An increasingly popular view in the literature on rationality attempts to vindicate the strong normativity of rationality by giving a unifying account of rational requirements and what one ought to do in terms of reasons that fall within one’s perspective. In this paper, I pose a dilemma for such a view: one’s rationality is determined by a narrower set of reasons, such as the set of reasons that one is attending to, whereas what one ought to do is determined by a broader set of reasons that comprises reasons one is not attending to. Thus, no single set of reasons can play the dual role of determining what one is rationally required to do and determining what one ought to do: either it is too broad to determine what one is rationally required to do or it is too narrow to determine what one ought to do.
Is rationality normative, in the sense that you ought always to do what you are rationally required to do? It might seem that the answer is No. For it might seem clear that what you are rationally required to do can come apart from what you ought to do. If you mistakenly believe that the liquid in front of you is fresh water when it is in fact an odorless, transparent poison, it can be rational for you to drink it, although you ought not. More generally, it seems that rationality is a matter of what makes sense from your perspective, whereas what you ought to do is determined by how things really are, independently of your perspective. When there is a mismatch between your perspective and reality, therefore, it is natural to expect them to diverge. Indeed, many prominent theories of rationality, in recognizing the intimate connection between one’s perspective and rationality, are committed to denying that we always have reason to do what rationality requires.Footnote 1
Recently, Kiesewetter (2017) and Lord (2018) have forcefully argued that what you are rationally required to do just is what you ought to do, which, if true, would vindicate the strong normativity of rationality.Footnote 2 On their view, both what you ought to do and what you are rationally required to do are a matter of what makes sense from your perspective. In particular, there is one and the same set of reasons, reasons available to you or reasons you possess, which is supposed to play two roles: (1) determining what you ought to do; and (2) determining what you are rationally required to do.
The aim of this paper is to show that this view is mistaken: there is no single set of reasons which can plausibly play both roles. Even when we restrict ourselves to available or possessed reasons, there is still a distinction between the kind of reasons that are relevant to your rationality and the kind of reasons that bear on what you ought to do. Thus, what you are rationally required to do might diverge from what you ought to do: the intuitive gap between them remains.
The paper proceeds as follows. Section 1 explicates Kiesewetter and Lord’s views of the relationship between reasons you possess, rationality, and what you ought to do, in the central sense of ‘ought’. Section 2 considers the idea that the set of reasons you possess determines what you are rationally required to do and then show that it is mistaken. To determine what you are rationally required to do, the set of possessed reasons must be much more restricted than what Kiesewetter and Lord make it out to be, such as the set of reasons you are attending to. In Sect. 3, however, I shall argue that the kind of ought determined by such a restricted set of reasons cannot play the role the central ought is expected to play.
1 The perspectivist account of reasons, rationality, and ought
This section explicates the argument for the view that what you are rationally required to do is equivalent to what you ought to do. Both Kiesewetter and Lord argue for the alleged equivalence on the basis of the idea that only reasons which fall within your perspective in some relevant sense determine both what you are rationally required to do and what you ought to do. For convenience, I shall call their view perspectivism in what follows.
1.1 How (possessed) reasons determine rationality
Let us begin by considering the following principle linking rationality and reasons:
Equivalence: You are rational if and only if you respond correctly to reasons.
I shall standardly assume that a reason is a fact or a true proposition which counts in favor of a particular response, where a response is understood broadly to include actions, attitudes like belief and intention, and the absences thereof. I shall also assume that a reason comes with a weight, or the degree to which it supports a response, and that sometimes a set of reasons is weighty enough to decisively count in favor of a response, making it the case that you ought to have it. Thus, to respond to a set of reasons correctly is partly a matter of responding in a way that those reasons decisively count in favor of.Footnote 3
Equivalence faces what John Broome calls the Quick Objection, however. Consider:
Fish. The fish on the plate in front of you contains salmonella. This is a reason for you not to eat it, and let us assume all your reasons together require you not to eat it. But you have no evidence that the fish contains salmonella. Then you might eat it even though your reasons require you not to, and nevertheless you might be rational. (Broome 2007: 352)
Fish shows that you can fail to do what you have decisive reason to do without being irrational, when it falls outside of your perspective. In response, perspectivists either (1) deny that facts completely outside of your perspective are your reasons, or (2) agree that they are genuine reasons and yet deny that they affect your rationality.Footnote 4 The upshot is a restricted version of the view that rationality consists in responsiveness to reasons: rationality consists in responding correctly to reasons that fall within your perspective. Such reasons are called available reasons (Kiesewetter) or possessed reasons (Lord). I will arbitrarily pick the term ‘possessed’. This leads us to:
Equivalence*: You are rational if and only if you respond correctly to reasons that you possess.
What exactly is it to possess a reason? First, it is partly to stand in a positive epistemic relation with it, such as believing, being justified in believing, knowing, being in a position to know, etc.Footnote 5 Second, it is partly to grasp, in the relevant sense, the favoring relation holding between the reason and the response that it supports. For example, even if you know the (non-normative) fact that the fish contains salmonella, you might still be ignorant of the (normative) fact that this is a reason not to eat the fish, in which case you might rationally eat it (cf. Broome 2007: 352). In light of this, perspectivists hold that fully possessing a reason in the way relevant to your rationality involves grasping, in some way, the relevant favoring relation.Footnote 6 Thus understood, Equivalence* appears to be a plausible analysis of rationality in terms of reasons, which captures the intuitive idea that rationality is a matter of doing what makes sense from your perspective.
1.2 How (possessed) reasons determine what you ought to do
Reasons have yet to play another role if the perspectivist project is to succeed: they have to determine what you ought to do. For one might agree that rationality consists in responsiveness to possessed reasons but deny that you ought to do what is rational in this sense. The question is: isn’t what you ought to do plausibly determined by facts outside of your perspective? For example, isn’t there a clear sense in which, in Fish, you ought not eat the fish, objectively speaking, and regardless of whether it is rational for you to eat it?
Perspectivists argue that what you objectively ought to do is not what you really ought to do, or what you ought to do full stop, as opposed to what you ought to do in some qualified sense or from a particular standpoint (e.g. from the standpoint of morality or prudence).Footnote 7 Perspectivists call the relevant concept of ought the deliberative ought (Kiesewetter 2017: 9–10) or the ought of deliberation (Lord 2018: 210), on the grounds that it is often what you ought to do in this central sense of ‘ought’ that you try to figure out when deliberating about what to do.Footnote 8 In what follows, I shall refer to this, central concept of ought unless otherwise indicated. One prominent way of identifying such a sense of ought is as the concept of ought that figures in the following principle:
Non-Akrasia: It is irrational for you to believe that you ought to φ but fail to intend to φ.Footnote 9
Perspectivsts’ argument that the relevant concept of ought is the one that is determined by reasons you possess, rather than a fully objective one that is determined by all relevant facts, is based on what might be called three-options cases. Here is one such case:
Three Envelopes. You are given a choice between three envelopes: envelope A, envelope B, and envelope C. You know for certain that A contains $1,000. You also know that either B or C contains $1,500 and the other one is empty. But you have no evidence as to which of the two contains $1,500.Footnote 10 (cf. Ross 2006: 174)
Intuitively, you ought to choose A. It would be silly to choose either B or C. But what is meant here cannot be that you objectively ought to choose A: it is clear that, relative to all facts, you ought not choose A. Thus, the sense in which you ought to choose A must be the one that takes your epistemic situation (or perspective) into account. What is meant, for example, might plausibly be that you ought to choose A, given what you (are in a position to) know: that you are going to win $1,000 by choosing A; that there is a 50% chance that you will get nothing by choosing B or C. Such facts seem to make it the case that you ought to choose A. By contrast, facts that you aren’t in a position to know, such as the fact that B contains the $1,500, don’t feed into what you ought to do in this perspective-relative sense of ‘ought’.Footnote 11
Perspectivists argue that what you ought to do is precisely what you ought to do in this perspective-relative sense, rather than in the objective (or fact-relative) sense. Recall that the relevant kind of ought is the one that figures in Non-Akrasia. In Three Envelopes, however, it is implausible that your belief about what you objectively ought to do plays this role: even if you correctly believe that you objectively ought not choose A, it isn’t irrational for you to intend to choose A. It rather seems that it is irrational for you to fail to intend to choose A in line with your belief that you ought to choose A. If so, what you ought to do is what you ought to do given your perspective (rather than given all facts).
I shall grant to perspectivists that you ought to φ in this perspective-relative sense just in case your possessed reasons decisively support φ-ing. We have already seen that, on perspectivism, rationality is partly a matter of doing what your possessed reasons decisively support. On perspectivism, therefore, the following equivalence holds, in virtue of the two roles played by the reasons you possess: you are rationally required to φ if and only if you ought to φ.
2 Do possessed reasons determine your rationality?
This section argues that not all of the reasons you possess, in the sense specified by perspectivists, determine your rationality. For you can fail to do what your possessed reasons decisively support without any rational failure. I suggest that only a narrower set of reasons, such as the set of reasons you are attending to, is relevant to your rationality. If this is right, the proposed analysis of rationality in terms of possessed reasons fails unless the notion of possession is more restricted.
2.1 Possession without attention: a problem
Let us first consider the following thesis, which is implied by Equivalence*:
Possession-Rationality Link: You are rationally required to φ at t if and only if the set of the reasons you possess at t decisively supports φ-ing.
Recall that on perspectivism, possessing a reason r (to φ) requires meeting an epistemic condition and grasping the relevant favoring relation (1.1). Plausibly, these two individually necessary conditions are jointly sufficient for possessing r. While there is a debate about what this epistemic condition precisely amounts to, it can be plausibly assumed that knowledge is sufficient for satisfying it. Thus, if (1) you know r and (2) you grasp the favoring relation, then you possess r. With this in mind, consider the following case:
Blanket: You moved into a new apartment and want to get some household items this evening after work. You have many things to buy, but you expect not to have much time for shopping, since you also want to be back at home to watch your favorite TV show on time. For an efficient use of time, you make a shopping list with your phone. You know that, in the past, you have managed to get everything you need by relying on such a list. You also know that you need a blanket to stay warm, so you put ‘blanket’ on the list. Due to an extremely unlikely technical glitch, however, ‘blanket’ is undetectably erased from the list. In the evening, you go shopping, make sure that you get everything on the list. You notice that it’s almost time for the show to begin, and head back home. A cold, sleepless night awaits you.
Blanket, I submit, is a counterexample to Possession-Rationality Link. Let us stipulate that the following facts give you decisive reason to get a blanket: the fact that you need a blanket to stay warm; the fact that you need it to get a good night’s sleep. Call the set of these reasons R1. You possess R1, in the sense that you know its members, as well as grasping the relevant favoring relation. Thus, you possess decisive reason to (intend to) continue shopping rather than heading back home, which you fail to do.
However, you seem to be rational in all respects throughout the story. Let us begin by considering the synchronic rationality of your responses: the rationality of attitudes you have (or lack) at a given time. First, it is implausible that you rationally ought not have intended to rely on a list. You knew that your phone had worked perfectly well in the past and had no reason to doubt its reliability. Indeed, making a list seems to be a sensible thing to do, given the number of items you need to purchase. Second, it is implausible that you have been irrational for (unrealistically?) intending to both get household items and watch the show in the first place. It can be easily stipulated that you had strong evidence that you would be able to do both successfully, at the moment when you set up the plan. Indeed, you would have done so were it not for the glitch.
Third, it is implausible that you are rationally criticizable when you don’t check again if your list contains everything, when you are at the store. The very point of having such a list is to make an efficient use of your time and resources so that you don’t have to go through what you need all over again. Moreover, you lack good reasons to suspect that your list is flawed. Rather, you possess a good enough reason to think that it is a normal list: the fact that you had written down everything you need. With that said, it seems rationally permissible, if not required, for you to rely on your list.
Finally, it is implausible that you are irrational when you form the intention to head back home. Consider the set of reasons you are attending to (or noticing), when your shopping draws to a close: that the TV show is about to begin; that you’ve got everything on your list; that relying on a list has proved reliable in the past, etc. Call the set of these reasons R2. R2 doesn’t decisively support heading back home, relative to the total body of reasons you possess, since it is outweighed by R1. However, R2 does decisively support heading back, relative to the (sub)set of all reasons that you are attending to at the moment. Thus, you are correctly responding to the (possessed) reasons that you are attending to. Moreover, it is plausibly one of the things you are rationally required to do. For it would be irrational for you to fail to do what the set of reasons you are attending to decisively supports: it would not make sense if you decided to continue shopping, given that you are attending only to R2 (which decisively supports the opposing course of action) and not attending to any other reason that defeats it.
Let us also consider the diachronic rationality of your responses: the rationality of the ways in which your attitudes are maintained over time. It is difficult to identify a plausible diachronic requirement of rationality that is violated in Blanket. Broome (2013: Ch. 10), for example, argues that rationality requires persistence: rationality requires of you that if you intend (or believe) at t1 that p and no cancelling event occurs between t1 and a later time t2, you still intend (or believe) that p at t2 or consider whether p, where a ‘cancelling event’ refers to any event that licenses dropping the intention (or belief), such as considering whether p or realizing that your intention cannot be satisfied, etc. Your attitudes seem to satisfy such a persistence requirement. You begin by forming the intention to both do the shopping and watch the TV show, form the intention to rely on the shopping list, and no relevant cancelling event occurs until you carry them out. And you retain the intentions in the meantime, just as required.Footnote 12
One might object that you could still be criticized for failing to attend to (or notice) the decisive reason bearing on your decision to head back home: R1. On this line of thought, you correctly respond to all of the reasons you are attending to, and so merit some rational credit, but you fall short of full rationality because you violate yet another rational requirement on attention, which might be formulated as follows:
Attention Requirement: If you possess a reason R relevant to φ-ing, then you are rationally required to attend to R (within the context of considering whether to φ).
In response, I shall consider two available interpretations of Attention Requirement and argue against each. On the first interpretation, attending is a mental action that you can choose or intend to perform for reasons, which happens, for example, when you deliberately focus on something for the reason that it is important. Attending to a proposition you know, understood in this way, isn’t an action you can perform immediately, at least when your knowledge is buried in your memory. It is rather an action you can perform by performing other actions, such as asking yourself relevant questions and thereby retrieving or activating the knowledge from your memory.
So understood, however, Attention Requirement is implausibly demanding. Consider the following:
Ten Years Ago: You need to book a flight very soon for your upcoming conference. Knowing that one of your friends has long been a highly reliable trip advisor for you, you ask him which airline would be best between X, Y, and Z, and he recommends Z. But in fact, he told you, ten years ago, that he would play a bad trick on you on the second leap day of the 21st century (and you know that today is the second leap day of the 21st century).
The fact that your friend told you that he would fool you is surely relevant to booking a flight with Z: in the least, it is a good reason not to take what he says today at face value. Suppose you possess this fact as a reason: you know it and you grasp the favoring relation. Attention Requirement implies that you are rationally required to attend to this fact, where attending is understood as a mental action. But this is overdemanding. First, you might not know how to activate this knowledge. Although you would easily recall it if someone else asked you, ‘but didn’t he say that he would play a trick on you ten years ago?’, you might not be able to formulate this question by yourself. More specifically, you might not be able to recall it upon wondering about the questions you could be expected to ask yourself, such as: ‘has he recently given me any bad advice?’; ‘are there any other reasons why I shouldn’t take Z?’. Second, even when it is psychologically possible for you to activate the knowledge on your own, by perhaps taking up an unduly skeptical stance toward your friend and going over every conversation you had with him in the past, this might take a long enough time to thwart your original plan to book a flight in time. More generally, for cognitively limited creatures like us, satisfying Attendance Requirement might, in many cases, rule out the possibility of rationally making a quick decision.Footnote 13
Let us then consider the alternative interpretation of Attendance Requirement, on which attending to reasons is a mental event that simply occurs to you, rather than an act you can perform for a reason. On this interpretation, it is not as if you need to perform a further mental act in order to activate relevant information. As you consider the question of whether to φ and look for reasons, some of the reasons you possess simply come to your attention, and you satisfy Attendance Requirement just in case all of the relevant reasons happen to come to your attention. This version of Attendance Requirement would avoid the problem of demandingness, as it doesn’t call for any mental activity on your part.
However, this version of Attendance Requirement is implausible. For there is no rational process through which you can satisfy it.Footnote 14 On this understanding of attendance, the process of coming to attend to relevant reasons looks very much like an arational process. When you attend to the right reasons in a given situation, all that happens is that they occur to you: you are not following any rational rules; nor are you correctly responding to reasons to attend; nor are you attending to relevant reasons through a process of reasoning. This means that whether you comply with Attendance Requirement is simply left up to chance, from the rational point of view: you comply with it if the arational process luckily delivers the right reasons and fail if you are unlucky. But it is implausible that your rationality can depend on a process which is arbitrary from the rational point of view, just as it is implausible that you can be irrational on account of the malfunctioning of your digestive process. If so, this version of Attendance Requirement cannot plausibly be seen as a rational requirement.
It might be objected that it is irrational, at least to some extent, to violate Attendance Requirement, on the grounds that an ideally rational being would not fail to attend to any of reasons they possess. Perhaps an ideally rational being would, at any given moment, have every reason they possess activated in their working memory. Thus, they would attend to or notice R1 and so would correctly respond to R1 by continuing shopping and getting a blanket. However, one might continue, you are rationally criticizable to the extent that you fail to approximate an ideally rational being.
This objection equivocates between the capacity sense and the normative sense of ‘rational’.Footnote 15 Cognitively limited creatures like us lack the capacity to activate all of our knowledge instantaneously in our working memory. In this capacity sense, we are not fully rational. But it is an open question whether it is irrational, in the normative sense, for us to fail to approximate someone who is fully rational in the capacity sense. In fact, it seems deeply implausible that we rationally ought to approximate a being who is fully rational in the capacity sense, since there are things that we are required to do precisely because we lack the relevant capacity, or given our cognitive limitations, such as creating a shopping list (which is clearly a stupid thing for the ‘ideally rational being’ to do).Footnote 16 The idea that you rationally ought to approximate someone who is fully rational is only plausible if ‘fully rational’ is used to mean: satisfying all rational requirements that apply to you. The objection at hand, however, does nothing to show that you fail to be fully rational in this sense.
2.2 Possession requires attention: a solution for perspectivists
How should perspectivists respond to Blanket, then? First, they might appeal to some version of ought-implies-can principle and deny that you are required to intend to buy a blanket: it might be argued that you are required to respond to a reason only if you have the specific ability to respond to the reason, but you lack the relevant kind of ability to respond to R1 (cf. Lord 2018: 236). This doesn’t seem to be a promising way out of the problem, however. For it is unclear why you must be lacking the specific ability to respond to R1. This claim would be plausible if R1 were (at least momentarily) inaccessible to you. However, it seems that R1 is easily accessible: if you had asked yourself, ‘Wait—did I also get a blanket?’ then you would have easily recalled R1 and continued shopping on its basis. Indeed, your belief in R1 is clearly different from paradigmatically inaccessible beliefs, such as beliefs one doesn’t consciously avow but in some sense retains at the sub-personal level, or beliefs the contents of which one cannot recall despite one’s continued effort. Thus, R1 is plausibly accessible: it is just that you are not rationally required to access it in Blanket.
Alternatively, perspectivists might deny that your possessed reasons in Blanket favor continuing shopping over heading back home, on the grounds that the fact that it seems to you that you have everything you need gives you decisive reason to head back home and watch the show on time. But it doesn’t seem to be a promising way out of the problem, either. Even if it is granted that the fact about how things seem to you gives you a reason to head back, it is implausible that it outweighs the reason provided by R1: the fact that you need a blanket to stay warm. First, if you were asked to weigh R1 against the reason provided by the seeming, you would treat the former as outweighing the latter. And this seems correct: given R1, and the fact that you did not get a blanket yet, the seeming-reason only misleadingly supports going back home. Second, facts about how things seem to you, at least in general, aren’t what you treat as reasons when you decide what to do: deliberating agents often take (first-order) facts about the world as reasons in deciding what to do and cite them as their reasons for action, rather than how things seemed to them.Footnote 17
The most promising way out for perspectivists, in my view, is to restrict the set of reasons that determines your rationality further. The core idea of perspectivism is that only what falls within your perspective is relevant to your rationality, where your perspective consists of everything you (are in a position to) know. What an example like Blanket shows, however, is that such a broad conception of your perspective undermines the core idea. If the set of reasons you possess and the set of reasons relevant to your rationality are to be coextensive, therefore, some further condition has to be added for possessing a reason.
One possibility is that your perspective is constituted by everything you are attending to: what you are not attending to falls outside of your perspective, even when you know it. On this view, possessing a reason requires, in addition to meeting the epistemic condition for possession and grasping the relevant favoring relation, attending to the reason.Footnote 18 Call this more restrictive notion of possession Possession*. Perspectivists could then link rationality with possessed* reasons as follows:
Possession*-Rationality Link: You are rationally required to φ at t if and only if all of the reasons you possess* at t decisively support φ-ing.
Possession*-Rationality Link is consistent with the judgment that there is no identifiable rational failure in Blanket. At every point, you are correctly responding to the reasons that you are attending to. This is so even if you know R1 and possess it in some sense, when you decide to go back home. Since you are not attending to R1, you don’t possess* it. Given Possession*-Rationality Link, it isn’t irrational for you not to respond to it. Moreover, Possession*-Rationality Link explains why it would be in some way irrational for you to continue shopping: you are attending to R2, which, in the absence of R1, decisively favors going back home, and so you are required to respond to R2 by intending to do so.
Importantly, it is unclear how any notion of possession that is more inclusive than possession* could both (1) explain why you aren’t rationally required to respond to R1 and (2) explain why it would be irrational for you to fail to respond to R2, at the moment when you are attending only to R2. Regarding (1), if the set of reasons you possess includes more than what you attend to, then it is difficult to see how a reason that you can attend to as easily and justifiably as R1 could be ruled out, which in turn makes it difficult to deny that you are rationally required to respond to it. Regarding (2), if the set of reasons you possess includes more than what you attend to, then in every case where reasons you attend to decisively support φ-ing but all of the reasons you possess decisively support not φ-ing, it must be rationally permissible for you not to φ, which makes it difficult to see why you are irrational in some way when you fail to respond to reasons you are currently attending to.
Finally, Possession*-Rationality Link doesn’t implausibly imply that there is no rational constraint on what you attend to. For it entails that if you have decisive possessed* reason to attend to something, you are rationally required to do so, so long as attention is understood as a mental act you can perform for a reason. Thus, it allows us to rationally criticize agents for inattention when they notice (or are aware of) decisive reason to attend to something but fail to do so.
Still, Possession*-Rationality Link is intended only as a possible view that avoids the problems for Possession-Rationality Link, and it is not my goal to fully defend it here. What my argument really shows is that the set of reasons that affect your rationality is far more restricted than the set of reasons you know, and so any plausible analysis of rationality in terms of reasons should make recourse to such a restricted set of reasons. In what follows, I shall argue that any such analysis seriously undermines the perspectivists’ project of vindicating the normativity of rationality and, in doing so, use Possession*-Rationality Link simply for illustrative purposes. Readers with qualms about the principle might replace it with their preferred principle, so long as it links rationality with the set of reasons narrower than the set of reasons you know.
3 Do possessed* reasons determine What you ought to do?
Crucial to the perspectivist project is the idea that the same set of reasons determines both what it is rational for you to do and what you ought to do, full stop. So, if what determines your rationality is the set of the reasons you are attending to, then what you ought to do must be also determined by the same set of reasons. This section argues that what you ought to do is not plausibly determined by the set of reasons you possess*.
3.1 The argument from Non-Akrasia
Recall that the main argument for perspectivism about ought relies on the idea that it is the perspective-relative ought, determined by possessed reasons, that is central to one’s deliberation (1.2). One problem, however, is that the kind of ought determined by the reasons you possess* fails to play the central role in your deliberation: when you deliberate about what to do, the question you are after is, plausibly, ‘what ought I to do, relative to everything that I (am in a position to) know?’, rather than ‘what ought I to do, relative to what I am attending to?’.
Recall that a decisive test for whether an ought is of this kind is to see if it figures in Non-Akrasia (1.2). It is clear, however, that your belief about what you ought to do relative to some subset of your reasons, at least in general, doesn’t make it irrational for you to lack a corresponding intention. You might believe that, relative to some reason you have for smoking, e.g. a momentary pleasure, that you ought to smoke a cigarette now. But it doesn’t seem irrational for you not to intend to smoke. First, you might be also aware of other reasons not to smoke, which outweigh the stated reason for smoking. Second, even when you aren’t currently aware of any other reasons, you might still keep considering, thinking that there might be other reasons relevant to smoking. It is only when you believe that you ought to φ, relative to everything you (are in a position to) know, that Non-Akrasia applies to you. The reasons you are attending to fare no better in this regard. There is nothing irrational about not intending what you believe you ought to do, relative to what you are attending to. To see this, consider:
Advisor: You have an advisor, whom you find so reliable that you unconditionally trust whatever she says. You reason together with her about whether to watch a movie, informing her of all of the pros and cons you attend to, concluding on the basis of those reasons that you ought to watch it. Your advisor carefully follows your reasoning, tells you that relative to the reasons you are attending to, it is true that you ought to watch it. As usual, you firmly believe what she says. But you still wonder if there are any other reasons you have overlooked, considerations that you have failed to attend to. When you are about to ask your advisor about it, however, she has disappeared.
In Advisor, you conclude that you ought to watch the movie, in some sense. Intuitively, however, your deliberation has not been fully settled, and so it isn’t irrational for you to fail to intend to watch it. Suppose, however, that your advisor returns to you and convinces you that this is what you ought to do, given everything you (are in a position to) know. It is plausible that it is then (and only then) irrational for you not to intend it. This suggests that the kind of belief that is rationally incompatible with your lack of intention is your belief about what you ought to do relative to everything you (are in a position to) know, rather than what you ought to do relative to what you are attending to: from your own point of view, the set of reasons that determine what you ought to do must be exhaustive, even when it is in fact not.
3.2 The argument from mistake
Another argument that what you (really) ought to do cannot plausibly be determined by the reasons you are attending to begins with the following principle:
Deontic Mistake: You make a deontic mistake only if you fail to do what you (really) ought to do.
Deontic Mistake is plausible and perspectivists about ought are committed to accepting it. To see why, recall Three Envelopes (1.2). The reason why such a scenario makes a highly compelling case for perspectivism is that it is relevantly different from a garden-variety example (like Fish) in which you fail to do what you objectively ought to do because you aren’t in a position to know all of the relevant facts. The latter doesn’t give us compelling reason to favor perspectivism over objectivism.Footnote 19 Regarding such a case, objectivists can defend their view by driving a wedge between the deontic and the hypological: you fail to do what you ought to do, which is still a mistake; it is just that you are not blameworthy or criticizable as a person, since you fail unknowingly. Objectivists could then plausibly make a case for their view in the following way: “Look, it would be appropriate for you to regret what you have done by saying, ‘Oh no—I ought not have eaten the fish!’, which would be a clear way to admit to a (deontic) mistake. And it is precisely because what you ought to do is what you objectively ought to do that you make a deontic mistake: why would you think you were mistaken if you weren’t at all up to the business of doing what you objectively ought to do?”
What makes Three Envelopes distinctive is that there is no identifiable sense in which you make a deontic mistake in choosing Envelope A, even in full knowledge of the fact that it isn’t what you objectively ought to do. Here, the distinction between the deontic and the hypological doesn’t work, since you knowingly do what you objectively ought not to do. Moreover, there is nothing regrettable about your choice, although you fail to choose what you objectively ought to choose, since you know from the beginning that it isn’t what is at stake in your choice. Even when you later realize that the $1500 is in B, it wouldn’t be appropriate for you to regret by saying, ‘Oh no—I should’ve chosen B!’. And this is plausibly because you did precisely what you ought to have done. Here, the objectivist cannot plausibly make the same move and claim that you still make a deontic mistake.
This shows that if the argument from a three-options case is (as is likely) the best available argument for perspectivism, then perspectivists can only avail themselves to the best argument for their view if they accept Deontic Mistake. For the argument from the premise that you make no deontic mistake in a three-options case to the conclusion that objectivism is false is valid only if they are mediated by Deontic Mistake.
But if Deontic Mistake and Possession*-Rationality Link are true, then it is extremely implausible that what you ought to do just is what you are rationally required to do. The latter is determined by what you are currently attending to rather than everything you (are in a position to) know, given Possession*-Rationality Link. However, it seems that you can make a deontic mistake even when you do what is decisively supported by the reasons you are attending to.
To see this, let us consider Blanket again. Suppose you arrive back home to find out that you didn’t get a blanket. It seems natural for you then to judge that you failed to do what you really ought to do: it seems perfectly in order for you to express your regret by saying, for example, ‘Oh no, I made a terrible mistake—I should’ve remembered that I need a blanket!’. Let us also consider Ten Years Ago. Let us further suppose, in this case, that the friend did play a trick today: he recommended Z to you when X is in fact the best option. If you were to find out later that Z is worse than X, and to be reminded of the conversation you had with your friend ten years ago, you would naturally think that you made a mistake (‘Man, I shouldn’t have trusted him!’). Moreover, it is plausibly because what you really ought to do is what you ought to do relative to everything you (are in a position) know that you make a mistake in each of the above cases: you knew that you needed a blanket; you knew that your friend was intending to fool you. And given these reasons, you ought not have done what you have done.
In both cases, however, you do what you ought to do relative to what you are attending to, which, together with Possession*-Rationality Link, entails that you do what you are rationally required to do. Given Deontic Mistake, what you are rationally required to do is identical to what you ought to do only if you make no deontic mistake when you do what you are rationally required to do. Since you make a deontic mistake, it follows that what you are rationally required to do is not what you ought to do.
4 Conclusion: a dilemma for perspectivism
The ambition of perspectivism has been to bridge the intuitive gap between what you ought to do and what you are rationally required to do, by imposing a perspectival restriction on what determines the former. I have argued that there is a structural reason to doubt that the strong normativity of rationality can be vindicated in this way, even with such a restriction in place: the set of reasons that plausibly determines what you ought to do turns out to be overly demanding to account for rationality; or conversely, the set of reasons that plausibly determines your rationality turns out to be too weak to account for what you ought to do. One might be able to define each in terms of reasons, in line with the spirit of so-called Reasons First approach to normativity,Footnote 20 but one cannot do so in a way that vindicates the normativity of rationality, that is, by analyzing both properties in exactly the same terms.
According to the structuralist (or coherentist) views of rationality, endorsed by Scanlon (1998, 2007), Broome (2013), Way (2018) and Worsnip (2018), rationality consists in avoiding incoherent combinations of attitudes, which you can achieve even when you take up an attitude that is not supported by reasons. The idea that you always ought or have reason to be rational in this structuralist sense has been famously challenged by Broome (2007, 2013) and Kolodny (2008). According to the subjective reasons account, espoused by Schroeder (2009) and Parfit (2011), rationality consists in responding correctly to apparent reasons, or contents of one’s beliefs which, if true, would be reasons. On this account, you can rationally fail to do what you ought to do in case your relevant beliefs are false, and so apparent reasons and real reasons fail to line up. On the transparency account, advanced by Scanlon (1998, 2007) and Kolodny (2005, 2008), rationality consists in doing what you believe you have reason to do. Again, on this view, if you have false beliefs about what you have reason to do, you can be rationally required to do what you have no reason to do. For an elegant overview of how each view fails to vindicate the normativity of rationality, see Kiesewetter (2017: Ch.5).
It is common to distinguish between two claims about the normativity of rationality: strong and weak. Rationality is weakly normative just in case there is a reason for you to φ if rationality requires you to φ. Rationality is strongly normative just in case you ought (or have decisive reason) to φ if rationality requires you to φ. The primary focus of this paper will be the strong normativity claim, since the weak normativity claim has never been defended in print and is not theoretically interesting in any case. For, as many philosophers note (Kiesewetter 2017; Way 2010), the weak normativity claim is in a poor position to account for the strictness of rational requirements.
Only partly, since it is plausible that correctly responding to reasons also requires that there be an appropriate basing relation between the reasons and your response. See Lord (2018) for a detailed discussion.
See Lord (2018: Ch. 3) for a taxonomy of possible views, and his argument for the view that the relevant epistemic relation is being in a position to know. See Kiesewetter (2017: Ch. 7) for criticism, and for the view that knowledge is sufficient for meeting the epistemic condition.
Lord (2018: Ch. 4) argues that, unlike the first condition, this need not be an epistemic condition: you need not explicitly believe that the fact is a reason to do such-and-such. He calls it the practical condition, which he thinks is a matter of having an ability (or knowledge how). Roughly, you meet the practical condition for possessing a reason r to φ just in case you are in a position to manifest your knowledge of how to use (e.g. in your reasoning) r as a reason to φ. However, there is room to doubt that Lord’s practical condition is plausible, as Fogal and Worsnip (ms.) argue. I shall not take a stand on the nature of the second condition, while assuming that some such condition is needed.
One might call into question the assumption that there is a single concept of ought that you are after when you are deliberating in this way, though I won’t press this worry here. See Sepielli (2018), however.
Both Regan’s (1980) miners case and Jackson’s (1991) drugs case have the same structure and are often discussed in the same context. My reason for focusing on Three Envelopes is that both Jackson’s and Regan’s cases primarily concern our moral obligation, and judgments about what the agents in such cases ought to do could easily be interpreted as judgments about what they morally ought to do.
An anonymous reviewer has insightfully pointed out that Blanket can be a counterexample not only to Possession-Rationality Link, but also to Non-Akrasia. For one could easily imagine a variant of the example in which you believe that you ought to get a blanket, rather than believing just that you need a blanket, and reach the same verdict about your rationality. This might complicate the dialectic, since I have identified the central ought in terms of Non-Akrasia (1.2) and one of my arguments against perspectivism to be presented in Sect. 3 also depends on this principle. I have two reservations. First, there is room to resist the idea that Blanket is a counterexample to Non-Akrasia. For it might be argued that explicitly believing that you ought to get a blanket is one thing and possessing R1 is another. They might turn out to be near-equivalent if grasping the relevant favoring relation is a matter of explicitly believing that R1 counts decisively in favor of getting a blanket. But it could instead be a matter of having an implicit belief or meeting what Lord calls the practical condition for possessing R1, in which case possessing R1 would be different from explicitly believing that you ought to get a blanket. This in turn might make a rational difference between being akratic and the failure to respond to R1. Second, even if Blanket turns out to be a counterexample to some unrestricted version of Non-Akrasia, it does nothing to show that clear-eyed akrasia, or the failure to intend what you occurrently or consciously believe you ought to do, can be rational. Hence, a restricted version of Non-Akrasia, which applies only to occurrent beliefs, still seems true. So long as such a restricted version of Non-Akrasia holds, there is no serious threat to my argumentative moves. Lee (2020), for example, argues that local structural requirements of rationality (if there are any) apply only to occurrent attitudes.
Podgorski (2016: 1933) briefly argues, on similar grounds, that the requirement to respond to your total evidence is overly demanding so long as you are a cognitively limited agent.
See Kiesewetter (2017: 78–79) for what he calls Rational Process Constraint on state-requirements of rationality.
For detailed arguments for these points, see Littlejohn (2019).
Feldman (1988) presents a similar conception of having evidence, according to which someone has p as available evidence if and only if he is currently thinking of p. But on Feldman’s view, it is possibly false contents of one’s beliefs, rather than what one knows, that count as evidence..
Broome, J. (2007). Does rationality consist in correctly responding to reasons? Journal of Moral Philosophy, 4(3), 349–374.
Broome, J. (2013). Rationality through reasoning. Hoboken: Wiley-Blackwell.
Feldman, R. (1988). Having evidence. In D. F. Austin (Ed.), Philosophical analysis (pp. 83–104). Dordrecht: Kluwer Academic Publishers.
Fogal, D., & Worsnip, A. (ms.) Which reasons? Which rationality? http://www.alexworsnip.com/wp-content/uploads/2020/07/Which-Reasons-Which-Rationality.pdf/.
Jackson, F. (1991). Decision-theoretic consequentialism and the nearest and dearest objection. Ethics, 101(3), 461–482.
Johnson, R. N. (1999). Internal reasons and the conditional fallacy. Philosophical Quarterly, 49(194), 53–71.
Kiesewetter, B. (2017). The normativity of rationality. Oxford: Oxford University Press.
Kolodny, N. (2005). Why be rational? Mind, 114(455), 509–563.
Kolodny, N. (2008). Why be disposed to be coherent? Ethics, 118(3), 437–463.
Lee, W. (2020). The real myth of coherence. Erkenntnis. https://doi.org/10.1007/s10670-020-00239-y.
Littlejohn, C. (2019). Being more realistic about reasons: On rationality and reasons perspectivism. Philosophy and Phenomenological Research, 99(3), 605–627.
Lord, E. (2018). The importance of being rational. Oxford: Oxford University Press.
Parfit, D. (2011). On what matters 1. Oxford: Oxford University Press.
Podgorski, A. (2016). Dynamic permissivism. Philosophical Studies, 173(7), 1923–1939.
Regan, D. (1980). Utilitarianism and co-operation. Oxford: Oxford University Press.
Ridge, M. (2014). Impassioned belief. Oxford: Oxford University Press.
Ross, J. (2006). Acceptance and practical reason. Ph.D. Dissertation, Rutgers University.
Scanlon, T. M. (1998). What we owe to each other. Cambridge: Harvard University Press.
Scanlon, T. M. (2007). Structural irrationality. In G. Brennan (Ed.), Common minds: Themes from the Philosophy of Philip Pettit (pp. 84–103). Oxford: Clarendon Press.
Schroeder, M. (2009). Means-end coherence, stringency, and subjective reasons. Philosophical Studies, 143(2), 223–248.
Sepielli, A. (2018). Subjective and objective reasons. In D. Star (Ed.), Oxford handbook of reasons and normativity (pp. 784–799). Oxford: Oxford University Press.
Smith, M. (1994). The moral problem. Oxford: Blackwell.
Way, J. (2010). The normativity of rationality. Philosophy Compass, 5(12), 1057–1068.
Way, J. (2018). Reasons and rationality. In D. Star (Ed.), Oxford handbook of reasons and normativity (pp. 485–503). Oxford: Oxford University Press.
Wedgwood, R. (2007). The nature of normativity. Oxford: Oxford University Press.
Worsnip, A. (2018). The conflict of evidence and coherence. Philosophy and Phenomenological Research, 96(1), 3–44.
Zimmerman, Michael. (2008). Living with uncertainty. Cambridge: Cambridge University Press.
For helpful comments and suggestions, I thank the anonymous referees for Philosophical Studies and members of the Philosophical Anthropology and Ethics research group at the University of Duisburg-Essen.
Open Access funding provided by Projekt DEAL.
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.