In what precedes, we have seen a positive case for the Erotetic view. Our discussion has been focused on how well the Erotetic approach respects constraints and desiderata for a theory of reasons in comparison to alternative accounts, and how well it integrates with general observations about normative ‘Why?’ questions. The overall aim of the present chapter is to provide an additional argument in favour of the Erotetic view. The argument here appeals to the advantages of adapting the Erotetic view with respect to a further debate about reasons. More specifically, the Erotetic view can shed new light on the well-known debate about pragmatic reasons for belief. This further theoretical application of the view adds an additional argument in its favour.

6.1 Pragmatism–Evidentialism Debate

An increasingly popular debate within contemporary epistemology concerns the question of whether there can be pragmatic reasons to believe a proposition – that is, whether pragmatic (e.g. practical, moral, eudaimonic) considerations may be genuine normative reasons for one to believe certain propositions. According to what until recently seemed to be a clear majority view in epistemology, only truth-conducive considerations – that is, considerations that somehow indicate, entail, or probabilify the truth of a proposition (and may or may not contain the proposition itself) – can be genuine reasons to believe the proposition. This view goes under the name of evidentialism within the current debates. For it is common to assimilate truth-conducive considerations in favour of p with evidence for p.Footnote ¹ Explicit defences of this view can be found, for example, in Kelly (Reference Kelly2002), Shah (Reference Shah2003), Shah and Velleman (Reference Shah and Velleman2005), Engel (Reference Engel2007, Reference Engel and Chan2013, Reference Engel2019, Reference Engel2020b), and Way (Reference Way2016), among others. Evidentialism is opposed to an increasingly popular view, pragmatism, according to which, roughly, it is not the case that only truth-conducive considerations can be reasons to believe.Footnote ² Contemporary pragmatism comes in two forms. On the one hand, there is radical pragmatism, according to which, strictly speaking/ultimately, only pragmatic considerations can be genuine normative reasons to believe.Footnote ³ Versions of this view have been defended in Rinard (Reference Rinard2015, Reference Rinard2017, Reference Rinard2018, Reference Rinard2019) and Maguire and Woods (Reference Maguire and Woods2020). On the other hand, there is moderate pragmatism (or pluralism), according to which, roughly, there can be both pragmatic and evidential genuinely normative reasons to believe. Versions of this appear in Foley (Reference Fogal, Sylvan and Ichikawa1992), Reisner (Reference Reisner2008, Reference Reisner2009, Reference Reisner and Star2018), McCormick (Reference McCormick2014), and Leary (Reference Lassiter2017), among others. At present, the debate has evolved to a stage where some philosophers take pragmatism to be the default position (see Maguire and Woods Reference Maguire and Woods2020, quote (6) reproduced in Section 6.2). Clearly, evidentialism has lost its absolute majority status. And the debate seems to be in a stalemate.

This raises a natural question: what should we really think about the possibility of there being pragmatic reasons amid the current disagreement within the field? Are there pragmatic reasons to believe a proposition or not? The debate between evidentialists and pragmatists seems to be in a deadlock. Both camps have provided positive arguments for their views, offered considerations against the opposite view, and elaborated some strategies to reply to the objections from their opponents. Yet the debate continues, as if it were based on some deep disagreement. Neither side seems to be convinced by the arguments from their opponents. It appears that we have reached a stalemate with no clear way out in sight.

A concrete objective of the present chapter is to show that the Erotetic view of reasons can provide a way to overcome the deadlock in this debate. The key element of the proposal is that both camps within the debate are operating with a somewhat defective understanding of normative reasons. In a sense, both camps are wrong. Yet both are also right about something important. The Erotetic view enables us to give due respect to what is true in both of these approaches with respect to pragmatic reasons for belief. Finally, the implicit mistaken assumptions about the nature of reasons in these views also allows us to provide a plausible error theory about why both of the approaches may initially appear attractive. In what follows, I elaborate the details of this proposal. If successful, the proposal will constitute a concrete demonstration of how the Erotetic view can be theoretically fruitful. Before arriving there, however, we need to add some clarifications about the present dialectical situation of the evidentialism–pragmatism debate and about what plausible constraints for overcoming the deadlock in this debate might look like. We do just that in Section 6.2. Section 6.3 is devoted to presenting details of the (dis)solution of the evidentialism–pragmatism dispute by appeal to the Erotetic view. Section 6.4 contains replies to some possible objections. Finally, I conclude by drawing some remarks on the theoretical fruitfulness and further potential applications of the Erotetic view of reasons.

6.2 Present State of the Debate and Options for Breaking the Deadlock

What are the criteria that a successful approach to the debate about the possibility of pragmatic reasons to believe has to satisfy? This is the first thing that we must clarify to have any hope of moving forwards and overcoming the stalemate within the dispute over pragmatic reasons for belief. What are the constraints that every party would agree to put on a theory that could break the deadlock? Both evidentialists and pragmatists seem to put a lot of weight on being able to best explain our pre-theoretical judgments about the relevant cases. Thus, in what follows, I suggest that we take on board the assumption that all parties in the debate seem to accept, namely that being extensionally adequate (or at least having the highest degree of extensional adequacy together with a viable error theory about the rest) constitutes a key constraint for a satisfactory solution to the dispute about the possibility of pragmatic reasons to believe. In other terms, the approach that respects most of the pre-theoretical judgments about the relevant cases – that is, constitutes the best explanation thereof – should be preferred to its competitors, everything else being equal (e.g. that it doesn’t lead to a contradiction).

It should be noted, of course, that while extensional adequacy is the gold standard in this debate, it is also a live possibility that no one single approach can respect all or even most of the pre-theoretical judgments in this context. If this proves to be the case, then we are warranted to follow a standard move of requiring that a successful approach be able to provide a plausible error theory of the cases that it cannot account for directly. Considerations of simplicity and theoretical fruitfulness may play a role in further assessment of competing error theories.

Now, it is one thing to agree over the theoretical constraints that should guide the discussion, but quite another thing to actually agree on how to measure whether the constraints are met or who is meeting them best. Both evidentialists and pragmatists claim that their approaches fit best with the pre-theoretical judgments about some relevant cases. Oddly enough, some of the key cases that both parties put forward are the same or, at any rate, share a disturbing number of similarities. But even if we put aside the apparently similar cases, the mere fact that our pre-theoretical judgments about some cases involving possible pragmatic reasons to believe depart in opposite directions as radically as the participants of the debate maintain is somewhat worrisome. Of course, we should not expect our best theories to fit our pre-theoretical judgments perfectly. The days of ordinary language philosophy are over. But still, such a disparity in judgments demands an explanation. In order to illustrate how dramatically evidentialists and pragmatists differ in their suggestions about what our pre-theoretical judgments are, I propose to look at some revealing passages from both camps. Let us start with three cases from evidentialists:

Note that when Thomas Kelly talks about considerations on which a belief is based (in (1)), what he really has in mind is normative reasons to believe, given that he defines these considerations by appeal to basing relations and their function of rationalising a belief. Assuming that rationalising of belief is one of the jobs of normative reasons for beliefs, these basing considerations can be seen as normative reasons (and not, say, mere motivating or explanatory reasons). Kelly’s (Reference Kelly2002: 176) suggestion that the relevant aspects of the case indicate ‘that the belief is not based on considerations of utility’ can be understood as a suggestion that this case (or, presumably, our pre-theoretical judgments about it – for example, that we find it plausible that the religious belief is not abandoned on the basis of a mere change in expected utility for the subject) is best explained by the evidentialist view, according to which nothing apart from evidential considerations can constitute normative reasons for beliefs. In other terms, we can interpret Kelly as suggesting that if pragmatism were right, then it should be possible that considerations of utility rationalise (e.g. constitute normative reasons for the subject in this case for abandoning the religious belief. But it doesn’t seem possible that mere utility considerations rationalise abandonment of the belief in this case). Thus, pragmatism (and not evidentialism) entails that our pre-theoretical judgments are wrong in this case and they owe us an explanation (an error theory) of why we are wrong on this account, according to this line of thought. Thus, this constitutes, according to the present reading of Kelly, a prima facie case in favour of evidentialism.

Similarly, we should note that Williams’s talk of ‘motives’ in the quoted passage (2) should really be understood as being about normative reasons. Recall our discussion on the Reasoning view of reasons earlier (Chapter 2), where we identified Bernard Williams as an early proponent of the contemporary Reasoning view of reasons, since on his account, roughly, one has a normative reason to F only when there is ‘a sound deliberative route’ from one’s motivational set to F-ing (cf. Williams Reference Williams and Millgram2001: 91). The aspect that is particularly relevant for our present discussion is that this is a definition (or characterisation) of what Williams calls ‘internal reasons’ which are, according to him, the only reasons that can have a normative importance for us (the relevant contrast here is with ‘external reasons’). And, crucially, Williams introduces the relevant phenomena (i.e. internal reasons) by characterising these as motives. He proposes to paraphrase the relevant reading of ‘A has a reason to ϕ’ as ‘A has some motive which will be served or furthered by his ϕ-ing’ (Williams Reference Williams and Harrison1979: 101, emphasis added). Thus, we are warranted to read Williams’s remarks in (2) as being about reasons. Crucially, these remarks have to be about normative reasons, since these are, again, the considerations that have to play a role in rendering a project, a belief, an attitude rational. Williams considers two possible (normative) reasons for the project of acquiring the relevant belief (e.g. that one’s son is alive), what he calls ‘truth-centred’ and ‘non-truth-centred’ motives. These are naturally interpreted as corresponding to evidential normative reasons and pragmatic reasons for the project of acquiring belief (or, let us say, indirect pragmatic or evidential reasons for belief). The bottom line, then, of (our interpretation of) Williams’s suggestion here is that while it is incoherent to even conceive of evidential indirect reasons for belief that one’s son is alive such that it is essentially constituted by one’s desire that one’s son is alive, it may be conceived that one has pragmatic indirect reason for the belief that one’s son is alive (constituted by the desire in question). But such reasons, while possible, cannot be normative reasons, since they cannot possibly render (even in the indirect sense) the belief that one’s son is alive rational (or, alternatively, they cannot render one’s project of acquiring the relevant belief rational). It appears that Williams sees this latter judgment as being our pre-theoretical judgment about the desperate father’s case. This reading seems appropriate given the following remark from Williams: ‘I think that most of us would have a very strong impulse against engaging in a project of this kind however uncomfortable these truths were which we were having to live with’ (see quote (2)). Again, a proper unpacking of Williams’s remarks seems to amount to the suggestion that only evidentialism can explain our pre-theoretical judgments about cases like the desperate father, and pragmatism has the burden of explaining why we are wrong in our pre-theoretical judgments.

Shah’s proposal seems to follow a similar line of reasoning, but it is expressed in terminology closer to our own. According to Shah, we cannot process intelligibly the idea that there might be pragmatic genuinely normative reasons for belief. Again, the point seems to be a point about a possible case and our pre-theoretical reaction to it: the talk of genuine normative reasons to believe that are constituted by pragmatic considerations – for example, that believing in the faithfulness of one’s partner will be comforting – is misusing the very concept of ‘reason to believe’. We have to conclude then that, following this line of thought, pragmatists owe us an extra explanation of what is going on here: since their view is that pragmatic considerations are reasons to believe, they have to conclude that we are wrong in our pre-theoretical judgments that someone like the husband in Shah’s case doesn’t master the concept of ‘reason to believe’ and, crucially, they have to explain why unbeknownst to us we are wrong about that.

One might object at this point that moderate pragmatists – that is, pluralists about reasons for belief – actually can account for judgments about these cases and that they owe no extra explanation here (see, for example, Reisner Reference Reisner2008, Reference Reisner and Star2018; McCormick Reference McCormick2014). More specifically, the idea is that moderate pragmatists will maintain that only when one’s evidence doesn’t favour p over not-p – that is, when evidence is equivocal/equipotent – one may have a pragmatic reason to believe that p. And, crucially, proponents of moderate pragmatism of this sort would insist that the aforementioned cases (1–3) are all cases of non-equivocal evidence. Evidentialists are right, they will insist, that in these cases the reasons are evidential and indeed our pre-theoretical judgments capture just this. But they will also insist that this is not all there is; that there are other cases, equivocal evidence cases, where judgments are different, and pluralism can directly explain the intuitions about both the non-equivocal evidence cases and equivocal evidence cases.

While such a version of moderate pragmatism may appear more promising in this context (compare also to William James Reference James1896 and his suggestions about live options), it is also unclear whether evidentialists would agree that it does fit our pre-theoretical judgments (see later for more theory-driven worries for this pluralism). It’s true that key cases from evidentialists are cases of non-equivocal evidence in the sense that in these cases evidence does support p over not-p. But one might also expect that if evidentialists were to consider specifically the cases of equivocal evidence, they, or at least some of them, would insist that our judgments are the same in the cases of equivocal/equipotent evidence.Footnote ⁶ Indeed, this is a plausible expectation, given, for example, Williams’s remark that ‘a man may want to believe something not caring a damn about the truth of it, but because it is fashionable or comfortable or in accordance with the demands of social conformity to believe that thing’ (see quote (2)), where the project of acquiring such a belief is described as ‘deeply irrational’ (Williams Reference Williams and Williams1973: 149–150). The vague expression of ‘not caring a damn about the truth of it’ may well be understood in a general sense, where a subject who is not suspending her belief about p in a case of equivocal/equipotent evidence for p falls within the category of ‘not caring a damn about the truth of’ p. If so, then according to evidentialists (e.g. Williams), the moderate pragmatists’ cases of alleged pragmatic reasons in situations of evidential equipoise would also fall within the category of cases where our pre-theoretical judgments qualify the subject’s project of acquiring such a belief as deeply irrational. But we will return to the pluralist view later. Let us focus now on three telling passages from the pragmatist camp:

It’s straightforward that pragmatists are suggesting in all these passages (4)–(6) that their view fits best with our pre-theoretical judgments about the relevant cases. They see our pre-theoretical judgments there as constituting a prima facie case for pragmatism, indeed as grounds for claiming that pragmatism is the default view and that actually evidentialists owe us an extra explanation of what is going on in these cases if we are to take evidentialism seriously. An interesting observation is how close pragmatist cases actually come to cases proposed by evidentialists. Consider, for example, the religious belief case in (4). Belief that God exists would make Mary happier, according to the description of the case by Leary. She suggests that it is up to evidentialists to demonstrate that this eudaimonic consideration is not a normative reason for Mary to believe that God exists (the comparison with exercise is supposed to help us see the parallel). Similarly, Rinard in (5) suggests that it does seem that belief in one’s survival contrary to evidence (see also Reisner Reference Reisner2008) and belief in one’s success as an athlete do constitute normative reasons to believe. Again, Rinard seems to be in the business of putting forward what appear to be our pre-theoretical judgments about the cases.Footnote ⁸ Maguire and Woods sum up a number of cases from the literature that seem to speak in favour of pragmatism – that is, that appear to fit best within the pragmatist framework. Note, however, a complication involving claims about friendship, faith in humanity, and promise cases. Initially, the authors that introduced these cases within the literature didn’t conceive of them at all as prima facie cases for pragmatism – actually, quite the contrary. When Sarah Stroud first introduced considerations about friendship requiring beliefs against evidence, she was well aware and made it explicit that such beliefs commonly appear to have some irrationality (cf. Stroud Reference Stroud2006). Similarly, when Marušić (cf. Marušić Reference Marušić2013, cited in a footnote to the passage from Maguire and Woods reproduced earlier) discusses promises and related statistically informed cases, he is clear that he is putting forward an argument towards the conclusion that beliefs that seem to go contrary to evidence but are nonetheless necessary for sincere promises and so on are to be seen as rational after all. But his consideration here is based on an elaborated argument. He acknowledges that there is an initial tension that arises once we consider the relevant promise case. But he doesn’t seem to be in the business of putting forward a simple prima facie claim that starts from an allegedly common-sense view that beliefs contrary to evidence in such cases are rational. This is, rather, his conclusion. Thus, Maguire and Woods, and also Rinard elsewhere, are mistaken in listing friendship, promises, and similar cases as prima facie cases for pragmatism, at least when they clearly refer to Stroud’s and Marušić’s versions and treatment of these cases. Now, having this in mind, we can still accord to pragmatists that other cases that they propose are such that they seem to indicate the possibility of pragmatic reasons for belief.

What should we make of this apparent disagreement about which position enjoys the default status and is in line with our pre-theoretical judgments? An important point to note, I think, is that a charitable reading of the debate doesn’t allow us to rule out either the evidentialist claim of being in line with pre-theoretical judgments or the pragmatist claim of respecting pre-theoretical judgments and having a good prima facie case. We just don’t have grounds for favouring affirmations of one side at the expense of the other on the present grounds. So, let us assume that both sides are sincere when they report that it is plausible, apparent, and so forth, that the relevant cases speak in favour of the possibility or the impossibility of pragmatic reasons for belief.

With this assumption on board, we are left with three theoretical options. First, and most radically, we can just throw the baby out with the bathwater and consider that such a disparity constitutes a defeater against any view within this debate that claims to be in accord with the pre-theoretical judgments. One might think that given that the judgments are in such disarray here, it indicates that our pre-theoretical conception is just too confused to be accounted for. There is no argumentative value in trying to fit our theory with our pre-theoretical judgments in this context, according to this line of thought. Now, this is surely a radical approach, and I think, while it is of course an option, we should regard it as the last resort. For giving up totally on the ambition to at least respect our pre-theoretical commitments might backfire in ways that might lead to a more general scepticism and the generation of theories that are completely out of touch with what commonly matters to us.

The second option is to admit that while both the authors of passages (1)–(3) and the authors of passages (4)–(6) are sincere in their affirmations, and the relevant cases do trigger the pre-theoretical judgments that the authors claim they do, one side here is nonetheless mistaken. That is, some of the pre-theoretical judgments are wrong, and there is a plausible error theory of why we might be led to hold these judgments.

So, for instance, a popular line among pragmatists is to acknowledge that evidential considerations often do appear to be the only possible candidates for the role of normative reasons in favour of believing a proposition, but such pre-theoretical judgments are explained away (i.e. explained without concluding that evidentialism is correct), according to this line of thought, by the fact that it is often practically advantageous for us to believe on the basis of truth-conducive considerations, since it is often advantageous for us to have true beliefs. The following passage illustrates this version of pragmatist error theory:

Presumably, then, one might claim that at least some of the evidentialist cases might fit into this sort of error theory (see also Maguire and Woods Reference Maguire and Woods2020 for another sophisticated approach that basically treats believing as on a par with an activity of a game that has its own constitutive standards which lack the genuinely authoritative normativity). However, it is not clear whether this error theory really fits the bill with respect to all the cases in (1)–(3), for these seem to be precisely cases where one is supposed to gain something from believing against one’s evidence. Hence, it is not clear how the expected utility of believing truth and the expected utility associated with believing against truth-conducive considerations should be balanced in these cases. Alternatively, a proponent of pragmatism may merely assert that the apparently evidentialism-supporting judgments about cases like (1)–(3) are wrong since evidentialism is wrong (a conclusion for which pragmatists have independent argument). On this view, they are just relics of wrongheaded evidentialist thinking. Compare Rinard:

On the other side of the debate, a popular error theory among evidentialists is an appeal to the idea of there being the ‘wrong kind’ of reasons for belief. Roughly, some considerations are not reasons to believe that p, but rather considerations that speak in favour of getting oneself to believe, or desiring to believing, or undertaking an indirect project of acquiring the belief that p. The proposal here is parallel to a well-known proposal by reasons-first theorists with respect to the ‘wrong kind’ of reasons problem, where a threat is considered to be a reason to get oneself to admire the despicable demon (i.e. the threatener), but not a reason to admire the demon (see Section 1.4). Evidentialists suggest that while considerations, such as the considerations in (4)–(6) and the like, are reasons to get oneself to believe that p, they are not reasons to believe that p. But, given that getting oneself to believe that p is, on the face of it, quite close to believing that p, it is not surprising that one might mistake the reasons to do the former for reasons for the latter. Versions of this sort of error theory or, at any rate, very similar thoughts may be found in Parfit (Reference Parfit2011), Engel (Reference Engel2019, Reference Engel, Langlet and Monnoyer2020a), and Hieronymi (Reference Hieronymi2005), among others.

I don’t intend to assess here the merits and pitfalls of the error theories introduced earlier. My objective lies elsewhere. What I want to observe at this point is that there is a third option that we may take with respect to the apparent disagreement about pre-theoretical judgments. The third option is to recognise that both those who claim that cases (1)–(3) are best explained by the impossibility of pragmatic reasons to believe and those who claim that cases (4)–(6) are best explained by the possibility of pragmatic reasons to believe are right. Of course, the trick for holding such a position, indeed for holding the view that pre-theoretical judgments both support and don’t support the possibility of pragmatic reasons for belief, consistently is that ‘pragmatic reasons for belief/to believe’ has two possible readings. There are two sorts of normative reasons, and in one sense, there can be pragmatic reasons to believe, but in another there cannot. This option is what I elaborate in Section 6.3. Let me conclude the present section by two final remarks.

First, it should be noted that the third option is optimal, given our initial agreement on what should constitute a key criterion for overcoming the deadlock within this debate. Recall that we agreed earlier that a theory that respects most of the pre-theoretical judgments should be preferred, ceteris paribus, to its competitors. That is, a view that doesn’t need to explain away our common intuitions about the cases (e.g. (1)–(6)) has an advantage over views that need to resort to error theories.

The second point is that the proposal to be elaborated in Section 6.3 is, of course, not the only possible way to try to respect all our pre-theoretical judgments. In particular, as observed earlier, some seem to think that versions of moderate pragmatism or pluralism with respect to reasons for belief can do just that (cf. Reisner Reference Reisner2008, Reference Reisner and Star2018). On such a view, roughly, there are two sorts of normative reasons to believe that p. One sort is evidential or truth-conducive – considerations that constitute evidence in favour of p. Another sort is pragmatic – considerations that make believing p somehow practically, eudaimonically, or morally advantageous. Crucially, according to this view, pragmatic reasons kick in only in contexts where evidential reasons don’t favour believing over disbelieving p, and rather recommend suspension of judgment.

With respect to this option, recall from the aforementioned that we may have some doubts about whether such an approach can really vindicate evidentialist intuitions. Evidentialists may well claim that going for anything else than suspension of judgment in cases of equipotent evidence is still irrational on such a view, given the rationality-normative reasons connection. They may claim that our pre-theoretical judgments indicate that no pragmatic consideration can constitute genuine normative reasons to believe even in evidential equipoise cases (recall Williams’s complaint about reasoners who don’t respect truth). But even if we put this worry aside, two further issues are looming in. First, such an account faces the challenge of explaining how exactly we are supposed to combine or weight pragmatic normative reasons against evidential normative reasons. It is a fundamental assumption of the pluralist proposal that they can and indeed have to be weighted against each other. I am not claiming here that no story can be provided about such weighing (see Reisner Reference Reisner2008 and Steglich-Petersen and Skipper Reference Steglich-Petersen and Skipper2020 for recent attempts to account for combinations). I am noting merely that it is a challenge and not everyone is convinced that this sort of project can be completed successfully (cf. Berker Reference Berker2018, who provides arguments for doubts and maintains that pragmatic and evidential considerations are ‘like water and oil’ and don’t really mix). The account that I am going to present avoids this issue altogether, since the two sorts of reasons I am going to introduce are not supposed to combine. Second, this sort of view also has the challenge of explaining why these two sorts of reasons are indeed two sorts of the same thing and what exactly their common factor is. Again, I am not claiming that no successful way of meeting this challenge can be provided, but merely that it constitutes a challenge that asks for a more substantive explanation of why there are two sorts of reasons for beliefs. Merely postulating that there are two sorts of reasons is ad hoc. One needs to provide an explanatory story of why there are two sorts of reasons. After all, postulating more entities (or relations) of something unified calls for an extra explanation. We should not, of course, postulate the existence of entities or relations beyond necessity. Every new distinction we make had better be grounded in independently motivated considerations. The view that I am about to elaborate avoids this challenge, since it has an independent substantial back-story about why there are different sorts of normative reasons.

Nonetheless, at the end of the day, I don’t intend here to try to prove that this sort of pluralism is definitely wrong. My purpose rather is to put on the table a new option, a new proposal that can vindicate independent theoretical motivation and provide a substantive account of why there are two fundamentally different sorts of reasons that correspond to what we might call ‘evidentialist’ and ‘pragmatist’ reasons within this debate. Let us presently turn to considering this new proposal in further detail.

6.3 Explaining Pragmatic Reasons for Belief: Insight from the Erotetic View

The aim of this section is to develop and defend a new positive proposal with respect to the dispute over the possibility of pragmatic reasons for belief. This proposal applies the Erotetic view of reasons to explore the sense in which pragmatic reasons for beliefs are possible and the sense in which they are not possible. Let us elaborate this in more detail.

On the Erotetic view of reasons, developed in Chapter 5, a normative reason to F just is, roughly, an (appropriate) answer to the normative ‘Why F?’ question on one of its two possible readings. The two readings here are an explanation-requiring reading and an argument/reasoning-requiring reading. More precisely, when asked ‘Why F?’, one might be prompted to provide an explanation of why one ought to F, or alternatively one might be prompted to provide an argument/a piece of reasoning for the conclusion that one ought to F. These two readings of the normative question ‘Why F?’ give rise to two sorts of possible normative reasons to F, namely normative reasons as (partial) explanations of why one ought to F and normative reasons as premises of arguments/patterns of reasoning towards the conclusion that one ought to F (alternatively, fitting premises in fittingness-preserving arguments/patterns of reasoning concluding in F-ing). The two sorts of reasons are distinct and cannot be reduced one to another. We know already from our preceding discussion that some considerations can only be reasons of one of these sorts. In particular, some considerations can constitute only explanation-providing answers to ‘Why F?’ questions. The example that we presented to illustrate this feature was the case already discussed in the context of objections to the Reasoning view of reasons (see Chapter 2). Namely, it was the case involving Moore-paradoxical considerations of the sort ‘the building is on fire but S doesn’t know that the building is on fire’ in situations where S is unaware of the fire in the building where she finds herself. Plausibly, such considerations still speak in favour of S’s checking the state of the building, verifying whether she is safe, and so on. The fact that she is ignorant of the fire does not matter, in a sense. Assuming that pre-theoretical judgments about a consideration speaking in favour of F-ing does indicate that the consideration in question is a normative reason, we can conclude that such Moore-paradoxical considerations are indeed normative reasons in the relevant contexts. Crucially, however, there is no way for the subject to use appropriately Moore-paradoxical considerations speaking in favour of F-ing in her reasoning or arguing towards the conclusion that she ought to F. After all, she can neither know nor properly believe such considerations. Thus, the Moore-paradoxical considerations cannot be normative reasons pertaining to reasoning; that is, they cannot be normative reasoning reasons – reasons in the sense of being appropriate answers to the ‘Why F?’ question in its argument/reasoning-requiring sense.

The proposal that I would like to put forward with respect to the possibility of pragmatic reasons to believe, then, is that we can respect pre-theoretical judgments of all sides and need not explain away some of these by appeal to more or less sophisticated error theories, by simply generalising the lessons from the Erotetic view’s treatment of the Moore-paradoxical considerations. More specifically, we can deploy the distinction between normative reasoning reasons and normative explanatory reasons to the case of pragmatic considerations that, in a sense, sometimes seem to speak in favour of believing a proposition and, in a sense, don’t seem to speak in favour of believing a proposition. In short, pragmatic considerations are another example of considerations that can be normative reasons just in one sense, but not in the other sense predicted by the Erotetic view. The thought is that pragmatic considerations, similarly to Moore-paradoxical considerations, cannot be normative reasoning reasons, but may still constitute normative explanatory reasons, as considerations that partly explain why one ought to believe that such and such is the case.

On the present proposal, pragmatic considerations cannot be appropriate answers to the argument/reasoning-requiring reading of the ‘Why believe p?’ question. Thus, they cannot be normative reasoning reasons to believe. And there is truth in evidentialism; this is the insight we can take from evidentialists. The problem with their view, though, is that they don’t distinguish between these two sorts of normative reasons – normative reasoning and normative explanatory reasons. Thus, strictly speaking, their proposal is wrong. Evidentialism can only be the right theory for normative reasoning reasons. Note also that, interestingly, some of the paradigmatic proponents of evidentialism also seem to endorse the Reasoning view of reasons (cf. Williams Reference Williams and Prior1989, and elsewhere, arguably, Hieronymi Reference Hieronymi2005). This is not surprising at all. If anything, this only makes our proposal even more plausible. For if your theory of normative reasons in general defines reasons as premises in good patterns of reasoning/arguments, then, of course, you will also reject the possibility of pragmatic normative reasons for belief. Pragmatic considerations are just the wrong kind of inputs for arguments/patterns of reasoning for believing that p. (This last observation also perfectly fits the observation that the main focus of evidentialist arguments is on the role of reasons in doxastic deliberation or reasoning towards belief.) Our view predicts that, of course, if you focus on one sort of normative reasons at the expense of the other, then some of the considerations that one might want to qualify as reasons, in a sense, will not come out as reasons on your view. We observed this with Moore-paradoxical considerations that presented a genuine worry for the Reasoning view, and we see it again with respect to pragmatic considerations that are excluded from the category of possible reasons for belief according to the Reasoning view of reasons. Thus, the present view vindicates a reinterpreted evidentialist insight: some considerations can never be normative reasons to believe, in a sense – namely, pragmatic considerations can never constitute normative reasoning reasons.

But, of course, there is another sort of normative reasons, normative explanatory reasons. And nothing prevents some pragmatic considerations from being normative reasons to believe in this sense. It is entirely conceivable that, similarly to the case of Moore-paradoxical considerations, some pragmatic considerations may in some contexts constitute appropriate answers to the explanation-requiring reading of the ‘Why believe p?’ normative question. That not believing p would cause tragic consequences may speak in favour of believing p. It might well be the case that in such a situation where not believing that p brings about tragic consequences, you ought (in a sense) to believe that p. That not believing brings about tragic consequences, then, explains why you ought to believe. In this sense, then, these purely pragmatic considerations about the tragic consequences of not believing p constitute normative reasons for you to believe that p. Of course, pragmatists, or at least radical pragmatists, are wrong that all normative reasons are ultimately pragmatic considerations. But the present view also takes on board their insight (or at least the insight of a reinterpreted version of pragmatism) that sometimes pragmatic considerations matter for belief. They matter, on the specification of the present view, only when (and because) they explain why one ought to believe the relevant proposition. They may constitute only one sort of reasons to believe in some situations. Note also that, interestingly, some contemporary pragmatists also appear to endorse the Explanation view of reasons in general (see Maguire Reference Maguire2016 in particular; Reisner also seems to be favourable towards the Explanation view but stops short of endorsing the view that reasons can be analysed, by an appeal to oughts in particular; cf. Reisner Reference Reisner, Hirose and Reisner2015: 191, fn 10). Again, this is predictable on our view. For as we saw earlier, the major motivation for evidentialism seems to come from their focus on the role of reasons for belief in doxastic deliberation, reasoning, but if from the outset you don’t think that there is any essential or definitional connection between reasons in general and reasoning, then you will be less moved by the considerations that evidentialists accumulate in favour of their view. In a sense, both are right: the connection to reasoning is crucial for evidential reasons, but the connection to explanation of oughts is crucial for another sort of considerations, considerations that may well be pragmatic and constitute explanatory reasons for belief. Thus, the present view also vindicates a (reinterpreted) pragmatist insight: pragmatic considerations may sometimes constitute normative reasons to believe of a specific sort – namely, they may constitute normative explanatory reasons.

Note also that the Erotetic view contains resources for explaining why both evidentialists and pragmatists are wrong. As we saw in Chapter 5, it is relatively easy to overlook the existence of two readings of the general normative ‘Why F?’ question by focusing only on one of these. Sometimes we have to pay attention to discern whether one is in the business of arguing/reasoning or in the business of explanation. It is not that rare to confuse the two activities. Thus, it is not that difficult to be led to focus only on one of the readings of the ‘Why believe that p?’ question and relatedly on only one of the two possible sorts of answers to this question (which, of course, need not involve any conscious activity of directing one’s mind to one reading of the normative question only), especially if one starts with an assumption that all normative reasons are premises in good reasoning or, alternatively, are parts of a deontic or axiological explanation. Focusing on explanatory reasons to believe at the expense of reasoning reasons, then, might easily lead one to endorse pragmatism. Focusing on reasoning reasons exclusively might easily lead one to endorse evidentialism.

Let me conclude this section by rapidly reviewing the key examples that we’ve introduced in Section 6.2 and that were proposed as constituting prima facie cases for evidentialism and pragmatism accordingly. Consider the example in the quote (1) from Thomas Kelly where he imagines the case of an agnostic who realises at time t1 that the expected utility of believing in the existence of God at that time is very high and thus undertakes the project of becoming a believer, and at time t2, she effectively believes that God exists, but the expected utility changes unexpectedly: it is no longer advantageous to be a believer. Kelly observes that the believer would not abandon her belief on the basis of the change in expected utility. We call this observation a pre-theoretical judgment about the case. Indeed, it seems to be a common-sense judgment. According to Kelly (Reference Kelly2002: 176), this observation ‘indicates that the belief is not based on considerations of utility’. As we saw in detail earlier, what Kelly means by this is that pragmatic considerations (e.g. considerations of utility) cannot be normative reasons to believe or to disbelieve, given that the job of normative reasons is to rationalise a belief (make it rational) by being the (appropriate) basis of belief (i.e. the considerations on which a belief is appropriately based via the basing relation). The conclusion that Kelly draws is that only evidential considerations can be reasons to believe, since only these can be (appropriate) bases and thus rationalise belief. The Erotetic view can vindicate the pre-theoretical judgment that Kelly puts forward here without endorsing the conclusion that only evidential considerations can be reasons for belief. The trick is to restrict the role of rationalising only to normative reasoning reasons to believe or rather to some proper subset of these. With this restriction in place, we can clearly and wholeheartedly agree with the observation that indeed the believer will not disbelieve in the existence of God on the basis of pragmatic considerations. We need no error theory here. But this doesn’t indicate that there are no pragmatic reasons in this case for the subject to disbelieve, since this only indicates that there cannot be pragmatic considerations that rationalise a belief by being its basis. One might think that there might be pragmatic reasons to disbelieve in this case that are not bases in Kelly’s sense. Nothing in the example rules this out. Moreover, this proposal receives independent support if we assume that the rationality of a doxastic state is tied more closely to the way an agent arrives at it, through, say, reasoning, rather than to an explanation of why an agent ought to be in such a state. For, arguably, explanation of why one ought to be in a state, but not one’s reasoning towards that state, can be independent – that is, unconnected to one’s perspective. Recall the discussion of the Monty Hall case in Chapter 1. An alternative way of seeing the case (not necessarily a way that conflicts with our initial assessment, though) is that there is an explanation of why one ought to switch in the Monty Hall situation. This explanation constitutes a reason in the explanation sense of normative reasons, to switch, even though it may not be rational from the subject’s own perspective to switch; one just cannot reason towards that conclusion given what one believes and disbelieves. Thus, tying rationality to reasoning rather than to explanation of deontic facts may well be independently motivated. If so, we can see, then, that restricting the observation in the case (1) about the possible bases of a doxastic state to reasoning reasons only gives us room to admit the possibility of pragmatic reasons to believe, in the sense of normative explanatory reasons, while still accepting straightforwardly the pre-theoretical judgment that the believer will not disbelieve in this case on the basis of pragmatic considerations. On the present proposal, no pragmatic considerations can constitute a rationalising basis for a doxastic state – that is, a normative reasoning reason to believe, disbelieve, or suspend judgment.

Similarly, the Erotetic view can easily account for the cases from passages (2) and (3). Again, the move is the same; the Erotetic view can explain why we would find the desperate father’s belief that his son is alive when he has strong evidence to the contrary but also has pragmatic considerations (e.g. it would make him feel less miserable) that seem to speak in favour of his belief in his son being alive. That is, we assume the content of our pre-theoretical judgment of the case. The explanation is that rationality is tied to how one arrives at a belief (i.e. basing of belief), but pragmatic considerations can never be the appropriate inputs for reasoning towards belief (they cannot be reconstructed as premises of a valid argument or pattern of reasoning towards an appropriate belief). Thus, arriving at a belief indirectly from pragmatic considerations alone can never make the belief rational. All this can be respected by the Erotetic view. But it also entails that this doesn’t yet mean that there are no pragmatic reasons for belief – there might be if we understand them as restricted to explanations of why the father should/ought to, in a sense, believe that his son is alive. (Note that if everyone agrees that in this case there is no sense in which the father ought to believe that the son is alive, the present view predicts that indeed there are no reasons, not even normative explanatory reasons, to believe that the son is alive, which I think is another advantage of the view, since it gives us flexibility for different cases.) If there is a sense in which the father ought to believe that his son is alive, then the present view can clearly explain why the pragmatic considerations do speak in favour of the father’s belief in his son being alive; these are normative explanatory reasons. Perhaps these are normative explanatory reasons in an objective or overall sense, given that his grief is so severe and produces so much suffering that, in a sense, he ought to maintain the belief that his son is alive. In the case discussed by Shah (in (3)), of a husband’s belief that his wife is unfaithful, the explanation is the same. The present view accommodates the pre-theoretical judgment that it is unintelligible to think that the pragmatic considerations constitute a reason for the husband to believe that his wife is not unfaithful if we restrict ‘reasons’ to the normative reasoning reasons here. This, again, doesn’t entail that these pragmatic or eudaimonic considerations cannot be normative explanatory reasons for why the husband ought to/should, in a sense, believe that his partner is faithful.

With respect to the cases from pragmatists, the explanation goes the other way but is essentially the same. Consider Leary’s take on religious belief in passage (4). She seems to suggest that it is pre-theoretically plausible to see the eudaimonic considerations about the benefits of belief in God as a normative reason to believe in God, just as eudaimonic considerations about physical exercise constitute normative reasons to do physical exercise (e.g. that exercising would make one happier is clearly a reason to exercise). The Erotetic view can accommodate this intuition without endorsing pragmatism and without appealing to an error theory about pre-theoretical judgments here. If it’s true that one should/ought to be happy (perhaps other things being equal), and that believing that God exists would make one happy, then that the fact that believing that God exists would make one happy is a (partial) explanation of why one ought to believe that God exists. This is, then, in a normative explanatory sense, that the eudaimonic considerations can be normative reasons for one to believe that God exists. This is not to say that there may be normative reasoning reasons of the eudaimonic variety. Indeed, admitting that eudaimonic considerations can be normative explanatory reasons for why one ought to believe something doesn’t entail that all reasons to believe are ultimately pragmatic, neither that the thesis according to which no pragmatic consideration can constitute a normative reasoning reason is false. The example of believing that one will survive (in (5)) in a situation where that belief is not supported by evidence but would nevertheless slightly raise the chances of survival can be explained without accepting pragmatism. This consideration is a reason to believe that one will survive in the sense that it partly explains why one ought to believe that one will survive – one will thereby increase one’s chances of survival. This need not be a normative reasoning reason for one to believe that one will survive. And the fact that an athlete’s belief that they will succeed increases the chances of success may also explain why the athlete ought to, in a sense, believe that they will succeed and thus be a normative explanatory reason without being a normative reasoning reason for the athlete to believe. The cases presented in (6) undergo the same treatment. Indeed, in all cases where one claims to have an intuition that a pragmatic consideration is a normative reason to believe some proposition, the Erotetic view can respect that intuition provided that it is a case in which the subject ought to/should believe the proposition in question and the relevant pragmatic considerations partially explain why the subject ought to/should believe it. All the cases that we have seen appear to fall into this category. The Erotetic view can explain the pragmatist intuitions without giving up evidentialist insight. There are two sorts of normative reasons to believe corresponding to an explanation of a normative fact about belief and to a premise in a reasoning/argument towards the relevant belief correspondingly. All the pre-theoretic judgments about the possibility or not of pragmatic reasons to believe seem to correspond to one or the other of these two categories of normative reasons. The mistake is to assume that there is only one category of normative reasons.

Thus, application of the Erotetic view of reasons to the question of the possibility of pragmatic reasons for belief enables us to vindicate all the pre-theoretical judgments. It has, hence, an important advantage over alternative treatments of the question. It appears that it may well constitute a constructive way to overcome the present deadlock within this debate. This theoretical fruitfulness, in turn, constitutes another argument in favour of the Erotetic view of reasons.

6.4 Replying to Potential Worries

The aim of the present section is to provide replies to some potential worries about our proposal with respect to the possibility of pragmatic reasons for belief. In providing these replies, we are also further elaborating some surprising aspects of the Erotetic view as applied to reasons for doxastic states.

My hope is that the present proposal will be seen as a serious contender for overcoming the deadlock in the debate about pragmatic reasons for belief, but, of course, we may reasonably anticipate that some proponents of evidentialism and of pragmatism will question some aspects of the view. In what follows, I consider five potential objections in particular.

First, pragmatists might actually question the assumption that we cannot believe on the basis of pragmatic reasons. In fact, some of the new pragmatists have suggested exactly this in response to older evidentialist arguments from Kelly, Shah, and Velleman, among others (cf. McCormick Reference McCormick2014; see also Reisner Reference Reisner and Star2018). If they are right, then one might doubt whether normative reasoning reasons to believe cannot be themselves constituted by pragmatic considerations. That is, a pragmatist might accept our fundamental distinction between normative reasoning and normative explanatory reasons but question the claim that normative reasoning reasons to believe cannot be pragmatic considerations.

The first thing that I would like to note with respect to this potential worry is that, of course, strictly speaking, the Erotetic view can be made compatible with pragmatism – that is, the core claim about two sorts of reasons at any rate. However, this is not the option taken up here. On the current view, pragmatic considerations indeed cannot constitute (appropriate) bases that could be normative reasoning reasons for belief. As noted in Section 6.3, one line of thought here is that the rationality of a belief seems to be connected to the ways in which a given subject comes to hold that belief. You might, well, possess the best possible evidence in favour of a claim, but if your belief is formed by a mere hunch or via a motivated reasoning, your belief doesn’t qualify as rational in the standard sense of rationality. Focusing on beliefs arrived at via reasoning or doxastic deliberation, the evidentialist suggestion that I am taking on board here is that only truth-conducive considerations can play an appropriate role in such reasoning towards beliefs. And only considerations that play an appropriate role in reasoning can be properly understood as bases of that belief. Recall also that evidentialism as it is understood in the present discussion in no way is committed to the claim that all rational beliefs are based on evidence – that is, on truth-conducive considerations (including its own content or not). Maybe there are beliefs that are not arrived at through reasoning (e.g. arguably, perceptual beliefs); evidentialism, in our sense, is not the stricter form of evidentialism that would entail that beliefs not based on independent evidence can be justified. In no way can pragmatism present the fact that possibly some beliefs are not based on evidence as an argument against our version of evidentialism (even if some pragmatists seem to appeal to this line of thought against the versions of evidentialism they are targeting). These, possibly rational, ‘groundless’ beliefs are, I would say, by definition not based on pragmatic considerations, since they are based on no considerations at all and hence don’t constitute an argument for pragmatism. Thus, the debate about the bases has to focus on beliefs arrived at through reasoning as being based on truth-conducive considerations.

Ultimately, the motivation for excluding non-truth-conducive considerations from possibly playing an appropriate role in reasoning towards belief, as I see it, comes from the idea that good patterns of reasoning have to parallel good/sound arguments. That the fact that God exists will bring someone happiness doesn’t seem to be the right kind of consideration to plug into an argument in favour of the existence of God. That the chances that a patient survives rises from 10 per cent to 15 per cent doesn’t seem to constitute an appropriate input, ceteris paribus, into a sound argument towards the conclusion that the patient will survive (cf. Reisner Reference Reisner2008: 18). The examples can be multiplied. If this line of thought is on the right track, then it can explain why pragmatic considerations cannot be normative reasoning reasons to believe. They cannot be appropriately plugged into an appropriate doxastic deliberation and hence cannot be bases that would rationalise the relevant belief. This conclusion is vindicated by a plausible reading of the main claims of the Erotetic view. According to the Erotetic view, normative reasoning reasons are appropriate answers to the reasoning/argument-requiring reading of the ‘Why F?’ question. Assuming that only truth-conducive considerations can constitute appropriate answers of this sort applied to the case of belief leads to the conclusion that only truth-conducive considerations may constitute normative reasoning reasons for belief. Thus, I think that, even if the core commitments of the Erotetic view don’t entail that, strictly speaking, pragmatism is false, a plausible reading of it does, and it does so on what appear to be independently plausible grounds, while also vindicating some of the key insights from broad pragmatism.

Could a pragmatist object to my argument by insisting that I endorse a too narrow view of what good reasoning is? The idea would be that good patterns of doxastic deliberation need not necessarily follow a premise–conclusion pattern but could be similar, say, to a mere list of pros and cons. Crucially, such lists would be similar to, if not a kind of, practical reasoning towards (intention to) adopting a belief. If so, practical considerations could figure in such lists and be part of appropriate inputs in doxastic deliberation (see McCormick Reference McCormick2014, Reference McCormick, Bondy and Carter2019, for an argument along these lines).Footnote ⁹

In reply to this further suggestion, I would say, first, that I am not entirely sure why such mechanisms would count as doxastic deliberation, as opposed to alternative belief acquisitions (e.g. groundless beliefs). And, second, I would like to insist that good patterns of theoretical reasoning – that is, of doxastic deliberation – should be connected to arguments for a hypothesis; whereas practical deliberation is connected to patterns of reasoning leading ultimately to intention/action. And it is not clear how inputs into a practical list of pros and cons for intending/undertaking the effort to get into the state of believing could be parts of the argument for the relevant hypothesis, corresponding to the content of the relevant conclusion-belief. In Logins (Reference Logins2021), I argue, on a similar basis, that one worry with pragmatism is that it seems to lead to the conclusion that practical incentives (e.g. bribes) could constitute good arguments for, say, a philosophical theory, a conclusion that clashes with our very understanding of how philosophy should be done (and is arguably self-undermining for pragmatists who would accept that a bribe for believing that pragmatism is mistaken can be a good argument). See also Way (Reference Way2016) for a similar argument.

More generally, there is an important question we must ask when thinking about good patterns of reasoning. The question to ask is: what’s the point of reasoning? As McHugh and Way (Reference McHugh and Way2016: 588) have convincingly argued, it is reasonable to think that the point of reasoning is to get things right and arrive at fitting responses. We value good reasoning since it is a way of arriving at true beliefs or knowledge, appropriate intentions, appropriate regret, admiration, and so on, assuming we also begin the reasoning from good, appropriate premise-responses. This kind of general picture provides an additional motivation for the idea that only truth-conducive considerations could play a role in doxastic reasoning.

The second worry might come from the evidentialist camp. Some evidentialists might object that contrary to what the present proposal admits, pragmatic considerations cannot be normative reasons of any sort. Be they explanatory or reasoning, pragmatic considerations just don’t have any normative force to count as normative reasons to believe, according to this line of objection. One way to see the worry is that the very notion of normative reasons is tied to normative force, a force that a believer who believes according to her evidence respects. But it is not clear in what sense pragmatic considerations, even if they are restricted only to being possible normative explanatory reasons, could have any normative force with respect to someone who believes. At this point, one might object that either there are no normative explanatory reasons with respect to belief in general or that there are no normative explanatory reasons of the pragmatic variety, for there is no normative force attached to pragmatic considerations with respect to beliefs. The normativity of reasons to believe is, on this picture, essentially attached to the rationality of belief, which in turn depends on how one reasons towards believing on the appropriate bases, bases that can only be constituted by truth-conducive considerations. A relative worry or perhaps a way to develop the general worry further is to stress that the job of normative reasons to believe is to provide material for the assessment of beliefs. But pragmatic considerations can never ground the assessment of a doxastic state. It’s not the right sort of thing to ground the assessment of states of belief, disbelief, or suspension.Footnote ¹⁰

To this worry, I would like to reply by putting forward an inference to the best explanation. Here I don’t have a knock-down argument to show that strict evidentialism cannot be right. Yet I think my proposal provides the simplest overall view about the relevant phenomena. Here is my reasoning. As we have already seen in the case of Moore-paradoxical considerations, one might be tempted to exclude these (and other similar considerations which cannot constitute an appropriate basis for our reasoning) from counting as normative reasons of any sort, especially if one is attracted to a sort of theory of normative reasons that ties normative reasons to good patterns of reasoning. However, the problem there was that if one excludes Moore-paradoxical considerations from the category of normative reasons, one still has to explain why these considerations do seem to speak in favour of the relevant F-ings. In other terms, one is left with the burden of providing a plausible and independently well-motivated error theory about our pre-theoretical judgments that would explain why we are wrong. This is not an easy task. I am not saying that it cannot be done satisfactorily; it is rather that it will necessarily involve complicating the overall picture, arguably, beyond necessity. One can, for instance, invoke the possibility of a distinct normative category that would correspond to Moore-paradoxical considerations that seem to speak in favour of some F-ing but that fall short of counting as normative reasons. The burden, then, is to explain in a non-ad hoc way why then there is such a distinct normative category at all; what the common normative element is between these ‘merely speaking in favour of F-ing’ considerations (e.g. Moore-paradoxical considerations) and genuine normative reasons; and, finally, what the substantive difference is between these categories. My positive proposal was that an alternative approach that admits that Moore-paradoxical considerations are genuine normative reasons is clearly simpler. We should not postulate normative entities and categories beyond necessity. Thus, the suggestion, an inference to the best explanation indeed, was that a view that treats Moore-paradoxical considerations as reasons is preferable. The Erotetic view treats these as genuine normative reasons, more specifically as reasons of the normative explanatory sort. Now, this detour concerning Moore-paradoxical considerations is only supposed to suggest that something similar is also happening in the case of pragmatic considerations to believe. Of course, an evidentialist can dig in his heels and maintain that normative explanatory reasons, in our sense, are not really normative reasons or, at any rate, that normative explanatory reasons of the pragmatic variety are not genuine normative reasons. But then this evidentialist owes us an independently warranted story about why it still appears that in certain situations specifically pragmatic considerations seem to speak in favour of specifically believing the relevant propositions (and not speaking in favour of merely desiring to believe, or trying to get oneself to believe, or undertaking an indirect strategy to get oneself to believe, and so on). Or one has to explain in a non-ad hoc manner why speaking in favour of believing that p is different from being a normative reason to believe that p. Again, I am not suggesting that there are no ways to come up with something like an appealing story of why this happens. The problem is that every possible way of doing it will involve complicating the overall picture by postulating distinctions and categories that don’t need to be introduced on competing views. Indeed, the reasoning here, again, is based on an inference to the best explanation. The present proposal offers a simpler overall picture that doesn’t need to appeal to error theories and look for independent motivation of further distinctions and categories (e.g. some non-normative category for pragmatic considerations that seem to speak in favour of believing some propositions). Of course, it is not a knock-down argument against strict evidentialists. But I think it puts pressure on the thought that pragmatic considerations can never be genuinely normative reasons, in a sense.

Another worry that one might raise against the present proposal is a version of a standard objection to any pluralist/contextualist account within the normative domain. This worry concerns the view’s prospects for dealing with situations where the two reasons – that is, normative reasoning reasons and normative explanatory reasons – appear to conflict; that is, when they speak in favour of incompatible doxastic states. More specifically, the problem is not such tension per se but that the view should provide a credible story of how to combine the two reasons of different sorts – that is, of how to deal with the conflict. But on the face of it, the Erotetic view doesn’t provide any instructions on how to combine reasons of these two distinct sorts. In other terms, if both of these reasons are genuinely of the same sort, one might expect that they have to be commensurable and should be available to be ‘weighed’ in some overall assessment, say, a sufficient overall reason to believe (or overall ought to believe) that depends on/is grounded in the combined sum of all the normative reasons that one has for and against believing in the relevant situation. However, it is not at all clear how one might combine normative explanatory and normative reasoning reasons. They are, to use Selim Berker’s expression again, like ‘oil and water’; they just don’t combine. This worry is also parallel to the objection to moderate pragmatist accounts (refer to our earlier discussion). See Berker (Reference Berker2018) for a thorough and complete discussion of this problem for pragmatist accounts. So, applying this combinatory problem to the Erotetic view’s treatment of the debate on the possibility of pragmatic reasons, one can ask: granted, that one will be practically/eudaimonically/and so forth better off if one believes that p (e.g., say, that one will survive) is a consideration that speaks in favour (in a sense) of believing that one will survive, and that one’s evidence indicates that one will not survive speaks in favour (in a sense) of disbelieving that one will survive, but what should one believe at the end of the day? It still seems meaningful to ask: what is the doxastic attitude that one has to have tout court? It makes sense to think that there is a simple, plain question that we can ask about how one is supposed to decide the matter about believing or not that p. But, then, if this is the case, when one is trying to figure out this plain ‘what should the subject believe at the end of the day?’ question sans phrase, which considerations ‘weigh’ more, reasoning reasons or explanatory reasons, or do they perhaps have the same ‘weight’? One might worry that the Erotetic view cannot explain how to measure these two sorts of reasons against each other and, indeed, how to manage situations of conflict between them in view of answering the plain what should the agent believe question sans phrase.

In reply to this worry, I would like to observe, first of all, that contexts in which one might reasonably ask ‘what should the subject/I believe?’ sans phrase are precisely the contexts where one doesn’t know that one should/ought to believe that p (or not-p, or suspend judgment, for that matter). These are (genuine) inquiry contexts, contexts where one seeks to learn what is the right doxastic attitude for one to have. The second thing to note is that, given the specifics of our distinction between reasoning and explanatory reasons, it should also be clear that in such (genuine) inquiry contexts, there can only be normative reasoning reasons for one to believe/disbelieve/suspend judgment about p. For in such contexts, where it is not known (i.e. one doesn’t know it) that one ought to believe/disbelieve/suspend judgment about p, there cannot be explanatory reasons why one ought to believe/disbelieve/suspend judgment about p. The only considerations that can be part of an appropriate answer to the argument/reasoning-requiring reading of ‘Why believe/disbelieve/suspend judgment about p?’ question can function as normative reasons in such contexts. Thus, the answer to the aforementioned worry is that, of course, it can be meaningful to ask the question ‘What should the subject believe?’ sans phrase, but only in contexts where it is not known that one ought to believe that p (or ought to disbelieve p or suspend judgment). But, one might object, how about the claim that the Erotetic view can vindicate pragmatist intuitions that in some cases pragmatic considerations seem to constitute normative reasons to believe (or disbelieve or suspend judgment)? If there are no normative explanatory reasons in contexts where it is reasonable for us to ask ‘What should the subject believe?’ sans phrase, and pragmatic considerations are identified with normative explanatory reasons, then how could the Erotetic view plausibly vindicate the pragmatist intuitions that pragmatic considerations (e.g. the survival case) can constitute genuine normative reasons in favour of believing in such cases? The key here is to note that observers of the case, say, we who discuss it, may well be in a context where it is known that the subject ought to believe that p (e.g. that she will survive) even if the observers also know that the subject’s evidence speaks against p. We may well be in a context where there are normative explanatory reasons for the subject to believe that p. The pragmatist intuition is vindicated in this context. But, again, in these contexts, it doesn’t make sense to ask the plain ‘What should the subject believe?’ question sans phrase. In these contexts, there are no normative reasoning reasons for the subject to believe that p since it is known what the subject ought to do. In sum, this line of reply to the worry endorses the claim that indeed normative reasoning and normative explanatory reasons are like ‘oil and water’; they never mix, hence there is no genuine conflict. Crucially, this move is theoretically motivated by the existence of two mutually exclusive readings of ‘Why F?’ normative questions, the two readings that give rise to two distinct sorts of reasons – explanation providing and argument/input in reasoning providing sorts. But how about someone who does seem to experience the conflict in the relevant sort of case? The conflict may well appear real in such cases. Consider the agent in the survival case – she knows that believing that she will survive can boost her chances of survival a bit, but she also definitely knows (given her evidence) that even with such a boost her chances of survival are meagre. The agent in such a situation may well feel the inner conflict between the boost-in-chance considerations, speaking in favour of believing that she will survive and the evidential considerations speaking against believing that she will survive. Do we really want to say that she switches between inquiry/explanation contexts seamlessly and perhaps without noticing it as she changes the focus from one to the other of these two considerations? Surely, we sometimes do pass from the inquiry context to the explanation context and back, but arguably not in such a smooth and seamless way. I would like to suggest that the appearance of a conflict arises from elsewhere in such cases. Let’s assume the context in which the subject of the survival is in the context of an inquiry. She really seeks to know what she should believe. Strictly speaking, there are for her only normative reasoning reasons (with respect to forming a doxastic state about whether she will survive). However, she is also aware that it might very well be the case that she ought to believe that she will survive. In other terms, she may realise that the possibility that she ought to believe that she will survive is a very real one. And realising this, she may also realise that the fact that believing that she will survive boosts somewhat the chances of her survival would in such a case explain why she ought to believe that she will survive. She is thus aware that this consideration is a potential explanation of the potential deontic fact that she ought to believe in her survival. But, arguably, she also realises that such a consideration is contrary to the reasons she presently has with respect to believing that p. The tentative suggestion here, then, is that the salient possibility for her that it might well be the case that she ought to believe in her survival because it would boost her chances of survival in the light of her still being in the context of inquiry, and thus having evidential considerations against believing in her survival, is what gives rise to the feeling of inner conflict in this case (and admittedly, in similar cases, modulo specifics of the situation). Indeed, not only is the Erotetic view compatible with the intuition of conflict; it seems it has a theoretical advantage over its rivals, since it provides a more substantial account – compared to, say, pragmatists – of why the conflict arises (and is able to vindicate the pragmatist pre-theoretical judgment that pragmatic considerations sometimes speak in favour, in a sense, of believing). The conflict arises on our proposal because of two fundamentally different activities one might undertake – the activity of inquiry and the activity of explanation. It is not some odd feature about belief or action that explains why there is an apparent conflict in the relevant cases. The appearance of the conflict is there because one is still in the context of an activity of inquiry whether to F, while realising that one option might be particularly salient as a reply to the explanation-requiring reading of the question of ‘Why F?’, since the very possibility that one ought to F becomes salient. Inquiry and explanation are pulling one in different ‘directions’, so to say; hence, the feeling of conflict.

The preceding discussion leads us naturally to the fifth and final worry. Given our response that there is really no genuine conflict between evidential (e.g. normative reasoning reasons) and pragmatic considerations (normative explanatory reasons), and hence that we should not expect that these can be combined, one might wonder in what sense then these two can still be things of the same sort. What’s common to both reasons if they are not the same sort of thing? Why think that both are normative reasons? In other terms, what is the common ‘normative’ element in both? We know from what precedes that it’s probably not rationality, it’s not being grounds of deontic facts, but if so, what is it then?

To this I would like to reply that the common normative element is that both are sorts of appropriate answers to the normative question ‘Why believe/disbelieve/suspend the judgment?’. The fundamental element here is the normative ‘Why F?’ question applied to doxastic states. In general, it’s a key aspect of the Erotetic approach to normative reasons that the normative ‘Why F?’ question is more fundamental than normative reasons – normative reasons of all sorts are reduced to the sorts of answers to this normative question, and hence the unifying element among all sorts of normative reasons is that they are just the possible appropriate answers to that question in one or another of its readings. One might wonder, though, in what sense the ‘Why F?’ question is really normative. In general, in what sense can questions be normative? For one thing, as we have seen already, the ‘Why F?’ question can be translated as ‘Why should/ought S to F?’. Thus, we see that on such an interpretation, ‘Why F?’ questions are about oughts and have in their content a normative element. But is this enough to make questions normative in the relevant sense? It is common to think that normative notions/properties/statuses exercise a certain force on us, a normative force. Typically, this is understood in the sense that they are capable of guiding us. Can the ‘Why F?’ questions have this sort of normative force to count as genuinely normative? As it happens, I think that they can. It seems to be a common idea that guidance comes in the form of commands or imperatives. On such an understanding, statements of reasons or oughts or values, or virtues, or indeed anything normative (if there are other normative things beyond reasons, values, oughts, and virtues), have to encode a command or an imperative (conditional, hypothetical, or categorical), probably by communicating an illocutionary force of commanding, permitting, prohibiting, suggesting, and so forth (cf. Potsdam and Edmiston Reference Potsdam, Edmiston and Aronoff2015). In short, a common understanding of guidance seems to hide an implicit assumption that guidance is to be understood on the model of a command. But why should this be the case? That is, more precisely, why should we think that normative guidance is always in the form of a command (suggesting, permitting, prohibiting)? An alternative way of thinking would be to admit that sometimes guidance happens by questioning and not by a command. One might well guide someone in their quest by asking questions. The Socratic method, as it appears in Plato’s dialogues, is one obvious example that would illustrate our point. Education is another, arguably less colourful, example. Sometimes the best way to ensure that my kids eat vegetables is by questioning them about it. In a situation where they don’t want to eat vegetables and indeed ignore that they should eat them, a strategy that works (sometimes) is to ask whether they prefer to eat, say, broccoli or carrots, which might be understood, without too much of a stretch, as a question of the form ‘Why eat carrots rather than broccoli?’, which fits our general form of normative ‘Why F?’ questions. Similarly, to get them to better understand why they should eat vegetables in the case where they already know that they should, asking some questions helps. By asking questions, I can influence them in their quest for a better understanding through explanation. In short, it seems that genuinely normative guidance may happen through questioning and not only through command. But questioning happens through raising questions, and ‘Why F?’ is a question that can perfectly fit the bill on this account. This is, of course, only a sketch. But the point is that it is not obviously wrong to think that questions can be genuinely normative, even if normativity is tied to force and guidance. In sum, the common normative element in all sorts of normative reasons according to the present proposal is that they are all answers to the normative ‘Why F?’ question; the ‘Why F?’ question is genuinely normative, at least according to a common understanding of normativity that ties normative notions and properties to guidance, for normative ‘Why F?’ questions can genuinely guide one in one’s quest to know what to do and to better understand and explain why one ought to do what one ought to do.

6.5 Concluding Remarks and Further Potential Applications

This concludes our exploration of the possibility of pragmatic reasons for belief. We observed that the present state of the debate about whether some pragmatic considerations may (sometimes) constitute genuine normative reasons to believe (or disbelieve, or suspend judgment for that matter) is in serious deadlock. On one side, we have evidentialists, who maintain that only truth-conducive considerations may be reasons to believe. On the other side, we have pragmatists who insist either that there are also pragmatic reasons to believe, or that all genuinely normative reasons are ultimately pragmatic. Both sides insist that their proposals perfectly fit our pre-theoretical judgments about some cases. We have seen that, surprisingly, they even maintain different interpretations of our pre-theoretical judgments about almost identical cases. In such a situation, it is very tempting to conclude that at least one side of the debate is wrong and, in particular, what they take to be pre-theoretical judgments in favour of their approach actually don’t really support their view. An even more dramatic conclusion that one might be tempted to draw from such an entrenched disagreement is that no one is right here, and we should not look into pre-theoretical judgments when theorising about reasons for belief at all.

The dialectical line that was undertaken in the present chapter was to grant that actually both sides might be right, in a sense. Indeed, the suggestion was that our pre-theoretical judgments about certain cases do support the conclusion that nothing but truth-conducive considerations can count as normative reasons to believe, in a sense. And at the same time, our pre-theoretical judgments about some cases seem to support the conclusion that sometimes pragmatic considerations might be normative reasons for belief, in a sense. The key claim was admitting that this doesn’t lead to a contradiction, if we accept the Erotetic view of reasons. Applying the Erotetic view of reasons to this debate leads to a conclusion that only truth-conducive considerations can be normative reasoning reasons to believe, the reasons that are relevant in the context of inquiry – that is, when one is seeking to reply to the argument/reasoning-requiring reading of the ‘Why believe p?’ question. And it also leads to a conclusion that some pragmatic considerations may constitute normative explanatory reasons for belief, the reasons that are relevant in the context of explaining why one ought to believe/disbelieve/suspend judgment – that is, in the context where it is taken for granted (including by the subject) that one ought to believe (etc.). In other terms, pragmatic considerations are relevant in the context where the ‘Why believe?’ question asks for an explanation-providing answer.

Applying the Erotetic view of reasons to the question of the possibility of pragmatic reasons to believe has a theoretical advantage of fitting with all the relevant pre-theoretical judgments about the cases (i.e. cases where we think that a consideration is a normative reason for a belief). That is, contrary to its rivals (i.e. evidentialists, pragmatists, and eliminitivism – that is, who would deny that there are normative reasons to believe at all), the proponents of the Erotetic view don’t need to appeal to any error theory to ‘explain away’ the intuitions – to tell a story about why our pre-theoretical judgments about some cases should not be taken at face value. Thus, the main argument in this chapter is an inference to the best explanation, and as such it doesn’t represent any ambition to ‘knock out’ alternative proposals. Yet, absent further counter-arguments, the present proposal presents itself as a viable option within the debate. Indeed, given its explanatory power (i.e. that it explains more with less, since we don’t need any error theory on our proposal), I would like to suggest that it is a bit more than a viable option for overcoming the current theoretical deadlock concerning the possibility of pragmatic reasons for belief. We may well be warranted in thinking that it is to be preferred to existing alternatives of an evidentialist, pragmatist, or eliminitivist sort.

Now, the more general lesson that I would like to draw from this explanatory success is that it constitutes a further line of argument in favour of the Erotetic view of normative reasons. Chapter 5 contained a positive (also abductive) case in favour of the Erotetic view, from the considerations about the extensional adequacy of this view and its simplicity in explaining the dual life of normative ‘Why F?’ questions. We saw there that the Erotetic view seems to do better than its rivals on this account. The present chapter can be seen as an additional argument in that it contains a demonstration of a theoretically fruitful application of the Erotetic view to the problem of the possibility of pragmatic reasons for belief.

Interestingly, the application of the Erotetic view to the question of the possibility of pragmatic reasons to believe provides us with insights that might lead to further potentially fruitful applications of the view. One venue to explore would be the general issue of epistemic rationality (or justification). More specifically, the well-entrenched dispute between so-called internalists about rationality/justification, on one side, and externalists about rationality/justification, on the other side, might get a new treatment given the Erotetic view of reasons. Roughly, according to the internalists, rationality/justification of a doxastic state for one supervenes on how things appear from one’s internal perspective (e.g. one’s non-factive mental states); whereas according to the externalists, it (also) depends on external factors to the subject (e.g. the reliability of one’s belief formation mechanisms, one’s abilities and competence, or what one knows). The line that might prove to be fruitful to explore in this context would be to apply the distinction between normative reasoning and normative explanatory reasons to internal and external factors in internalist and externalist theories or rationality accordingly. Maybe we can see internal factors (as in internalist theories) as constituting normative reasoning reasons, and external factors (as in externalist theories) as constituting normative explanatory reasons. If this is on the right track, then the opposition between internalists and externalists can be overcome. However, the exact details of how this would work will be left for another occasion. Here we only register that paying attention to the double nature of normative reasons, as tied to reasoning and explanation, may provide still further insights and fruitful applications with respect to well-known problems in epistemology and meta-ethics.