2. Framing the debate

It is the endorsement of a requirement of public justification that most clearly demarcates public reason liberalism from other versions of liberalism. While the precise demands of a requirement of public justification will vary from one public reason liberal to another, a generic requirement of public justification roughly states that a coercive law is justified only when each individual subject to the law has sufficient reason to agree to the law.Footnote ⁴

The debate between consensus and convergence theories of public reason is a debate about what sorts of reasons can enter into the justificatory domain. Consensus models of public reason endorse a public-reasons-only (PRO) requirement, while convergence models of public reason reject the PRO requirement. Public reasons are often described as ‘shared’ reasons: they are not grounded in one’s personal comprehensive moral, philosophical, or religious conceptions, but are instead acceptable to all no matter what personal comprehensive doctrines one endorses.Footnote ⁵ I will refer to reasons based in one’s personal comprehensive doctrines as non-public reasons. These reasons are not necessarily private, since others may in fact accept them. Yet, non-public reasons do not transcend all comprehensive doctrines, hence they should not be considered public. To adopt the PRO requirement – as consensus theorists do – is to hold that only public reasons may ground an individual’s acceptance or endorsement of some proposed coercive law.

Importantly, the convergence model does not require that for a law to be publicly justified that all citizens must agree to the law on the basis of non-public reasons. To the contrary, convergence models merely allow for citizens to agree to or endorse laws on the basis of non-public reasons, but they may certainly also agree to or endorse them on the basis of public reasons.

I have provided here only a coarse-grained distinction between consensus and convergence models of public reason. As I have remarked, there are some other differences, but they are not crucial to drawing the distinction at hand. For example, consensus theorists adopt an accessibility requirement along with the PRO requirement while convergence theorists have thus far rejected both. But, as I will argue below, while there is good reason to abandon the PRO restriction, convergence theorists should retain the accessibility requirement.

3. Against the public-reasons-only (PRO) restriction

If the debate between consensus and convergence theorists is primarily a debate about whether public reason liberalism should incorporate a PRO restriction, then one needs to take a stance on this issue prior to developing one’s more nuanced and particular version of either the consensus or convergence model. Since Jonathan Quong has recently constructed what is arguably the most sustained and complete defense of the consensus model, I will take as my focus his argument in favor of the PRO restriction. Given that the PRO restriction supplements a requirement of public justification, my aim is to defend the convergence model by arguing that Quong’s case for such supplementation is unconvincing.

Quong argues that public reason liberalism must incorporate a sincerity requirement. The sincerity requirement dictates that ‘[w]e should sincerely think that our view of the [political] matter is based on political values everyone can be reasonably expected to endorse’ (Rawls Reference Rawls1996, 241). Quong understands this to mean that we cannot ‘merely aim at getting others to assent to our proposal, we must sincerely believe our proposal can be justified to them’ (Reference Quong2011, 265). This prompts Quong to construct the following principle of sincerity:

• (1) A reasonably believes he is justified in endorsing X.

• (2) A reasonably believes that B is justified in endorsing X

• Furthermore, following Rawls’s duty of civility,

• (3) A may only (in the political domain) offer arguments in favour of X to B that he reasonably believes B would be justified in accepting. (266)

Quong asserts that PJS ‘undermines the convergence view of public reason’ (Reference Quong2011, 266). His critical move is to argue that the only way we can sincerely believe others are justified in endorsing our proposals is if those proposals are grounded solely in public reasons. As such, it is condition two in particular that summons the PRO restriction (266–267). The general assertion, then, is that public reason liberalism must incorporate a PRO restriction because it must incorporate a sincerity requirement. Given this, two strategies are immediately available to dissenters. First, one might argue that a sincerity requirement is not a requirement of public reason liberalism, thus a PRO restriction cannot be appealed to on such grounds. Second, one might argue that even if some form of a sincerity requirement is necessary, such a requirement does not call for a PRO restriction. It is this second approach that I will adopt.

Quong maintains that the sincerity requirement has two important functions. First, it ‘helps to distinguish public reason from rhetoric and manipulation.’ Second, ‘it is also important due to the role it plays in sustaining… the value of civic friendship or respect.’ For A to be in a state of civic friendship with B is for A and B to respect ‘each other as free and equal citizens’ (Quong Reference Quong2011, 265). On Quong’s account, then, a principle of sincerity – a principle that is satisfied only when we sincerely believe that others can accept the (public) reasons we offer in favor of our preferred positions – is necessary to respecting each other as free and equal citizens and, hence, is necessary for civic friendship (265–266).

I maintain that Quong’s position is too strong. Offering public reasons is not a necessary condition of respecting each other as free and equal citizens. While it may be true that satisfying Quong’s principle (the second condition in particular) – and hence only offering public reasons – will increase the probability that public deliberation furnishes agreement (call this this practical claim), Quong thinks the claim that ultimately undercuts the convergence model is a moral claim. He believes that offering non-public reasons constitutes a failure to respect one’s fellow citizens as free and equal. The idea is this: A fails to respect B as free and equal when A argues on behalf of p and it is the case that A is not in the epistemic position to sincerely believe that B can accept the reason(s) in favor of p that he or she offers to B. But, according to Quong, A cannot be in the epistemic position to sincerely believe that B can accept the reason(s) in favor of p unless those reasons are public (Quong Reference Quong2011, 267). Quong, however, does not adequately explain how a failure to be in such an epistemic position has the significant moral consequences that he attributes to it. A is not coercing B. A is not even demanding that B act in accordance with A’s proposal. Rather, A, as a free and equal being herself, is merely presenting an argument for why she thinks a proposal should be accepted. In other words, she is simply expressing the very considerations that shape her own agency. It remains unclear why such an act should be construed as an instance of A failing to respect B as a free and equal citizen.Footnote ⁶

On the face of it, it appears that consensus theorists have traditionally run together two different political practices that should be carefully delineated: (1) the practice of A proposing a policy to B; and (2) the practice of A demanding that B adhere to some policy. As Nicholas Wolterstorff has noted, the public reason liberal may think that ‘to advocate for some proposed piece of legislation is thus, implicitly if not explicitly, to place a moral demand on one’s fellow citizens’ (Reference Wolterstorff and Cuneo2012, 60). If the consensus theorist does intend to deny that there is any morally-relevant difference between proposing a policy and demanding that others adhere to it, then we need a less-puzzling account of how A’s proposing a policy to B amounts to A’s demanding that B adhere to that policy.

What is perhaps morally problematic about a case of one person, Ari, demanding that another, Sky, adhere to some policy (without consideration for whether Sky has a sufficient reason, even counterfactually, to agree to that policy) is that Ari does not care that Sky is, to borrow a phrase from Thomas Scanlon, ‘a locus of reasons’ (Reference Scanlon1998, 105) – he merely wants to claim authority over her agency. This, then, is an instance in which it might plausibly be argued that Ari fails to respect Sky as a free and equal citizen.Footnote ⁷ But no such degree of disrespect occurs when Ari merely argues on behalf of a policy – i.e. when he elucidates the reason(s) he has to endorse some particular policy.Footnote ⁸ To propose or advocate for a policy – even one in which others do not have sufficient reason to agree to – is neither to impinge on their free and equal status nor to fail to recognize them as a locus of reasons. When Ari advocates on behalf of a policy, he publicizes what he takes to be a worthwhile argument for the policy. Not only does this not amount to any sort of obstruction of agency, but it has the virtue of bringing into the public Ari’s genuine (we suppose) feelings about the matter. As Brian Carey argues, the admission of non-public reason into public debate may actually increase participation in public deliberation and ‘yield more stable agreements’ (Carey Reference Carey2018, 52).Footnote ⁹

It should be emphasized that public reason liberals, in recognizing citizens to be free and equal, argue that citizens have defeater power. Citizens are free insofar as they are not naturally subject to the authority of others; and they are equal to one another with regard to their free status (Quong Reference Quong2013; Gaus Reference Gaus, Ellen Frankel and Paul2007, 90). As such, if a citizen does not have sufficient reason – even counterfactually – to endorse some proposal, then absent the citizen’s consent, the proposal is defeated.Footnote ¹⁰ In light of this, when I propose reasons, even those grounded in personal comprehensive doctrines, I am not engaged in any sort of moral imposition. The structure of a requirement of public justification protects against such an imposition. In acknowledging that each of my fellow citizens is free and equal and a locus of reasons, I recognize that they have defeater power. If this is right, then it makes sense to argue from personal comprehensive doctrines when one is unsure about how widespread his or her comprehensive views are. After all, sharing one’s arguments may benefit the political process when either (1) those sharing similar comprehensive views realize that they have good reasons to support a particular proposal; or (2) one’s arguments based on comprehensive views are shown to be poorly constructed.Footnote ¹¹

Still, the consensus theorist’s emphasis on sincerity has its place. Let us suppose that Ari is advocating for p. Suppose further that, when confronted with Sky’s skepticism, Ari argues that Sky’s existing beliefs, when coupled with one or more essential inferences, commits Sky to endorsing p. But now imagine that Sky’s beliefs are poorly-formed and the inferences that Sky would need to make are invalid inferences. Suppose further that Ari recognizes all of this, and thus recognizes that Sky does not have sufficient reason to endorse p, but nevertheless continues to press his line of reasoning to Sky. This sort of case captures the consensus theorist’s worries about manipulation. Ari is not treating Sky as a locus of reasons. Rather, Ari merely views Sky as someone to be persuaded by any means necessary. Although this case captures genuine normative concerns regarding the role of sincerity, cases of this type cannot ground the sort of robust sincerity requirement that Quong is after for they cannot ground a PRO restriction. Manipulation is not a practice that is uniquely (or even necessarily) tied to non-public reasons.Footnote ¹² As such, an anti-manipulation requirement does not call for a PRO restriction. Once more, then, the concern is that Quong has not adequately explained why one must be in the epistemic position of sincerely believing that others can accept R for p in order to avoid failing to treat them as free and equal. If we set aside instances of deliberate manipulation, my not being in such an epistemic state plausibly provides me with a reason to publicize my argument – not to withhold it.Footnote ¹³

While I do not deny that there may be a place in public reason liberalism for a requirement of sincerity, I claim that such a requirement is weaker than Quong contends. Importantly, such a requirement does not appear to implicate a PRO restriction. When arguing from comprehensive views, it is not evident that any moral harm is done so long as the relevant legislative proposal is not simply imposed on one’s fellow citizens and so long as one is not resorting to mere manipulation.

I am not alone in denying that an individual necessarily fails to respect his fellow citizens when he advocates for a coercive law that he cannot sincerely believe all others have sufficient reason to endorse.Footnote ¹⁴ Christopher Eberle argues that it is a mistake to ground a doctrine of religious restraint (essentially a narrow version of the PRO restriction) in the claim that failure to adhere to such a doctrine amounts to a failure to respect others (Reference Eberle2009, 152). According to Eberle, what is required of citizens and officials is that they adhere to the ‘ideal of conscientious engagement’ – i.e. they should sincerely attempt to construct persuasive arguments in support of those coercive laws which they advocate for (Reference Eberle2009, 167–168).Footnote ¹⁵ Eberle believes that one needs an overly ‘thick’ conception of what is required to show respect for others in order to claim that any citizen who is thoughtful and genuine in developing her arguments – even if those arguments rely solely on religious reasons – is guilty of failing to respect her fellow citizens (see Eberle’s case of the Agapic Pacifist: Reference Eberle2009, 152ff.). However, it does not appear that Eberle makes his case strictly in the context of a requirement of public justification. As such, he is not working under the assumption that each citizen has defeater power. In this respect, Eberle’s argument is much more ambitious than mine.Footnote ¹⁶ If one departs from a requirement of public justification (which incorporates a unanimity requirement) in favor of, for example, a simple majority rule, then new complications will indeed arise as a result of the rule’s unique characteristics.Footnote ¹⁷ Such complications may require reconsidering the case for a PRO requirement.

4. Distinguishing political practices

As explained in Section 3, a defense of a robust principle of sincerity may provoke consensus theorists to run together two different political practices: the practice of A demanding that B adhere to some policy and the practice of A proposing a policy to B. However, some public reasons liberals have also supposed that there is no morally-relevant difference between one agreeing to (or endorsing) a policy and one proposing (or advocating for) a policy. When this claim is united with the claim that there is no morally-relevant difference between proposing a policy and demanding that others adhere to that policy, it yields the corollary that there is no morally-relevant difference between my agreeing to a policy and my demanding that others adhere to that policy. The consequence is that some public reason liberals have been prone to treat the practice of agreeing to a policy as morally on par with the practice of demanding that others adhere to that policy. For Rawls, this conflation begins with the claim that ‘the ideal of public reason does hold for citizens when they engage in political advocacy in the public forum… [and] holds equally for how citizens are to vote in elections when constitutional essentials and matters of basic justice are at stake’ (Rawls Reference Rawls1996, 215). As Samuel Freeman has commented in discussing Rawls’ view:

On this view, to vote for a law that cannot be justified to others is to disrespect others. As such, the robust version of the principle of sincerity (and, hence, the PRO restriction) is said to hold even when merely selecting policies that will have jurisdiction over others. As Boettcher emphasizes, ‘liberal-democratic citizens also disrespect their compatriots when they attempt through voting or other advocacy in the public political forum to settle constitutional questions solely on the basis of religious doctrine’ (Reference Boettcher2012, 172–173). Our problem remains: it takes an exceedingly thick conception of what is required to show respect for others in order to claim that the act of registering one’s genuine preferences necessarily entails disrespecting one’s fellow citizens in some morally salient way.Footnote ¹⁸ It is true, as Rawls remarked, that ‘[w]e recognize that our own [comprehensive] doctrine has, and can have, for people generally, no special claim on them beyond their own view of its merits’ (Rawls Reference Rawls1996, 60). As such, perhaps I cannot demand that you accept p on the basis of my comprehensive beliefs when you do not have sufficient reason to accept those beliefs. But it does not follow from this that I cannot agree to p on the basis of such beliefs without disrespecting you.

When one morally conflates the political practices of agreeing, proposing, and demanding, it leads to some perplexing and unfortunate circumstances. For example, let us suppose that Sky agrees to p on the basis of comprehensive reason R_c. Let us suppose further that a vote is taken to determine whether p is publicly justified. According to the standard consensus approach, Sky fails to respect her fellow citizens as free and equal so long as she votes in favor of p solely (or even substantially) on the basis of R_c. But notice the unusual implication that follows. The consensus theorist must say to Sky: ‘I know you agree to p, but I cannot let that stand. You must agree to p for the reasons that I have deemed permissible.’ Assuming that p is in fact agreed to by all citizens, the public reason liberal undercuts the public justification of p by insisting that the PRO restriction applies to the political practice of agreeing to policies. If, as I have argued, the robust version of the principle of sincerity should be rejected in favor the weaker version, then the weaker version has no jurisdiction over the political practice of agreeing to policies. Instead, the principle of sincerity only applies in deliberative situations between two or more parties in which at least one party intentionally seeks to manipulate another.Footnote ¹⁹

5. Shared evaluative standards

The PRO restriction is not the only restriction that consensus theorists place on citizens. In addition to the PRO restriction, consensus theorists also endorse an accessibility requirement which dictates that for a reason to qualify as justificatory, it must satisfy at least some shared (agent-neutral) evaluative standards. Evaluative standards are what Rawls referred to as ‘guidelines of inquiry’ – the ‘principles of reasoning and rules of evidence…’ (Rawls Reference Rawls1996, 223–224).Footnote ²⁰ These are the principles and rules that one appeals to when determining (a) whether a reason supports a particular claim; and (b) whether the reason itself is well grounded.

Convergence theorists, such as Gerald Gaus and Kevin Vallier, reject both the accessibility requirement and the PRO restriction. However, a convergence theorist might endorse the accessibility requirement while rejecting the PRO restriction. Importantly, it is not the condition of requiring shared evaluative standards that limits justificatory reasons to public reasons. Some reasons may be public, but not satisfy the accessibility requirement. Alternatively, there may be non-public reasons that do satisfy the accessibility requirement. As such, the requirement of shared evaluative standards is independent of the PRO restriction.

But why think public reason liberalism needs an accessibility requirement? One worry is that if there are no shared standards of reasoning and evidence, then our judgments about whether a reason qualifies as sufficient will be drastically and adversely impacted. Let us consider an example. Suppose that society S seeks to institute a law that prohibits littering in public places. But let us also suppose that citizen C rejects the law on the basis that he believes that littering (1) tends to improve property values; (2) is good for the environment; and (3) never poses a danger to individuals (no matter the substance of the litter). The crucial question is this: does C have a sufficient reason to reject the law (alternatively, does C possess a defeater that undermines the claim that C has a sufficient reason to accept the law)?

In normal circumstances, we might reasonably expect that those engaging C would call on him to provide evidence in support of his claims. We might also expect that those engaging C would require him to appeal to accepted standards of evidence. As such, if C were to say that his claims were based on mere intuitions, then we should not be surprised if his fellow citizens found his claims to be poorly supported. Assuming a good argument has been furnished on behalf of the law, and assuming that C’s present claims are his only potential defeaters, then C’s fellow citizens may continue to insist that C has a sufficient reason to agree to the law. Importantly, it is not the case that C must in fact agree to the law, but rather that (moderately) idealized C would find reason(s) to agree. It is arguably the case that acceptable levels of idealization would lead actual C to recognize that his claims rely on overly-weak standards of evidence.

If one rejects the accessibility requirement, then this sort of response to C is unacceptable since to reject the accessibility requirement is to reject the claim that standards of reasoning and evidence have some agent-neutral basis.Footnote ²¹ But if such standards do not have an agent-neutral basis, then barring an explanation to the contrary, any criticism of C’s standards is unwarranted.Footnote ²² Consequently, one would have to say that C – even idealized C – does possess a defeater reason. If so, it would be morally wrong to impose the law in question on C.

Identifying what the evaluative standards are is an additional and cumbersome challenge. Freeman has suggested that

For the consensus theorist, the accessibility requirement functions as a constraint mechanism alongside the PRO restriction. According to the standard consensus model, only reasons based on shared political values can enter into the pool of reasons that are potential justificatory reasons (hence the PRO restriction). However, as Rawls stresses, without guidelines of inquiry – i.e. without shared evaluative standards – ‘substantive principles cannot be applied’ (Rawls Reference Rawls1996, 223).Footnote ²⁴

As for convergence theorists, whether one endorses or rejects the accessibility requirement has significant ramifications for one’s account of public reason liberalism. Let us look at another example of what sorts of reasons could be appealed to on a convergence model if the accessibility requirement were rejected. Vallier is a convergence theorist who construes evaluative standards as agent-relative and hence rejects the notion of shared evaluative standards (Reference Vallier2014, 107). This allows for divine revelation, for example, to be considered a legitimate evaluative standard. Even more, contradictory standards of divine revelation are said to be equally legitimate. So, the Christian’s standard of divine revelation is as legitimate as the Muslim’s standard. According to Vallier, it is permissible for ‘reasons supported by one of these agent-relative evaluative standards alone to figure into a public justification for a law’ (Reference Vallier2014, 107).

It does not follow from this that Vallier does not impose any sort of restriction whatsoever on the sorts of reasons that may be appealed to. To the contrary, he endorses an intelligibility requirement in which ‘A’s reason R_A can figure into a justification for (or rejection of) a coercive law L only if it is intelligible to all members of the public.’ Moreover, ‘A’s reason R_A is intelligible for members of the public if and only if members of the public regard R_A as epistemically justified for A according to A’s evaluative standards’ (Vallier Reference Vallier2014, 106). So, it is perfectly acceptable for the religious citizen to appeal to reasons based on divine revelation as long as those reasons are intelligible to all others in light of the religious citizen’s acceptance of divine revelation.

Framed as such, the intelligibility requirement is ambiguous. On the one hand, one might reject the idea of shared standards of evidence, but endorse the notion of shared standards of inference. But if there are no shared standards of evidence, then even if one draws valid inferences, some positions that one reaches will clearly be unjustified. For example, on the approach under consideration, one does not criticize P for believing that a pack of unicorns has formed a colony on the summit of Mt. Everest. One only criticizes the way P makes inferences based on that belief. On the other hand, if all evaluative standards are to be seen as agent-relative, then different individuals can also have different standards of inference. On this approach, it would follow that S can criticize P’s beliefs and inferences only if P has not adhered to his or her own standards.

On the issue of whether Vallier endorses the weaker or the stronger of the two positions, the clearest passage we get is this:

But once again, are so-called ‘poor inferences’ poor because they fail to meet some agent-neutral standard; or, are they poor because the appropriate agent-relative standard was improperly applied? If Vallier intends to endorse agent-neutral standards of inference, but agent-relative standards of evidence, then an argument needs to be offered for why the two types of evaluative standards can permit dissimilar evaluative criteria. Otherwise, it is plausibly the case that any argument in favor of agent-neutral standards of inference will undercut any claim to agent-relative standards of evidence.Footnote ²⁵ The worry, then, is that the intelligibility requirement is too weak; some comprehensive views may have to be deemed unjustified on the basis of agent-neutral evaluative standards.Footnote ²⁶

A rejection of the intelligibility requirement need not drive us to the consensus model. This is partly because I doubt that the PRO restriction does any worthwhile work beyond the work that is already done by the accessibility requirement. After all, if one’s comprehensive views meet shared standards of evidence and inference, then what is the point of further requiring that one’s reasons be public?Footnote ²⁷ The consensus approach, then, appears to be too strong.

There is a viable middle ground to be defended, one that entails a rejection of the PRO restriction but an endorsement of shared standards of evidence and inference. As I have implied, an endorsement of shared evaluative standards is frequently associated with a notion of accessibility. While accessibility might be understood in a variety of ways, I take P’s reason R to be accessible if and only if R (including P’s rationale for R) is supported by shared evaluative standards (Cf. Vallier Reference Vallier2011a, 369; Reference Vallier2014, 108). Moreover, R must be accessible in order for it to play a justificatory role with regard to legislative proposals. This notion of accessibility will clearly restrict what sorts of reasons can count as justificatory reasons – reasons that can be appealed to in order to justify (or defeat) a particular proposal.Footnote ²⁸

The accessibility requirement is widely-discussed in the public reason literature. Importantly, while it has traditionally been associated with consensus models of public reason, Footnote ²⁹ there is nothing that dictates that if one employs an accessibility requirement, then one must be a consensus theorist. One plausible explanation of why the accessibility requirement has typically been attached to consensus theories is that some consensus theorists may believe that the accessibility requirement would block all religious reasons from entering into public justification. As Vallier has noted, ‘the accessibility requirement was practically crafted with religious reasons in mind’ (Vallier Reference Vallier2011a, 385).Footnote ³⁰ This has perhaps led some to think of the accessibility requirement as being necessarily linked to consensus theories. Vallier, for example, says that ‘[w]ithout accessibility, public reason will lean towards “convergence” conceptions of public reason…’ (389). But as I have stressed, the accessibility requirement can be incorporated into convergence models of public reason. I refer to such models as soft convergence models of public reason.

As I construe the term, accessibility incorporates the notion of adequacy. As Robert Audi writes, ‘[a]n adequate reason (for something) is one that, in rough terms, evidentially justifies the belief, act, or other element it supports. We might also say that adequacy implies that an action or belief based on the reason is thereby rational’ (Reference Audi2011, 67–68). Audi proceeds to say that an adequate reason must ‘be in a certain way accessible to rational adults: roughly, appraisable by them through using natural reason in the light of facts to which they have access on the basis of exercising their natural rational capacities’ (Reference Audi2011, 70). In other words, adequate (or justificatory) reasons must pass agent-neutral evaluative criteria. I have described such criteria in terms of shared evaluative standards.

The consequences of not accepting at least a minimalist accessibility requirement are high: in the absence of such a requirement, defeater reasons will be radically unconstrained. For this reason, I take the issue about what degree of accessibility is most appropriate to be a fundamental aspect of the on-going debate not just between consensus and convergence theorists, but amongst convergence theorists.

Regarding the issue of defeaters, it is important to note that a justified reason is not a sufficient reason (and hence, not an undefeated reason). As such, even if a reason is accessible, it might well be defeated by a weightier accessible reason.Footnote ³¹ So, the accessibility requirement merely places a constraint on which reasons may be considered justificatory reasons. To re-emphasize, such reasons need not be available to actual agents, but only to their moderately idealized counterparts.

It might be objected that the view presented here is engaging in a bit of rhetoric by conveniently associating the accessibility requirement with rationality or reasonableness. This is a result of speaking about the accessibility requirement in terms of the notion of adequacy. When we deliberate about whether reason R evidentially justifies some belief or action, we are deliberating about whether R satisfies some evaluative standards – whatever those standards may be. There may be a temptation, then, to go further and label that which satisfies such standards as ‘reasonable’ or ‘rational.’ This is not problematic so long as we avoid extracting our evaluative standards from antecedent assumptions about what is reasonable or rational. As such, I make no assumptions here as to what sorts of appeals would be ruled out by an accessibility requirement. For example, I do not assume that an accessibility requirement will constrain all religious reasons. Nor do I assume that any religious reasons will be vindicated.Footnote ³² Such issues must be the subject of on-going debates about whether such appeals are adequate to evidentially justify the claims, actions, or policies, they are intended to support. What is of crucial importance is that the accessibility requirement applies equally to religious reasons, and other non-public reasons, as well as public reasons.Footnote ³³ At the end of the day, what matters is whether our reasons (and the beliefs they are based on) are well-grounded.Footnote ³⁴

Part of settling debates about the role of religious reasons will require putting extensive effort into better clarifying and defining the notion of shared evaluative standards that is essential to the notion of accessibility. Viewed as such, the issue of how to best characterize the relationship between a sufficient reason and an individual member of the public is an issue that must draw from working theories of rationality. To gain traction on many of the outstanding issues in political philosophy (and ethics more generally), we must further confront the problem of how to best distinguishing between ‘good’ and ‘bad’ reasons. This is no easy task, but it is a problem that profoundly impacts the vast majority of ethical and political theories.

6. Concluding remarks

I have focused on the case of citizens who are not official agents of the state. Specifically, I am concerned with whether citizens fail to respect each other when they either agree to or advocate for coercive laws on the basis of reasons that others may not be able to accept. In doing so, I have adopted a particular form of pubic reason liberalism. To see this, let us distinguish between two modes of public reason. The first mode is what I will call top-down public reason. On this mode, ‘decisions regarding constitutional essentials and matters of basic justice should be decided by political authorities…’ (Freeman Reference Freeman2007, 329). In other words, it is up to public officials to determine which laws satisfy the requirement of public justification. This mode of public reason can be contrasted with bottom-up public reason, in which ordinary citizens express their preferences – e.g. through voting – in order to determine whether a proposed law satisfies the requirement of public justification. The difference between the two modes is important. On the top-down mode, public officials are the ones to decide whether individual citizens have sufficient reason to agree to a proposal. On the bottom-up mode, it is the citizens themselves (or perhaps we should say their suitably-idealized counterparts) that determine whether they have a sufficient reason to agree to a proposal.

It may well be the case that the epistemic or normative constraints that one places on the political practice of proposing a law will depend on which mode of public reason is under discussion. If public reason liberalism is meant to be nothing more than a decision procedure for public officials to follow – i.e. purely a matter of top-down public reason – then one might plausibly argue that public officials should be confined to speaking only in terms of public reasons (Freeman Reference Freeman2007, 386; Rawls Reference Rawls1996, 216).Footnote ³⁵ After all, perhaps the most practically expedient way to satisfy the requirement of public justification, as a public official, is to appeal only to political values that all citizens would accept. For if officials propose a law that can be justified only on the basis of comprehensive reasons, it is extremely likely that at least one citizen will have a different, but reasonable, comprehensive view which serves as grounds to reject the proposal.Footnote ³⁶ Importantly, I leave open the issue of whether an appeal to practical expediency could help to normatively justify a PRO restriction on a top-down account of public reason.

This qualification aside, I have argued against the claim that citizens fail to respect one another when they advocate for coercive laws on the basis of non-public (but genuine) reasons. Because of this, I have rejected the call for a public-reason-only restriction. However, I have maintained that public reason liberals should remain committed to an accessibility requirement that is framed in terms of shared (agent-neutral) evaluative standards. The resulting view occupies a viable middle ground between the consensus and convergence models of public reason that have recently been proposed.