Sharing reasons and emotions in a non-ideal discursive system

This paper critically evaluates two aspects of Maxime Lepoutre's important book, Democratic Speech in Divided Times. First, I examine Lepoutre's approach to the shared reasons constraint—the requirement to offer shared reasons within public deliberation—and the place of emotions in public discourse. I argue that he, and indeed all who adopt such a highly inclusivist approach, face a dilemma that pushes him either to apply the shared reasons constraint more widely than he desires or to abandon it completely. I chart a course through this dilemma, but one that involves significant revisions to Lepoutre's position, particularly regarding the need for idealization. Second, I consider Lepoutre's use of the systemic approach to public discourse, which is central to many of his arguments, including his responses to critics of the discursive democratic ideal. Using his arguments regarding angry speech and dogmatic group cognition as illustrative, I highlight the somewhat speculative nature of these systemic arguments, which often rely on conjectures about how the system might operate, how its parts fit together, and how the system as a whole might attenuate seemingly problematic features of its component parts. This limits the ultimate persuasiveness of Lepoutre's responses to skepticism about democratic speech in our divided times.


Maxime Lepoutre's Democratic Speech in Divided Times
confronts a crucial question: can the importance and democratic value of inclusive public discourse be defended even in the face of inequality, antagonism, and political polarization? Should our non-ideal reality lead us to reject or minimize the role for public discourse within politics? Or can a normatively attractive and realistic model of democratic speech be presented even within our divided times? Lepoutre aims to do just that, and thus to convince us that inclusive public discourse can serve crucial normative functions even under non-ideal circumstances, and indeed that many of those circumstances present less of an obstacle to this than we might have thought, while others can be overcome through appropriate norms of public discourse and wise public policy.
Part I of the book identifies and defends norms of public discourse, in terms of both the content of that discourse-what kinds of things we should say-and the form of that discourse-the way that we say those things. These norms are explicitly shaped by, and thus intended to apply within, non-ideal circumstances, in which anger, hate speech, and misinformation are prevalent. Some might worry, however, that our times are even less ideal, or even more divided, than Lepoutre imagines, in ways that make fruitful public discourse impossible. Part II considers three such worries-the problems of (lack of) goodwill, political ignorance, and fragmentation-and argues that they can be defused, and indeed that the norms defended in Part I can be part of the solution to these problems. Lepoutre's direct engagement with many concerns about how public discourse functions in actual, polarized, societies, rather than simply operating at the level of ideal theory, is one of the features that makes the book a major contribution to democratic theory.
Indeed, Democratic Speech in Divided Times is a truly wonderful book. Lepoutre writes beautifully and lucidly, constructing arguments with rare skill and control. Real-life examples and empirical evidence are seamlessly woven into the argument, illustrating and supporting Lepoutre's claims. And Lepoutre draws from a wide range of domains of philosophy and empirical research, far beyond what one might expect in a work of political theory. My reflections here certainly cannot do justice to the depth and scope of the book. Instead, I want to accomplish two things.
First, I will examine in some detail Lepoutre's approach to the shared reasons constraint-the requirement to offer shared reasons within public deliberation-and the place of emotions within public discourse. Lepoutre's account contains crucial insights concerning the application of ideas of public reason to non-ideal conditions, but also some unresolved tensions. Indeed, I will argue that he faces a dilemma that might force him either to apply the shared reasons constraint more widely than he desires or to abandon it completely. This dilemma confronts all those who wish to adopt an "inclusivist" approach to public reason, which permits citizens to offer nonpublic reasons, and especially confronts those sensitive to real-life circumstances. I will chart a course through this dilemma, but one that involves revisions to Lepoutre's position, particularly regarding the use of idealization, which I will argue cannot be avoided in the way Lepoutre desires.
Second, I will critically assess Lepoutre's use of the systemic approach to public discourse. One of the major developments in deliberative democratic theory in recent years has been the "systemic turn" (see Mansbridge et al., 2012). While earlier theories of deliberative democracy tended to be presented as if the public forum was one place, where all citizens deliberated together, more recent theories have emphasized the idea that public discourse occurs across multiple sites, which together form the overall discursive system. 1 Lepoutre (2021: 30-32) locates himself squarely within this approach, and indeed demonstrates its fecundity throughout the book. However, his arguments are often somewhat speculative, relying on conjectures about how the system might operate, how its different parts might relate to one another, and particularly how the system as a whole might attenuate seemingly problematic features of its component parts. I illustrate this tendency in the final section of the paper, and argue that it limits the ultimate persuasiveness of Lepoutre's responses to skepticism about public discourse in divided times.

The shared reasons dilemma
The requirement to offer shared reasons in democratic debate is of course familiar from the literature on public reason and public justification, which contains innumerable formulations, defenses, and objections to this constraint. 2 Many of these are highly abstract and idealized, focusing on the conditions of legitimacy and duties of reason-giving within a "well-ordered society." In keeping with his general approach, Lepoutre instead considers the place of a shared reasons constraint within non-ideal conditions. Lepoutre's (2021: 1-2) approach to the constraint is shaped by the two central roles that he sees inclusive public discourse as playing within democratic politics. First, an accountability role. Public discourse enables citizens to hold decision-makers accountable, to both shape and challenge their decisions, and to demand justifications from them. It thus "constitutes a bulwark against domination," where to be dominated is to be subjected "to arbitrary power-to power that is not forced to consider and track the concerns of those subject to it" (Lepoutre, 2021: 2). 3 Second, an epistemic role. Diverse citizens contributing their varied concerns, experiences, and arguments enables good policy-making. It enables both the identification of social problems that need addressing and the formation of appropriate policy responses. "It is a crucial resource for pooling the politically relevant knowledge that is distributed across different social groups" (Lepoutre, 2021: 239). Lepoutre (2021: 19-21) endorses the shared reasons constraint, on the ground that it contributes to citizens' non-domination by ensuring that they can recognize the normative force of the reasons offered in support of laws and policies, and thus can critically engage with those reasons. This contributes to non-domination because it enables citizens to contest the political power that they are subject to, thus making the exercise of that power more responsive to their concerns and interests. This kind of contestation is much more difficult if political debate is conducted based on reasons that a citizen considers to be deeply misguided. In such a situation, "because the power she is subjected to is justified in terms she profoundly rejects, she is unable to force it to track the considerations she deems fundamentally important" (Lepoutre, 2021: 21).
However, Lepoutre also worries that the exclusiveness of the shared reasons constraint can itself cause some citizens to be dominated-including citizens from vulnerable and marginalized groups. Such citizens often cannot present their concerns in terms that others can accept, so the constraint threatens to prevent them from contesting the exercise of political power on the basis of their group-specific experiences and perspectives (Lepoutre, 2021: 22-25). Given the importance of such contestation for the realization of non-domination, this results in their being dominated.
We thus face a dilemma: "reasons grounded in non-domination support the shared reasons constraint in some respects, and pull against it in others" (Lepoutre, 2021: 25). Further, the exclusiveness of the constraint comes at an epistemic cost, since it means that crucial policy-relevant information concerning the experiences of marginalized groups is overlooked.
The key task, therefore, is to show how we can maintain the value of the shared reasons constraint while reducing its exclusiveness. Lepoutre rejects one possible approach, via idealization of the justificatory constituency, before defending two others: a limit on the sites in which the shared reasons constraint applies and an embrace of non-argumentative, emotionally charged speech. I will consider each of these three arguments over this and the next section, in order both to identify a dilemma that Lepoutre's arguments themselves create, and to suggest a way through it.

Sites of reason-giving
Lepoutre's first approach, which limits the sites in which individuals are required to offer shared reasons, is consonant with arguments made by various theorists in recent years. The general thought is that there is important value in laws and policies ultimately being justified by certain kinds of reasons-reasons that all (reasonable 4 ) citizens can recognize as having normative force 5 -but this requirement need not place obligations on all citizens to present such reasons in their own political advocacy. John Rawls's final view was that citizens are permitted to present non-public reasons "at any time" as long as proper public reasons are offered "in due course" (Rawls, 2005: 462). There is some ambiguity here as to whether the citizen offering non-public reasons must themselves fulfill this provison, or whether it is sufficient for someone to do so. Other theorists have gone further, explicitly holding that it is only officials or representatives who need to offer shared reasons; the discourse of "ordinary citizens" is unconstrained in this respect (see Bonotti, 2017: 124-51;Habermas, 2006;Hertzberg, 2019: 24-62;Laborde, 2017: 125;North, 2012). Lepoutre goes further still, holding that the constraint only applies in "the most formal deliberative forums," and naming "the opinions of Supreme Court justices, and the final stages of legislative justification" (2021: 33). His thought seems to be that as long as public reasons are used at the point at which binding political decisions are being made and justified, the aim of the shared reasons constraint is realized. All stages before this point can thus be unconstrained-from discussions in "barbershops, [to] newspaper opinion pieces, and even the opening stages of legislative debates" (Lepoutre, 2021: 33). 6 This being the case, however, it seems that Lepoutre's argument is misframed. He presents his view as defending a version of the shared reasons constraint on contributions to public discourse. But he limits the applicability of the constraint to such an extent that it is unclear that it is really about discourse at all, as opposed to being about official public justification. Only the official justifications for the exercise of state power-through legislation or court judgments-must be made in terms of shared reasons. This kind of constraint on official justification is quite different from a constraint on discourse. Indeed, it is endorsed by some theorists who reject the shared reasons constraint. For example, Chaplin (2012) offers various objections to the constraint, defending unconstrained speech by both citizens and representatives, before arguing that official justification should be offered in shared terms. From a very different perspective, Vallier (2014: 181-224) also rejects any requirement that citizens or representatives offer shared reasons, while nonetheless holding that laws must satisfy a public justification principle, and certain officeholders such as judges must offer shared reasons in order to ensure this. Vallier (2015) presents this view as involving a "divorce" of public justification from public deliberation. Laws must be publicly justified, 7 but this does not require the use of particular kinds of reasons in deliberative fora. Lepoutre seems to share this view, in a way that makes it misleading to suggest that he endorses the shared reasons constraint at all.
More generally, we can imagine a range of views here, all of which agree that laws must ultimately be publicly justified, but which differ with respect to the duties of reasongiving that they see as arising from this. 8 These views vary with respect to who has duties to offer shared reasons-ordinary citizens, candidates for office, members of the legislature, members of the government, senior officials, judges, and so on. They also vary with respect to the content of the duties borne by these parties. Can those with duties to offer shared reasons also offer unshared reasons alongside them? Must they always offer shared reasons, or only "in due course"? Must this "in due course" requirement be fulfilled by each individual, or just by some individuals at some point? The strictest view here would hold that everyone-from ordinary citizens to Presidents and Supreme Court justices-must offer shared reasons, and only shared reasons, whenever they publicly discuss political matters. The laxest would hold that only those offering the official justification for political decisions must give shared reasons, and even they are also permitted to offer unshared reasons alongside.
Lepoutre's position is at the laxer end of this spectrum, with fairly limited reasongiving duties 9 being generated by the public justification requirement. I am sympathetic to this view. But it faces an important challenge: why does the argument for the importance of public justification not also justify more extensive reason-giving duties? After all, if laws must be justified by shared reasons, then should all citizens not offer those reasons, in order to provide arguments for laws that could ultimately serve as public justifications? If citizens are seeking to persuade one another to support particular policies, and even voting based on their policy preferences, then surely they should offer reasons that can in fact serve as proper justifications for those policies (Watson and Hartley 2018: 62-87).
Whether one can offer a persuasive response to this challenge will turn on one's argument for endorsing the ideal of public justification itself. Some such arguments might entail that unshared reasons can have no legitimate role within public discourse, while others will leave more space for them. As we have seen, Lepoutre's argument appeals to the value of non-domination. One's ability to assess, respond to, and contest the justifications offered for laws is enhanced when that justification uses reasons one shares. "When one is given reasons one both understands and accepts as a justification for political power, one is better able to hold that power accountable and thereby make it less dominating" (Lepoutre, 2021: 21). In contrast, citizens experience a level of domination when the exercise of political power is not justified to them using reasons that they accept. In order for everyone to enjoy non-domination, the reasons offered for laws must thus be ones that all citizens can accept.
Is this argument consistent with greatly restricting the sites of the duty to offer shared reasons? Perhaps it is, since most public discourse does not involve the exercise of political power. Citizens do not exercise political power over one another when they offer arguments for their preferred policies, and even deliberations in the legislature do not directly exercise power. So these discourses can be unconstrained, with non-domination secured by the final, coercively-enforced, decision being justified using shared reasons.
On the other hand, ordinary citizens do exercise a modicum of political power when they cast their votes. So perhaps they should be able to offer shared reasons at that point, in order to avoid dominating their compatriots. Indeed, the idea that this is when even ordinary citizens must offer public reasons is a plausible interpretation of the "due course" of Rawls's proviso. Candidates for office certainly aspire to exercise political power, so perhaps they should offer shared reasons, to demonstrate that they will exercise that power in non-dominating ways. Deliberation within legislatures, meanwhile, even at its early stages, does fairly directly shape the use of political power. One might thus think that Lepoutre's non-domination argument would lead to more extensive duties to offer shared reasons than he supposes. Indeed, returning to ordinary citizens, even prior to voting citizens' deliberations ultimately have the exercise of political power in view. Even here, justifications are aimed at influencing decision-making, and thus arguably must be justification "addressed to others, and, as such, nonshared reasons simply cannot play the role of offering a justification to others" (Watson and Hartley, 2018: 75). These arguments would expand the sites in which the shared reasons constraint applies, and thus shift our attention back to public discourse as opposed to merely official justification.
But this would be costly for Lepoutre, since it would mean there are more forums in which the shared reasons constraint has the exclusionary implications that he wants to avoid. It would thus undermine his first means for alleviating that exclusion. To sustain his argument, he must explain in more detail why non-domination can be secured without reason-giving duties applying more widely.
One feature that such an explanation would need is the claim that citizens can have fruitful discursive interactions without appealing to shared reasons. More specifically, citizens are able to assess, respond to, and contest one another's contributions to public discourse-thus making that discourse (and the exercises of political power that it influences) responsive to the concerns and interests of each-even when those contributions appeal to unshared reasons. But if this is the case then it might lead us to question the need for shared reasons even at the point of official justification. If critical engagement and contestation are possible in relation to unshared reasons, then does non-domination really require that shared reasons are used within official justification?
One alternative to all views that endorse a public justification requirement is what I have called "argumentative democracy" (Billingham, 2016). On this view, citizens and representatives ought to openly and sincerely engage with one another, offering all of the reasons that they take to be relevant to the issue at hand, and seeking to respond to one another's arguments, but they then ultimately ought to vote for whatever policies they consider most justified, and enacted laws are legitimized by coming out of this democratic procedure, without any further requirement that particular kinds of reasons be offered in support. Advocates of argumentative democracy hold that it realizes non-domination, due to its participatory procedures, which allow all citizens to hold power to account.
Lepoutre worries that this view does not adequately secure non-domination, because some citizens might find that the arguments given for laws are ones they believe to lack any normative force, such that they cannot get any argumentative purchase on them. Citizens might even feel that the whole debate with respect to a law has been misframed through the use of considerations they reject, making it difficult for them to offer any response to those justifications (Lepoutre, 2021: 20-21). But argumentative democrats would hold that such citizens are free to explain why they consider these considerations irrelevant or unacceptable, and to highlight the reasons they think should be the focus instead. Others ought to sincerely engage with such arguments, enabling exchanges of reasons whereby decision-makers can be held accountable. Lepoutre (2021: 21) motivates his concerns about argumentative democracy by appeal to cases where a prohibition on same-sex marriage is justified by the unnaturalness of same-sex unions and a racially discriminatory policy is supported by the claim that racial equality is unholy. I fear that the force of these examples largely comes from the substantive policies in question, rather than the fact the considerations mentioned in their support are unshared. Consider instead the case of an environmental policy justified by the inherent value of nature, welfare policies justified by God's special concern for the poor, or subsidies for artistic events justified by the value of the arts. Some citizens will reject these arguments, but it is not at all obvious that this leaves them dominated, or unable to hold the exercise of political power to account by contesting the laws.
In sum, Lepoutre-and indeed all those who take the kind of position he advocatesfaces a dilemma here. Justifying the public justification requirement involves arguments showing that it is critical that citizens be offered justifications for laws that they can accept. But those same arguments might well seem to justify imposing reasons-giving duties at all stages of the deliberative process. Arguments showing why this is not necessary will be grist to the mill of argumentative democrats, who will hold that this really shows that justification using shared reasons is itself unnecessary, as long as we have robust and open democratic processes. The challenge, then, is to provide an argument that clearly justifies the public justification requirement, while showing why reasongiving duties can be limited to particular persons and forums. As I noted above, I think that the resulting position is an attractive one. But identifying the balance of arguments that justifies it is not easy, and I am not persuaded that Lepoutre's appeal to nondomination can do the trick. Lepoutre's (2021: 22-25) initial motivation for restricting the sites where reason-giving duties apply, recall, was to alleviate the concern that the shared reasons constraint excludes many of the considerations that vulnerable groups might otherwise appeal to.

Idealization
Another response to this concern, which Lepoutre considers, is idealization. Public reason theorists typically speak of reasons that are acceptable to or accessible to "reasonable citizens," where this picks out an idealized justificatory constituency, whose members are defined by their commitment to certain normative values and/or their compliance with certain epistemic norms. Vulnerable groups' reasons might well be shared by this constituency. But Lepoutre rejects, or at least seeks to minimize the place for, idealization, arguing that "shared reasons" should be defined as the reasons that all unidealized, reallife citizens can recognize as having normative force-or as close as possible. He offers two reasons for this, neither of which is persuasive.
The first is that idealization allows those deemed unreasonable to suffer domination (Lepoutre, 2021: 27-28). Reasons that they reject can count as shared reasons, and thus be used to justify laws. Political power can therefore be exercised over them without reasons being offered that they can accept. Lepoutre claims that his argument that this constitutes a form of domination applies to the unreasonable, as much as to the reasonable. 10 Whether this is true, however, depends on whether we moralize our account of domination in a particular way, namely by building normative judgments concerning those subject to power into our view of whether those individuals are dominated. For example, it might seem mistaken to consider someone with morally egregious views to be dominated when they are coerced in ways that they reject on account of those views, especially when the coercive law has been enacted through democratic procedures, and is preventing the individual from violating others' rights by acting on their immoral views. Is the prospective murderer really dominated when they are prevented from acting on this desire? Similar considerations can apply when it comes to epistemic failings. If my failure to recognize reasons for a law is due to some grievous epistemic failure on my part, then am I really dominated by that law? The problem here seems to be with me, rather than the law. If we moralize the account of domination in order to accommodate these kinds of cases, then those who exhibit the relevant moral and epistemic failings would not be dominated by laws the reasons for which they cannot accept. 11 Idealization deeming them unreasonable would thus no longer result in domination. The upshot of this is that the implications of Lepoutre's non-domination argument for our conception of reasonableness depends on the precise account of domination. 12 I take it that Lepoutre's objection to idealization depends on an account of domination that does not feature the kind of moralization I have suggested; but he does not defend this account. 13 Sometimes Lepoutre suggests a different way in which the unreasonable might be dominated. The domination might come not (only) from laws being justified by reasons one rejects, but by one being barred from offering the reasons one accepts. Non-domination is enhanced when citizens "can publicly appeal to the considerations they deem most significant" (Lepoutre, 2021: 90), even when those considerations are normatively or epistemically unreasonable, because this enables citizens to hold political power accountable. But Lepoutre's account already permits this, given that the duty to offer shared reasons does not apply to ordinary citizens. If non-domination requires permission to appeal to whatever reasons one desires then this is already secured in Lepoutre's view, independent of how reasonableness is conceptualized.
Indeed, once we narrow the sites where reason-giving duties apply, it seems plausible for the shared reasons constraint to refer to the reasons shared by idealized citizens.
Actual citizens are free to express their views and concerns, appealing to whatever considerations they wish. Laws must then ultimately be justified by reasons that would be shared by a relevantly idealized constituency-for example, a constituency committed to basic liberal values and whose reasoning is free of obvious errors. The shared reasons constraint thus operates to ensure that laws have the right kinds of justifications -justifications that all reasonable citizens could accept-while all actual citizens can fully participate in public discourse, unconstrained by duties to offer particular kinds of reasons. Lepoutre (2021: 28-30) also offers a second argument for rejecting idealization, however. In real-life, non-ideal, societies the unreasonable are often the people with power. Thus, Lepoutre claims, if our account of discursive norms permits us to ignore the views of the unreasonable then we will end up ignoring the views of those in power. This will make it difficult for the vulnerable effectively to hold the powerful to account. If "weaker agents frame their claims in terms of considerations that powerful agents are unresponsive to, or simply ignore powerful agents" (Lepoutre, 2021: 29), then public discourse will not serve its purpose of holding the powerful accountable. Idealization thus threatens to lead to the domination of vulnerable groups.
This argument is rather puzzling, in the light of Lepoutre's view that most citizens are not required to offer shared reasons, most of the time. Even if shared reasons are defined by appeal to an idealized conception of reasonableness, this does not place any restrictions on the ways in which the vulnerable can seek to hold the powerful to account, even if the powerful are unreasonable. Citizens are perfectly free to engage with the unreasonable, and to appeal to whatever reasons they want to within that engagement.
Further, the relevant contrast here is with a view that adopts an unidealized conception of the constituency who must accept a reason if it is to count as shared. But it seems implausible to think that the vulnerable can better hold individuals with objectionable views to account when those individuals are included in that constituency. After all, this would mean that unreasonable citizens can rule out vulnerable groups' concerns from counting as shared reasons, on the ground that the unreasonable cannot accept those reasons. It would thus limit the resources that the vulnerable have for contesting the exercise of political power. Importantly, this concern applies even with Lepoutre's narrow view of reason-giving duties, since his view still says that political decisions must not ultimately be justified using reasons that those within the justificatory constituency reject. Vulnerable groups' reasons would not be admissible in official justification. An idealized public justification requirement would do better on this score. It could ensure that vulnerable groups' concerns count as shared reasons, and are thus able to justify laws. This would also give vulnerable groups a basis for objecting to uses of power by the unreasonable that cannot be justified by shared reasons; the very fact the laws cannot be so justified is grounds for complaint. If anything, then, Lepoutre's concern with the domination of vulnerable groups seems to provide an argument for idealization.
This discussion also points toward a broader worry with an unidealized account of the justificatory constituency, and thus of shared reasons: it is not clear that there are many such reasons, in real-life democracies. One response to this is to appeal to ideals such as "freedom" and "equality," and to argue that everyone accepts these values, at some high level of abstraction, and thus laws justified by appeal to (some interpretation of) those values are justified by shared reasons. This response seems dubious, however. Even if you and I both accept the abstract idea of equality, it does not seem that you have justified a law using a reason I share if you appeal to an interpretation of equality that I do not see as having any normative force. 14 Lepoutre (2021: 35-36) appears sensitive to this concern, and implies that reasons must be accepted by all parties at a fairly granular level in order to count as properly shared. This does not mean that all parties must be persuaded by the specific argument made for the law. But it does mean that they must recognize the normative force of the considerations used as justification, rather than merely endorsing a highly abstract version of the values invoked.
This reinforces the worry that the pool of shared reasons might be extremely limited, or even empty. When it comes to actual legislative outcomes, Lepoutre's view seems to lead to a minimal state of the kind endorsed by right-libertarians, or even to there being no reasons that can justify state action at all. It is not clear to me how this implication can be avoided without endorsing some idealization. To put it bluntly, we have to deal with the fascist somehow, and having him determine which reasons regarding equality count as shared does not seem like an attractive approach. Now, as I will discuss shortly, Lepoutre offers an account of how currently unshared considerations can come to be shared, through the use of emotionally charged narratives. This might increase the pool of shared reasons. The vulnerable and marginalized sharing their experiences can certainly lead others to recognize normative considerations that they previously overlooked. Nonetheless, it seems implausible to think that this can create a sufficiently expansive set of reasons shared among actual people to legitimize crucial justice-promoting laws.
In sum, then, I do not think that Lepoutre offers persuasive reasons to reject idealization. Indeed, there are reasons internal to his account to embrace at least some idealization, both to enable the vulnerable to hold the powerful to account and to avoid unattractive implications with respect to legislative outcomes. Further, it does not seem like some idealization being deployed when defining the shared reasons that must ultimately be used to justify laws creates problems with respect to creating inclusive public discourse-at least if Lepoutre's narrow view of who is under duties to offer shared reasons can be maintained. To further examine whether that view can be maintained, we should consider the role Lepoutre assigns to emotional narrative.

Shared emotions to the rescue?
Those who allow unrestrained public discourse, with respect to the reasons that citizens offer one another, but who still believe that laws must ultimately be justified using shared reasons face something of a puzzle: what role does public discourse, and the reasons offered there, play, if many of those reasons cannot ultimately justify laws? Partly this is a question of citizens' motivation: why would they engage in public discourse if they know their reasons will be filtered out of formal arenas? Citizens do not simply want to express their concerns in informal forums; "they want the expression of those concerns to influence formal decision-making that produces coercive decisions" (Lepoutre, 2021: 33). But Lepoutre argues that this is also a question of accountability and non-domination. Appeals to unshared considerations "will not help voters contest political power (and thereby protect themselves from domination) unless they eventually penetrate formal decision-making arenas" (Lepoutre, 2021: 37). In other words, there must be some way for unshared considerations to become shared, such that they are able to shape formal decision-making.
A few other theorists have proposed ways in which unshared reasons offered in informal discursive forums might be able to productively contribute to democratic decisionmaking, often focusing on religious reasons (see Billingham, 2023;Habermas, 2006;Hertzberg, 2019: 63-102;Krause, 2008: 160-161). But these accounts tend to focus on rational argumentation. Lepoutre argues that this is insufficient, and offers an original account of how new considerations can be introduced into the pool of shared reasons, via the use of narrative and emotionally charged speech.
Personal narrative, testimony, and emotional appeals can "enable members of different social groups, who encounter radically different constraints and enablements, to communicate politically important experiences and perspectives to one another" (Lepoutre, 2021: 83). They can thus serve a distinctive epistemic function. Such speech "aims directly to introduce or make visible certain experiences or perspectives, and thereby to enrich the pool of publicly recognized considerations" Lepoutre, 2021: 39). It can show things that cannot yet convincingly be logically argued for, due to a lack of shared experiences or concerns (Lepoutre, 2021: 41). It allows listeners to appreciate considerations they would otherwise overlook, and enriches the conceptual resources available to interpret those experiences. It can thus bring to light reasons that come to be shared, and therefore can justify the exercise of political power.
Emotional speech can achieve this due to the phenomenon of emotional contagion, whereby listeners "catch" the emotions of others. Those emotions then act as sources of salience, highlighting evaluative properties that one might otherwise overlook, or even that do not fit within one's existing conceptual categories. Lepoutre develops this argument in detail in the case of anger. Lepoutre (2021: 52) characterizes anger as an emotion whose fitting object is injustice or wrongdoing. 15 It thus can make salient injustices that those who are not its victims could easily miss (Lepoutre, 2021: 59-69). He powerfully illustrates this using the examples of Frederick Douglass's denunciations of slavery and Malcolm X's attacks on racial injustice. Listeners who "catch" the anger of such speakers are enabled to understand the nature and extent of that injustice in a way that would not otherwise be possible. In this way, new reasons can become shared, and thus used to justify laws and policies.
There are several attractive features of this argument. First, as Lepoutre emphasizes, in non-ideal societies, where various kinds of injustices are common, many people feel a sense of frustration, upset, anger, and so on. It is appealing to find a place-and indeed a fruitful place-for the expression of these emotions within public discourse. Second, those who have defended the place of emotion in public discourse have generally opposed ideas of public reason, while those who endorse public reason have generally been skeptical of, or more often simply silent about, emotionally charged speech (Lepoutre, 2021: 39-40). Lepoutre shows how a public justification requirement can be combined with an openness to narrative and emotion. Third, as we have noted, it provides an additional solution to the puzzle of the role of non-shared considerations within views that endorse public justification.
Nonetheless, there are also some reasons for caution. An obvious concern is that the rhetoric of emotional narrative can be used to cover for inferential fallacies and other kinds of argumentatively dubious moves. The novel considerations added into the pool of shared reasons via emotional narrative might suffer from significant epistemic defects.
Consider the example that Lepoutre offers of same-sex marriage debates in France. Lepoutre describes the speech that Christiane Taubira, then French Minister for Justice, gave in the parliamentary debate on this issue. Taubira gave an impassioned description of marriage's historical evolution, whereby it has become increasingly inclusive and egalitarian, in a way that encouraged her audience to reconceptualize marriage as having a shifting meaning, and to see its extension to same-sex marriage as "the natural next step in marriage's march towards greater inclusiveness" (Lepoutre, 2021: 43). She thus opened the way for arguments for same-sex marriage appealing to newly shared premises.
My worry about this example is the claim that same-sex marriage is the "natural next step." Even if it is true that marriage has become increasingly inclusive, this does not make same-sex marriage the "natural next step" in any logical sense, or necessarily give reasons to increase marriage's inclusivity further. If representatives go on to support same-sex marriage based on being caught up in Taubira's passionate appeal then they would be making a mistake. Of course, this is not at all to say that such support cannot be justified by shared reasons. It is simply to raise a concern about reaching this conclusion via emotional rhetoric.
This kind of concern is especially pertinent given Lepoutre's reliance on the "involuntary, and largely unconscious" (Lepoutre, 2021: 61) process of emotional contagion. This process is not one that is subject to rational control, and thus does not discriminate between apt and inapt emotions or reliably track the reasons that apply. Emotional narrative can enable listeners to see things they would otherwise miss, but can also lead them to believe they see things that are not really there. Douglass and Malcolm X used angry speech to assist listeners to recognize genuine injustice; but we are equally familiar with political leaders riling up a crowd for less noble ends, including arousing anger toward vulnerable minorities. Now, Lepoutre (2021: 46) correctly notes that dispassionate argument can also involve fallacious reasoning. But the difference is that those fallacious inferences do not count as properly fulfilling the public justification requirement. Even if actual citizens are persuaded by them, we should not see them as truly providing shared reasons. Whereas it seems that on Lepoutre's view emotional narratives can directly create shared reasons, which might suffer epistemic defects.
To evade these worries, we might require that the novel considerations introduced through emotional appeals must be logically connected to existing shared reasons, and/or place other epistemic constraints on these considerations, such as requiring that the emotions be fitting, in the sense of accurately representing their object. 16 In other words, we might introduce idealization, with respect both to defining what count as shared reasons endorsed as a result of emotional narrative and what emotions can properly be expressed in public narratives, at least in certain settings. Lepoutre's resistance to idealization would make him averse to this move. But this lack of idealization is what makes the concern regarding rhetoric particularly pressing; it means that considerations that come to be accepted by actual citizens can fulfill the shared reasons constraint even if they suffer from epistemic defects.
Similar comments apply with respect to impartiality. Lepoutre (2021: 45-46) considers the objection that emotionally charged narrative undermines impartiality, due to conveying a particular, biased, perspective. He replies by noting that shared reasons are still required in the formal stages of decision-making, which should thus ensure a level of impartiality. But this reply does not address the worry that his view allows those shared reasons themselves to be biased, due to being considerations that people accept only as a result of emotionally charged rhetoric and emotional contagion. 17 Lepoutre (2021: 46) also offers a second reply to the impartiality worry, which is to argue that true impartiality involves allowing different groups' partial narratives to be heard. Impartiality is found in engaging with all such narratives, rather than seeking some objective perspective. But it is not clear how this line of thought is compatible with holding on to the ideal of public justification. That ideal seems committed to some idea of impartiality via objectivity. 18 Lepoutre's rejection of this idea seems to push him toward a different account of political justification and legitimacy, such as argumentative democracy. The best way to resist this, while maintaining Lepoutre's fruitful insights into the epistemic role of emotional speech, is to endorse an account of idealization, in order to mitigate the concerns about such speech.

Sites of reason-giving (again)
Lepoutre's arguments concerning emotional speech can provide an additional argument for his narrow view of the reason-giving duties arising from the public justification requirement. If those duties applied more widely, such that all citizens were under duties to offer shared reasons, then emotional narratives that do not include appeal to shared reasons, or are not offered alongside shared reasons, would violate those duties. Citizens refraining from engaging in such speech, however, would deprive the polity of the epistemic value that Lepoutre identifies. In order to harness that value, citizens must be permitted to engage in such speech, including when they do not also have shared reasons, since it is in precisely those circumstances when such narratives can be especially significant.
Similar arguments can be made with respect to other kinds of speech that does not appeal to shared reasons, such as argumentative speech appealing to unshared reasons. Habermas (2006) argues that religious contributions to informal public discourse can have epistemic value, since religious arguments can contain normative insights that even non-religious citizens can recognize and be moved by (see also Billingham, 2023). Hertzberg (2019: 44-47), meanwhile, emphasizes the information-gathering and -processing function of democratic discourse and decision-making, and the way in which unrestrained speech in a variety of forums can raise awareness of overlooked issues, and offer perspectives that give others helpful information.
If these kinds of arguments are persuasive, as I think they are, then they give a strong reason to allow a wide range of contributions within public discourse, and thus to limit the sites in which shared reasons are required. 19 This is not an argument that Lepoutre makes, but it is consonant with his views, and would provide a response to the claim that his nondomination argument pushes him toward applying reason-giving duties more widely. Even if that claim has some merit, the epistemic value of keeping those duties narrow would be overriding, especially since non-domination is still ultimately secured by final decisions being supported by shared reasons. Again, however, in order to be persuasive the resulting view of public justification and public reason would need to involve some measure of idealization.

Evaluating Lepoutre's systemic approach to discourse
Many of Lepoutre's arguments that I have been considering crucially rely on his systemic approach to public discourse, which emphasizes that that discourse occurs in a system composed of many, varied, arenas, which can play different functions. For example, binding decisions are made in formal arenas, while informal arenas enable citizens to form and share their opinions. An important implication of this approach is that "it is not necessary for each arena to satisfy all the desiderata of democratic public discourse" (Lepoutre, 2021: 31); what matters is that the system as a whole functions in a way that serves the values of epistemically sound and accountable policy-making. In other words, we evaluate each part of the system based on its contribution to the properties that we want at the systemic level, rather than expecting each part to display those properties or to operate according to the same norms (Lepoutre, 2021: 76). This is why unshared reasons and emotional narrative can play a fruitful role within public discourse, even if they should not constitute official public justification. Similarly, some arenas can exclude certain perspectives, or be devoted to developing and promulgating a particular viewpoint, as long as every viewpoint is included somewhere, and all can feed into the formal decision-making arenas. "On the systemic understanding of public discourse, public discourse is composed of many interconnected arenas of political talk, which perform complementary roles" (Lepoutre, 2021: 225).
This systemic perspective plays a key role throughout Democratic Speech in Divided Times; indeed, arguments appealing to it appear in almost every chapter. Lepoutre draws upon it to defend angry speech (2021: 74-81) and dogmatic group cognition (2021: 180-185), and to argue that public discourse requires less trust and goodwill than one might expect due to the possibility of networked communication (2021: 135-138), a possibility that also means that not every community needs to be integrated in order for residential integration policies to successfully overcome the fragmentation of public life and promote intergroup communication (2021: 202-206). In this final section I will evaluate a couple of these arguments, in order to motivate some more general closing reflections on Lepoutre's methodology.

Angry speech
Critics of public anger argue that it is corrosive of trust, makes cooperation between groups more difficult, and undermines the fellow-feeling and good will required for wellfunctioning democratic processes (Lepoutre 2021: 54-56). We have already noted one of Lepoutre's responses to such critics, which is to highlight anger's epistemic role. He also offers a second response: even if anger does have the effects the critics' claim, this does not mean that an overall system that includes some angry speech will lack trust or cooperation (Lepoutre, 2021: 74-81). Anger might be welcomed in some forums and discouraged in others. The interactions between these different forums can produce clear benefits in terms of both the understanding of injustice and coalition-building to fight against it, as compared to a discursive system where anger is everywhere unwelcome. When we approach this question from a systemic perspective, rather than thinking narrowly about dyadic interactions, new possibilities are opened up for anger to have a productive place within the overall system, while mitigating its negative effects. 20 This argument is promising, but points to many fine-grained questions that Lepoutre does not have space to consider. Lepoutre makes his argument by comparing a discursive system "with anger" to one "without anger," arguing for the superiority of the former. But this broad comparison does not get us very far. After all, there are many possible arrangements with respect to the places where anger might be, or not be, exhibited, and so to get a clear view of the proper place of anger within the discursive system we need to compare various alternatives. For example, perhaps anger is helpful when expressed within partisan settings, where it helps partisans to gain a vivid sense of the injustices to which they are opposed, but counterproductive in more diverse settings, where listeners are put off by the anger and feel that they are being attacked or blamed, rather than being seen as potential allies. Or perhaps it is the other way around: perhaps in partisan settings, where individuals already have a shared sense of their values and aims, angry speech simply riles people up without revealing anything that was not already known, while it is in mixed settings that the epistemic benefits of anger can be realised. Plausible competing hypotheses could be constructed here, and further research would be needed in order to identify the settings in which anger is beneficial. The distinction between partisan and mixed settings is just one example of dimensions in which public forums vary that such research might consider. It might turn out that some systems "with anger" are worse (in the instrumental sense we are focused on here-see footnote 20) than those "without anger," due to the anger being expressed in the wrong places and times, such that its costs with respect to undermining good will and cooperation outweigh its epistemic benefits. Now, Lepoutre does sketch some possibilities of how angry and non-angry deliberative forums could interact in fruitful ways, and in particular how any problems of trust and cooperation that anger creates could be dealt with elsewhere in the system. One example involves the "anxious ally," who is swayed by angry speech and feels moved by the injustices it identifies, but is part of the group targeted by the anger and thus feels unwelcome or afraid to coordinate with the speaker (Lepoutre, 2021: 74, 78). Such an individual can find other groups who are organizing against the injustice while refraining from expressing anger, and coordinate with them. Hearing the angry speech might be essential for the anxious ally become aware of the injustice-they simply would not have been moved without feeling the anger of the victims-but the presence of less emotionally charged forums is then necessary for them to engage in fighting the injustice. To put it another way, borrowing a central example that Lepoutre (2021: 78) uses to illustrate his arguments, the anxious ally is shown the injustice by Malcolm X, but works against it with Martin Luther King. Similarly, the "skeptical ally," who is somewhat convinced by an angry speech, but worries she would be shunned if she expressed her lingering doubts, can find another, non-angry, venue to discuss the issues further (Lepoutre, 2021: 74, 78). Lepoutre (2021: 79) also suggests that the presence of angry arenas can help speakers in other forums who express the same perspective without anger to acquire the trust of wary listeners, via this contrast. Seeing Malcolm X's anger can lead skeptics to be willing to engage with King, such that the latter gets a hearing that he otherwise would not. In these cases, then, the two kinds of speech and different public forums interact in highly productive ways.
These are all plausible suggestions. But there must be a worry that the systemic focus makes it too easy to tell these kinds of "just so" stories. Plausible stories can also run in the other direction. Some individuals might be so put off by the angry speech of someone like Malcolm X that they become less likely to listen to King, on the grounds that King is pointing to the same injustices. Their negative reaction to the anger leads them to be unwilling to engage with other speakers who are highlighting similar concerns, even a non-angry way. Or some individuals who are initially persuaded by King might come to have doubts due to worrying that accepting King's arguments inevitably leads to Malcolm X's more radical position, which they consider objectionable. Anecdotally, I have certainly met people who are suspicious of arguments concerning racial injustice due to being put off by the anger of Black Lives Matter protestors. The presence of angry forums thus might make some citizens less likely to give a hearing to those protesting against similar injustices in non-angry forums. Lepoutre emphasizes the possibility of complementarities across different kinds of forums, but there is also the possibility of competition, and even of the way that some forums function undermining what might otherwise be achieved elsewhere in the system. The overall balance of these effects is very difficult to predict.

Dogmatic group cognition
Theorists who are skeptical of the value of public discourse often highlight citizens' political ignorance, and in particular the prevalence of group cognition, whereby individuals' social group identities strongly condition what, and who, they believe (Lepoutre, 2021: 159-164). Even worse, many exhibit dogmatic group cognition, whereby evidence that fits their assumptions is accepted or given undue weight while evidence that conflicts with those assumptions is rejected or given insufficient weight (Lepoutre, 2021: 171-172). This makes their beliefs-including inaccurate beliefs-very hard to change. Lepoutre (2021: 179-185) draws on his systemic approach to defend dogmatic group cognition. Even though each group will have blindspots and biases, the information yielded through their presentation of their divergent experiences, beliefs, and values can lead to gains with respect to both the epistemic quality of outcomes and the accountability of political power. "A democratic community composed of many social groups, each of which defends its starting commitments with some (non-absolute) degree of dogmatism, is liable to generate more varied information that is relevant to forming sound political judgments" (Lepoutre, 2021: 181).
This argument crucially turns on the system as a whole bringing together the different pieces of politically relevant information that are unearthed and emphasized by different dogmatic groups. But it is not clear how "the system" is meant to combine these biased perspectives in order to reach an unbiased conclusion. What kinds of procedures and processes are needed in order for the evidence that each group identifies to be properly pooled, assessed, and weighed into decisions? And can those procedures function effectively in a situation where many agents have their perspectives shaped by their social group and exhibit dogmatic group cognition? Do we need some individuals to act as objective arbiters, who stand above group cognition and can provide balanced assessments? Or can democratic procedures themselves somehow fulfill this role? Lepoutre's (2021: 169-181) argument draws on evidence showing the place of dogmatism within scientific progress. But there are disanalogies. In the scientific case different groups can dogmatically pursue their research programs and present their findings to one another, without there being a point at which collective decisions need to be made and imposed on the whole scientific community. The relevant kind of information pooling here involves scientists reading others' work and seeking to take others' findings into account in their own research-which can include dismissing or explaining away those findings, in order to sustain their own preferred theories. All scientists do this within their own research programs, without any single approach to the evidence or final view being imposed. In the democratic case, by contrast, information pooling must involve that information being taken into account within collectively binding decisions. The pooling thus takes a different form and serves a different purpose. And this makes the processes by which dogmatic groups can share information and reach collective decisions crucial. Simply appealing to the system is not enough.

Limits of systemic arguments
The arguments of the previous two sub-sections illustrate more general concerns. I agree with Lepoutre that we should take a systemic approach to public discourse, and should not assume that the same norms or principles need to operate everywhere. Not every dyadic interaction involves individuals seeking to directly justify the exercise of political power. I also agree that recognizing the diversity of public forums and the different roles they can play within the overall system enables us to see that various phenomenon that might otherwise seem troubling or objectionable can in fact play a productive role (see also Mansbridge et al., 2012). But my concern is that this approach can make it too easy to tell plausible stories about the potential value of anger, dogmatism, and so on, and similarly about the way that chains of interaction can overcome distrust and lead to policy-relevant information being disseminated. The problem is that a critic could equally well tell stories that run in the opposite direction, and it is not clear which we should consider more likely to be accurate. The system might wash away apparently troubling phenomenon or it might amplify them.
An obvious response is that we can draw on empirical research to test various hypotheses, and indeed can seek to inform future empirical research. Lepoutre does both of these frequently, citing empirical results in order to support his arguments, and explicitly identifying hypotheses that could be the focus of future studies. He is well aware that the force of his arguments often depends on fine-grained empirical work, and sees the role of political philosophers here to be that of identifying relevant, normatively-informed, empirical hypotheses, which can then be tested by social scientists (Lepoutre, 2021: 58, 133, 140, 241).
But empirical evidence also has obvious limitations. Even if we have some evidence in support of one of Lepoutre's claims, we might well doubt whether things are likely to play out the same way in other settings or situations. Real-life discursive interactions are endlessly varied, and the discursive system is extremely complex. The fact that an angry speech had a certain effect at a certain time, or that we can identify a chain of mediated interaction in a particular case, does not show that it will always work out that way, given the many contextual factors at play. This makes it hard to know how much weight to place on this empirical evidence, or on the hypotheses that it purports to support.
This worry applies even in cases where we have some empirical evidence. But in many cases relevant to Lepoutre's arguments that evidence is still lacking, or at least minimal. Either way, we will often be left with somewhat speculative arguments, which rely on stories that could equally run the other way, or where the workings of the overall system are somehow meant to wash away problems within its individual parts. This places limits on the power of many of Lepoutre's arguments appealing to the systemic approach to public discourse.
The upshot of this is not that we should abandon the systemic approach, the development of philosophically-informed hypotheses, or the pursuit of empirical evidence. But it does leave me unsure what to ultimately make of Lepoutre's positive outlook regarding the possibilities for epistemically fruitful and accountability-promoting public discourse within our divided times. All of Lepoutre's arguments are beautifully constructed, subtle and persuasive. But they often leave me with a nagging doubt: do things really work like that? Will we really see this positive effect, as opposed to the negative effect that others have suggested? Isn't it just as possible, or even likely, that the negative effects will dominate? 21 Again, the fact that so many of Lepoutre's arguments ultimately depend on how empirical matters turn out is not an unattractive feature or weakness. Indeed, it is to some extent inevitable given his non-ideal, real-life, focus. As I noted at the outset, a central attraction of Lepoutre's book and reason that it deserves to be widely-read is its engagement with concerns about how public discourse functions in actual societies. But the very fact that his arguments depend on conjectures about how the discursive system can work within our messy reality means that it is not easy to be fully persuaded by his optimistic viewpoint. I suspect that the skeptics of public discourse will remain unmoved.
In response to this, Lepoutre might point us to Chapter 8 (2021: 213-238), where he provides a robust critique of "political minimalism," the label he uses for various theories that respond to our democratic malaise by advocating for alternative forms of government that do not require extensive public discourse, such as epistocracy and lottocracy. Lepoutre powerfully demonstrates that such theories face a dilemma: either they are equally affected by the problems that are taken to undermine ambitious democratic ideals, or they propose remedies that can be reconciled with, or incorporated into, the discursive ideal. Either way, they fail to identify a way forward that abandons the commitment to public discourse. Thus, even those who struggle to share Lepoutre's optimism cannot easily retreat into minimalism.
Perhaps what Lepoutre's book shows, above all, is that questions regarding democratic speech are complex and multifaceted. Lepoutre shows great awareness of these complexities, and is always careful not to overstate the force or implications of his arguments. In one sense the whole book is a call for critics of democratic discourse to recognize that things are not as simple as we might sometimes think. Voter ignorance does not automatically make democracy a bad idea or democratic discourse pointless. The fact that anger can be off-putting to many does not automatically mean that it is democratically destructive. The idea that we need shared reasons for laws does not automatically mean that all citizens must offer such reasons. The fact that many groups are dogmatic does not automatically mean that they cannot play a positive epistemic role within public discourse. The fact that we face widespread distrust does not automatically mean that discourse cannot function fruitfully, or itself provide the solution. And so on. Lepoutre skillfully and winsomely alerts us to the complexity and multiplicity of democratic speech in divided societies. This is a great service to everyone with an interest in this topic-which should be all citizens of such societies.