1 Introduction

Consider the problem of finding a justification of liberal democracy that meets the liberal principle of legitimacy: fundamental political principles, laws and institutions, to be legitimate, must be justifiable from the point of view of all reasonable citizens. At the same time, this justification must meet the challenges posed by the fact of reasonable pluralism: in a free society people will always subscribe to a plurality of reasonable yet conflicting comprehensive doctrines (Rawls 2005, p. 59), we should therefore not expect that any one comprehensive doctrine can be accepted by all reasonable citizens.

John Rawls famously drew the conclusion from this purported fact about free societies that since supporters of liberal legitimacy want our fundamental laws and political principles to be justifiable from all reasonable citizens’ points of view, no one comprehensive doctrine can legitimately ground the justification for a free society’s basic political structure.

Rawls proposes a solution to this problem, involving the idea that a justification must be ‘freestanding’ (Rawls 2005, p. 12). The justification of liberal democracy must avoid implying or relying on any specific comprehensive doctrine, and yet remain compatible with the full range of reasonable worldviews. Rawls developed a theoretical framework for this kind of political justification. Many commentators have, nevertheless, remained sceptical of the consistency and feasibility of Rawls’s political liberal solution. Much critique has been directed at the theory’s concept of reasonableness which is vague and, according to many critics, excludes too many or much from the realm of justification.

Recently, Robert Talisse has proposed the highly interesting suggestion that there is an epistemic justification of the core institutions of liberal democracy. Rather than basing political justification in our moral commitments we should, argues Talisse, ground the justification of liberal democracy in our common epistemic commitments. In so doing we will, according to this theory, expand the realm of justification and thus overcome some of the problems inherent in the political liberal solution. Roughly his idea is this: simply in virtue of holding beliefs one is rationally committed to certain basic norms of evidence and reason-responsiveness. In virtue of this commitment, one will also be rationally committed to endorse the institutional arrangements securing a free exchange of reasons and evidence. Such institutions are partly constitutive of liberal democracy. Hence, when reflecting on our most basic epistemic commitments, we see that we are rationally committed to upholding these liberal democratic institutions. This holds no matter which moral or religious worldview one subscribes to. The argument thus purports to be a public justification, by which we mean a justification that can be endorsed from all reasonable citizens’ points of view. Talisse categorizes this kind of view as social epistemic liberalism, and argues that this is a kind of liberalism that breaks with central strictures of Rawlsian political justification, but is nonetheless still able to accommodate the fact of reasonable pluralism in virtue of being a public justification that crosses the boundaries of reasonable moral, philosophical and religious worldviews (Talisse 2008).

Our main focus of this paper will be Talisse’s epistemic justification of liberal democratic institutions, and we will argue that while interesting and important, it ultimately fails as a successful public justification. We attribute this failure to the obstacle of deep epistemic disagreements: a term which in the context of this discussion will refer to disagreements that individuals may have concerning which basic epistemic norms and principles are truth conducive in a given domain.Footnote 1

Before engaging with this argument, it is worth noting how Talisse’s justification differs from similar approaches. The general idea that institutions facilitating open reason-exchange can be defended epistemically on part of their corrective features was emphasised by John Stuart Mill, who argued that human judgment can be relied on only when the’means of setting it straight are kept constantly at hand’ (Mill 1870, p. 12). Here Mill refers to the institutions of freedom of thought and discussion, facilitators of proper inquiry under which ‘wrong opinions and practices gradually yield to fact and argument’ (Mill 1870, p. 12). The gist of Mill’s epistemic argument is that in the collision of adverse opinions, that will take place under free discussion, false beliefs will gradually be corrected, half truths will become whole and whole truths, withstanding thorough scrutiny, will be better understood.

However, for Mill, the pursuit of free and proper inquiry ultimately matters because it is crucial to developing individuality, and the development of individuality is an essential human good. Mill’s theory is thus perfectionist in the sense that it asserts a value that, in Mill’s view, applies to all. According to Rawls, Mill’s theory therefore ’fails to satisfy, given reasonable pluralism, the constraints of reciprocity, as many citizens […] may reject it’ (Rawls and Freeman 1999, p. 586). Thus, from a political liberal perspective the problem with Mill’s argument is that it cannot be reasonably expected to be endorsed by all reasonable citizens.

Allen Buchanan has offered a socio-epistemic justification of core liberal democratic institutions that he takes to be compatible with political liberalism (Buchanan 2004). Buchanan argues that any rational citizen will be concerned with reducing the risks associated with socially instilled false beliefs and therefore will endorse key liberal institutions for their epistemic benefits. The general idea in Buchanan’s argument is that we are, due to our limited cognitive abilities, socially dependent on other people and institutions for true beliefs. This dependency puts us at risk of adopting false beliefs. Now Buchanan argues, quite reasonably, that avoiding false beliefs and acquiring true beliefs about the world is important if we want to ‘ensure that we act appropriately on prudential or moral principles’ (Buchanan 2004, p. 108), whatever these principles may be. Therefore, any rational citizen concerned with furthering her moral principles and well-being will want to minimize the epistemic risks associated with our social epistemic dependency. This epistemic risk is best managed when liberal institutions securing free flow of information, positions of epistemic authority based on merit, equal opportunity, and accountability on part on the government and epistemic authorities in society, are in place. Therefore, whatever our moral commitments may be, if we take them seriously, we will also commit to these key liberal institutions. Buchanan claims that this socio-epistemic argument for key liberal institutions is compatible with the strictures of political liberalism, since the argument can be made without relying on any comprehensive moral conception. Buchanan’s socio-epistemic argument relies only on the assumption that one cares about reducing prudential and moral risk, not upon any particular specification of what welfare (or happiness) is or any particular conception of morality (Buchanan 2004, p. 100).

Talisse and Buchanan agree on this starting point: that a justification based on epistemic commitments can cross the boundaries of diverse and conflicting moral worldviews. However, Talisse points out that to the extent that the socio-epistemic justification assumes that we all have the same, overriding epistemic reason to endorse liberal democracy, the justification does transgress constraints of political liberalism. According to Talisse, Buchanan’s justification must be based on a (at least partially) comprehensive philosophical doctrine ‘about the epistemic good and its relation to moral judgement’ of the kind that the political liberal must seek to avoid (Talisse 2008, p. 114). So, Buchanan’s justification is not ‘freestanding’ in the sense that individuals find reasons from within their respective comprehensive worldviews to support liberal democratic institutions. Rather, individuals are offered the one and same epistemic reason to endorse these institutions. In Rawls, a worldview (or doctrine) is comprehensive as soon it transgresses the political and includes non-political values and virtues (Rawls 2005, p. 175) and the socio-epistemic justification must include such a value: to remain forceful Buchanan’s justification must commit itself to the overriding value of epistemic risk reduction (Talisse 2008, p. 114). Buchanan’s argument thus fails, according to Talisse, in its attempt to seek compatibility with political liberalism.

Whether Buchanan’s argument is really open to this criticism or not, Talisse proposes an epistemic justification of liberal democracy that does not strive for compatibility with political liberalism, but nonetheless purportedly accommodates the justificatory challenge that accompanies a recognition of the fact of reasonable pluralism. Talisse’s proposed justification in fact purports to be a counter-example to Rawls’s crucial premise in political liberalism: that no one comprehensive doctrine can be endorsed by all reasonable citizens (Talisse 2008). Talisse’s claim is that an epistemic doctrine based merely on a commitment to certain very basic epistemic norms can be endorsed by all reasonable citizens, and thus form a basis for a public justification of the core liberal institutions. In this paper we argue that Talisse fails to provide a successful public justification of core liberal institutions.

2 The Socio-Epistemic Justification of Liberal Democracy

Let us first consider Talisse’s argument. Here is our reconstruction:

  1. (T1)

    Rational believers operate in compliance with a set of basic epistemic norms (EN). Roughly EN comprises the idea that rational believers aim at believing truly, and that in doing so, they need to be responsive to evidence and reason. EN further holds that to hold a belief involves incurring an obligation to articulate one’s reasons, evidence, and arguments when incited to do so. Finally, to articulate one’s reasons is to enter into a social process of evidence and reason-exchange.

Sometimes we will refer to the feature of rational believers asserted in (T1) as reason-responsiveness. So, at least for rational believers, beliefs are reason-responsive. Talisse emphasizes that he assumes what he calls a folk conception of epistemology when he refers to the norms of EN. This implies that the detailed philosophical specification of’reason’,’evidence’ and ‘argument’ is left undetermined, just as when philosophers routinely refer to our folk-psychological notions of beliefs and desires and leave the exact nature of these notions undetermined. Focus is thus on the everyday use of terms like ‘reason’, ‘evidence’ and ‘argument’ (Talisse 2009, p. 85).

We qualify (T1) to concern believers that are rational, while Talisse tends to merely talk about believers, that is, subjects who hold beliefs. We nevertheless find this qualifying restriction to be illustrative and implicit in Talisse’s thinking. Note that all that is comprised in the term ‘rational’ in this case is something like being coherent or moderately epistemically responsible; the requirements for rationality are thus minimal. When Talisse claims to have proposed a justification for liberal democracy that can be endorsed by all reasonable citizens, he thus has a basic epistemic sense of reasonableness in mind. What is really meant is that only citizens who are somehow epistemically debilitated or cognitively dysfunctional would not count as rational believers in the sense relevant to (T1).

Now, as EN involves a willingness to enter into a social process of reason-exchange, a commitment to EN carries with it a commitment to facilitate such reason-exchange, or so Talisse asserts. More precisely, Talisse assumes the following:

  1. (T2)

    Anyone who, upon reflection, sees themselves as operating in compliance with EN is rationally committed to endorse the social institutions facilitating the epistemically best social process of evidence and reason-exchange.

The point where this assumption about rational agents gets in contact with liberal democracy is in the following premise:

  1. (T3)

    There is a set of key liberal democratic institutions that are epistemically best with respect to facilitating the social process of exchange of evidence and reason.

The institutions in question are the ‘Institutions associated with the First Amendment [… freedom of speech, freedom of the press, freedom of association, liberty of consciousness …] and protections for critics, skeptics, dissidents and whistle-blowers’ (Talisse 2009, p. 123). We will refer to these institutions as the core liberal institutions, and they will be the primary target of our discussion. Talisse assumes that if the core institutions are to truly foster open-reason-exchange, they must be supported by other democratic institutions securing political equality, equal voting rights, governmental accountability and the like. Hence, what must be endorsed by someone who sees herself as complying with EN is more like the whole package of norms and institutions typically associated with liberal democracy, or liberal democracy in full. Granted this, we arrive at the conclusion of the argument:

  1. (T4)

    Rational believers are rationally committed to endorsing liberal democracy.

Obviously, without (T4), the argument at most shows that we have a rational reason to support the core institutions of liberal democracy directly associated with reason-exchange, not liberal democracy in full. For the purpose of the ensuing discussion, however, we grant (T4), as well as the other premises of the argument, so we do not need to distinguish between an argument supporting the core institutions and an argument supporting liberal democracy in full.

We will refer to (T1)-(T4) as the master argument. Crucially, for most of the following discussion, we will not question that the master argument is in fact sound: for all we say, the premises are true, and the conclusion indeed follows from the premises, or can be made to do so by adding uncontroversial extra premises. So, our criticism of Talisse does not depend on denying any of the premises in the master argument, or the validity of the argument.

Yet, despite being sound, we will argue that the master argument fails as a successful public justification of the core institutions of liberal democracy. How can this be? The crucial point to observe is that for the master argument to provide a successful public justification of liberal democracy in a given polity, soundness is not enough. The following two additional assumptions about the members of the polity in question need to be true as well:

  1. (A1)

    All (or almost all) members of a given polity come out as rational believers in the sense relevant for (T1), irrespective of their moral and religious outlook, or other parts of their reasonable comprehensive views.

  2. (A2)

    No-one (or almost no-one) among the members of the polity can reasonably reject the premises or the inferences of the master-argument, no matter what their religious and moral outlook or other parts of their (reasonable) comprehensive view is.

Note why these two additional assumptions are necessary for the master argument to constitute a successful public justification: If (A1) were false, it could be the case that the master argument were sound (as we have indeed granted that it is), and yet some agents would not be rationally committed to endorse liberal democracy simply because they were not rational believers in the sense required in (T1). In a sense, the argument would not apply to them. Something similar holds if (A2) were false, while (A1) true. In that case, all or almost all agents would be rational believers in the sense relevant for (T1), and yet they could reasonably reject one or more premises of the master argument, in which case they would not be rationally committed to endorse liberal democracy. In that case, the conclusion of the argument would not be binding for them, since they could reasonably reject it.

Our main claim is going to be that (A2) is false and this is why the master argument, despite being sound, fails to constitute a successful public justification of liberal democracy.Footnote 2 But before turning to our reasons why this is so, let us first assess (A1). Is it really true that everyone, or almost everyone, is a rational believer in the sense required for (T1)? Do we all fall under the scope of (T1)?

In Talisse’s view this is so, because it is simply near impossible not to aim at believing truly. We can take ourselves to hold a range false of beliefs, but it is absurd for us to hold that any particular belief that we have is false. ‘We take each individual belief we have, considered one by one, to be true’ (Talisse 2009, p. 91). Moreover, it is also near impossible for us to believe that we aim to believe truly and at the same time hold that we are not responsive to evidence and reasons, just as we cannot believe that we do not have any reasons for believing the way we do. ‘To say that a proposition is true is to say that it will square ultimately with the best reasons, evidence and argument. When one believes that p, one typically takes oneself to have sufficient reasons or evidence for taking p to be true.’ (Talisse 2009, p. 92)

Consider a case that might pose a problem for (T1). Suppose there is a religious believer who holds firm beliefs about the existence of God, and whose beliefs are ‘properly basic’Footnote 3 in the sense that they are not based on other beliefs providing them with evidential support. Since it is not obvious that these beliefs are responsive to reason and evidence, it is questionable whether this person qualifies as reason-responsive in the sense relevant for (T1), as far as the aforementioned religious beliefs are concerned. Either these are not genuine beliefs, or the person holding them is not minimally rational with respect to these beliefs.

In response to a challenge of this sort, Talisse points out what is no doubt true: namely, that any such individual inevitably will have a host of other beliefs that are responsive to reasons in the right way. If there are non-responsive religious beliefs, they will only comprise a tiny fraction of anyone’s total belief system. The claim that Talisse makes is merely that “for the most part when we believe, we take ourselves to be responding to reasons.” (Talisse 2009, p. 101). A religious believer of the kind we imagine will therefore still, in virtue of all these other reason-responsive beliefs, be within the scope of (T1). Consequently, it still holds that (T1) applies to most subjects as far as the vast majority of their beliefs is concerned. So, Talisse might argue that (A1) is close enough to being true for the master-argument to succeed as a public justification.

When assessing such a response, it is important to observe the way in which certain religious beliefs, while few in number, might still have a great impact on what an individual believes about many other questions; say questions about abortion or gay rights. Moral or even factual non-religious beliefs may be reason-responsive, and yet crucially depend on certain religious core beliefs that are non-responsive. In consequence, a subject’s views about these other matters may to a lesser extent be all in all reason-responsive; after all, to a considerable extent, these other beliefs depend on core beliefs that we assume to be non-responsive.

Obviously, it is difficult to decide where this leaves agents with local non-responsive beliefs with respect to their all things considered reason-responsiveness. But we will leave discussion of this issue for another occasion, and simply grant (A1), the assumption that all ordinary believers are reason-responsive in the sense relevant for (T1).

Turn now to (A2). As we pointed out, if (A2) were false (and A1 true), the master argument could be sound, and yet fail as a public justification. All agents would then be rational in the sense relevant for (T1), and yet some agents could nonetheless reasonably reject that liberal democracy should be endorsed for the type of reasons that Talisse outlines. As we will explain in a moment, these agents could reasonably reject (T3) of the master argument (maybe some agents could also reject (T2) or (T4), but we don’t see any reason to assume that this is so, and we will therefore discuss only the case of (T3)).

A word about how Talisse uses the idea of reasonable rejectability. For Talisse, the idea of reasonable rejectability ‘invokes the epistemic norms of reason-responsiveness’ (Talisse 2007, p. 86). So, reasonably rejecting an argument means rejecting the premises or validity of the argument for reasons. We take this to mean that to reasonably reject the premises or validity of an argument, one has to first engage with the particular claim and then, from a first-personal point of view, one has to consider oneself to have good reasons for rejecting it. Hence, for a rational believer to be in a position to reasonably reject (T3), is for this believer to reject (T3) for a reason that, from this agent’s own point of view, is a good reason.

3 The Basic Objection

In the following, we want to suggest that some agents may reasonably reject (T3) and consequently that (A2) is false. We argue that a believer can be minimally rational in the sense of operating in accordance with EN, and could indeed explicitly see herself as operating in compliance with EN: Yet, such a believer could reasonably reject (T3), the assumption that the core institutions of liberal democracy facilitate the epistemically optimal exchange of evidence and reason; indeed such a believer may rationally hold that core liberal institutions put her in an epistemically worse off position. Again, this is not because (T3) is in fact false, but because some agents may, given their other factual beliefs about the world, reasonably think that it is. To see how this may happen, consider Steve:

Steve is a man of faith, a devout believer. Steve is a sincere and intelligent man, he is concerned with getting things right and he is reason-responsive in the sense that he is sensitive to new relevant reasons and evidence that may impact his beliefs and he is always willing to elaborate his standpoint with reasons. Steve holds beliefs about God and God’s relation to the world, and these beliefs are central to his entire belief system and formative of his moral outlook. In Steve’s view, these beliefs are supported by ample evidence and reasons that he is willing to share. But Steve also holds a particular view about which sources of evidence support his beliefs about God. According to Steve’s particular set of epistemic norms, what a holy book says in a certain interpretation is a highly reliable source of evidence for truths in a number of domains. Steve is part of a substantial network of fellow citizens and (what for him are) epistemic authorities who are committed to epistemic norms that are roughly similar to his own. He is thus able to engage in reason-exchange with individuals who share these norms. In society at large, however, individuals who are committed to these norms are in the minority.

Borrowing a term from Alvin Goldman, we will call the various ‘sets of norms, standards, or principles for forming beliefs and other doxastic states’ (Goldman 2010, p. 187) that epistemic agents adhere to epistemic systems (or E-systems). Among other things, our respective E-systems typically harbor norms for gathering and evaluating evidence. Now, given Steve’s particular E-system, there are important domains of belief in which Steve might not find the sort of exchange of evidence and reason promoted by liberal democracy epistemically helpful. Why? Because Steve will not have reason to think that exchange of reasons and evidence with individuals not sharing his views on proper sources of beliefs is epistemically valuable. That does not mean that Steve would need to think that people who do not share his epistemic commitments are stupid, irrational or unreasonable. Indeed Steve may be willing to engage respectfully in debate with these people, but in so far as the disagreement about proper E-systems that divide them are deep, he may have no reason to hope that such exchange will lead to any kind of epistemic progress. From Steve’s perspective, it may even be worse than that. It may even be, that Steve can rationally believe that a constant flow of unreliable information and open discussion with inherently misguided subjects will have a distorting effect on his belief system.

A devout believer such as Steve may thus come out as a rational believer in the sense relevant to (T1), and Steve may also for that reason be committed to facilitate what he takes to be the best social processes of reason-exchange. But he may, nonetheless, not be rationally committed to (T3), the idea that key liberal institutions are the kind of institution that facilitates the epistemically best exchange of evidence and reason.

Whether Steve should accept (T3) may, among other things, depend on his beliefs about the composition of views among his fellow citizens. If Steve believes that enough participants in the exchange of evidence and reason do not share his E-system, then Steve will not, from a first-personal perspective, have reason to expect that his own epistemic position will improve by being exposed to these other perspectives. In other words, if the dominant set of E-systems in the society in which Steve lives is significantly at odds with central elements of Steve’s own E-system, then Steve may have no reason to seek exposure, or may even rationally seek to avoid exposure. Consequently, Steve may, from his point of view, have good reasons not to endorse the institutions that facilitate this exposure.

Imagine, for example, a situation in which news channels, the political debate, influential political organizations etc. almost solely provide information about a Godless world and routinely neglect or reject that evidence relevant to moral and factual questions can come from religious sources. Steve may rationally reject that such an environment will enlighten religiously grounded beliefs. Furthermore, he can rationally fear that his beliefs become more vulnerable to distortion. He may become confused about matters that seemed clear to him before and come to doubt things that he should not doubt, according to the norms and principles of his own E-system, doubt. From Steve’s perspective it thus becomes rational to avoid exposure to reasons stemming from E-systems that are significantly at odds with his own. After all, in Steve’s view, individuals who offer such reasons must have violated important norms of evidence. One may ask how Steve could be sure that his E-system is superior and the answer is, of course, that he can’t. But our concern here is the first-personal perspective and it is important to note that this perspective is not a kind of Archimedean point. It is a perspective that is already furnished with beliefs and norms of inquiry. Steve evaluates potential interlocutors according to the basic norms of his particular belief-system. He may be wrong, but this is all he has to go on.

Here is another way of stating the crucial point. The crux of the matter is what commitments follow merely from being reason-responsive. This much seems clear: being reason-responsive requires being responsive to any relevant evidence and reasons brought forth by inquirers that one deems to be at least minimally competent. It seems clear that this is the kind of reason-responsiveness that Talisse has in mind. We see this, for instance, in the examples he uses to illustrate the absurdity of non-responsiveness from a first-personal perspective. To emphasize the unreasonable epistemic attitude that accompanies non-responsive beliefs, Talisse asks us to consider the following cases:

  1. (a)

    I believe that p, but I am unaware of what competent opponents say about p.

  2. (b)

    I believe that p, but whenever I state my reasons for p, otherwise intelligent, sincere, and competent people are unmoved.

  3. (c)

    I believe that p, but I always lose fairly-conducted argumentative exchanges with competent interlocutors who reject p. (Talisse 2010, p. 48)

It does indeed seem obvious that it is unreasonable to deny subjecting beliefs to examination or opposition from competent interlocutors i.e. inquirers that we deem able to recognize proper sources of evidence and use suitable methods to evaluate and weigh evidence in any given circumstance. But it is not obvious that, from the first personal perspective, it is unreasonable to deny subjecting beliefs to opposition from incompetent interlocutors. We can be reasonable in denying the need to revise or reconsider our existing beliefs when encountering what we deem to be clearly irrelevant evidence brought forth by what we deem to be incompetent interlocutors. We should keep in mind that we are finite epistemic agents with natural limits to our time. Being able to pick the right battles or to determine which challenges to any given viewpoint are serious, and which are not, is part of what it means to be an epistemically responsible inquirer. Of course, we should acknowledge that we can be wrong about who the competent interlocutors are and what the relevant evidence is. We can be wrong in several ways: it is possible that we are wrong objectively, while subjectively, from the first-personal perspective, we can have robust evidence and good reason to believe that reason-exchange in certain circumstances is epistemically non-beneficial or damaging. It is also possible that we are wrong according to our own belief-system, we may have made mistakes according to our own norms of inquiry and evidence. We can acknowledge this risk and still be reasonable in rejecting the epistemic benefits of reason-exchange with certain interlocutors. In this sense it is a kind of cost-benefit analysis: we can be reasonable in holding that the epistemic costs involved in engaging in reason-exchange outweigh the benefits.

This would apply to the perspective of Steve surrounded by atheists as well as atheists surrounded by Steves. Granting that the domains in which these believers disagree constitute an important part of their entire belief-system (as in the case with Steve) and granted that truths that are important to these believers are dependent on these controversial beliefs, these believers have, from their own point of view, no reason to support the kind of reason-exchange that will be facilitated by core liberal institutions on epistemic grounds. Of course, we could easily add to Steve’s attributes that he, for moral reasons, is a warm supporter of liberal democracy but that is another matter, which does not directly concern the question of whether or not we should accept the master argument.

The problem for Talisse’s argument arises when there is a significant clash of E-systems in a given society. In pluralist democracies citizens may be divided not only by deep moral disagreements, but also by what we have called deep epistemic disagreements. When discussing abortion, gay rights, science education, euthanasia and stem cell research, we are not merely engaged in moral disagreements, but sometimes also in deep epistemic disagreements. These disagreements may, for example, pertain to how and where to gather evidence about the origins and evolution of life on earth. Many controversies about abortion, stem cell research and euthanasia turn on an underlying factual disagreement about the existence of God. The belief in God’s existence or non-existence can significantly shape the way we understand and discuss these important political issues. Think, for example, of how influential this factual belief is in shaping the way believers and non-believers discuss the question of when life begins. Of course, it is not only politically controversial issues that may feature deep epistemic disagreements. Less political disagreements may, for example, pertain to such matters as the existence of an afterlife or divine intervention: factual viewpoints that may not necessarily be immediately relevant to public policy, but that nevertheless are tremendously important to our understanding of the world we live in and thus also important to our general aim of believing truly.

4 Formal Epistemic Norms and Substantive Epistemic Norms

We think there are some lessons to be learned from a general version of the objection. The objection trades on a distinction between what we might call formal epistemic norms and substantive epistemic norms.Footnote 4 The set of norms we labelled EN, comprising the idea that one should aim at believing truly and that one should respond to reasons and evidence, are formal epistemic norms. These norms are formal in the sense that they do not specify what good or relevant evidence is, or what a proper response to a piece of evidence would be.

Substantive epistemic norms, by contrast, are norms that identify specific types of evidence as good, proper, or relevant in any given area of inquiry. Substantive epistemic norms also specify methods or practices for acquiring such evidence, or for identifying sources of evidence, and norms for proper accommodation of evidence and reason in one’s belief system. Here are two crude examples of substantive epistemic normsFootnote 5:

  1. (SEN 1)

    Empirical methods such as those used in evolutionary biology are generally reliable regarding the question of the origin of life.

  2. (SEN 2)

    What the Bible says about the origin of life is weighty evidence regarding the question of the origin of life, and may outweigh other sources of evidence.

Clearly, our full set of epistemological commitments cannot consist merely of formal epistemic norms. They must also contain many substantive epistemic norms. The important point now is that while we may all be committed to the same basic set of formal epistemic norms, say in virtue of being rational believers, this does not entail that we are all committed to the same set of substantive epistemic norms. So, given that EN is a set of formal epistemic norms, two individuals might both accept (and operate in accordance with) EN, and yet accept very different substantive epistemic norms, like those we mentioned above. And, importantly, they may do so without being irrational in the sense that they violate formal epistemic norms. In other words, formal epistemic norms, such as the norms of reason-responsiveness, underdetermine which substantial epistemic norms one is rationally committed to.

For rational individuals, the acceptance of substantive epistemic norms will critically depend on their factual beliefs, and may also depend on religious and moral beliefs. We should, therefore, in some cases expect individuals to commit to very different substantive epistemic norms as a result of having very disparate other commitments.

The importance of this is that the acceptance of (T3) - the premise that core liberal institutions indeed facilitate the epistemically best exchange of reason and evidence - depends on certain substantive epistemic norms and importantly on the composition of substantive epistemic norms held by individuals in a given society. Holders of certain substantive epistemic norms that are in the minority in a society can come to reasonably reject (T3). This is so even if we grant that (T3) is in fact true. The wider implication is that the truth of (A2) now is hostage to fortune.

If we now draw upon our example from the previous section, consider again religious beliefs and beliefs about the world that are based on religious beliefs (call these direct and indirect religious beliefs R-beliefs). Suppose that these beliefs support a certain set of substantive epistemic norms (call them R-norms) such that the following holds:

  1. (S1)

    A substantial number of people have R-beliefs about the world.

  2. (S2)

    Subjects who hold R-beliefs and rationally accept R-norms can be reasonable, i.e. abide by the norms of reason-responsiveness.

  3. (S3)

    Subjects who accept R-beliefs and R-norms can reject (T3) above without being unreasonable.

In the scenario depicted by (S1)-(S3) there is a substantial number of people for whom there is no epistemic justification of liberal democracy of the kind Talisse has outlined. In this scenario, (A2) fails. Though the premises in the master argument may indeed be true, the epistemic justification of liberal democracy would still not succeed as a public justification.

Our examples have drawn upon religious believers and the way they can come to reject (T3). It is, nonetheless, crucial to observe that the example could be turned around such that non-religious believers may find vital parts of their E-systems to be in the minority in a society (a society dominated by R-beliefs and R-norms). In this case, non-religious believers could reasonably reject (T3). The decisive point is that believers and non-believers can be reasonable and subscribe to E-systems that clash at the level of substantive epistemic norms in such a way that the party in the minority can come to reasonable reject (T3).

5 How Might One Respond to Our Objection, on Behalf of the Socio-Epistemic Justification of Liberal Democracy?

One option would be to deny (S1), which amounts to holding that there are no or only a few individuals around who hold R-beliefs and R-norms. If this were the case, the clash between R-norms and R-beliefs and the substantive epistemic norms of other E-systems would not pose a serious problem for the socio-epistemic justification, which takes as its starting point the ordinary believer. We have, however, modeled Steve to be an ordinary believer, and we suggest that as a matter of empirical fact that in our actual democracies you may find many religious believers who resemble Steve in the relevant respects. What we assume is merely that in our actual democracies there is a significant number of religious people for whom religious beliefs are central to their belief-system and formative of their moral outlook, and depending on the epistemic make-up of their fellow citizens, this may permit them to reasonably reject (T3).

Alternatively, one might reject that our ordinary believer is reasonable, i.e. reject (S2). There are several ways of pressing this option. Targeting our particular example, one may deny the claim that holders of R-beliefs and R-norms are reasonable. One way to argue this would be to show that a commitment to merely formal epistemic norms inevitably excludes commitment to R-beliefs and R-norms. Such a view would entail that those who hold such beliefs and norms (R-beliefs and R-norms) must have made a mistake in rationality somewhere, making them unreasonable. Alternatively, one might want to define reasonableness partly in terms of some specific set of substantive epistemic norms that excludes commitment to R-norms.

For the reasons already mentioned, we do not find either of these options promising; it simply does not seem that a commitment to formal epistemic norms entails commitment to any particular belief about the world, or any particular set of substantive epistemic norms. And holding R-beliefs and committing to R-norms does not appear unreasonable per se, so excluding these from the realm of the reasonable by mere stipulation does not seem warranted.

In his recent book, Philip Kitcher contests this latter point. Kitcher argues that holders of what we have called R-norms and R-beliefs tend to be committed to what Kitcher labels chimeric epistemologies: religious believers will accept the reliability of certain established scientific methods in some domains, but they deny the reliability of the very same methods in other apparently similar domains in favour of entirely different and potentially conflicting methods. These believers thus endorse ‘methods of certifying that can deliver opposing verdicts about acceptance and rejection’ (Kitcher 2011, p. 187). Due to the apparent arbitrary way that these believers apply epistemic norms and reject scientific reasons, one might with Kitcher suggest that believers are epistemically irresponsible and not properly reason-responsive. Kitcher, at least, seems to conclude that such beliefs and norms must be contained in private domains and bracketed from public reasoning.

We do not think that this is an argument that can save Talisse’s socio-epistemic justification. First, it is far from clear that it is impossible to be rationally committed to chimeric epistemologies. As we said above, rational commitment to substantive epistemic norms depends on which factual beliefs one accepts. So, roughly it only takes a sufficiently chimeric view of the world to be rationally committed to chimeric epistemologies, or more precisely, to applying very different methods of inquiry to apparently similar, but really quite dissimilar, domains of inquiry. Such a chimeric view may, for example, support that some domains in biology can be investigated by scientific methods and evidence whereas in other domains these are outmatched by religious methods or evidence. In other words, a proponent of chimeric epistemologies will surely reject the universal applicability of scientific methods, but need not do so ungrounded in reason. From the first personal-perspective, the perspective that Talisse is concerned with, these believers need not be violating rules of rationality. They are being reason-responsive and are operating in accordance with the epistemic rules and principles of their own particular E-system.

When opting for the strategy of denying (S2), it is also worth observing that, if (S1) is true then one would have to concede that there is a substantial number of unreasonable citizens. In turn this would imply that there is no epistemic justification for liberal democracy for these many citizens. This is unfortunate for the prospects of providing a public justification of liberal democracy that seeks to accommodate a diversity of moral, philosophical and religious worldviews.

There is a third line of response, which amounts to rejecting (S3). It can be argued that in most non-moral and non-religious domains, individuals operate within roughly similar E-systems, that in turn commits them to (T3), and accordingly that any R-beliefs and R-norms that our ordinary believer holds will be limited to certain moral and religious domains. Deep disagreements are thus secluded in moral and religious domains whereas in the non-moral domain epistemic agents operate on common ground. Talisse seems to subscribe to a version of this view, and he further argues that we can use this common ground to make progress on our deep moral disagreements. According to Talisse, many deep moral disagreements will harbour non-moral components, and interlocutors will benefit greatly from reason-exchange on this component of the disagreement. The idea is that reason-exchange on the non-moral components of deep moral disagreements can either eventually solve the disagreement or at least advance the disagreement by enlightening the contending parties with regard to relevant factors unknown to them before (Talisse 2008). This means that even though certain parts of individuals’ E-systems in a given society may be irresolvably at odds, these individuals may still have epistemic reasons for retaining a commitment to the core liberal institutions on behalf of their non-moral beliefs as well as their moral and religious beliefs. Individuals can enlighten each other with regard to non-moral facts relevant to their deep moral commitments and thus make epistemic progress in both domains.

In response to this, we note again that the wider impact of deep epistemic disagreements seems to be overlooked in this line of reasoning. Many familiar persistent moral disagreements seem to involve disagreement about basic moral and religious questions, which may turn precisely on differences regarding substantive epistemic norms, particularly about norms of proper evidence and reasons relevant to the disagreement. To illustrate this point, return for a moment to the disagreement about evolution in public education or the abortion controversy and ask: What factual (non-moral) components of these disagreements could be resolved with more reason-exchange that would make a significant difference for the controversy? If I believe a holy book and certain religious authorities are vital sources in determining the origins and evolution of life on earth and you believe that neither the book nor these authorities can contribute in any way to settling this matter, then we will be so epistemically divergent that it may be hard to think of a reason that the other part could produce which could make either of us change our mind. If my starting point in the abortion debate is that God says that life is sacred and starts at conception, and you barely agree to even consider this (in my view weighty) evidence, then from my perspective you are hardly a competent and worthy interlocutor on this issue. This assessment on my part may be entirely reasonable. That is despite the fact that we may be able to make some progress on certain non-moral, empirical facts bearing on the issue (such as the relation between illegalization of abortion and the incidence of unsafe abortions).

6 Concluding Remarks

It is important to stress that there is a socio-epistemic justification of core liberal institutions - it is just that it depends on ordinary believers’ agreeing on factual assumptions about what the world is like and how evidence is best gathered and assessed (that is, substantive epistemic norms), and on contingent facts about the composition of epistemic commitments among ordinary individuals in a given society. Talisse has pointed to a set of shared formal epistemic norms (EN). However, this set of norms turns out to be too thin to establish a commitment to reason-exchange in a liberal democratic setting. We have illustrated that in democratic societies, the existence of deep, sometimes irreconcilable, disagreements on substantive epistemic norms may cause individuals to reasonably reject that they are epistemically committed to reason-exchange with fellow citizens. The socio-epistemic justification is thus open to reasonable rejection. This is why Talisse’s socio-epistemic justification fails as a public justification.

We conclude that Robert Talisse has presented us with a fascinating and persuasive justification of core liberal institutions. Unfortunately, it is not a public justification that meets the persistent challenges that accompany a recognition of the fact of reasonable pluralism.