5.1 Taking Stock

The purpose of the present chapter is to elaborate and motivate a new account of normative reasons. The view in question takes on board the insights from the views discussed earlier. In particular, it relies on the pre-theoretically plausible claims that (i) reasons, in a sense, are somehow connected to reasoning and (ii) reasons, in a sense, are somehow connected to explanation. Before going into details of the view, however, let us take stock.

We began with a pre-theoretically plausible assumption that reasons are considerations that speak in favour of an action or an attitude (cf. Scanlon Reference Scanlon1998). So, for instance, if I promised my friend to come to the party at his place tonight, then my promise – that is, the fact that I promised – is a reason for me to (intend to) go to his place tonight. Or if Michael tells me that he saw Thomas at the concert last Wednesday, Michael’s testimony is a reason for me to believe that Thomas went to the concert.

We also agreed that reasons in this sense are normative and have to be distinguished from considerations that are mere bases on which we act or form/maintain an attitude. In this sense, reasons for one to F (e.g. to act or to hold an attitude) are to be distinguished from reasons for which one acts or holds an attitude. Our focus is on the former, the normative, and not on the latter, the motivating, reasons.

A fundamental question in the debate on reasons is whether there is anything more substantial and positive that can be said about normative reasons. This is also the guiding question of the present work. In the stage-setting chapter (Chapter 1), we assumed that the reasons-first approach (cf. Scanlon Reference Scanlon1998; Schroeder Reference Schroeder2007; Skorupski Reference Skorupski2010; Parfit Reference Parfit2011), according to which we cannot say anything more substantial – for example, analyses about normative reasons (by appeal to normative properties) than the aforementioned mere characterisation but can appeal to reasons in defining/explaining other normative notions – is problematic. For one thing, the reasons-first programme cannot really avoid the so-called wrong kind of reasons problem. Part of the reasons-first programme is a buck-passing account of value, according to which values are defined/explained in terms of reasons. For instance, according to this account, that someone is admirable just amounts to there being sufficient reasons for one to admire the person. The buck-passing account is an important part of the reasons-first approach since it is supposed to show how reasons are more fundamental than values. However, as the wrong kind of reasons problem shows, there are cases where one seems to have considerations speaking in favour of admiring someone while, clearly, the person in question is not admirable (e.g. a demon who threatens to kill everyone unless one admires it; cf. Rabinowicz and Rønnow-Rasmussen Reference Rabinowicz and Rønnow-Rasmussen2004). The problem is that in order to explain why the threat is not a reason, while still being a consideration that speaks in favour of admiring the deplorable person (i.e. the demon), a proponent of the reasons-first programme undermines her account by accepting that there has to be, after all, something more substantial that can be said about reasons in order to distinguish these from considerations that merely speak in favour of F without being reasons. In other words, it is unclear how one can maintain at the same time the idea that normative reasons are undefinable and fundamental while also avoiding the counterintuitive consequences of the wrong kind of reasons type situations.

Moreover, it has been increasingly acknowledged that the reasons-first approach is less parsimonious than its alternatives, and hence, if a viable, more parsimonious account of reasons becomes available, it should be preferable to the reasons-first view. A recent wave of reasons anti-fundamentalism doubts the need for an undefinable notion of reasons as a mysterious entity in the normative realm.

The bulk of the present work, then, was to explore the most promising reductionist alternatives to the reasons-first approach, according to which normative reasons can be reduced to a combination of some normative property and other elements. We have classified the existing reductionist views into two categories. The difference between the two amounts to how exactly the ‘other elements’ are understood here. Two competing views seem to emerge. According to the approach of the first broad category, the other elements here appeal to the function of reasons in reasoning. Thus, according to this sort of approach, r being a reason for one to F just amounts to, roughly, r being part (of a premise) of good/fitting/appropriate [depending on the preferred normative property] reasoning for S that concludes in F-ing. According to the second broad category, normative reasons roughly are parts of an explanation of why one ought to F (or why it’s fitting for one to F) or, alternatively, it is an explanation of why promoting F-ing would be good/valuable/fitting.

In assessing the existing views, we have been guided by what many take to be constraints on or desiderata for a plausible account of normative reasons. Some of these desiderata are implicit in the debates. Different authors focus on different aspects of these. Here is a brief reminder about what might be seen as the main desiderata for a theory of normative reasons in the literature. First, a theory of normative reasons should be extensionally adequate.Footnote ¹ That is, a theory of normative reasons should count as normative reasons all and only considerations that are normative reasons. This, of course, raises the question of how we should go about determining what counts and what doesn’t count as a normative reason. The risk is that in assessing whether something is a normative reason, we are already influenced by our background theory of normative reasons and so we don’t have a theory-neutral ‘method’ for establishing what counts and what doesn’t count as a normative reason. A promising proposal here is to focus on the pre-theoretic, rough characterisation of reasons to F as considerations that count in favour of F-ing (see Scanlon Reference Scanlon1998). Thus, a plausible theory of normative reasons should fit our pre-theoretical judgments about considerations that speak in favour of F-ing.

Second, a viable theory of normative reasons has to respect the motivating–normative reasons distinction while also being able to explain their supposed connection. Often this is taken to imply that there has to be some sort of connection between reasons and reasoning, as we saw in Chapter 2.

Third, a theory of normative reasons has to be tolerant/general enough. That is, it has to apply to reasons to act, but also to reasons to believe, to fear, and other attitudes.

Fourth, the most economical theory is to be preferred. In other words, a simpler theory (i.e. a theory that appeals to a lesser number of entities and principles) that can explain all reasons (reasons to act, to believe, to fear, etc.) is, ceteris paribus, preferable to more complex views.

Fifth, a theory of normative reasons has to respect the ‘currency’ aspect of reasons. Namely, it has to fit well with the observation that normative reasons are useful tools for rough, prima facie characterisation of normative notions in a domain. Think, for instance, of the naturalness of describing a moral ought in terms of what one has most (moral) reasons to do (which is not to say that moral oughts can be fully reduced to what one has most (moral) reasons to do). According to Parfit (Reference Parfit2011: 269), ‘[i]n the conflict between […] various [meta-ethical] theories, reasons provide the decisive battlefield’. This can be understood as a claim that whether, say, naturalism is the right normative theory can be determined by debating the nature of normative reasons. We need not be reasons-first proponents to accept the general point that reasons seem to have this ‘currency’ aspect that enables us to express (roughly) and debate various normative theories in somewhat neutral terms. The currency aspect is also illustrated in the context of domain-specific debates. For instance, appeal to normative reasons seems to allow us to present clearly the possible view of pragmatism in epistemology (as opposed to evidentialism). Arguably, this currency aspect can also be understood as the idea that reasons have to be somehow connected to the explanation of normative properties/facts. If there are explanations of why, say, one ought to F, then, plausibly, such an explanation is best provided in terms of normative reasons.

Sixth, a theory of normative reasons is supposed to respect the ‘gradable’ aspect of normative reasons. It is easy to observe that reasons admit of comparisons, combinations, outweighing, and so on (for example, my promise to go to the party is a ‘weightier’ reason than my desire to binge-watch BoJack). Many think that this comparative aspect is best captured by the talk of ‘weight’ of reasons or, at any rate, by an appeal to some sort of scale and the idea that reasons are scalar properties.

In the previous chapters, Chapters 1–4, we have seen how exactly the Reasoning view and the Explanation view cope with the aforementioned constraints/desiderata. We have seen that the best versions of both approaches fit easily with some of the constraints, while failing to respect others. So, for instance, versions of the Reasoning view can account straightforwardly for the motivating–normative reasons connection and distinction (the reasons–reasoning connection is at the very core of that approach). They are also tolerant and economical (and, arguably, can deal with the currency aspect). Yet they (at least some of them) have a hard time accounting for the gradable aspect of reasons. Some sophisticated accounts manage to deal with outweighed reasons, and yet the more fundamental question of what the ‘weight’ of reasons amounts to exactly remains largely unexplained by even the most promising versions of the Reasoning approach.

The Explanation views are tolerant and economical, and they respect the currency aspect. They also respect, at least on the face of it, the gradability aspect. But they are not really apt to account for the reasons–reasoning connection.

Moreover, neither the Reasoning approach nor the Explanation approach fully manages to respect the extensional adequacy constraint. They both, in their own ways, over-generate and under-generate normative reasons. The Reasoning approach has a hard time explaining why mere enabling conditions (e.g. that one is able to F) and entailing conditions (e.g. that one has a reason to F) are not normative reasons. It also predicts that some considerations that speak in favour of F-ing are not normative reasons (e.g. Moore-paradoxical considerations and self-undermining beliefs). The Explanation views that distinguish pro toto from pro tanto reasons as sorts of explanation of why one ought to F, but accept that both are considerations that count in favour of F-ing, are bound to double-count reasons that there are for one to F. They also predict that considerations that don’t seem to speak in favour of F-ing are nonetheless partial explanations of why I ought to F (e.g. against-F considerations in weighing explanation of why one ought to F). Other versions of the Explanation approach (e.g. value-based versions) have a hard time explaining how there might be normative reasons for attitudes (e.g. on some of these views, there are no reasons to fear, or to be happy, and they have difficulties in explaining reasons to believe). They predict that there are no reasons where our pre-theoretical judgments imply the contrary and thus they face the ‘right kind’ of reasons problem. Thus, we have seen that both the Reasoning approach and the Explanation approach in their most promising forms capture important aspects of our pre-theoretical notion of reasons, while both also have some substantial and, in my view, insurmountable difficulties.

This dialectical situation, then, leads us back to square one. We had better not accept reasons-first and look for a plausible reductionist account, but, at the same time, we don’t have any satisfying reductionist account available. The best existing reductionist accounts of reasons are problematic. Is there any other viable approach to reasons available at all? At this point, one might think that we should really be sceptical about reasons altogether. One might be tempted at this point to endorse an eliminitivism about normative reasons. One can observe that there is no one single thing that would respect all the constraints/desiderata that philosophers have imposed on normative reasons. There is no one single sort of thing that could satisfy all the supposed functions of the concept of normative reasons. Thus, an eliminitivist might conclude that it’s not really meaningful to talk about normative reasons. On such a view, there are no normative reasons, reducible or not. Talk about reasons just is not meaningful on such an extreme approach.

In this chapter, I propose to elaborate a view that avoids such a radical eliminitivist/sceptical conclusion while sharing some of its premises. I agree with the eliminitivist on the observation that there is no one single sort of thing that corresponds to normative reasons. Indeed, we should reject the monist assumption that dominates the debate about normative reasons. And yet we should not be eliminitivists. It’s still meaningful to talk about normative reasons. On the view elaborated a little further, there are two fundamentally distinct and yet somehow connected sorts of normative reasons – connected in that both are normative reasons. There is a unified element of the two that has been overlooked in the debates. But this element is essential to understanding reasons and to understanding why the two sorts of normative reasons are indeed two species of a genus. The key element of our account is a shift in the focus on normative questions and reasons as possible answers to a normative question. The aim of the next section, Section 5.2, is to put the new proposal on the table.

5.2 The Erotetic View of Reasons

According to the view developed in this chapter, a normative reason to F is, roughly, a possible appropriate answer to the normative question ‘Why F?’. This view, I suggest, is able to respect the desiderata/constraints on a theory of reasons, introduced earlier. It integrates the best aspects of both the Reasoning and the Explanation views while avoiding their respective pitfalls. Before looking at the details of the view and the arguments in its favour, let me take a step back and describe two major insights on which the present account builds.

The first insight comes from Pamela Hieronymi’s treatment of reasons. In her influential paper on the wrong kind of reasons problem, Hieronymi (Reference Hieronymi2005) notes that we should approach the question of what reasons are by focusing on reasons as considerations that bear on a question. According to Hieronymi, the reasons-first view about reasons cannot be maintained. The reasons-first proposal that reasons just are considerations that count in favour of an action or an attitude generates robust ambiguity in certain sorts of cases. The problem, according to Hieronymi, is that the proponents of the reasons-first approach are focusing on the wrong kind of relation in characterising reasons. Instead of thinking of reasons (exclusively) as considerations that count in favour of an action or an attitude, Hieronymi (Reference Hieronymi2005: 438) suggests that ‘we would do better to think of a reason as a consideration that bears on a question’. Now, as we saw earlier when we classified Hieronymi as a proponent of the Reasoning view, she also thinks that reasons are things that play a role in reasoning. She writes: ‘To start reflection, we can note that, most generally, a reason is simply an item in a piece of reasoning’ (Hieronymi Reference Hieronymi2005: 443). Hence, the relation of bearing on a question, according to Hieronymi, is to be understood in a reasoning-centred sense. The bearing on a question relation that captures normative reasons, according to Hieronymi, is a relation of bearing on a question that one aims to answer in a piece of reasoning (or deliberation). Bearing on a question is closely related, if not equal, to bearing on a conclusion (of reasoning), according to Hieronymi (Reference Hieronymi2005: 444, fn 16): ‘One could say [that a reason is], “a consideration that bears on a conclusion.” I do not think there would be any relevant difference [between this proposal and the claim that a reason is a consideration that bears on a question], though I find the idea of answering a question more intuitive for capturing the activities of rational agents’. Now, the relevant point for us, the foundational insight for our new account, is that we can take on board Hieronymi’s point about reasons being considerations bearing on a question without accepting her other point about reasons always being items in a piece of reasoning. How exactly this can be done will appear more clearly in what follows.

The second insight on which the present account builds comes from a well-known, indeed classic, observation in theory of argumentation, informal logic, and rhetoric. According to this observation, a ‘Why?’ question can be interpreted in at least two different ways. When someone asks ‘Why is such and such the case?’, one may be in the business of asking for an explanation of why such and such is the case, or, alternatively, one may be in the business of asking for an argument for why such and such is the case. For a locus classicus on this distinction, see Whately’s erotetic logic (cf. Whately Reference Whately1827, Reference Whately1828; see also Prior and Prior Reference Prior and Prior1955 on Whately’s erotetic logic). For example, when one asks why dolphins are not fish, one may be in the business of asking for, say, an evolutionary explanation of how swimming mammals evolved and so on. In such a case, typically, one is not challenging the assumption that dolphins are indeed not fish. One is only asking for an explanation of this fact. Presumably, in most typical cases, the aim of asking for an explanation is to gain a better understanding of the relevant fact (except, of course, in the context of exams, quizzes, and suchlike). Alternatively, one may ask the question ‘Why are dolphins not fish?’ in a more challenging way. By posing such a question, one may ask for an argument that would establish and support the conclusion that dolphins are not fish. In such cases, one is typically challenging the relevant supposition. In our case, one would be asking in such a context for premises that deductively, inductively, probabilistically, or abductively would support the conclusion that dolphins are not fish. A proper answer to this reading of the ‘Why are dolphins not fish?’ question might consist, for example, in one replying that dolphins are not fish because they are not cold-blooded, because they are mammals, or because that’s what biologists have shown to be the case. In short, every ‘why is such and such the case?’ seems to be interpretable as either an explanation-requiring question or an argument/reasoning-requiring question.

This observation is closely linked to a well-known distinction in philosophy of science, namely the orthodox assumption that there is a substantial distinction between explanations and arguments/reasoning. Pace Hempel (cf. Hempel Reference Hempel1965) and logical positivists, it is widely admitted that scientific explanations and scientific arguments/reasoning are not the same. One well-known argument comes from Salmon (Reference Salmon and Salmon1971), who notes that while one can, in principle, add any number of true premises to a sound argument without spoiling it, the same cannot be done in the case of an explanation (see also Section 3.3 for a discussion). An explanation containing, say, irrelevant but true claims (say, necessary truths) is not as good as a simpler explanation that doesn’t appeal to the irrelevant truths.

For a recent treatment of the distinction in the theory of argumentation, see McKeon (Reference McKeon2013), who interestingly enough sees the distinction between explanation and argument as less substantial than has sometimes been assumed in the post-positivist literature. Interestingly, the common element between explanation and arguments for McKeon is that they both offer reasons: ‘For purposes of this paper, arguments and explanations are taken to be products of reason-giving activities: in presenting an argument or an explanation, one gives reasons for a proposition’ (McKeon Reference McKeon2013: 284). Another interesting observation is that McKeon doesn’t doubt at all the distinction between the speech acts of explaining and the speech acts of arguing, which is, again, entirely in accord with what we suggest here.

It has to be stressed that the explanation–argument distinction is not limited to the scientific domain only. It seems pretty commonsensical to assume that there are two distinct speech acts involved here. The speech act of explaining anything seems to be quite different from the speech act of arguing (presenting a pattern of reasoning/argument). One is in a different sort of linguistic business when one is aiming for one’s interlocutors to understand something and when one is in the business of aiming to convince one’s interlocutors of something. These are distinct intentions, corresponding to two distinct speech acts: explaining versus arguing (presenting premises in an argument/pattern of reasoning).

Now, putting together these two aforementioned insights – that is, question centrality and the explaining–arguing distinction – a new proposal about normative reasons emerges. On this account, normative reasons are to be understood as appropriate answers to the normative ‘Why F?’ question in one or the other of its possible readings (as per usual ‘F’ stands for verbs referring to actions or attitudes that can have reasons in their favour). Hieronymi is right in focusing on questions in thinking about reasons. But her account fails to integrate the crucial observation that some instances of questions require as an answer an explanation, while others require an argument/reasoning. Once we integrate this observation about the duality of ‘Why?’ questions, a new version of a question-centred (erotetic) account of normative reasons emerges. To see the view in detail, let us start by putting on the table a succinct general statement of the view:

This formulation needs some unpacking and further specifications. The first thing to note is that for a consideration to be a normative reason on this account, it has to be (a part) of the content of a response to a question, but it is not required that the response has been actually given or even entertained by someone. It is not required that the question be actually asked by anyone. There only needs to be an appropriate response to the relevant ‘Why F?’ question. We are focusing on questions and possible responses as context-sensitive abstracta. A subject need not be aware that there is any possible ‘Why F?’ question directed at her.Footnote ² (In this respect, the proposal here is similar to the proposal of the proponents of the Reasoning view; cf. McHugh and Way Reference McHugh and Way2016 who suggest focusing on the patterns of reasoning, the abstract entities.) We could similarly say that the focus here is on patterns of questions, not the concrete pieces, episodes of asking questions. So, for instance, that there is a drowning child in the pond next to me is a reason for me to jump into the pond. I may be unaware of the fact that there is a drowning child in the pond. Thus, arguably I don’t possess the reason to jump into the pond. Yet this doesn’t change that that a child is drowning next to me is a reason for me to jump into the pond. The present proposal can account for this aspect easily: there is a normative question that can be addressed to me, namely ‘Why jump in the pond?’ and that the child is drowning in the pond is an appropriate response to at least one reading of the question. That no one concretely asks me this question doesn’t alter the fact that there is a question that someone could ask me: ‘Why jump in the pond?’

The aforementioned statement of the view appeals to ‘Why F?’ questions. But we should not put too much weight on the exact formulation of the question. As far as I can see, nothing substantial would be lost if we paraphrased the question in the statement of the view as ‘Why should/ought S to F?’ question. In what follows, I take ‘Why F?’ as addressed to S to be roughly equivalent to ‘Why should/ought S to F?’Footnote ³

We’ve observed earlier that ‘Why?’ questions admit of two distinct readings. Consequently, there are two distinct possible appropriate ways to answer ‘Why?’ questions. Plugging this observation into our account of normative reasons, we now have a consequence that there are two possible readings of the normative ‘Why F?’ question and consequently two distinct ways to answer the normative ‘Why?’ question. Let’s start with questions.

According to one reading of the ‘Why F?’ question, the addressee is required to provide (or to point to) an explanation of the normative fact, namely of the fact that the subject of the question should/ought to F. One cannot provide an explanation of X unless X. Thus, the ‘Why F?’ question on this reading presupposes that it is the case that the subject of the question ought to/should F. In conversational contexts where this reading is the relevant one, both the addresser of the question and the addressee take it for granted that the subject of the question ought to/should F. That the subject ought to F is not challenged and stands in no need of defence. The addresser of the question is asking for an explanation of what makes it the case that(alternatively, on what grounds) the subject ought to F. In standard explanatory ‘Why F?’ contexts, the addresser of the question aims to arrive at a better understanding of the normative fact that S ought to F.

According to the other reading of the ‘Why F?’ question, the addressee is required to provide a (appropriate) premise of a good argument/pattern of reasoning that would support as a fitting conclusion S’s F-ing or the claim that S ought to F. Roughly, in contexts of conversation where this reading is the relevant one, the addresser of the question is asking for any (appropriate) consideration that S could use in a sound reasoning towards S’s F-ing (the conclusion that S ought to F). The ‘Why F?’ question on this reading doesn’t presuppose that it is the case that S ought to F. That is the thing that needs to be shown to be the case. Typically, the addresser of the question is either challenging or in a position to challenge the claim that F-ing is fitting (or that S ought to F). Arguably, in typical contexts of conversation where this reading of the ‘Why F?’ question is the relevant one, the addresser aims to know whether S ought to F or not by asking for an argument that would support such a conclusion.

This dual life of the ‘Why F?’ question leads naturally to the view that there are two distinct sorts of appropriate answers to the ‘Why F?’ question. One way (pattern) is to provide an appropriate answer to the explanation-requiring reading of the ‘Why F?’ question. The other way (pattern) is to provide an appropriate answer to the argument/reasoning-requiring reading of the ‘Why F?’ question. These are distinct ways (patterns) of replying, since, as we have seen earlier, providing an explanation and providing an argument are two distinct things (corresponding to two distinct speech acts).

Given these specifications, we can provide a more explicit statement of the Erotetic view of normative reasons:

We can also read the first rough statement of the aforementioned Erotetic view as a way to point to the common element in considerations that count as normative reasons, and the second, more explicit statement as a way to specify the exact differences between the two sorts of normative reasons.

Finally, a very general comparison to the existing approaches is in order (we will see more specific comparisons in a moment). The Erotetic view agrees with the reasons-first view that the best way to pre-theoretically characterise normative reasons is by focusing on considerations that speak in favour of F-ing, something that we seem to be able to grasp without a theory. This is the best way to be clear about the phenomena that we are interested in theorising about. Yet the Erotetic view rejects the reasons-first presupposition that normative reasons cannot be explained/defined in more basic terms. There is a reductionist account of reasons available; hence, reasons are not prime. The Erotetic view agrees with the Explanation approach in that normative reasons are, in a sense (parts of), explanations of deontic facts (e.g. that S ought to F). Yet the Erotetic view disagrees with the Explanation approach in that it doesn’t assume that normative explanatory reasons are the only sort of normative reasons. The Erotetic view agrees with the reasoning reductionist approach in that normative reasons are, in a sense, contents of premises in good patterns of reasoning. Yet the Erotetic view disagrees with the Reasoning approach in that it doesn’t assume that normative reasoning reasons are the only sort of normative reasons. The Erotetic view agrees in a sense with the reason sceptics/eliminitivists that there is no one single notion of normative reasons that could play all the theoretical roles that one might think are the roles of our common-sense concept of reasons to act/have an attitude. Yet the Erotetic view disagrees with the sceptics/eliminitivists in that this would show that the very concept of normative reasons is incoherent or that there is no meaningful way to appeal to it in our (meta-)normative theorising. According to the Erotetic view, there is a common core in the two sorts of normative reasons, and this common core makes it the case that we are not lumping together two completely unrelated concepts under a common label of ‘normative reasons’. The common core is that of being an appropriate answer to the normative ‘Why F/Why ought S to F?’ question. Now, given that there are two ways of understanding this normative question, it is inevitable that there are two appropriate ways of replying to it, depending on which is the relevant reading. We have two sorts of reasons, because we have two possible readings of the normative, fundamental question ‘Why F?’. This duality of readings of the fundamental normative question, I think, has been largely overlooked in the contemporary debates about normative reasons. But it is precisely because of this duality that all the trouble and opposition arises. The Erotetic view not only shows what is right and what is mistaken in existing views but also has the resources to provide a reasonable error-theory of why philosophers might have accepted mistaken conceptions. Roughly, it is because the distinction between asking for/providing an explanation and asking for/providing an argument can be easily overlooked, and one may easily focus only on one of these. The proponents of the Reasoning view have focused on the providing arguments aspect, whereas the proponents of the Explanation view have focused on explanation at the expense of reasoning. But it suffices to think about the difference between these two to see that the two are distinct and cannot be reduced one to another. Thinking about the non-normative ‘Why?’ questions might prove itself to be especially helpful for grasping the robust difference between requiring and providing an explanation and requiring and providing an argument. There is no reason why this robust distinction should not apply to the normative ‘Why?’ question. The Erotetic view works out the implication of this distinction for our theorising about normative reasons.

Before turning to some more detailed arguments in favour of the Erotetic view, let me first provide some toy examples to illustrate the two sorts of the normative ‘Why F?’ questions and correspondingly two sorts of normative reasons that the Erotetic view describes.

Soup. Consider the normative question ‘Why eat soup tonight?’ directed at my kids. That is, consider the question that one (for instance, the kids themselves) might ask about whether my kids ought to or should eat soup tonight. The Erotetic view predicts that there are two possible readings of this possible question. On the first one, the question is about the explanation, indeed the grounds of the normative fact that my kids ought to eat soup tonight. On this reading, one is asking what makes it the case that my kids ought to eat soup. This reading of the question takes it for granted that it is the case that my kids ought to eat soup tonight. Typically, this reading of the question is relevant in contexts of conversation where one wants to acquire a better understanding of the grounds of the normative fact that my kids ought to eat soup tonight – namely, in contexts where one wants to know: where does this ought come from, or in virtue of what other considerations/facts does this ought hold?

On the second reading of the question, it is about an argument/reasoning for the conclusion that my kids ought to eat soup. On this reading, one is asking whether it really is the case that my kids ought to eat soup tonight. It is in no way assumed that my kids ought to eat soup tonight. Typically, this reading of the question is relevant in contexts where one (perhaps the kids themselves) is challenging the claim that my kids ought to eat soup.

These two readings of the ‘Why eat soup tonight?’ question give grounds to two distinct sorts of reasons for my kids to eat soup tonight. The first corresponds to the explanation-requiring reading of the ‘Why eat soup?’ question. The normative explanatory reasons for my kids to eat soup tonight correspond to appropriate answers to this explanation-requiring reading of the question. What are these? Note that, arguably, mere appeal to healthy aspects of soup, vegetables, and vitamins will not work. For there are tons of healthy foods that contain vegetables, vitamins, and so on. What makes it the case that my kids ought to eat soup and not, say, salad, broccoli, or kale tonight? In virtue of what ought my kids to eat soup and not plain kale? It seems that an appropriate answer to the explanation-requiring ‘Why eat soup?’ question here will appeal to my parental decision to buy soup for the kids’ dinner (yes, I buy rather than prepare soups). The fact that I have authority over and responsibility for what my kids eat for dinner (and, presumably, that eating soup isn’t anything bad for my kids) constitutes the relevant background conditions for this explanation to be appropriate. Thus, the fact that I chose and bought soup for the kids’ dinner is a normative reason for my kids to eat soup tonight. (Another consideration that can be a normative reason of this sort in this context is that there is fresh soup at home, and unless we eat it, it will be wasted, or that there is simply nothing else at home apart from the soup, and so on.) That I bought soup speaks in favour, in a sense, of the kids eating it tonight. This reason is a normative explanatory reason, the sort of reason that is salient in contexts where participants of a conversation are not challenging the claim that my kids ought to eat soup. It is a content of an answer to an explanation-requiring question of ‘Why ought my kids to eat soup tonight?’.

The second sort of normative reasons for my kids to eat soup corresponds to an argument/reasoning-requiring reading of ‘Why eat soup?’. The normative argumentative/reasoning-centred reasons for my kids to eat soup correspond to appropriate answers to the argument/reasoning-requiring reading of the ‘Why eat soup?’ question. What are potential examples of these? Note, first, that, typically, that I bought soup will not constitute an appropriate answer to the ‘Why eat soup?’ question in its argument-requiring reading. If kids are in the mood to challenge the claim that they ought to eat soup, then bringing up the fact that I bought the soup will not provide any support for that conclusion, and rightly so. It is not (pace special circumstances) a premise in a good pattern of argument/reasoning towards the conclusion that the kids ought to eat soup. That I bought it doesn’t support on its own the claim that they have to eat it. That it contains vegetables and is good for their health seem to be more appropriate responses to the argument-requiring question. It is the argumentative/reasoning reason for my kids to eat the soup. That the soup is good for their health is a consideration that can be a content of a fitting premise-response in a good pattern of reasoning/argument from this premise-response (together with other premise-responses, presumably, such as that my kids desire be healthy) to the conclusion-response of them eating soup (or the conclusion of a sound argument that they ought to eat the soup). Thus, there are two sorts of reasons for my kids to eat soup tonight. One is explanation-centred, and the other is argument/reasoning-centred. Both correspond to patterns of appropriate answers to the normative ‘Why F?’ question.

Army.Footnote ⁴ Consider the question ‘Why go North?’, as concerning soldiers in a battle – that is, the question ‘Why ought soldiers to go North?’. On the explanation-requiring reading, this question asks for considerations that explain, ground, or make it the case that soldiers ought to go North. It is assumed that soldiers ought to go North. An appropriate answer to the question on this reading may be that a general has ordered soldiers to go North. That a general has authority on a battlefield over soldiers is part of the background considerations. That the general orders soldiers to go North seems to be a reason for them to go North. The Erotetic view explains this. That the general orders soldiers to go North is the content of (a pattern of) an appropriate answer to the explanation-requiring reading of the question ‘Why ought soldiers to go North?’.

On the argument/reasoning-requiring reading, the question ‘Why go North?’ asks for considerations that would constitute premises in a good pattern of an argument/reasoning towards the soldiers going North (towards the conclusion that the soldiers ought to go North). It is not assumed in this reading that it is the case that soldiers ought to go North. An appropriate answer to this reading of the question might appeal to the fact that going North will surprise the enemy and will put the army in a better position to win the battle. That the general has decided that the soldiers should go North on its own need not always be an appropriate answer in this context (e.g. this is clear in a context where the general’s authority is questioned). That going North will surprise the enemy and will put the army in a better position to win the battle is a reason, in a sense, for soldiers to go North. The Erotetic view explains this. That going North will surprise the enemy and will put the army in a better position to win the battle is content of an appropriate premise in a pattern of good reasoning from these premise-responses with this content to the conclusion of soldiers going North (to the claim that the soldiers ought to go North).

Hijacked flight. Consider the following example from a different context that nevertheless seems to point to the same distinction that we are drawing. ‘[i] Tom is worried that his wife was on the two o’clock flight, because that’s the one that was hijacked (attitudinal). [ii] Tom is worried that his wife was on the two o’clock flight, because she said she’d be arriving early in the evening (epistemic)’ (Gordon Reference Gordon1987: 35). This example comes from the context where Gordon discusses two sorts of reasons for emotions (and we’ve seen in Section 3.3 that this example may be used to introduce the fittingness–epistemic justification distinction with respect to possible assessments of emotions). Here I would like to suggest that this example illustrates equally well the explanation-reasons versus reasoning-reasons distinction while stopping short of a full endorsement of the idea that the two distinctions in the assessment of emotions (fittingness-justification and reasoning-reasons-explanation-reasons) are about one and the same thing. This might be so, but this is not the place to pursue this issue. The point here is a simpler one. Consider the question ‘Why worry that my wife is on the two o’clock flight?’ as considered by Tom. It seems that the reason provided in (i) the aforementioned is an answer to an explanation-requiring reading of that question. In this context, it is known that Tom should worry that his wife took the two o’clock flight. But the reason provided in (ii) seems to be more of an argument/premise in a reasoning sort (of course, again, the reasoning-argument here is to be understood broadly enough to encompass processes by which we arrive at emotional attitudes). In the context where (ii) is an appropriate answer to the ‘Why should Tom worry?’ question (‘Why should I worry that my wife took the two o’clock flight?’ as considered by Tom), it is not known that Tom should indeed worry. That his wife said she’d be arriving early in the evening is something that supports the conclusion that she was on the two o’clock flight. And thus it supports, in a pan-argument sense, Tom’s being worried.

P therefore p. Finally, consider a case of belief. Consider the question ‘Why believe that it is raining outside?’ that applies to me at this moment. On one reading of ‘Why should/ought I to believe that it is raining outside?’, one is asking for an explanation of the fact that I ought to believe that it is raining outside. (Let us postpone the question about the nature of the ought that arguably applies to the doxastic case. Also, if you think there are no positive oughts for beliefs, the proposal here might be easily adapted by an appeal to negative obligations – for example, that it is not the case that one is required to believe that such and such is the case.) An appropriate answer to this reading of the question ‘Why believe that it is raining outside?’, arguably, might just be that it is raining outside (with the assumption that truth is the norm of belief or, alternatively, that knowledge is the norm of belief). On the other reading of the question, one is asking for a premise towards the conclusion that it is raining outside. An appropriate answer to this second reading might be that through the window, I see raindrops falling (that it is raining outside, arguably, will not be an appropriate answer to an argument-seeking question, the question that doesn’t assume that it is raining outside). Notice an intriguing aspect of our view already (and we will come back to it later), that there is a reading on which p is a reason for one to believe that p, and there is a reading on which p is not a reason to believe that p.

The aforementioned examples (or most of them) were such that the considerations that constituted normative explanatory reasons were distinct from the considerations that were normative reasoning/argumentative reasons. These examples were supposed to make the point about the genuine distinction between the two sorts of normative reasons. Yet, of course, in many ordinary cases, the same consideration may count as both a normative explanatory reason and a normative reasoning reason in different contexts. Consider, for instance, a promise. The fact that I promise to go to my friend’s party may well play a role in an appropriate answer to an explanation-requesting reading of the ‘Why go to the party?’ question as well as in an appropriate answer to an argument/reasoning-requiring reading of the question, depending on common knowledge and intentions of the locutors. That I made that promise is something that can play a role in explaining, grounding the fact that I ought to go to the party, as well as playing a premise content role in an appropriate pattern of reasoning from, say, my belief that I made the promise plus some other fitting attitudes (say, intention to respect the promise) towards the conclusion-response of going to the party (or the claim that I ought to go to the party). However, as the aforementioned examples showed, the fact that the same considerations can often play both roles should not mislead us into thinking that there is only one sort of normative reasons. Compare the fact that in some cases p is a proper answer to a normative explanation-requesting ‘Why?’ question and also a proper answer to an argument-requesting non-normative reading of the ‘Why?’ question doesn’t show that there is no genuine distinction to be drawn between arguments and explanations. The same applies to answers to normative ‘Why F?’ questions.

Arguments in favour of the view. Let us now turn to arguments in favour of the Erotetic view. First, and almost platitudinously at this point, the Erotetic view fits perfectly with the observation that there are, in general, two readings of ‘Why?’ questions and corresponding appropriate answers to these. There are explanation-centred and argument/reasoning-centred ‘Why?’ questions. If this duality of ‘Why?’ questions were to somehow disappear when we focus on normative ‘Why?’ questions, it would certainly be somewhat mysterious and would stand in need of a further explanation. The Erotetic view has here a clear advantage in that it doesn’t postulate any miraculous change of ‘Why?’ questions at the level of normative ‘Why?’ questions. All the ‘Why?’ questions have two readings and not just non-normative ones. The Erotetic view fits perfectly with this natural observation about ‘Why?’ questions in general.

Second, turning to the abovementioned constraints or desiderata for a theory of reasons, the Erotetic view can account for the observation that there seems to be some connection between motivating and normative reasons. As observed earlier, this intuitive link is often understood in terms of reasons being able to play a role in reasoning. The Erotetic view has it that one sort of normative reasons are appropriate contents of premise-responses in patterns of good reasoning. It is so where these (reasoning/argument-centred) contents are what is required by the corresponding ‘Why F?’ questions. Thus, on the Erotetic view, some normative reasons are essentially connected to reasoning (broadly construed). Assuming that motivating reasons are considerations that play the (content of the) premise role in reasoning tout court, the Erotetic view has it that one sort of normative reasons just are considerations that could be possible good motivating reasons. (A corollary that we will see later is that the connection between motivating and normative reasons should appear plausible only with respect to reasoning/argumentative normative reasons, which seems to be the case.)

Third, the Erotetic view satisfies the tolerance/generality constraint. That is, it can be easily applied in the case of reasons to act, but also in the case of reasons to believe, to fear, and to hold other attitudes. The aforementioned examples constitute an illustration of this point.

Fourth, the Erotetic view satisfies the desiderata of economy. That is, the view achieves great explanatory power (i.e. explaining all normative reasons) with a relatively small number of initial assumptions and without postulating new entities. The view only appeals to two possible readings of ‘Why?’ questions as applied to the normative (deontic) domain (e.g. oughts/shoulds), an explanation-requiring and an argument/reasoning-requiring reading and two corresponding possible patterns of answers to the readings of this question. It is a reductive view of reasons, and as such, it ultimately explains reasons in terms of other properties and entities and is not committed to any additional ontological assumptions about reasons being irreducible entities.

Fifth, the Erotetic view satisfies the gradability aspect of reasons. My promise to go to a restaurant with a friend seems to be more important than the fact that I will get some satisfaction from binge-watching BoJack instead of going to the restaurant. The proposal here is that we need not appeal to ‘weight’ to account for this apparent gradability (and outweighing). The Erotetic view inherits the strategy used by proponents of the Reasoning view who appeal to the defeasibility of good reasoning. Recall that on that view, roughly, when p is a reason to F and r is a reason to G (incompatible with F), p outweighs r, just in case adding p to the reasoning from r to G triggers the ceteris paribus clause and functions as a defeater but not the other way round (i.e. adding r to the reasoning from p to F doesn’t figure as a defeater for that piece of reasoning). The variation in the apparent ‘weight’ of reasons for one and the same F-ing can be explained on the present view in a similar way. In a situation where p and r are both reasons to F, but p is ‘weightier’ than r, the reasoning from p to F is more resilient in the face of potential defeaters than the reasoning from r to F. The Erotetic view can thus account for our pre-theoretical judgments about the gradability of (the importance of) reasons by endorsing the existing appeal to defeasible good patterns of reasoning.

Sixth, and this is a crucial point in comparing our view to the reductionist accounts discussed earlier, the Erotetic view satisfies the extensionality constraint. That is, it classifies as reasons all and only things that we take pre-theoretically to be reasons. (Recall that our rough guide here in our pre-theoretical assessment of considerations as normative reasons appeals to the idea that reasons are considerations that count somehow in favour of an action or an attitude.) This is where the Erotetic view really has an advantage over its rivals. Both the Reasoning view and the Explanation view over-generated and under-generated reasons. They implied in their various versions that some considerations are reasons where they don’t seem to be, and they failed to account for considerations that seemed to be reasons to act or have an attitude. The Erotetic view does better on this account. It is able to avoid the pitfalls of previous accounts in over-generating and under-generating reasons. Let me provide some details on this crucial point.

We observed earlier that the Reasoning view predicts that considerations that count as mere enabling conditions are normative reasons. Yet, of course, the enabling conditions, such as the fact that I am able to F, cannot (always) count as normative reasons to F. And it seems that entailing conditions, such as ‘I have a reason to F’, are not reasons to F. These are cases where the Reasoning view over-generates normative reasons. The Erotetic view avoids this by restricting normative reasoning reasons only to these contents of possible premises of good reasoning/argument that are appropriate answers to the relevant readings of ‘Why F?’ questions. The Reasoning view is committed to the claim that all contents of fitting premise-responses of a good pattern of reasoning to F (where the subject has at least some of the relevant premise-responses) are normative reasons for the subject in question. The Erotetic view is not committed to this claim. Consider the case of abilities. Certainly, there is a good pattern of reasoning such that the fact that I am able to meet my friend for lunch is a content of a premise-response of such reasoning. Yet it is not clear that the fact that I am able to meet my friend for lunch is always a content of an appropriate answer to the question ‘Why meet my friend for lunch?’. Maybe sometimes it is. For instance, when I am the only one in a group of several friends who is actually able to go for lunch. We have all promised to meet him for lunch, but all except me are sick and incapable of going outside. In such a context, the fact that being unable to go for lunch is an excuse for not going is made salient. Thus, it may be appropriate to say that the fact that I am able to go is a reason for me to go even if all the other friends are unable to go. I am not relieved from my promise even if the other friends are. Maybe in such a context that I am able to go for lunch is a reason for me to go. If so, the Erotetic view can account for it: this consideration is part of a good pattern of an answer to the ‘Why go for lunch?’ question. But most often, appeal to the mere ability to F will not be part of an appropriate answer to ‘Why F?’ questions. Typically, in ordinary deliberative contexts where the salient thing is to find premises for a sound argument/good reasoning towards going for lunch with a friend, mentioning one’s being able to go for lunch won’t be part of the relevant considerations. Typically, in standard cases, that I am able to go for lunch with a friend will not be part of an appropriate answer to the argument/reasoning-requiring reading of the question ‘Why should I go for lunch with my friend?’. It is crucial to stress that the Erotetic view explains reasons in terms of appropriate answers to one of the two possible readings of the normative ‘Why F?’ question. One of these readings is the argument/reasoning-requiring reading. And in this last aspect, it has some similarities with the Reasoning view (and thus it can vindicate the insights that come from appealing to the reasoning in explaining reasons, in a sense). But it is also fundamentally different from the Reasoning view. Its appeal to good patterns of reasoning in explaining reasons is not unrestricted, so to say. On the Reasoning view, roughly, any consideration that can serve as an appropriate input in a good pattern of reasoning towards F-ing can be a normative reason. Not so, on the Erotetic view. Only those considerations that are appropriate answers to the reasoning/argument-requiring reading of the ‘Why F?’ question will count as normative reasons on the present view. These considerations surely have to be contents of premises of an argument/premise-responses in reasoning. But being a premise is not a sufficient condition here. These premises have to be appropriate answers. Which answers are appropriate is also context-sensitive, and depends on the further aims and presuppositions in a conversation (which need not be concrete; recall that we are focusing on abstracta, the patterns of questions and answers). Thus, the Erotetic view doesn’t treat mere enabling conditions as normative reasons (but it also allows for considerations about, say, one’s abilities to be normative reasons in special contexts). The Erotetic view neither over-generates reasons here, nor is it too restrictive. It is an improvement on the Reasoning approaches. Similar conclusions also apply with respect to mere entailing conditions.

How about the cases where the Reasoning view under-generates reasons? We saw earlier (Chapter 2) that the Reasoning view had a hard time explaining cases like the Surprise Party and Ice Cream cases, and it entailed by definition that there are no reasons in the Moore-paradoxical and self-undermining belief cases. And yet these cases also appeared pre-theoretically to be cases of genuine normative reasons. The Erotetic view has resources for treating these cases as cases of genuine normative reasons. Take, for instance, the Moore-paradoxical considerations. That there is a fire in the building but John doesn’t know that there is a fire in the building may well be part of an appropriate answer to the question ‘Why check on the building?’. It is an appropriate answer to the explanation-requiring reading of that question (the question ‘Why ought John to check on the building?’). Answers on our account are abstracta, and neither the inquirer nor the addressee of the question needs to be John. The mere fact that there is a pattern of a normative ‘Why check on the building?’ question that has John as a subject and that has a possible appropriate answer ‘that the building is on fire and John doesn’t know that it is’ is enough for the Moore-paradoxical consideration to be a reason for John to check. Sure, it is not a consideration that John could reasonably use in a fitting piece of reasoning. And thus, it is not a normative reasoning reason. But it may still be a normative explanatory reason. Thus, the Erotetic view preserves the pre-theoretical claim and is extensionally more appropriate than the Reasoning view. It doesn’t under-generate the reasons in Moore-paradoxical cases. Similar considerations also apply in the case of self-undermining beliefs (e.g. a memory-erasing pill), the Surprise Party case, the Ice Cream case, and others (see Chapter 2). A consideration may still count as a genuine normative reason, namely a normative explanatory reason for why one ought to F even if the consideration is not or cannot be a premise in one’s reasoning. The distinction between normative reasoning reasons and normative explanatory reasons explains how this is possible.

In a similar way, the Erotetic view also avoids the pitfalls of the Explanation view. That is, it avoids over-generating and under-generating reasons in the way that explanatory views do. We observed earlier that some explanatory views distinguish pro tanto and pro toto normative reasons (cf. Broome Reference Broome, Jay Wallace, Smith, Scheffler and Pettit2004, Reference Broome2013), where the pro tanto reasons are supposed to be considerations that play a role in the special normative weighing explanations and the pro toto reasons just are explanations of why one ought to F. One worry there was that postulating the existence of two distinct sorts of normative reasons to F over-generates normative reasons (see Chapter 3 for details). The Erotetic view easily avoids this problem. On the view defended here, there are simply no such things as pro tanto reasons. There are normative explanatory reasons for S to F, on this view, but these are explanations (grounds) of why S ought to F and hence would rather correspond (in a sense) to pro toto reasons. Another worry with the Deontic Explanation view (cf. Broome Reference Broome, Jay Wallace, Smith, Scheffler and Pettit2004, Reference Broome2013) was that it was unclear why considerations that appear to speak clearly against F-ing (e.g. that I can have free holiday accommodation by the sea at my disposal as speaking against staying at home during the vacation) are still on this view parts of explanation of why I ought to F (e.g. stay at home). Again, the Erotetic view avoids this worry, by not postulating the existence of pro tanto reasons and weighing explanation. Yet it is still able to preserve the attractive features of the Explanation view, namely it respects the intuition that reasons do appear to be connected, in a sense, to explanation. The Explanation value-centred view had a hard time explaining how reasons for attitudes are possible (some even go as far as to endorse an outright denial of reasons for emotions). Yet this is contrary to our pre-theoretical views. On the Erotetic view, there is no problem with that. There might be both normative reasoning reasons as well as normative explanatory reasons to believe and to fear and so on.

Overall, then, the Erotetic view has a crucial advantage over its reductionist rivals. It is extensionally adequate. And it also satisfies other constraints/desiderata for a theory of reasons. Moreover, it has still further considerations speaking in its favour. It makes sense of reasons talk, it makes best sense of the dual life of normative ‘Why F?’ questions, and it fits best with the observation that ‘Why?’ questions in general can be understood either by an question-requiring or by an explanation-requiring reading. Thus, we may conclude that these considerations make the Erotetic view superior to its rivals. Section 5.3 considers some possible objections to the Erotetic view. Chapter 6 looks at an application of the Erotetic view to a well-known debate. The theoretical fruitfulness of the Erotetic view demonstrated there will constitute yet another argument in its favour.

5.3 Further Clarifications

This section anticipates some possible objections to the Erotetic view. I provide here some considerations that might seem to speak against the Erotetic view, and I reply to all of them.

First, back to enabling conditions (once more). One might worry about the strategy that we have proposed in Section 5.2, where it was claimed that the Erotetic view, contrary to the Reasoning view, doesn’t predict that all enabling conditions (e.g. that I am able to go for lunch) are normative reasons even though they can be premises of good patterns of reasoning/argument. The Erotetic view avoids counting enabling conditions as reasons (in the case of normative reasoning reasons) by restricting the relevant considerations not merely to premises in good patterns of reasoning/argument but more specifically to good premises in good patterns of reasoning/argument that are appropriate answers (in a context) to the argument/reasoning-requiring ‘Why F?’ question. In short, which considerations will count as normative reasoning reasons depends not only on their standing as premises in good patterns of reasoning but also on the context-sensitive (yet abstract) features of conversation – whether or not they are possible appropriate answers to the relevant normative question. Now, the worry here is that one might suggest that proponents of the Reasoning view could equally appeal to context-sensitive aspects of reasoning/argument in order to restrict the considerations that would count as reasons. The idea would be that what counts as a good pattern of reasoning partly depends on the aims and presuppositions of a reasoner, and thus is also context-sensitive. In this way, one might hope, the proponents of the Reasoning view could rule out enabling considerations from counting as reasons. If so, the Erotetic view doesn’t have an advantage over the Reasoning view.

In reply to this worry, I would like to note that, if successful, such a move indeed would put the Reasoning view and the Erotetic view on a par with respect to enabling conditions. But I don’t think that this is so. It is not clear how exactly the appeal to the context sensitivity of good patterns of reasoning could rule out enabling conditions from being appropriate premises on the Reasoning view. It is not to say that the reasoner’s aims and presuppositions shouldn’t be taken into account in determining the pattern of reasoning that the reasoner is engaging in/proposing. Surely, the reasoner’s intentions count. Without taking the reasoner’s intentions into account, it is often difficult to establish whether the proposed argument is, say, the fallacy of affirming the consequent or an inference to the best explanation (IBE). The fact that a reasoner is aiming to provide a deductive argument, to establish the conclusion by guaranteeing its truth, will matter in classifying the argument as the fallacy or an instance of an IBE in such a case. But it is difficult to see how aims and presuppositions (or indeed other factors) could exclude enabling conditions from patterns of good reasoning. It just seems to be a core feature of any good argumentative structure that adding a true premise doesn’t invalidate, ceteris paribus, the soundness of the argument. The proponents of the Reasoning view should show that adding a true premise expressing enabling conditions triggers the ceteris paribus clause. But it is not clear why this should be so. Merely postulating this would amount to an ad hoc move and doesn’t have any independent theoretically satisfying motivation. It would only be motivated by avoiding the problem of enabling conditions, which is not an independently plausible theoretical consideration. The Erotetic view, on the other hand, has resources to motivate the move of the appeal to aims and presuppositions. It is a core observation about answers that their appropriateness depends on the context of the conversation in which they are proposed. They depend on the aims and presuppositions in place in the conversation. It depends on what the inquirer is asking for and on what is presupposed and so on. It depends, for instance, on the fact that typically the inquirer is asking for considerations that would convince her in endorsing the claim that S ought to F. Citing mere enabling conditions (that S can F) would not count as things that could convince the inquirer that S ought to F. Thus, the Erotetic view has an independently plausible account of why enabling conditions will not count as appropriate answers to the argument/reasoning-requiring ‘Why F?’ questions and can avoid the problem of enabling conditions. The Erotetic view can satisfactorily account for why mere enabling conditions are not (typically) reasons to F, whereas the same move is not available to proponents of the Reasoning view.

Second, one might have some further thoughts about the way the Erotetic view treats the question of the ‘weight’ of reasons and comparative judgments about reasons. The worry here is that, on the one hand, the Erotetic view postulates the existence of two different sorts of normative reasons, but, on the other hand, it explains the apparent ‘weight’ of reasons only in terms of defeasible reasoning. It seems, then, that on the Erotetic view, only normative reasoning reasons can have ‘weight’ or, more precisely, can allow for genuine comparative judgments. For by definition, normative explanatory reasons cannot be explained in terms of reasoning, defeasible or not. Yet, the worry goes, one might think that the apparent ‘weight’ of reasons is a common property of all sorts of normative reasons. How does the Erotetic view account for the ‘weight’ for all reasons?

I would like to respond to this worry in two stages. First, I would like to acknowledge that indeed, it is a consequence of the Erotetic view that genuine direct comparative judgments about normative reasons make sense only in the context of normative reasoning reasons (‘direct’ here excludes comparisons in terms of which normative explanatory reasons explain better the relevant ought fact – for example, explain in a simpler and more elegant way). Second, I would like to defend the proposal that this is not, contrary to what the worry suggests, an unwelcome consequence of the view. Consider, for instance, a clear case of normative explanatory reasons. Consider: (a) ‘the building is on fire but John doesn’t know that it is on fire’. We agreed that it is a reason for John to check (again) the building. Now, what could be an appropriate target for a comparison for (a)? That is, what could be another normative explanatory reason that we could compare to (a) with respect to their relative ‘weight’ or ‘strength’? A proponent of this line of worry needs to find a reason that would have, for example, a lesser ‘weight’ than (a) but would still qualify as a normative explanatory reason. My suggestion is that there isn’t really any reason that fits the bill. One might think that (b) ‘that the fridge contains John’s preferred ice cream, but John doesn’t know that it contains it’ is a reason for John to check the fridge rather than the building. But it doesn’t seem to be. After all it is not the case that John ought to/should check the fridge rather than the building in the present case. Explanation is assumed to be factive. Nothing can explain p, when p is not the case. Thus, there is no genuine comparison here. In a situation where John ought to check the building and not the fridge, only (a) is a normative explanatory reason for him and not (b).

Fundamentally, in thinking about the apparent ‘weight’ of reasons, we should ask the question: but what’s the point of comparative judgments about reasons? What do these judgments help us with? A natural answer to this question is, I think, the proposal that these comparisons help us sort things out in complicated deliberative situations. They help us order options, in particular, in situations of uncertainty. These are precisely the situations where we don’t know whether we ought to F or not (or at least situations where it can make sense to challenge such ought claims). If so, then it is easy to see why talk of the apparent ‘weight’ of reasons makes sense only in the context of normative reasoning reasons. That one ought to F is not challenged in situations of normative explanatory reasons. Thus, I would like to suggest that the Erotetic view’s prediction that the normative explanatory reasons don’t have ‘weight’ is not problematic. Quite the contrary, it seems to be the correct consequence given the ultimate purpose of comparative judgments about reasons.

The third potential worry concerns normative questions that differ from ‘Why F?’ questions. Consider, for instance, the question ‘Where to eat for lunch?’ It is clearly a normative question, since it is best understood as a question involving an ought/should. It can be easily paraphrased as: ‘where should/ought we eat for lunch today?’. The Erotetic view identifies reasons as answers to the ‘Why F?’ normative question. But is there an independent, theoretically well-grounded explanation of why it is limited only to ‘Why F?’ and not all sorts of normative questions? Isn’t focusing on ‘Why F?’ questions ad hoc? Is there any further theoretical motivation for the choice of the specific normative question within the Erotetic approach? Note that on the face of it, it doesn’t seem plausible to treat the answers to these further normative questions (e.g. ‘Where should we eat?’, ‘What should I do tonight?’) as normative reasons (e.g. ‘We should eat at that new ramen place’ is not straightforwardly a normative reason). Thus, the worry is not an idle one. If the proponents of the Erotetic view cannot provide an independent and theoretically well-motivated explanation of why we should focus on the ‘Why F?’ questions in defining reasons, the view is either ad hoc or leads to an implausible over-generation of normative reasons (this is the horn of the dilemma where all answers to all normative questions are treated as normative reasons).Footnote ⁵

To this worry, I would like to reply that there is indeed a well-motivated distinction between the normative ‘Why F?’ on one hand and other normative questions on the other hand. We can think of ‘Where to F?’, ‘What to F?’, and similar questions as questions requiring specifications about an (alleged) ought to F. When we are asking ‘Where should we eat?’, we are asking for precisions about the general should/ought to eat that is (allegedly) already in place. ‘Why F?’ questions, on the other hand, are, so to say, more fundamental: they either question the relevant ought claim in general (or, at any rate, ask for additional arguments in their favour) or are asking for explanation of the relevant ought. They are not about asking for specification of an (alleged) ought. Of course, there might be situations where by asking a ‘Where to F?’ question, we are ultimately asking a ‘Why F?’ question, such as in a case where, after having spent an hour trying to find a parking spot, and thereby already being late for the theatre, one might utter an exasperated ‘Where to park?’, meaning to ask whether it is really worth continuing to search for a parking spot rather than just go back. But, typically, the other non-‘Why F?’ questions are about inquiring into ways to specify some general (or vague) ought into more operationisable oughts or shoulds. This, then, appears to be a major difference between the ‘Why F?’ questions and other normative questions. Where the latter are requiring specification or precisification of a general or vague ought, the former is asking for arguments or explanations (grounds). Thus, the focus on ‘Why F?’ questions seems to be justified by this crucial difference.

Alternatively, one could also explore the idea that ‘Where to F?’, ‘What to F?’, and other similar questions can, after all, be reduced to ‘Why F?’ questions. The reduction, according to this line of reply, would have a contrastivist flavour (cf. Snedegar Reference Snedegar2017 on contrastivism about normative reasons in general). The idea would be that when we are asking, say, ‘Where to eat tonight?’, we are asking several distinct ‘Why F?’ questions that are lumped together, presumably, for reasons of economy and easiness of expression. The presumed ‘Why F?’ questions here would be of the following sort: ‘Why eat at A rather than at B?’ ‘Why eat at B rather than at A?’ ‘Why eat at C rather than at A or B?’, and so on. The idea would be that there is a (contextually determined) partition of options for eating out, and there are arguments in favour and against all these options, and, presumably, one option has better arguments in its favour than others. By asking ‘Where to eat?’, one is merely shortcutting the longer and more tedious line of questioning about arguments in favour of each and all the relevant options for eating out tonight. If this is the case, then ‘Where to F?’ and similar questions can, after all, be integrated into the Erotetic view. Of course, the answer ‘We should eat at that new ramen place’, while completely appropriate, is also to be taken as a shortcut, a proxy to the arguments that speak in favour of eating at the new ramen place (these are the arguments that make the ramen place the winner among other options). Now, these arguments in favour of the ramen place, then, can be properly seen as normative reasons speaking in favour of eating at the new ramen place. More specifically, they can be seen as normative reasoning reasons. While this is still a very rough sketch of how this line of answer could go, it appears to be another plausible strategy that a proponent of the Erotetic view might appeal to in order to deal with the worry of non-‘Why F?’ normative questions.

Fourth, and somewhat connectedly to what precedes, one might think that there are normative ‘Why F?’ questions that cannot have an appropriate answer. More precisely, one might think that some questions are abominable and any answer to these would be insensible. Think, for instance, about questions such as ‘Why kill innocent civilians with gunfire rather than kill them by torturing them to death with some archaic torture instrument?’. It’s sensible to think that there is no really appropriate answer that we could give to such a question. One might think that the appropriate thing to do in the face of such a question is to say nothing and thereby reject the question entirely. Some questions just have to be refused to be answered. One should not kill innocent civilians, neither by torturing them to death nor by gunfire. Yet one might also think that in a situation where one is going to kill civilians anyway, one has reasons to kill them by relatively less-painful means – that is, by using gunfire rather than by torturing them to death. The worry, then, is that the Erotetic view predicts that there are no such reasons, since the question ‘Why kill innocent civilians with gunfire rather than torturing them to death?’ doesn’t seem to have any appropriate answer.Footnote ⁶

In response to this question, I would like to observe first that this question appeals to the classical and well-worn debate over the lesser evil. Whether and in what sense one ought to do the less-wrong thing is a debate that I think goes beyond the scope of the present discussion, and indeed beyond the scope of the Erotetic view itself. Many intricate issues pertaining to scope and sense of ought in cases of lesser evil need to be taken care of. That being said, and this is my second point, if after a careful consideration we still think that there are reasons for one to kill the innocents by gunfire in a situation where one’s only alternative is to kill the innocents by torturing them to death, then there is also a reading of ‘Why kill the innocent by gunfire?’ that can be appropriately answered in such a context. The question may certainly appear inappropriate to us. But it is important to remember that the question is not necessarily directed at us. We are not in a situation where we are going to kill innocent civilians. Yet if one is indeed in a genuine situation where one is guaranteed to kill innocent civilians, the question may, after all, be meaningfully addressed to such a subject. One difficulty in seeing how such a question can be appropriately answered might come from our difficulty in conceiving of such a situation where there really is no other genuine option for the subject in such a case (e.g. freezing). But I would like to suggest that if it is meaningful to talk about reasons for the killer to kill by gunfire rather than by torture, then it should also be meaningful for the killer to consider the question ‘Why kill by gunfire rather than by torture?’. If this is so, then contrary to what the worry suggests, the Erotetic view predicts that there may be reasons for one to kill by gunfire rather than by torture. In other words, I would like to suggest that the killer having reasons to kill by less-painful means and the appropriateness of the question ‘Why kill using the less painful means?’ as having the killer as its subject either stand or fall together. Thus, the Erotetic view can avoid the problem by either (i) explaining that there are neither genuine normative reasons to kill by gunfire nor an appropriate answer to the question ‘Why kill by gunfire?’ or, alternatively, (ii) that there are reasons for the killer to kill by gunfire (rather than by torture), and the question ‘Why kill by gunfire rather than by torture?’ can be appropriately answered when it has the guaranteed killer as its subject. In either way, the Erotetic view seems to avoid the alleged problem.

The fifth and final worry concerns the place of rationality within the new Erotetic view. It is a common view that rationality (or justification) in the case of both action and attitudes is the affair of reasons. For instance, one might think that it is rational for you to F just in case you possess most reasons to F. And one might think that it is rational (or justified) for you to believe that p just in case your reasons, overall, support believing that p. Now, the worry is that it seems that determining what is rational for one to do or believe (or indeed which attitude to have) is an essential feature of normative reasons. It is, according to this line of thought, a central function of reasons to determine what’s rational for us to do, believe, and so on. However, the Erotetic view seems to imply that determining rationality is not that essential to reasons after all. For it postulates that there are basically two sorts of normative reasons, normative reasoning and normative explanatory reasons. But at best, only one of these may be said to play a role in determining rationality. More specifically, it is hard to see how explanatory reasons can determine what is rational for one to do, believe, and so on. It may be the case that reasoning reasons are connected to rationality. But it is difficult to see how explanatory reasons could also be linked to rationality. Think again of Moore-paradoxical cases, where one is, by definition, in no position to act for the relevant reason; one cannot even possess the relevant explanatory reason. Thus, one might think there is no way in which such reasons might be connected to what is rational for one to do or believe, at least not in the common-sense of rationality. Disconnecting normative reasons and rationality, then, seems to be a worrisome implication of the Erotetic view. The essential link between reasons and rationality seems to be lost on the Erotetic view.Footnote ⁷

To this worry, I would like to respond by suggesting that the problem is perhaps less worrisome than it might first appear. Indeed, perhaps, the observed loosening of the link between rationality and reasons in general is not a bad feature of the view at all. This is, however, not the place to develop what a full-blown theory of rationality could look like according to the Erotetic view. Some aspects of this task (with respect to belief) will be undertaken in Chapter 6. But let me suggest only that the view according to which all normative reasons have to be essentially connected to rationality may just be a mere byproduct of focusing on one sort of normative reasons only, namely normative reasoning reasons. Once one recognises the reality of normative explanatory reasons, one might see that the connection is not essential for all reasons. A proponent of the worry here could nevertheless protest. Actually, one might suggest that normative explanatory reasons shouldn’t really be considered as genuine reasons if they don’t have a link to rationality. One could suggest that they are not reasons, precisely because they are not connected essentially to rationality. But such a protest might come as a bit of a surprise. For such an alleged reasons–rationality connection seems to be something that is under question in our discussion. Surely we don’t have here a knock-down argument for the conclusion that the connection is not essential, at least not for all sorts of normative reasons. Nevertheless, I would like to draw attention to something we already touched upon in the chapter on the Reasoning view (Chapter 2) when we discussed Moore-paradoxical cases (and similar ones). We observed that Moore-paradoxical considerations do appear to speak in favour of the relevant F-ing. If one is denying them the status of being normative reasons, then one has to provide an independent theoretical motivation for introducing a new, distinct category that would encompass such normative considerations (e.g. the relevant Moore-paradoxical considerations). But such a move seems to be superfluous. Seeing both normative explanatory and normative reasoning considerations as normative reasons is more economical and better motivated. For both explanatory and reasoning reasons share the basic feature, the feature by which we identify reasons pre-theoretically – namely, they both appear to speak in favour of F-ing, in a sense. It is, thus, theoretically preferable to consider normative explanatory reasons as genuine normative reasons of a sort. If so, the Erotetic view can be maintained here. All that is required is that we can provide a plausible story about rationality within the framework of the Erotetic view. This task will be partly considered in Chapter 6 with respect to reasons for belief. Let me conclude this section by comparing very briefly (a fuller comparison will have to wait for another occasion) the present proposal to an existing view that seems to rely on somewhat similar distinctions.

In a recent article, Ralph Wedgwood (Reference Wedgwood2015) suggests that there are two distinct categories of concepts of normative reasons. One is that of concepts of ‘normative-explanation’ reasons (Broome’s account of reasons is a paradigm of this category of concepts of normative reasons). The other category is that of ‘ideal-motivation’ reasons (Setiya’s ideal dispositionalist account as well as various ideal reasoner models of reasons are taken to be paradigmatic of this second category of normative reasons). Wedgwood proposes several cases and arguments to show that there are two functions of normative reasons, and that these cannot be fulfilled by one single concept. There have to be two distinct categories of concepts of reasons. If we abstract from Wedgwood’s specific focus on ideal or good reasoners or dispositions (and focus instead on the more general category of good patterns of reasoning), then his classification of concepts of reasons (and by extension of corresponding reasons) comes very close to our own positive proposal.

There are, however, two important differences between our proposal and Wedgwood’s proposal. First, Wedgwood’s main target in that article are the reasons-first views of reasons, which would like to account for both normative-explanation and ideal-motivation aspects of reasons. Wedgwood’s point there is that it’s impossible to maintain such a position coherently – that is, a reasons-first view that also accounts for both apparently central aspects of reasons, the normative-explanation and ideal-motivation aspects. I agree with such a verdict. However, I also suspect that proponents of the reasons-first approach would not commit themselves to taking the aforementioned two aspects, the two functions of the concept of normative reasons, as crucial or essential to what reasons are. After all, reasons are undefined according to the reasons-first approach. More importantly, we are not aiming here to demonstrate that the reasons-first account cannot be maintained by presenting some further problem cases. Rather our aim here is to provide a new, positive reductionist account of reasons. The second difference between our proposal and Wedgwood’s is that the two proposals differ in the explanation of what the common element between the two categories of concepts is. Wedgwood appeals to a confusion of two levels: the level of a theoretician and the level of an agent. He writes: ‘that is, a confusion between (a) what must be known or grasped by the theorist who is giving an account of a certain sort of agent, and (b) what must be known or grasped by the agent herself’ (Wedgwood Reference Wedgwood2015: 137; Wedgwood indicates that this distinction comes from Alston). This might come close to what we propose here, but the present proposal says something more than that there is a confusion (and that proponents of the reasons-first approach who want to maintain the two aforementioned aspects of reasons are confused). On our account, there is a fundamental dual nature of normative ‘Why F?’ questions. And this gives rise to two distinct sorts of normative reasons. Maybe we can think of our argument-requiring ‘Why F?’ questions as corresponding to Wedgwood’s agent’s level and our explanation-requiring ‘Why F?’ questions as corresponding to Wedgwood’s theorist’s level. If so, maybe the two proposals are close and indeed compatible. However, details of the two-level distinction might need to be further specified to see how far our proposals can fit together (in particular, if the agent’s level is always a level of reasoning towards a deliberative ought and never a level of explanation for the agent herself of why she ought to F).