1.1 Decision-Making and Deliberation

I understand deciding (or choosing) as a matter of forming an intention—more precisely, as “a momentary mental action of intention formation” (Mele Reference Mele, Beckermann, McLaughlin and Walter2009: 693). Agents can form intentions on the basis of other attitudes, for example, on the basis of their beliefs and other intentions. This capacity is essential to agents’ abilities to engage in practical reasoning. Practical reasoning is reasoning which concludes in the formation of an intention (Broome Reference Broome2013: 250; although cf. Raz Reference Raz and Huer2022; Dancy Reference Dancy2014; see Brunero Reference Brunero2021 for discussion). And, as a species of reasoning, practical reasoning involves basing a certain attitude (the conclusion attitude) on other attitudes (the premise attitudes) (Kauppinen Reference Kauppinen and Star2018: 410; Boghossian Reference Boghossian2019: 59–60; Broome Reference Broome, Jackson and Jackson2019: 32–33).

Moving on to deliberation, deliberation is a conscious, voluntary activity aimed at making up one’s mind about some issue or question (Owens Reference Owens and Hatzimoysis2011: 262; Shah Reference Nishi2008: 11; Arpaly and Schroeder Reference Arpaly and Schroeder2012: 211; Malmgren Reference Malmgren2019: 201–203). Practical deliberation is deliberation aimed at settling what to do. Agents settle what to do by making a decision—i.e., by forming an intention. The question of ‘What to do?’ is sometimes framed as the question ‘What shall I do?’ (Parfit Reference Parfit2011: 386–387). However, this way of framing the question is potentially misleading. By asking ‘What shall I do?’, an agent might be asking what it is they will in fact do—i.e., attempting to make a prediction about their future action. Deliberation about what to do is not deliberation about what one will do (Shah Reference Nishi2008: 6; Southwood Reference Southwood, Jones and Schroeter2018: 94). Not only is deliberation about what to do distinct from deliberation about what one will do, we make a mistake if we view evidence about the likelihood of our own future actions in the same way in deliberation as in prediction, namely “the mistake…of treating something that is up to us as if it isn’t…We treat ourselves as strangers” (Vavova Reference Vavova2016: 534). Deliberation is also distinct from instrumental reasoning which starts from an intention to achieve some end e, involves the formation of beliefs concerning means to e and concludes in an intention to take some (believed) means to e. Deliberation involves adjudicating between competing ends (Kolnai Reference Kolnai1961).

1.2 Clarifying the Normativity Thesis

According to the Normativity Thesis, necessarily, if one deliberates about what to do, one employs normative concepts, specifically reason, ought, or the concept expressed by ‘best’ as the word features in the sentence, ‘The best thing for me to do now is to φ’. Saying that some option is best or good in this sense—in what we might call the ‘goodness as choiceworthiness sense’—seems equivalent to saying that one has reason to do it or ought to do it (Wedgwood Reference Wedgwood2009: 502).Footnote ¹ I think that some philosophers will find the Normativity Thesis so obvious that it needs no defence. However, as I’ll shortly illustrate, the thesis is disputed. And, as I’ll also go on to discuss, such objections raise an interesting question about how best to develop the thesis.

1.3 The Normativity Thesis in Contemporary Metaethics

The Normativity Thesis plays an important role in contemporary metaethical theorizing. It is central to David Enoch’s indispensability argument for normative truths, which suggests that if such truths are indispensable in deliberation about what to do we are justified believing in them (2011: Ch. 3), and it features prominently in Allan Gibbard’s development of his ‘plan-expressivist’ brand of non-cognitivism (Reference Gibbard2003: Ch. 1). Something in the vicinity of the Normativity Thesis also appears in discussions of the question of what makes concepts normative. An increasingly prominent answer to this question points to such concepts’ ‘action guiding’ or ‘practical’ role in thought, which some proponents of this view tie to deliberation. For instance, Matti Eklund (Reference Eklund2017: 38) suggests that “it is characteristic of normative predicates that they are fit to be used in practical deliberation about what to do”. Ralph Wedgwood (Reference Wedgwood and Star2018: 36–37) holds that normative concepts are to be identified by their role in ‘reasoning problems’ and that what he calls the ‘practical ought’ is characterized in terms of its role in the “practical reasoning problem—the problem of deciding what to do in a certain situation” (2018: 37; 2007: Ch. 4). In a similar vein, philosophers have argued that we can identify simpliciter or all things considered normative concepts through the role of such concepts in deliberation. J.L.D. Brown (Reference Brown2024: 504; cf. McPherson Reference McPherson and Russ-Shafer-Landau2018) claims that “our concept of ought simpliciter essentially has the functional or conceptual role of non-arbitrarily settling what to do in practical deliberation”. I’m sympathetic with the project of distinguishing simpliciter normative concepts by identifying their role in thought. I see this article as contributing to this project by helping to pin-down how such concepts figure in deliberation (see Section 3.3).

3.1 Deliberation and Desire

The first challenge to PDND maintains that one can deliberate about what to do by deliberating about what one wants. Derek Baker (Reference Baker2017: 589; see also Olson Reference Jonas2014: 172–177) suggests that “While my deliberation about what to do next often begins with the question, ‘What should I do now?’ it just as frequently begins ‘What do I want to do now?’” Perhaps then, at least some of the time, the considerations agents take to bear on the options in deliberation are (putative) facts about what they desire or facts about what would promote their desires. The truth of this claim doesn’t show that PDND is false if desires are themselves to be understood in normative terms or if one must be taking facts about one’s desires to be reasons in deliberation. I have more sympathy with the latter claim than the former claim, though the former claim has its supporters (Scanlon Reference Scanlon1998: 37–41; Quinn Reference Quinn1993; Raz Reference Raz1999b; Gregory Reference Gregory2013). To bring out the plausibility of the idea that one must be taking facts about one’s desires to be reasons in deliberation, consider the fact that deliberation about what to do presupposes that some answers to the question of what to do are better or worse than others. This assumption is plausibly constitutive of deliberating about what to do (Schueler Reference Schueler, Deonna and Lauria2017: 307; Enoch Reference Enoch2011: 72–73; Finlay Reference Finlay2014: 142). If the idea of answers being better or worse is understood in terms of answers being closer or further from the truth, a proponent of the view that agents can deliberate about what to do by deliberating about what they desire (without taking facts about their desires to be reasons) can accommodate the idea by suggesting that agents are presupposing that there is a correct answer to what it is that they desire. However, this interpretation of the idea that deliberation presupposes a commitment to better or worse answers is unsatisfying. When one deliberates about what to do one brings to mind considerations which one takes to bear on the correctness of an answer to this question by helping to render an action choiceworthy. Of course, it might be that desire facts help render an action choiceworthy in a particular agent’s eyes. However, this is just to acknowledge that this agent is taking facts about their desires to provide reasons in deliberation about what to do. In the background of my discussion is the important point that agents can step back from their desires and critically evaluate them. The reflective distance that agents have from their desires is plausibly a condition of deliberation; without this capacity there would be no deliberation, only reasoning about how to fulfill one’s momentarily strongest desire (Wallace Reference Wallace2006: 192; see also Enoch Reference Enoch2011: 75–76).Footnote ⁵

An important piece of evidence for the claim that one must take one’s desires to be reasons, insofar as they figure in one’s deliberation about what to do, comes from reflecting on why it is that agents sometimes deliberate about what to do by considering what they want and sometimes don’t. (I assume that, at least some of the time, desires are ‘backgrounded’ in practical deliberation so that they don’t explicitly feature in such deliberation.)Footnote ⁶ A natural answer is that sometimes we take our desires to settle which course of action is most choiceworthy or best, but not always. The proponent of the view that we can deliberate about what to do by considering what we want without taking facts about our desires to be reasons will presumably try to explain this phenomenon by suggesting that agents choose to deliberate about what to do in different ways on different occasions, and sometimes they choose to deliberate about what to do by deliberating about what they want. Granting for the sake of argument that agents can choose how to deliberate, I think that this explanation does not ultimately help one avoid the conclusion that one must take one’s desires to be reasons in deliberation. Notice that we can’t decide to deliberate about what to do by considering any question whatsoever. For instance, I cannot deliberate about whether to go to law school or take a job in philosophy by considering what Napoleon would do in my shoes. But what explains the limitation on how we can choose to deliberate? The answer is that we take the answers to some questions to be relevant to settling what to do but not others. But now the challenge is to spell out what this relevance amounts to in a way that isn’t normative—i.e., which isn’t a matter of some answers either being reasons for a particular action or telling us which action we ought to perform. If relevance is understood in normative terms, deciding to deliberate about what to do by thinking about what one wants involves taking facts about one’s desires to be reasons.

3.2 Silverstein on Deliberation

Matthew Silverstein (Reference Silverstein2017) develops our next challenge to PDND. Silverstein considers a case where an agent judges that they ought to have lunch at home tomorrow without having decided where to eat tomorrow. Silverstein—who seems to think that this is not simply a standard case of akrasia (2017: 357) —uses this case to illustrate the way that reaching a judgment about what one ought to do and deciding what to do (which he thinks of as forming an intention or performing an action) can come apart. However, he also thinks that the case shows that practical deliberation, which concerns the question of what to do, and normative deliberation, which concerns the question of what one ought to do, are distinct. He characterizes the case as one where “normative or ethical deliberation has concluded even though practical deliberation has yet to begin” (2017: 358). Silverstein (Reference Silverstein2017: 358) adds that “Of course, as I deliberate about where to eat I will weigh various considerations against one another” but he claims that “I can weigh the relevant considerations against one another and come to a decision about what to do without deploying any normative concepts”.

I find it hard to make sense of the activity of consciously weighing considerations for and against various actions without this simply being a case of weighing what one takes to be reasons. Suppose one makes a list of all the considerations one thinks bear positively or negatively on whether one ought to eat at a particular location. Then suppose one tries to list considerations which bear on whether to eat at that location but don’t also belong on the former list. There won’t be any (cf. Schueler Reference Schueler, Deonna and Lauria2017: 318–319). The items on the two lists will be the same. For instance, the fact that restaurant A is the cheapest option might be on both lists, along with the fact that it’s closer than the other restaurants one is choosing between. If this is right, deliberation about what to do takes the same inputs as deliberation about what one ought to do, namely, what one takes to be reasons.

Reflection on Buridan’s donkey cases also supports the idea that deliberation about what to do takes the same inputs as deliberation about what one ought to do. In such cases (e.g., choosing between multiple packages of the same brand of cereal in a supermarket aisle) it appears that one is unable to deliberate about which option to select precisely because one believes that there are no reasons to choose any option over any other (Enoch Reference Enoch2011: 74; Finlay Reference Finlay2014: 142). In such cases one must ‘just pick’.

Silverstein relies on an analogy with deliberation about what to believe to make his case that deliberation about what to do does not involve deploying normative concepts. He (2017: 358 footnote 10) tells us that “I can reason my way to a belief without deploying normative concepts because theoretical reasoning is governed by a substantive, nonnormative aim: truth. All I need to do in order to believe on the basis of reasons is believe on the basis of my evidence. If—as I argue…practical reasoning is also governed by a substantive, nonnormative aim, then I should also be able to reason my way to an intention or action without deploying normative concepts.” Silverstein is committed to claiming that thinking of something as evidence for a belief is distinct from thinking of it as a reason for belief. This claim isn’t obviously false, though it is controversial (cf. Kelly Reference Kelly2007). Granting the claim for the sake of argument, it provides Silverstein with a way of making sense of how we could weigh considerations in deliberation about what to believe that doesn’t collapse into weighing reasons for or against a belief. What we are doing is something like weighing degrees of evidential support (2017: 538). In the practical case, what Silverstein needs is a story about how we can make sense of weighing considerations in deliberation about what to do that isn’t to be understood in normative terms—something which he doesn’t provide. In my view, such a story won’t be forthcoming because, as I’ve argued, it appears that whatever considerations we take to bear on whether we ought to φ we also take to bear on whether to φ (and vice versa).

I also want to put some pressure on Silverstein’s doxastic deliberation analogy. Can deliberation about what to believe and deliberation about what one ought to believe come apart? If the answer is ‘No’, then doxastic deliberation doesn’t look like a useful analogy for Silverstein’s purposes. Suppose then that the answer is ‘Yes’. Given Silverstein’s comments about doxastic deliberation, he may say that this is because deliberation about what to believe doesn’t involve deploying normative concepts while deliberation about what one ought to believe does. Now consider the fact that in deliberation about what one ought to believe agents frequently consider evidential considerations. (One should accept this even if one holds that there are in fact no evidential reasons for belief but only practical reasons for belief (Rinard Reference Rinard2015: 291). The truth of this claim doesn’t preclude one from holding that people frequently mistakenly think of evidential considerations as reasons for belief.) Given that deliberation about what one ought to believe involves deploying normative concepts, we sometimes represent facts that we take to be evidence for p as reasons for believing p. Consequently, it appears that Silverstein is required to say that if one thinks about evidential considerations when considering whether to believe p one is not thinking of them as reasons, but also that when one thinks of these very same evidential considerations when deliberating about whether one ought to believe p one does think of them as reasons for believing p. Put differently, Silverstein looks committed to saying that the way an agent thinks of evidence as bearing on the belief that p is different depending on whether the agent is thinking about whether they ought to believe p or about whether to believe p. This isn’t an incoherent position, but it does strike me as implausible; I don’t see any grounds for holding that the way we think of evidence as bearing on the belief that p changes depending on which of these questions we ask ourselves in deliberation.

Suppose Silverstein were to try to distinguish deliberation about what to believe from deliberation about what one ought to believe by saying that the latter, unlike the former, can involve broadly pragmatic (non-evidential) considerations. Again, I’m not sure why one would think that deliberation about what one ought to believe and deliberation about what to believe differ in this way. Silverstein (Reference Silverstein2017: 376) appears to accept the ‘transparency’ thesis according to which deliberation about whether to believe p collapses into deliberation about whether p. However, influential explanations of transparency—such as Nishi Shah’s (Reference Shah2006: 488) view that the concept of belief includes a constitutive standard of correctness according to which a belief that p is correct if and only if p is true—look as though, if successful, they will have the consequence that deliberation about whether one ought to believe that p similarly collapses into deliberation about whether p is the case.Footnote ⁷ To sum up my discussion of doxastic deliberation, since it’s not obvious that deliberation about what one ought to believe and deliberation about what to believe are in fact distinct, doxastic deliberation is arguably unhelpful for Silverstein’s purposes.

I admit that there is an important gap in the response to Silverstein I’ve developed on behalf of PDND. My response claims that deliberation about what to do takes the same inputs as deliberation about what one ought to do, namely what one takes to be reasons. However, this leaves open the possibility that one might weigh these reasons differently depending on whether one is focused on the question of what to do or the question of what one ought to do. For instance, an agent might assign the fact that restaurant A is the cheapest option a different weight depending on whether they are deliberating about where they ought to eat or where to eat. Consequently, deliberation about what to do might still be distinct from deliberation about what one ought to do. This possibility involves positing that we take the same considerations to lend different levels of support to the same actions, depending on the question we are considering in deliberation. This appears to commit us to making a mistake in thought because it is mysterious why the same consideration would lend a different level of support to an action depending on a fact about which question an agent considers. However, one might suggest that there may not be a mistake here if it’s the case that reasons are not in fact things which bear on responses (i.e., actions and attitudes) but rather things which bear on questions, a view defended by Pamela Hieronymi (Reference Hieronymi2009; Reference Hieronymi2014; Reference Hieronymi, Chang and Sylvan2021). The truth of this claim would, I agree, make the position we are considering here less puzzling. However, it still requires one to provide some explanation for why we take one and the same considerations to bear on these questions with different weights.

3.3 A Challenge from Normative Scepticism?

The next set of objections to PDND I’ll examine arise from considering the practical deliberation of normative sceptics. Nicholas Southwood (Reference Southwood, Jones and Schroeter2018: 95) appears to reject PDND by appealing to the possibility of a consistent normative nihilist, a figure who is capable of consistently deliberating about what to do despite explicitly thinking that there are no normative truths. I’m sceptical that such a figure is possible, given the points I’ve made about how deliberation about what to do appears to involve bringing to mind and weighing what one takes to be reasons; what are the inputs into such an agent’s deliberation supposed to be if not what they take to be reasons?

Stan Husi (Reference Husi2013) discusses deliberation about what to do in the context of defending the view that: while there are truths about what various norms recommend or require, no norms have authority. To get a grip on the notion of a norm lacking authority, think of some norm like the touch nose rule according to which one must touch one’s nose whenever one takes a step. Although we can sensibly talk about what one ought to do or has a reason to do according to this norm, intuitively one does not make a mistake if one ignores this norm. This norm—in contrast with the norms of, say, prudence—lacks authority; it doesn’t genuinely or really tell one what one ought to do, rather it (and any ought or reason it issues) is merely ‘formally’ or ‘generically’ normative (McPherson Reference McPherson and Russ-Shafer-Landau2018; Wodak Reference Wodak2019; see Baker Reference Baker, McPherson and Plunkett2018a for discussion). Husi takes the view that we need normative truths to deliberate but that such truths can be provided by norms which we regard as lacking authority. As he puts it (2013: 425), he wants to show “how to deliberate in a sceptical fashion”—i.e., how to deliberate once we realize that no norms are “authoritative, true, and correct, period” but only endorsed by us or correct according to some other norm (2013: 424–425). Husi would, I think, accept PDND read as claiming that deliberation about what to do is deliberation about what one ought do according-to-some-norm-or-another.

Which norms are involved in practical deliberation according to Husi? The relevant norms are ones to which we are committed. Husi (Reference Husi2013: 443) explains that “Commitments are stable psychological dispositions, endorsed by reflection, to follow some norm’s dictates, incorporating a readiness and willingness to be guided by it without regret”. Without the appeal to the idea of commitment, the claim that we deliberate about what to do by thinking about what norms recommend has no plausibility. To see this, observe that I might judge that according to some norm (or set of norms)—e.g., the norms of Leviticus—I ought to φ but think that this has as much bearing on whether to φ as the fact such as that sugar maples are deciduous. Husi (Reference Husi2013: 443) is sensitive to something like this point, explaining that when we commit to a norm, we “endow it with a special kind of motivational force…with the consequence that we empower the norm to assume an action-guiding capacity, as a map we steer by.” The picture Husi appears to offer is one according to which, what makes it the case that the sceptical deliberator deliberates about what to do by considering certain norms, is that the deliberator desires to comply with these norms. These desires to comply with the relevant norms explain why these norms resonate with the deliberator in a way that other norms do not.

But, in reply, consider the point I developed in Section 3.1 to respond to the view that deliberation about what to do is deliberation about what one desires. I argued that deliberation induces a kind of reflective distance from one’s desires that requires one to take a stand on whether one’s desires render courses of action choiceworthy. This point obviously extends to a desire to comply with some norm. For the sceptical deliberator to structure their deliberation by thinking about what some norm recommends involves them taking a stand (at least implicitly) on whether their desire to comply with the norm provides a reason to do so and, importantly, this reason doesn’t seem to be merely a reason according to yet some further norm. Here is another way to make the point I’m developing: It seems arbitrary to deliberate using one norm rather than another if one regards this norm as no more justified than any another. What we are trying to do in practical deliberation is to avoid just picking arbitrarily (Enoch Reference Enoch2011: 75). The answer to this arbitrariness worry offered by Husi (cf. Husi Reference Husi2013: 436–437) is that the norms we use in deliberation are ones to which we are committed. But if committing to a norm is desiring to comply with it, we are faced with the question of how we think of our desires in the context of deliberation. And, in such deliberation, we think of these desires as providing us with (not merely standard-relative) reasons.

Husi might reply that his view of commitment to a norm involves ‘reflective endorsement’ of the norm, which goes beyond merely desiring to comply with it. However, Husi doesn’t tell us much about the state of reflective endorsement. Reflective endorsement is either a normative or evaluative state, or it is not. If it is not, I don’t see how it will be anything other than a state constructed from materials like desiring or intending. Consequently, it will not help Husi respond to the worry that we need to take a stand on the justificatory force of our mental states in deliberation about what to do.Footnote ⁸ On the other hand, if reflective endorsement is normative or evaluative, Husi owes us an explanation of which normative or evaluative notions are essential to reflective endorsement, subject to the constraint that they can’t imply a commitment to (non-standard-relative) reasons and oughts.

The points I’ve developed in response to Husi in this subsection can be used to raise a challenge to normative authority scepticism—the view that not only are there no authoritative reasons and oughts (as Husi thinks) but that we cannot even make sense of the idea of normative authority; talk of authoritatively normative reasons and oughts is confused, incoherent, or empty (Tiffany Reference Tiffany2007; Baker Reference Baker2018b). The response is this: Authority sceptics cannot capture the way normative concepts figure in practical deliberation because it is authoritatively normative concepts which feature in such deliberation (cf. McPherson Reference McPherson and Russ-Shafer-Landau2018).

Authority scepticism entails conceptual normative pluralism which denies that we can make sense of an ought simpliciter and corresponding reasons simpliciter, maintaining instead that there are only truths about what one ought to do or has a reason to do according to some normative system or standpoint—i.e., according to some set of norms or another. David Copp, who is probably the most prominent proponent of conceptual normative pluralism, claims that we can’t make sense of an ought simpliciter because the notion is subtly incoherent (Reference Copp2007; Reference Copp, Chang and Sylvan2021) but he rejects authority scepticism, suggesting that certain systems of norms (e.g., moral norms) generate “genuinely normative” reasons and oughts while others (e.g., the touch nose rule) do not (Reference Copp2009: 30). He (Reference Copp, Chang and Sylvan2021: 418–419) tells us that these genuinely normative reasons and oughts are such that one is making a “mistake” if one knowingly ignores them when deciding what to do. Copp’s view raises a challenge to arguing from the claim that we need something more than merely formally normative reasons and oughts in practical deliberation to the conclusion that we are employing simpliciter normative concepts in such deliberation. However, in my view, the most promising way of defending simpliciter scepticism is through adopting authority scepticism.Footnote ⁹ If this is right, my discussion in this subsection can help to support the conclusion that there are simpliciter normative concepts.

3.4 Three More Objections

The next three objections to PDND I’ll consider are grouped together because they accept that deliberation about what to do involves bringing to mind and weighing what one takes to be (simpliciter) reasons, but they dispute that this shows that such deliberation is a matter of thinking about what one ought to do, understood as thinking about what is most favoured by the balance of reasons. The first of these objections focuses on the fact that an agent might entertain a question like ‘Granted I ought to φ but I’m not going to do that, so what is the next best option?’. It may seem that the thinking an agent entertaining this question engages in is naturally described as deliberation about what to do. I think that the strongest response available to the proponent of PDND is to say that this agent is deliberating about what they ought to do—it’s just that they are deliberating about what they ought to do, given that they are not going to φ. To put it another way, they are engaging in deliberation about what they ought to do over a constrained set of options.

The second objection suggests that when agents engage in deliberation about what to do—unlike when they deliberate about what they ought to do—they are concerned with identifying a sufficiently good option rather than the most favoured option. This objection picks up on the fact that in deliberation about what to do it seems that agents are often content to satisfice rather than try to identify what is best. However, it isn’t obvious that this marks deliberation about what to do as distinct from deliberation about what one ought to do. Even if one explicitly considers what one ought to do in deliberation, one might settle for determining that an option is sufficiently good because of, say, the costs in time and mental effort associated with continuing to consider the matter. To make this objection work it needs to be shown that, in the case of deliberation about what to do, agents don’t set out with the goal of determining which option they ought to take.

The third objection to PDND claims that one’s belief that one ought to φ might feature as an input into deliberation about whether to φ alongside other considerations (cf. Silverstein Reference Silverstein2017: 358), so deliberation about what to do can’t be a matter of deliberating about what one ought to do. Here is an example: suppose David is thinking about whether to leave the party to go home and study. He believes that he ought to go home to study but that there are also things to be said for staying: he is having a good time and a friend he hasn’t seen recently had told him they will be coming later. Seemingly, David might weigh these considerations against his belief that he ought to go home. The proponent of PDND should reject the description of the case I’ve given here. They should maintain that this case involves either: (1) David reopening the question of whether he ought to go home and study or (2) rather than deliberating about whether to go home and study, David is finding his attention directed towards the considerations which support staying. This sort of mental process involved in (2), they can suggest, is the mental process commonly involved in agents’ choosing against their better judgment.