1. Ideal Theory and the Citizens of an Ideal Liberal Society

The editors of the Oxford Handbook of Political Psychology write, ‘both citizens and leaders exhibit distorted reasoning and a slew of cognitive and emotional biases … Partisan resistance to new information, ethnocentric reactions to immigrants, automatic and preconscious reactions to a political candidate's facial features … the list goes on’ (Huddy, Sears, and Levy Reference Huddy, Sears, Levy, Huddy, Sears and Levy2013: 4). Political liberals need not deny these facts. Indeed, Rawls writes, ‘we do not, of course, deny that prejudice and bias, self- and group-interest, blindness and willfulness, play their all too familiar part in political life’ (Reference Rawls2005: 58). ‘But’, he says, ‘these sources of unreasonable disagreement stand in marked contrast to those compatible with everyone's being fully reasonable’ (Reference Rawls2005: 58). Since they are incompatible with reasonableness, Rawls takes such causes of disagreement to be irrelevant to the project of political liberalism as an ideal theory. ‘Such explanations are too easy and not the kind we want’, he says; ‘We want to know how reasonable disagreement is possible, for we always work at first within ideal theory’ (Reference Rawls2005: 55, my emphasis). In a later footnote, Rawls reiterates that, though every society contains unreasonable views, he is ‘concerned with an ideal normative conception of democratic government, that is, with the conduct of its reasonable citizens and the principles they follow, assuming them to be dominant and controlling’ (Reference Rawls2005: 441, fn 3, my emphasis).

There are two different claims here that a critic of political liberalism could take issue with. The first is that we should ‘always work at first within ideal theory’ (Rawls Reference Rawls2005: 55). The second is that, as an ideal theory, political liberalism can assume that the citizens of an ideal liberal society would be overwhelmingly reasonable. Many philosophers reject the first claim by arguing that ideal theory is unnecessary (Sen Reference Sen2006), misguided (Farrelly Reference Farrelly2007), or even pernicious (Mills Reference Mills2005). Though my argument raises questions about the coherence of Rawlsian ideal theory, the first claim is not my primary target here. My argument is against the second claim: I contend that, even as an ideal theory, political liberalism cannot assume that the citizens of an ideal liberal society would be overwhelmingly reasonable.

To set up my argument, I first need to say a bit more about how political liberalism's status as an ideal theory is supposed to enable it to set aside the problems of unreasonable pluralism that plague actual liberal societies. As an ideal theory, political liberalism aims to describe the principles, norms, and institutions that would govern an ideal liberal society. To do this, political liberalism assumes that certain favorable conditions obtain and asks what a just society would look like under those conditions. For Rawls, such favorable conditions include full compliance with the principles of justice and the persistence of social and economic conditions suitable for sustaining a just society (Reference Rawls1999a: 8, 216; Reference Rawls2001: 47). The purpose of these assumptions is not to represent accurately the world as it is, but rather to model the world as it could be under ‘the best foreseeable conditions’ (Reference Rawls2005: xvii; see also Simmons Reference Simmons2010; and Stemplowska and Swift Reference Stemplowska, Swift, Mandle and Reidy2014). This is necessary to ensure that the vision of an ideal society developed by ideal theory is not hostage to contingent features of existing societies that could, and perhaps should, change. In this regard, developing a vision of an ideal society ‘requires the use of counterfactual assumptions, since the current state of our institutions—and the resulting social and moral condition of the persons living under them … —is so distant from the best we can realistically hope to achieve’ (Simmons Reference Simmons2010: 31).

Political liberals recognize that unreasonable views often predominate in actual liberal societies (for example, Rawls Reference Rawls2005: 126). However, as Jonathan Quong explains, the relevant population for the purposes of developing and assessing political liberalism as an ideal theory is not ‘the diverse constituency of persons that currently inhabit modern liberal societies’ (Reference Quong2011: 138). Rather, the relevant population is the ‘reasonable citizens who would populate a well-ordered society according to a liberal conception of justice’ (Reference Quong2011: 153). In other words, what matters is not society as it is, but rather society as it could be and would be in an ideal liberal society. In assuming, for the purposes of ideal theory, that reasonable citizens are ‘dominant and controlling’, Rawls (Reference Rawls2005: 411n3) is not trying to represent society as it is; he is trying to model the conditions that he thinks would obtain in an ideal liberal society. The extent to which actual citizens fall short of this ideal ‘can be attributed to the fact that our current society is less than ideal or well ordered’ (Quong Reference Quong2011: 154).

2. The Fact of Reasonable Pluralism as a Concession to Realism

I now come to a crucial question: If an ideal theory of a free and just liberal society can assume that the citizens of such a society would be overwhelmingly reasonable, why cannot such a theory also assume that the citizens of such a society would freely reach agreement about the correct answers to moral and religious questions? Political liberals contend that any normative vision of an ideal liberal society must assume the fact of reasonable pluralism. This gives rise to the central problem that political liberalism aims to solve, that is, to explain how a society can be just and stable given that its citizens are profoundly divided by reasonable moral and religious disagreements (Rawls Reference Rawls2005: 4). The question is, why is this a problem that ideal theory has to tackle? Why cannot ideal theory assume that such reasonable disagreements would be rare, perhaps even nonexistent, in an ideal society?

To understand why, and how, political liberalism needs to answer this question, it is useful to briefly consider the dialectic between political liberalism and one of its primary rivals, perfectionism. According to perfectionism, society ought to be arranged to promote a particular conception of the good life. Political liberals argue that, given reasonable disagreement about the nature of the good life, maintaining such a perfectionist society would require the state to prevent people coercively from acting on their own divergent, yet reasonable, conceptions of the good (Rawls Reference Rawls2005: 37). And, at least according to political liberals, such coercion is incompatible with basic liberal commitments such as respect for people as free and equal citizens (Rawls Reference Rawls2005: 61–62; Larmore Reference Larmore1999). The important thing to notice for present purposes is that political liberalism's argument against perfectionism depends on assuming reasonable disagreement about the good. The argument is that given reasonable disagreement about the good, a perfectionist society will inevitably treat people in ways that run afoul of core liberal principles and values. But why must our ideal theories take such disagreement as a given? Why cannot a perfectionist grant that such disagreement is common in actual, nonideal societies but insist that, in an ideal society, everyone would freely converge on the same conception of the good?

Consider a perfectionist who claims that a particular form of individual autonomy is a core constituent of the good life and that society ought therefore to be arranged to promote such autonomy. The political liberal objects that, given reasonable disagreement about the nature of the good life, such a society would fail to respect people as free and equal citizens. Now suppose that the perfectionist responds by emphasizing that she is articulating a vision of an ideal liberal society. She grants that, in actual liberal societies, people reasonably disagree about the value of autonomy, but, she says, this is just one of the many ways in which actual liberal societies are nonideal. In an ideal liberal society, she says, everyone would freely converge on the truth about the good life so everyone would recognize the value of individual autonomy. In the society our perfectionist envisions, state efforts to promote individual autonomy would not wrongly coerce or disrespect people because, by hypothesis, everyone would agree that such autonomy is essential to the good life and wholeheartedly consent to the state's efforts to promote it. What, if anything, can a political liberal say is wrong with this vision of an ideal society? Rawls's answer, simply put, is that the society our imagined perfectionist envisions is too unrealistic (Reference Rawls2005: xvi, 39). This brings us to the limits of ideal theory.

While ideal theory can—indeed, must—assume that an ideal society would be different from our own in many ways, even ideal theory cannot set aside certain facts about people and society. Ideal theory, at least in its Rawlsian form, is supposed to be realistically utopian. A theory is realistically utopian when it ‘uses political (moral) ideals, principles, and concepts to specify a reasonable and just society’ (Reference Rawls1999b: 14)—the utopian aspect—that ‘is feasible and might actually exist, if not now then at some future time under happier circumstances’ (Reference Rawls1999b: 12)—the realistic aspect. To be realistically utopian, a theory must operate within ‘the fixed constraints of human life’ (Reference Rawls1999a: 216), which include ‘general facts of moral psychology’ (Reference Rawls1999a: 126).

Political liberals insist that the fact of reasonable pluralism is one of the fixed constraints of human life that any vision of an ideal liberal society must take into account. As Rawls puts it, ‘[the] fact of reasonable pluralism limits what is practically possible’ (Reference Rawls1999b: 12). For reasonable pluralism is not just a feature of actual liberal societies; it is ‘the inevitable long-run result of the powers of human reason at work within the background of enduring free institutions’ (Reference Rawls2005: 4). In other words, reasonable pluralism is an inescapable feature of any liberal society, even an ideal one, because reasonable pluralism is the inevitable result of people freely ‘exercising their reason in good faith and to the best of their abilities’ (Larmore Reference Larmore, Sobel, Vallentyne and Wall2015: 68). Since the free exercise of human reason reliably leads to widespread reasonable disagreement about a range of moral and religious matters, a society where all citizens freely converge on a shared conception of the good life is, Rawls says, ‘impossible’ (Reference Rawls2001: 3). This is why the society envisioned by our imagined perfectionist is not a plausible vision of an ideal liberal society; it is not a society that is ‘feasible and might actually exist’ (Reference Rawls1999b: 12).

Why should we accept these claims? Why can we not hold out hope that, at least in the long-run, reasonable people will freely converge on the truth about how we ought to live (or, as the case may be, on the conclusion that there is no such truth to be had)? Rawls answers by pointing to what he calls the burdens of judgment: the many obstacles to discerning the truth and reaching agreement with others that are inescapable in the course of deliberating about moral, religious, and philosophical issues. These obstacles include the difficulty of assessing complex and conflicting evidence; disagreements about how to weigh conflicting evidence and considerations; the vagueness of our concepts and the resulting need for interpretative judgments; the many ways that different life experiences shape our reasoning and judgments; the difficulty of making all-things-considered judgments when there are competing normative considerations on either side; and the difficulty of prioritizing values when our institutions can only realize some at the expense of others (Reference Rawls2005: 56–57). Ultimately, ‘it will simply prove impossible, Rawls argues, for rational people to overcome the burdens of judgment and all arrive at some common ethical, religious, or philosophical framework’ (Quong Reference Quong2011: 37). Thus, we must assume that reasonable pluralism is an inescapable feature of any free society, even an ideal one.

My aim in the next two sections is to show than an analogous argument can be made with regard to unreasonable pluralism. I argue that human reason is beset by obstacles that make a kind of unreasonable pluralism an inevitable outcome of the free exercise of human reason. If I am right, then political liberalism cannot assume that the citizens of an ideal liberal society would be overwhelmingly reasonable, at least not without running afoul of its crucial commitment to be realistically utopian. The problems posed by unreasonable pluralism are thus problems that political liberalism, even as an ideal theory, must address head on.

3. Sources of Unreason and Types of Unreasonableness

When political liberals talk about what it takes to be reasonable, they usually focus on the motivations that are supposed to characterize reasonable people. Reasonable people are said to have the motivation ‘to propose fair terms of social cooperation and to abide by them provided others do,’ as well as a ‘willingness to recognize the burdens of judgment’ (Rawls Reference Rawls2005: 54). But possessing these dispositions is not sufficient to be reasonable; one must also reason about moral and political issues in a way that is free from bias, irrationality, and other such distorting influences. Rawls writes,

Indeed, political liberals often emphasize that reasonable pluralism is not the result of bias, irrationality, or other such causes precisely because they take such sources of disagreement to be inimical to reasonableness (for example, Rawls Reference Rawls2005: 55; Larmore Reference Larmore1994: 75; Quong Reference Quong2011: 37). I call such causes the sources of unreason. Following Rawls (Reference Rawls2005: 55, 58), I take them to include the following:

1. a. our prejudices and biases, both explicit and implicit, that skew our beliefs and perceptions in unjustifiable ways

2. b. the tendency of self- and group-interest to distort our judgments, especially those about justice and fairness

3. c. stubbornness and dogmatism that prevent us from changing our minds even when presented with sufficient reason to do so

4. d. irrationality and fallacious reasoning

The sources of unreason are like the burdens of judgment in that they are obstacles to discerning the truth and reaching agreement. But unlike the burdens of judgment, the sources of unreason are also obstacles to being reasonable.

Why are the sources of unreason inimical to reasonableness? One answer is that reasonableness includes an epistemic component: to be reasonable, one's beliefs must respond to reasons and evidence in (more or less) epistemically rational ways. This seems to be Rawls's view, at least at times (Reference Rawls2005: 59, 62). As Samuel Freeman notes, ‘Rawls defines ‘reasonable comprehensive doctrines’ epistemically, as doctrines that are responsive to evidence and possess certain other theoretical features’ (Freeman Reference Freeman2007: 346; my emphasis). Since the sources of unreason undermine responsiveness to evidence, they thereby undermine reasonableness as well.

However, some political liberals object to the epistemic component of reasonableness. They contend that views that are epistemically unreasonable—that is, views that are not sufficiently supported by reasons and evidence—nonetheless ought to be regarded as politically reasonable so long as they are consistent with the fundamental values of a liberal political regime. For example, Erin Kelly and Lionel McPherson argue that even ‘views that have little or no rational support or seem plainly irrational’ ought to count as politically reasonable so long as they are consistent with accepting fair terms of social cooperation among free and equal people (Reference Kelly and McPherson2001: 39). Similarly, Martha Nussbaum (Reference Nussbaum2011) argues that a wide range of popular doctrines, from astrology to Christianity, would wrongly be counted as unreasonable if assessed on their epistemic credentials.

Do the sources of unreason undermine political, as opposed to just epistemic, reasonableness? Yes. The sources of unreason not only undermine our responsiveness to evidence, but they also distort our sense of justice. When our beliefs about what is just, fair, or right are shaped by bias, self- or group-interest, dogmatism, or irrationality we are liable to support unjust (and often self-serving) laws, policies, and institutions because we erroneously believe them to be just. In such cases, our beliefs are unreasonable because they are practically inconsistent with accepting fair terms of social cooperation or with respecting others as free and equal persons. In other words, when our moral and political views are shaped by the sources of unreason, they are liable to be not only epistemically unjustified, but also at odds with any reasonable conception of justice.

It is important to recognize that the sources of unreason can cause people to hold unreasonable moral and political beliefs even when they possess a sense of justice and the moral motivations that political liberals typically ascribe to reasonable people. People can have ‘the capacity to understand, to apply, and to act from’ a conception of justice (Rawls Reference Rawls2005: 19), as well as the motivation ‘to propose and abide by fair terms of cooperation’ (Rawls Reference Rawls2005: 86), yet nonetheless hold deeply misguided beliefs about what constitute fair terms of cooperation because their reasoning is distorted by bias, irrationality, or the like. To appreciate this, it is useful to distinguish two different forms that unreasonableness can take.

First, people can be bereft of the basic moral commitments and motivations that are partially constitutive of reasonableness. Freeman describes unreasonable people this way; he says they ‘do not have a sense of justice or other moral dispositions’, have ‘no respect for others’, and are ‘hardly fit for social life’ (Reference Freeman2004: 2049). I call this dispositional unreasonableness to mark the fact that people who are unreasonable in this sense lack the basic moral dispositions in terms of which Rawlsian political liberals characterize reasonable people. When I say that the sources of unreason undermine reasonableness, I do not mean that they turn us into dispositionally unreasonable people of the sort that Freeman describes. My point is that one need not be such a person to hold unreasonable moral or political views.

One can have the basic moral commitments and motivations of a reasonable person—respect for others, a concern for justice, and so on—but nonetheless hold unreasonable beliefs about moral and political issues because one's reasoning is distorted by the sources of unreason. It is this kind of unreasonableness, which I call judgment unreasonableness, that I am primarily concerned with here. I emphasize that my point is not just that reasonable people can hold irrational or unwarranted beliefs—as argued by Joseph Raz (Reference Raz1998) and Gerald Gaus (Reference Gaus1999). The key point here is that otherwise reasonable people can hold beliefs that are not only epistemically problematic, but also politically unreasonable in the sense that they are at odds with any reasonable conception of justice (see Badano and Nuti Reference Badano and Nuti2018 for a similar point).

Judgment unreasonableness can manifest itself at many different levels, from interpretations of abstract principles and values to judgments about how those principles or values apply to particular laws, policies, institutions, decisions, or actions. One might believe that freedom, equality, and fairness are of paramount importance, but nonetheless interpret or weigh those values in unreasonable ways. Or one might affirm reasonable principles of justice but hold unreasonable beliefs about their practical implications or about whether particular laws, policies, or institutions satisfy them. Often (but not always) such judgment unreasonableness will be mediated by false nonmoral beliefs (see, in a different context, Buchanan Reference Buchanan2002). For example, one might affirm the importance of equality of opportunity, but deny that this requires taking further action to address racial inequality in the United States because one is among the nearly 50 percent of Americans who believe that ‘blacks who can't get ahead in this country are mostly responsible for their own condition’ (Pew Research Center Reference Center2017: 34).

Judgment unreasonableness, no less than dispositional unreasonableness, is a threat to the justice and stability of a liberal society. For a society is unlikely to realize stably a reasonable conception of justice if a large portion of the population holds misguided beliefs about what that conception requires in practice. I return to this problem in section 5, but first I explain why I believe judgment unreasonableness is bound to pervade any liberal society, even an ideal one.

4. The Obstinacy of Unreason

Political liberals need not deny that the sources of unreason—bias, irrationality, dogmatism, and the like—routinely permeate people's reasoning about politics. Their claim is not that the citizens of actual societies are overwhelmingly reasonable; rather, it is that the citizens of an ideal (or ‘well-ordered’) liberal society would be so. To assess this claim, we have to ask to what extent we can plausibly expect the citizens of an ideal liberal society to overcome the sources of unreason. In other words, to what extent can the bias and irrationality that people so often exhibit when it comes to politics be attributed to the fact that our current societies are unjust and nonideal in so many ways?

Rawls suggests that, though unreasonable views may be common in actual societies, their popularity would diminish over time in a society organized in accordance with a reasonable conception of justice (Reference Rawls2005: 158–68). He thinks that once a just basis for society is established ‘simple pluralism moves toward reasonable pluralism’ (Reference Rawls2005: 164). But Rawls does not tell us how life in a just society would reduce the extent to which the sources of unreason distort people's judgments about moral and political issues. He suggests that people will recognize the goods accomplished by liberal principles and values and thereby come to share a basic commitment to those principles and values (Reference Rawls2005: 163), but as I have argued, such a commitment is compatible with holding unreasonable views about how to interpret or apply those principles and values in practice. In other words, even if we accept Rawls's speculative story about how life in a just society would dramatically reduce dispositional unreasonableness, we are still missing a story about how life in such a society would tend to eliminate, or at least dramatically reduce, judgment unreasonableness.

One might suggest that distorted reasoning about moral and political issues would be rare if people were raised under just institutions, motivated to think through the issues more carefully, and equipped with the education necessary to do so. However, the empirical evidence provides several reasons to be skeptical of this conjecture. For one thing, there is considerable evidence that, far from being an antidote, education actually makes people more prone to distorted political reasoning. For example, several studies have found positive correlations between biased political reasoning and many of the core skills and competencies that education aims to impart including critical reflectiveness (Kahan Reference Kahan2013), knowledge of logic (Čavojová, Šrol, and Adamus Reference Čavojová, Šrol and Adamus2018), numeracy (Kahan et al. Reference Kahan, Peters, Wittlin, Slovic, Ouellette, Braman and Mandel2012; Reference Kahan, Peters, Dawson and Slovic2017), and scientific literacy (Kahan et al. Reference Kahan, Peters, Wittlin, Slovic, Ouellette, Braman and Mandel2012). Generally, greater knowledge and reasoning abilities seem simply to ‘make it easier for citizens to defend their political attitudes through motivated bias’ (Lavine, Johnston, and Steenbergen Reference Lavine, Johnston and Steenbergen2012: xiv). This helps to explain why the most educated people also tend to be the most politically polarized. Indeed, the empirical evidence suggests that ‘as education levels in the population continue to increase, we should expect … partisan-ideological polarization to increase as well’ (Abramowitz Reference Abramowitz2010: 127).

The root of the problem lies in human reason itself. Our reasoning capacities are beset by an array of built-in cognitive and affective biases that make it very difficult—often practically impossible—for us to think clearly and objectively about issues that affect our interests, divide us into competing groups, and arouse our passions, as politics inevitably does (Lodge and Taber Reference Lodge and Taber2013). Collectively, several of these biases result in what psychologists call motivated reasoning: the tendency to seek out, interpret, evaluate, and weigh evidence and arguments in ways that are systematically biased toward conclusions that we ‘want’ to reach for reasons independent of their truth or warrant (Kunda Reference Kunda1990; Ditto, Pizarro, and Tannenbaum Reference Ditto, Pizarro, Tannenbaum, Bartels, Bauman, Skitka and Medin2009).

Motivated reasoning is incredibly difficult to avoid because we are introspectively blind to it. Even when we look for the bias in our own reasoning, we rarely find it (Pronin, Lin, and Ross Reference Pronin, Lin and Ross2002; Frantz Reference Frantz2006). It typically seems to us that we are being evenhanded and objective even while we are twisting the arguments and evidence to make them fit the conclusions we ‘want’ to reach. Psychologists call this the illusion of objectivity (Kunda Reference Kunda1990: 482–83). Notably, this illusion is most robust under exactly the conditions created by the burdens of judgment: conditions where there is complex and conflicting evidence, where we must make interpretive judgments applying vague concepts, and where there are competing normative considerations on both sides of the issue. Such conditions ‘leave a motivated moral judge considerable flexibility to construct plausible justifications for preferred moral conclusions without offending their sense of their own objectivity’ (Ditto et al. Reference Ditto, Pizarro, Tannenbaum, Bartels, Bauman, Skitka and Medin2009: 314).

Hugo Mercier and Dan Sperber (Reference Mercier and Sperber2017) argue that human reason evolved to be systematically biased because collective reasoning is more effective and efficient when individual group members focus narrowly on articulating and defending different answers to collective problems. The trouble is that these beneficial effects generally occur only in conditions that are quite unlike many moral and political debates—conditions where people generally agree about what a correct answer would look like, and where they have few incentives to arrive at incorrect answers. Indeed, most of the evidence for such beneficial effects come from experiments in which groups are tasked with, and rewarded for, correctly solving logic or math puzzles. When what would count as a correct answer is in dispute, or when people have strong incentives to reach incorrect answers—as is often the case in politics—our biased faculties of reason regularly provide ‘at best, inert rationalizations and, at worst, excuses that allow the reasoner to engage in morally dubious behavior’ (Reference Mercier and Sperber2017: 314).

How can motivated reasoning lead people to hold views that are inimical to justice? As Hilary Kornblith observes,

To take just one simplified example: people whose financial self-interest would be better served by an unjust redistributive scheme are liable to favor that scheme and, through motivated reasoning, erroneously come believe that it is just. More generally, when acting justly would be costly to us or detrimental to our interests, our judgments about what justice requires often get warped. We are apt to judge that whatever promotes our own interests or the interests of our group is consistent with or even required by justice (Batson Reference Batson2016). Of course, interests—and especially financial interests—are far from the whole story. Indeed, financial self-interest tends to influence people's political views only on a narrow—though, from the standpoint of distributive justice, crucial—range of issues like taxation where personal financial costs are direct and apparent (Chong, Citrin, and Conley Reference Chong, Citrin and Conley2001). The trouble is that even when people's narrow self-interest does not exert much influence on their political views, their reasoning is typically no less biased; it is simply biased by other factors.

Among the most ubiquitous sources of bias are loyalties to in-groups and hostility toward out-groups. An abundance of evidence shows that people's views about a vast range of issues from welfare policies to healthcare to border control to climate change are shaped in troubling ways by their allegiances and animosities toward various social groups including racial groups, religious groups, and political parties (Gilens Reference Gilens1999; Cohen Reference Cohen2003; Kinder and Kam Reference Kinder and Kam2009; Kahan et al. Reference Kahan, Peters, Wittlin, Slovic, Ouellette, Braman and Mandel2012; Tesler Reference Tesler2012; Achen and Bartels Reference Achen and Bartels2016). One might suggest that this is indicative of unjust inequalities between groups or defective political institutions. While such factors surely play a role, the root of the problem is much deeper.

In The Federalist No. 10, James Madison observes that we are so strongly predisposed to divide ourselves into competing groups that even ‘the most frivolous and fanciful distinctions’ can become the bases of hostile factions (Reference Madison and Rossiter1961: 79). This observation is borne out by empirical work on so-called minimal groups that finds that even imagined trivial differences between groups—for example, being falsely told that one group prefers paintings by Kandinsky while the other prefers paintings by Klee—is sufficient to elicit not only preferential treatment of in-group members and but also hostility toward out-group members (Tajfel et al. Reference Tajfel, Billig, Bundy and Flament1971). More recently, evolutionary anthropologists and psychologists have argued that humans are a tribal species, and that many social and political divisions can be understood as products of an innate tendency to conceptualize our social world in terms of competing groups (Richerson and Boyd Reference Richerson, Boyd and Nesse2001; Kurzban, Tooby, and Cosmides Reference Kurzban, Tooby and Cosmides2001; van Vugt and Park Reference van Vught, Park, Stürmer and Snyder2009). While we can hope that some forms of hostility between groups, such as explicit racial prejudices, would diminish in a society where unjust social and economic inequalities between groups were absent and people were raised to see one another as equals, there is no reason to believe that partisanship and other triggers of group-based biases would disappear in an ideal liberal society.

In highlighting these features of human psychology, I do not mean to suggest that we cannot make social progress or overcome whatever unreasonable moral and political views happen to be popular at any given time. I am not defending the status quo or saying we cannot do better. My point is just that we face formidable obstacles to thinking rationally and impartially about moral and political issues. Since these obstacles arise from normal features of human psychology together with features endemic to politics, there is no reason to believe that they will disappear in an ideal liberal society. Nor, unfortunately, is there much reason to believe that the citizens of such a society will reliably overcome them—or, at least, there is no more reason to believe that such citizens would overcome the sources of unreason than there is to believe they would overcome the burdens of judgment.

Of course, people can and do overcome the sources of unreason some of the time; but the same is true of the burdens of judgment. The existence of complex and conflicting evidence does not make it impossible for us to sometimes converge on the truth. Nor do our differing life experiences preclude us from ever agreeing. The reason the burdens of judgment inevitably create reasonable pluralism is not that they are absolutely insurmountable in every case, but rather that we cannot overcome them reliably enough to settle many disagreements about how we ought to live. I submit that the same is true of the sources of unreason. Just as the sincere and conscientious exercise of our capacities of reason does not reliably lead to truth and agreement, it also does not reliably lead us to affirm views that are consistent with the demands of a reasonable conception of justice. To assume that unreasonable moral and political views would be rare or nonexistent in an ideal liberal society is therefore unrealistic for the same reasons it is unrealistic to assume that reasonable disagreement would be rare or nonexistent. To be realistically utopian, political liberalism must assume that the moral and political reasoning of citizens in any liberal society, even an ideal one, will often be distorted by the sources of unreason in ways that lead them to affirm unreasonable moral and political views. In other words, political liberalism must assume what I have called the fact of unreasonable pluralism.

5. Stability and Containment

I have argued that political liberalism cannot assume that the citizens of an ideal liberal society would be overwhelmingly reasonable without running afoul of its own constraints of realism. To what extent does this undermine political liberalism as a whole? I think the answer depends, in large part, on the answer to a question that political liberals have largely neglected: Can a liberal society effectively contain—that is, limit the influence and spread of—unreasonable moral and political views? Rather than attempt to answer this question here, I explain why I think the fate of political liberalism depends on this question and say a bit about the challenges that any affirmative answer must face.

Political liberalism aims to explain how a just and stable liberal society is possible ‘under realistic, though reasonably favorable, conditions’ (Rawls Reference Rawls2001: 13). Such conditions include the existence of widespread reasonable disagreement about morality and religion because such disagreement is an inevitable result of the free exercise of human reason. This raises the question that political liberals have focused most of their attention on: how is a just and stable society possible given such reasonable disagreement—that is, given the fact of reasonable pluralism? Rawls answers that such a society is possible by way of what he calls an overlapping consensus wherein all reasonable citizens affirm a shared conception of justice for reasons grounded in, or at least consistent with, their own reasonable moral, religious, and philosophical views (Reference Rawls2005: 133–72). But such an overlapping consensus only makes for a just and stable society if unreasonable views ‘do not gain enough currency to undermine society's essential justice’ (Rawls Reference Rawls2005: 39).

I have argued that, even in an ideal liberal society, citizens’ reasoning about moral and political issues will often be distorted in ways that lead them to affirm unreasonable views. This can undermine an overlapping consensus in two ways. First, it may lead citizens to reject reasonable conceptions of justice in favor of unreasonable conceptions that impose unfair (and perhaps self-serving) terms of cooperation. If enough citizens accept such unreasonable conceptions, then no reasonable conception of justice will be able to gain sufficient support to be the target of an overlapping consensus. Second, even if all citizens endorse a reasonable conception of justice at some abstract level, that conception is unlikely to be stably realized by their society insofar as many citizens hold unreasonable views about the institutions, laws, policies, and actions necessary to realize it. Simply put, the more citizens affirm unreasonable moral and political views, the less likely it is that an overlapping consensus can stably order a liberal society.

If, as I have argued, political liberalism's own constraints of realism require it to assume that unreasonable moral and political views will be common in any liberal society, then political liberalism must explain how a liberal society can prevent such views from gaining sufficient influence and support to thwart or undermine an overlapping consensus on a reasonable conception of justice. If political liberals cannot do this, then they cannot accomplish their goal of showing how a just and stable society is possible under realistic, but favorable, conditions. This would not, by itself, show that we should reject political liberalism outright, but it would show that the theory falls short of delivering all that it promises. And if political liberalism cannot deliver the robust stability it promises, then it may lose some of its appeal relative to alternative conceptions of liberalism, such as those that seek to base a liberal society on a modus vivendi or compromise between disagreeing parties (for example, Bellamy Reference Bellamy1999; McCabe Reference McCabe2010; and Horton, Westphal, and Willems Reference Horton, Westphal and Willems2019).

In a footnote, Rawls says that the existence of unreasonable views gives us ‘the practical task of containing them—like war and disease—so that they do not overturn political justice’ (Reference Rawls2005: 64n19). But, aside from a brief discussion in A Theory of Justice where he suggests that penal devices may play a role (Reference Rawls1999a: 502–5), Rawls never takes up the question of how the liberal institutions he thinks are incapable of preventing reasonable pluralism could effectively contain unreasonable pluralism. What I've just tried to show is that political liberalism cannot accomplish all that it sets out to without an account of how such containment can be reliably achieved. This is not the place to undertake a thorough assessment of the prospects of such an account; for now, I will simply say a bit about the challenges that any such account must overcome.

There are two ways a liberal society could contain unreasonable views: first, through state intervention, and second, through interventions by citizens or nongovernmental associations. Quong—one of the few political liberals to take up the task of articulating means and justifications for containment—focuses on the former. He argues that the state is at least sometimes justified in intervening to prevent schools from teaching unreasonable views and, more generally, intervening to prevent the expression of such views—especially when such expression constitutes hate speech (Reference Quong2011: 299–312). These are sensible proposals, and I grant that that such state interventions are at least sometimes justified and effective. However, I suspect that state interventions will not prove to be justified or effective in enough cases to prevent unreasonable moral and political views from gaining sufficient influence to undermine the justice of society. This is so for two reasons.

First, as Quong acknowledges, even when state intervention is justified in principle, the potential for the abuse of state power to suppress political opposition or oppress unpopular minorities should lead us to adopt ‘a strong presumption in favor of non-interference’ (Reference Quong2011: 305; see also Badano and Nuti Reference Badano and Nuti2018: 153–55). Second, relying on the state to contain unreasonable views seems to presuppose that control of the state is firmly in the hands of reasonable citizens, and thus that unreasonable views are already largely contained. People who hold unreasonable moral and political views are liable to vote for politicians who hold similar views. So, if unreasonable moral and political views are popular, the government of a democratic society is liable to reflect those views rather than act to contain them. The fact that government officials are liable to hold many of the same unreasonable views that are popular in society also gives us further reason to adopt a strong presumption in favor of nonintervention. This is not to say that state intervention cannot be part of the solution, but it seems to require that a large part of the problem has already been solved.

What about efforts by citizens and nongovernmental associations? Clayton and Stevens, focusing on unreasonable religious doctrines, suggest that reasonable religious citizens might be morally required to try to persuade fellow believers to adopt more reasonable versions of their religious doctrines (Reference Clayton and Stevens2014: 82). More recently, Gabriele Badano and Alasia Nuti have suggested that the citizens of a liberal society have a ‘duty of pressure’ which requires ‘that ordinary reasonable citizens press the unreasonable they know (e.g. relatives, friends, and colleagues) to change their mind and push them toward greater reasonableness’ (Reference Badano and Nuti2018: 157). These proposals are not susceptible to concerns about the abuse of state power, but I fear that their effectiveness will be severely limited by many of the same sources of unreason that give rise to unreasonable pluralism in the first place.

A wealth of empirical evidence shows that people tend to evaluate political arguments and evidence in systematically biased ways (Lord, Ross, and Lepper Reference Lord, Ross and Lepper1979; Taber and Lodge Reference Taber and Lodge2006; Kahan et al. Reference Kahan, Peters, Dawson and Slovic2017; Stanley et al. Reference Stanley, Henne, Yang. and De Brigard2019). People uncritically accept arguments and evidence that confirm their existing views, but when presented with counterevidence they ‘routinely rationalize the facts, figures, and arguments that they cannot effortlessly discount, depreciate, denigrate, or deny’ (Lodge and Taber Reference Lodge and Taber2013: 59). As a result, efforts to persuade people to change their moral and political views by presenting them with arguments and evidence are usually ineffective. As Leon Festinger and his colleagues observed in their pathbreaking study of religious fundamentalism, ‘even when presented with evidence, unequivocal and undeniable evidence, that his belief is wrong … the individual will frequently emerge, not only unshaken, but even more convinced of the truth of his beliefs than ever before’ (Reference Festinger, Riecken and Schachter1956: 3). Furthermore, efforts to persuade people by other means, such as calling their views ‘racist’ or ‘fascist’ (Badano and Nuti Reference Badano and Nuti2018: 160) may simply provoke defensiveness, retrenchment, and feelings of victimization rather than positive changes in people's beliefs. This is not to suggest that all efforts to persuade people to adopt more reasonable views are bound to fail, nor that we should give up trying. My point is just that there are formidable obstacles that are liable to severely limit the effectiveness of efforts to convert people to reasonable views through dialogue and debate (for more on this point, see Bagg Reference Bagg2018).

Much work remains to be done to determine whether a liberal society can reliably prevent unreasonable moral and political views from gaining enough support to thwart or undermine an overlapping consensus. My aim in this section has simply been to highlight some challenges that a liberal society faces in trying to do so, and to suggest that the success of political liberalism depends on our ability to overcome these challenges. If it turns out that we cannot overcome them, then a liberal society based on a stable overlapping consensus is not a society that ‘is feasible and might actually exist’ (Rawls Reference Rawls1999b: 12). In other words, in the absence of realistic and reliable means of containing unreasonable moral and political views, political liberalism's vision of an ideal liberal society cannot be realistically utopian.

6. Consistency and Realism in Ideal Theory

In addition to its implications for political liberalism's account of stability, my argument also raises questions about the coherence of Rawlsian ideal theory—that is, about the coherence of a theory that aims to be both realistic and utopian. I have argued that Rawlsian ideal theory's constraints of realism require it to assume that unreasonable pluralism rooted in bias and irrationality is an inevitable feature of an ideal liberal society. While such a society is realistic, it is unclear that it is utopian—that is, it is unclear that it is ‘a reasonable and just society’ (Reference Rawls1999b: 14). Insofar as unreasonable moral and political views are inimical to justice, how could a society in which such views are popular possibly be the kind of ‘perfectly just’ society that ideal theory aims to describe (Reference Rawls1999a: 216)?

This worry becomes more pressing if my basic argumentative strategy generalizes to other standard assumptions of Rawlsian ideal theory. Consider, for example, Rawls's assumption of full compliance with the demands of justice by all citizens. Rawls believes that full compliance is not unrealistic because

However, one could argue that a society where everyone complies with the demands of justice is no more realistic than a society where everyone overcomes the burdens of judgment and converges on the truth about morality and religion. In fact, the sources of unreason provide grounds for an argument that widespread noncompliance is an inevitable feature of any liberal society: Insofar as people's political reasoning is distorted by the sources of unreason, they are liable to hold misguided beliefs about what justice requires in practice. And insofar as people hold misguided beliefs about what justice requires in practice, they are unlikely to do what justice actually requires—that is, they are unlikely to comply with the demands of justice. Moreover, insofar as compliance is supposed to be driven by reciprocity—by the motivation to abide by fair terms ‘provided others can be relied on to do the same’ (Rawls Reference Rawls2005: 54)—noncompliance resulting from the sources of unreason will drive further noncompliance as reciprocity breaks down. So if, as I have argued, assuming that people will overcome the sources of unreason runs afoul of political liberalism's own constraints of realism, then so too may the assumption of full compliance.

The general problem here is that ideal theory's constraints of realism must be applied in a principled and consistent way. We cannot arbitrarily pick and choose which nonmoral facts to treat as constraints and which to idealize away. Once political liberals invoke constraints of realism to explain why ideal theory must assume reasonable pluralism, they must apply those constraints consistently to their own vision of an ideal liberal society. It may turn out that the constraints necessary to force ideal theorists to confront the fact of reasonable pluralism, when applied consistently, also force ideal theorists to confront much of what is usually thought to be the domain of nonideal theory. And if it turns out that ideal theory must deal with partial compliance, unreasonable pluralism, and a host of other seemingly ‘nonideal’ problems, then it is unclear in what sense ideal theory remains ideal theory.

At this point, one might suggest that the problem lies in trying to make ideal theory realistically utopian in the first place. Perhaps theorists like G. A. Cohen (Reference Cohen2008) are right in rejecting Rawls's constraints of realism and insisting that theories of justice need not accommodate any nonmoral facts. Accepting a view like Cohen's keeps the fact of unreasonable pluralism out of the domain of facts that our ideal theories must assume; but it keeps all other nonmoral facts out as well, including the fact of reasonable pluralism. The problem for political liberals is that one cannot do away with Rawls's constraints of realism without at the same time eliminating the need for ideal theory to yield to the fact of reasonable pluralism. For without Rawls's constraints of realism, there is no reason that an ideal theory of a liberal society cannot assume that all citizens will freely converge on shared answers to moral and religious questions. To put the point a bit differently, rejecting Rawls's constraints of realism leaves political liberalism with no way of rejecting the society envisioned by the perfectionist we imagined in section 2. Thus, rejecting the realistic aspect of realistic utopianism is not an option for political liberals, nor for anyone else who believes that ideal theory must take account of the fact of reasonable pluralism.

The challenge for political liberals is to show that there are justified and nonarbitrary constraints of realism that simultaneously require our theories to assume reasonable pluralism while also licensing the idealizations that political liberals must make in order to articulate a vision of a fully just society. I leave it as an open question whether this can be done.