the ideally coherent Caligula. Thus, in constructivism and among those who accept the internal reasons thesis, we find at least two separate verdicts on the notion of an ideally coherent Caligula. The Humean will accept that an ideally coherent Caligula is conceptually possible and that he has no reason to refrain from his cruelty. The Kantian will reject the conceptual possibility of the ideally coherent Caligula, insisting that rational agents all have sound deliberative routes from their subjective motivational sets to certain substantive conclusions about moral action, including that we have no normative reason to maximize the suffering of others. In virtue of our human nature and the features of our rational agency, we all have internal reasons to refrain from such cruelty.

These frameworks for thinking about the nature of normative truth are important for the fact-based view that holds that legitimacy is grounded in facts about what should be done because each framework suggests different sources for those facts themselves. Theorists who accept attitude-independence see these facts as somehow originating outside the practical standpoints of agents, while those who accept attitude-dependence see agents themselves or their rational features as the source of normative facts. Thus, in addition to Peter's notion of a fact-based view that assumes attitude-independence about normative facts, we can find in conceptual space a type of fact-based view that says this: Legitimacy is grounded in facts about what should be done, and those facts depend on the attitudes, reasons, or subjective motivational sets of the agents to whom they apply. If normative truth is attitude-dependent, then a factualist conception of political legitimacy will collapse into a will-based conception. The wills of citizens are the ground of the normative fact about what should be done, and those attitude-dependent facts are the ones that ground the legitimacy of political decisions.

The distinction between attitude-independence and attitude-dependence about normative truth has implications, too, for a belief-based conception of the grounds of legitimacy. While Peter assumes that the beliefs of a normative authority are aimed at attitude-independent normative truth, we can find in conceptual space a belief-based account according to which a normative authority's beliefs are aimed at attitude-dependent normative truths. Rather than seeking third personal justification, on this further conception, a normative authority in search of facts about what should be done must aim to form sufficiently justified beliefs about facts that spring from the first personal perspectives of citizens (who are the self-originating sources of valid claims).

We can thus revise the taxonomy of conceptions of the grounds of political legitimacy to include the following:

Attitude-independent fact-based conceptions: Political legitimacy is grounded in attitude-independent normative facts, and political decisions are legitimate in virtue of being warranted by those facts.

Attitude-dependent fact-based conceptions: Political legitimacy is grounded in attitude-dependent normative facts (sourced in persons as the self-originating sources of claims), and political decisions are legitimate in virtue of being warranted by those facts.

Attitude-independent belief-based conceptions: Political legitimacy is grounded in sufficiently justified beliefs about attitude-independent normative facts, and political decisions are legitimate in virtue of being responsive to our best beliefs about those facts.

Attitude-dependent belief-based conceptions: Political legitimacy is grounded in sufficiently justified beliefs about attitude-dependent normative facts (sourced in persons as the self-originating sources of valid claims), and political decisions are legitimate in virtue of how well they adjudicate between our best beliefs about those facts.

We may add to this taxonomy a conception that leaves out normative facts entirely—the facts themselves or our beliefs about them, attitude-independent or attitude-dependent. It is worth noting that this is plausibly what John Rawls (1985) has in mind for his principle of liberal legitimacy given his claim that his conception of justice as fairness is ‘political not metaphysical’, but for now I set this possibility aside in order to focus on the above conceptions. Although Peter presents will-based conceptions as viewing the grounds of legitimacy as the adjudication of the wills of citizens, rather than the normative facts or beliefs about normative facts, we can understand the reasons that come from our wills as constituting the normative facts. These are facts that are constructed through human agency rather than facts that exist independently of us. Thus we are left with two varieties of fact-based conceptions and two varieties of belief-based conceptions. The varieties differ with respect to the nature of normative truth.

3. In Defense of the Will as the Source of Legitimacy

Finally, I would like to give a response to Peter's arbitrariness objection to the will-based account. I argue that the purported arbitrariness of the will-based account does not threaten the notion that the wills of citizens may ground the legitimacy of political decisions. If it is true that normative truth is attitude-dependent, as many theorists contend, then attitude-dependent normative facts that come from the wills of citizens shall figure into the accounts that were supposed to compete with the will-based account—the fact-based and belief-based accounts. The theorists who accept attitude-independence about normative facts may have it right, but in case they do not, it is important to see that we can still arrive at legitimate political decisions that do not strike us as counterintuitive.

One such counterintuitive result that purportedly follows from a will-based conception includes Peter's suggestion that such conceptions fail to ratify the intuitive claim that the decision to separate parents from their children at the border is illegitimate. Peter also gives the example of a citizen who endorses the claim “my child should not be vaccinated” ‘in relation to a highly contagious infectious disease and a harmless vaccine’ (2020: 380). Peter gives these examples of purportedly incorrect judgments about what should be done that yet follow from the wills of some citizens, illustrating that there are some valid claims in politics that come from a third-personal perspective rather than from the first-personal perspective of citizens.

Peter considers a reply to her objection that says that in the will-based conception some decisions are illegitimate—such as the decision to separate parents from their children—because ‘they fail to treat at least some people as self-originating sources of valid claims, or . . . they are not the kind of decisions that rational or reasonable persons would support’ (2020: 380). One version of this reply denies that there are third-personal sources for the validity of claims about what should be done but regards certain decisions as illegitimate because they do not treat all persons as self-originating sources of claims. The reply insists that any reasonable will shall accept the constraints that other persons as sources of valid claims impose on them (see Rawls 1993). Decisions that tolerate or promote cruelty toward others are not reasonable and thus not legitimate. This sense of ‘reasonable’ is substantive: it requires not just that a person is rational in the sense of abiding by means-end coherence and consistency requirements, but it further requires that a person is committed to the respect due to others. Those who are committed to the respect due to others as the self-originating sources of claims would not support separating parents from their children at the border because they would respect the claims and important interests of those parents and children.

Peter's response states that the reply is of the wrong sort:

It might be able to generate the right answer—that such decisions are illegitimate—but for the wrong reasons. What makes such decisions illegitimate is not that reasonable wills will not will it, it is that they go against the correct judgment that those decisions are not what should be done. And we cannot rule out in other cases, the reply would give the wrong answer: what reasonable wills will need not be what they should will. (2020: 380)

I worry here that Peter begs the question with her insistence that what makes decisions illegitimate is not that reasonable wills will not will it, an insistence that perhaps owes to her assumption that what counts as a normative fact about what should be done is attitude-independent. But her response is correctly sensitive to the potential for a will that is constructed so as to will things that are counterintuitive or eccentric, a conceptual possibility highlighted by the Humean interpretation of internal reasons. On a Kantian conception of what counts as a correct (or true or factual) judgment about what should be done, the substantive reasonableness constraint will hold. But on other conceptions that see persons as the self-originating sources of claims, an ideally coherent and rational person need not be reasonable in this sense. She may be the source of self-originating claims herself, yet not contain in her subjective motivational set a commitment to the respect due to others as the self-originating sources of claims.

In response to Peter, we should consider what such an agent would be like and whether these sorts of agents are among our fellow citizens—is there a real threat to our intuitive thought that certain decisions are illegitimate no matter what some citizens think? Taking into consideration the internal reasons thesis, consider what must be true about an agent's subjective motivational set in order to land on the view that a citizen really has no reason to vaccinate her child or no reason to desire that parents and children not be separated at the border. This agent must not have anything in her subjective motivational set—commitments, desires, projects, cares—that could lead her to reject these reasons. It must not be possible that she would revise her view if she were to reason about these matters without any rational failure or inconsistency. She must not be mistaken about any of the non-normative facts that bear on these reasons. In the case of separating parents from children, she must lack the basic capacity for empathy that leads us to see such actions as abhorrent. But, in general, we do not think that it is true that antivaxxers or individuals who praise the separation of parents and children are perfectly rational, well-informed, or appropriately sensitive to their own capacity for empathy.

Take the example of antivaxxers. Antivaxxers are frequently mistaken about relevant non-normative facts that bear on their reasons to get vaccinated or vaccinate their children. These include the facts that vaccines are indeed harmless and healthful. Or we may think that antivaxxers have failed to reason about the issue correctly—they have failed to realize that despite whatever commitment motivates their antivax stance, they have a deeper commitment to their health and the health of their community. In order to conclude that the will of an antivaxxer truly is such that she has no reason for vaccinating her child that can be ratified by her own first personal perspective, we must assume that she is not misinformed about the non-normative facts in a way that has led her astray. We must assume that she has no deep commitment to her health or the health of her child. As with the ideally coherent Caligula, we might agree along with the Humeans that there is a conceptual possibility that such an agent might exist. But she would be an eccentric agent indeed, and she would be constituted quite differently than we generally suppose our fellow citizens are constituted. We tend to think that citizens who oppose important health measures are mistaken, that they are not fully informed and not fully rational. Further, on the Kantian picture, things look even more optimistic. We can rule out that there are such conceptually possible citizens.

Peter claims the will-based conception suffers a defect in that it sees persons as the self-originating sources of valid claims because some valid claims must be justified from a third-personal perspective. But when we think carefully about what it would take for those valid claims—such as the claim that we should vaccinate—to fail to be ratified from a first-personal perspective, we should find that the cases are eccentric or even conceptually impossible. When we think about the agents among us who in fact reject these claims, we find that our complaint toward those agents is usually not that they have failed to grasp a third-personal justification, but rather that they have failed to see how on their own terms they should conclude in favor of the claim. We may worry that they have been fooled by or manipulated by antivaxxer propaganda online or that they have not been adequately educated on the benefits of vaccines or that they have misplaced their trust such that they do not trust the opinions of scientific experts. Some of these agents may never have the opportunity to learn of the non-normative facts that lead to their mistaken belief, or they may never be afforded the time to introspect rationally on whether the mistaken belief is indeed consistent with the contents of their subjective motivational sets. However, we can still maintain that antivaxxers have an internal reason to vaccinate so long as they could with full information and perfect rationality revise that belief. If, to the contrary, antivaxxers are fully educated, are not misled about any non-normative fact, and have not suffered a rational failure, then we perhaps have a new worry to consider with respect to political decisions that coerce them to get vaccinated. For example, if we are following Raz's (1986, 2006) conception of legitimacy as serving the reasons citizens already have independent of the authority, we must ask how we are serving the reasons of citizens like this.

Conclusion

In this article, I did not aim to settle the dispute between a factualist and a belief-based conception of legitimacy—the dispute about whether it is normative facts or our beliefs about those facts in virtue of which a political decision has the property of legitimacy. But I have argued that the wills of citizens may have an important role to play in either of these grounding stories just in case normative truth is indeed attitude-dependent, as metaethical constructivists contend. We may worry that attitude-dependence about normative truth leaves us with a bleak picture of political legitimacy given Peter's arbitrariness objection against the will-based conceptions. I hope for my argument to have at least partially dispelled such worries, leaving open that the wills of citizens (or our beliefs about the facts that spring from our wills) serve as the grounds of legitimacy.