If you justifiably believe that you ought to Φ, you ought to Φ

Abstract

In this paper, we claim that, if you justifiably believe that you ought to perform some act, it follows that you ought to perform that act. In the first half, we argue for this claim by reflection on what makes for correct reasoning from beliefs about what you ought to do. In the second half, we consider a number of objections to this argument and its conclusion. In doing so, we arrive at another argument for the view that justified beliefs about what you ought to do must be true, based in part on the idea that the epistemic and practical domains are uniform, in a sense we spell out. We conclude by sketching possible implications of our discussion for the debates over what is wrong with akrasia and pragmatic encroachment on justified belief and knowledge.

1 Introduction

In this paper, we argue that justified beliefs about what you ought to do must be true. That is, we argue that the following principle holds necessarily:

$${\quad}{\text{Ought}}\,{\text{Infallibilism}}\quad \;\; {\text{If}}\,{\text{you}}\,{\text{justifiably}}\,{\text{believe}}\,{\text{that}}\,{\text{you}}\,{\text{ought}}\,{\text{to}}\,{\upvarphi },\,{\text{you}}\,{\text{ought}}\,{\text{to}}\,{\upvarphi }.$$

Although the claim is not that justification is in general factive, it might still appear surprising and contentious. Just as it is widely held that you can have false justified beliefs about non-normative matters, it is widely held that you can have false justified beliefs about what you ought to do. For instance, it will seem implausible to many that, if consequentialism is false, no-one is justified in believing consequentialism or that everyone on one side of, say, the debate over euthanasia must be not just wrong but unjustified.

In this paper, after some preliminaries, we argue for Ought Infallibilism by reflection on what makes for correct reasoning from beliefs about what you ought to do. We then consider a number of objections to this argument and its conclusion. In doing so, we arrive at another argument for Ought Infallibilism, based in part on the idea that that the epistemic and practical domains are uniform, in a sense to be spelled out. The two arguments are independent but mutually reinforcing.¹

2 Preliminaries

So as to clarify the principle we will be arguing for, we will make some remarks on the key terms which figure in it, starting with ‘ought’. The aim of this section is not to defend any substantive views but merely to explain how Ought Infallibilism is to be understood.

It is common to distinguish different senses of ‘ought’. We intend ‘ought’ as it appears in both the antecedent and consequent of Ought Infallibilism to be understood as concerning the ‘deliberative ought’ (cf. Broome 2013; Kiesewetter 2011; Lord 2015; Schroeder 2011; Zimmerman 2014): the ‘ought’ in play when you ask yourself, with the aim of making a decision, ‘What ought I to do?’²

One might ask whether this ‘ought’ should be understood ‘objectively’ or ‘perspectivally’. That is, does what you ought in this sense to do depend on the facts or on one’s epistemic position? We take this to be a substantive question: when you ask yourself what you ought to do, you employ a single notion of ‘ought’, one which is either determined by all of the facts, or by your epistemic position. Objectivists hold that this ‘ought’ is determined by all the facts; perspectivists hold that it is determined by your epistemic position. As we explain later, when discussing these views in more detail, this paper is neutral on this question.³

Another key normative notion which figures in Ought Infallibilism is that of justification. We take justification to be a general notion applying to decisions and actions as well as beliefs. In all cases, there is a difference between there being justification for φing and justifiably φing. A standard view, which we assume throughout, is that you justifiably φ if and only if there is justification for you to φ and your φing is properly based on this justification (or the considerations which provide it). In the jargon, the antecedent of Ought Infallibilism concerns a case of doxastic or ex post justification, not merely propositional or ex ante justification.

We make two further assumptions. First, justification has deontic implications: if there is justification for you to φ, you may φ (that is, you are permitted to φ).⁴ Second, ‘ought’ and ‘may’ are duals: you may φ if and only if it is not the case that you ought not to φ.

Just as there are different senses of ‘ought’, there are different senses of ‘justification’. And, just as there is a deliberative sense of ‘ought’, there is a deliberative sense of ‘justification’. The assumptions above guarantee this. If you ought not to φ in the deliberative sense, then in the same sense you are not justified in φing. Another way to introduce the relevant notion of justification is to note that, just as the question ‘What ought I to do?’ can govern or guide deliberation, the question ‘What am I justified in doing?’ can govern or guide deliberation. Plausibly, the answer to the second question will accord with the answer to the second. The notion of justification we are interested in is the one expressed in such contexts.

In view of this, it is unsurprising that, just as there is a debate between objectivists and perspectivists concerning what determines what you ought to do, there is a corresponding debate about what determines what you are justified in doing: for some justification depends on the facts, for others it depends on one’s epistemic position.⁵

We take it that these remarks pick out a single interpretation of Ought Infallibilism. We do not deny that ‘ought’ and ‘justifiably’ can also express other notions, so that alternative readings of Ought Infallibilism are possible. We will not consider such readings here. When interpreted in the way we intend, Ought Infallibilism is a significant thesis, insofar as the deliberative senses of ‘ought’ and ‘justification’ are central to ethics and epistemology: ‘what ought I to do?’ is widely taken to be the central question of ethics; many epistemologists have taken as their central concern a notion of justification which is deontic and plays an important role in guiding us in belief-formation and other epistemic practices. We consider the importance of our thesis in more detail below.

One final clarification. The variable in Ought Infallibilism should be understood as ranging over actions (only). Whether Ought Infallibilism could be extended to cover attitudes like beliefs and intentions is not an issue we address in what follows.

3 The argument from enkratic reasoning

We now turn to the principal argument for Ought Infallibilism, which turns on the role of normative beliefs in deliberation. To see the basic idea, suppose that you are deliberating as to whether to go the cinema this evening and you come to justifiably believe that you ought to do so. How should your deliberation proceed? On the face of it, you should go ahead and decide to go the cinema. But if you should decide to go to the cinema, you should also go when the time comes.

Our argument spells out and develops this idea in detail. We proceed in two stages.

3.1 Stage 1

Consider:

• (1) You justifiably believe that you ought to φ.

• (2) It is correct reasoning to move from the belief that you ought to φ to deciding to φ.

• (3) If you reason correctly from justified premise-attitudes, you will reach a justified conclusion-attitude.

• (4) So, there is justification for you to decide to φ.

• (5) So, there is justification for you to φ.

• (6) So, you may φ.

While this falls short of establishing Ought Infallibilism, it gets close to it:

$${\text{Almost}}\,{\text{Infallibilism}}\quad \quad {\text{If}}\,{\text{you}}\,{\text{justifiably}}\,{\text{believe}}\,{\text{that}}\,{\text{you}}\,{\text{ought}}\,{\text{to}}\,{\upvarphi },\,{\text{you}}\,{\text{may}}\,{\upvarphi }.$$

If this is as far as the argument gets, it is still an interesting and contentious conclusion. According to it, if you justifiably believe that you ought to go to the cinema, and act in accordance with that belief, you cannot go wrong—that is, you cannot find yourself doing something you ought not to do.⁶

Before considering how to move from here to Ought Infallibilism, some comments on this argument.

(2) is highly plausible and widely accepted (cf. Broome 2013; Wedgwood 2007). Indeed, since the ‘ought’ in play is deliberative, it is hard to deny. In deliberation you aim to make a decision by answering the question, ‘What ought I to do?’ It thus seems to be a presupposition of deliberation that a positive answer to this question will allow you to bring deliberation to a close—that is, that it is correct reasoning to move from the belief that you ought to φ to deciding to φ.

Note that (2) is not the claim that believing that you ought to φ makes it the case that you ought to decide to φ. (That would be ‘bootstrapping’). Rather, it is just the claim that the transition between these attitudes is correct.

(3) is highly plausible. It is part of the point of reasoning to allow us to come to new justified attitudes. Of course, reasoning cannot do this from nothing—correct reasoning from unjustified starting points cannot be expected to lead to justified attitudes. But if we begin from justified attitudes, we should expect correct reasoning to lead us to further justified attitudes.

That said, it might be thought that (3) fails to take account of the defeasibility of correct reasoning. For example, it might be claimed that it is correct reasoning to move from believing that you promised to φ to intending to φ. However, this piece of reasoning does not guarantee that, if you start from a justified premise-attitude, you will reach a justified conclusion attitude. For instance, you might also have a justified belief that by failing to φ you could do something more important. In that case, you would not be justified in intending to keep your promise.

Enkratic reasoning is, however, indefeasible. That is, given the justified belief that you ought to φ, there are no justified attitudes which could be added to your psychology which would make you unjustified in making the transition to deciding to φ (without also making you unjustified in believing that you ought to φ). To see this consider the following putative cases in which enkratic reasoning is defeated:

Case 1: As well as justifiably believing that you ought to go the cinema, you justifiably intend to get medicine for a friend. Given this intention, and assuming you cannot both go to the cinema and get medicine, you are not justified in making the transition from your belief to deciding to go to the cinema.

In this case, it seems that whatever considerations justify intending to get medicine, say, that your friend is ill, would defeat the justification for believing that you ought to go the cinema, say, that a new Coen brothers film is showing.⁷

Case 2: As well as justifiably believing that you ought to go the cinema, you justifiably believe that going to the cinema will prevent you from meeting an important deadline. So, you are not justified in making the transition from the belief about what you ought to do to deciding to go the cinema.

Typically, you would not be justified in believing that you ought to go the cinema while also justifiably believing that doing so would result in missing an important deadline. (The importance of the deadline would defeat any other reasons for going to the cinema.) But, in cases in which you do hold both beliefs justifiably you remain justified in making the transition from the normative belief to the decision.

More generally, it seems that, if you justifiably believe that you ought to φ, you cannot also have further justified attitudes which make the transition to deciding to φ incorrect. Thus we can avoid the worries about (3) by adding ‘indefeasibly’ in (3) and (2), or just by reading ‘correct’ as expressing an indefeasible notion.

The move from (4) to (5) is defensible too. Ordinarily, the kinds of considerations which justify a decision also justify the corresponding action. For example, if the fact that a new Coen film is showing justifies you in deciding to go to the cinema, it also justifies you in going to the cinema. More generally, if you are justified in deciding to do something, you are justified in doing it.⁸

One might think there are exceptions to this rule, namely, cases involving so-called reasons of the wrong kind (WKRs).⁹ Suppose that an eccentric billionaire offers you a reward for deciding to go to the cinema and on this basis alone you decide to go. In this case, one might think, you are justified in deciding to go the cinema but not in going.

However, the kind of case we are considering does not involve WKRs. The basis for the decision to φ is a justified belief that you ought to φ. A WKR for making a decision could not justify such a belief. If the billionaire’s offer justifies deciding to go the cinema, it does not justify you in believing that you ought to go; after all, we can make the case one in which, if you go, you will not get the reward.

Setting aside WKRs, the move from (4) to (5) looks fine.

The move from (5) to (6) follows from our earlier assumption that justification has deontic implications.¹⁰

So, from (1), the assumption that you justifiably believe that you ought to φ, plus additional premises which we have defended, we arrive at (6), you may φ. This establishes Almost Infallibilism.

3.2 Stage 2

Can we make the step from Almost Infallibilism to Ought Infallibilism? There are two lines of thought that support doing so.

First, as mentioned above, to say that you may φ is to say that it is not the case that you ought not to φ. It would be surprising if your justified normative beliefs guaranteed that you lack certain obligations but not that you have them. Indeed, if Almost Infallibilism were true but Ought Infallibilism false, there would have to be considerations which could prevent it being the case that you ought to φ which (i) did not prevent you being justified in believing that you ought to φ and (ii) could not make it the case that you ought not to φ. We consider below (Sect. 5) if there are such considerations of which (i) is true. But it would be very odd if there were such considerations and (ii) was also true of them. If there are considerations which are compatible with justifiably believing that you ought to φ that can make it permissible not to φ, why can’t there also be considerations of this sort which can make it impermissible to φ? Indeed, it seems it should be possible to turn considerations of the former sort into considerations of the latter sort, just by increasing their weight.

To make the point more concrete, consider:

You justifiably believe that you ought to go the cinema this evening on the grounds that a new Coen brothers film is showing. Unbeknownst to you, the pilot for the new Coen brothers TV series is also showing this evening. Given this consideration, it is not the case that you ought to go the cinema – you may go to the cinema but you may instead watch TV. So, your justified belief is false.

If this is possible, the following is surely possible too:

Unbeknownst to you, the pilot will show once and once only (whereas the film will be available on DVD in due course). Given this fact, you ought to watch TV and ought not to go the cinema. So, your justified belief is false.

In view of this, we think that, if Almost Infallibilism is true, Ought Infallibilism must also be true.¹¹

Here is a second line of thought which takes us from (4)—there is justification for you to decide to φ—to Ought Infallibilism:

• (7) If you justifiably believe that you ought to φ, you can have no other justified attitudes from which you could correctly reason to deciding not to φ.¹²

• (8) If you can have no other justified attitudes from which you could correctly reason to deciding not to φ, you lack justification for deciding not to φ.

• (9) So, you lack justification for deciding not to φ.

• (10) If there is justification for you to decide to φ and no justification for deciding not to φ, you ought to decide to φ.

• (11) So, you ought to decide to φ.

• (12) So, you ought to φ.

Cases 1 and 2 above suggest that (7) is true. In Case 1, it seems that you could not be justified in both believing that you ought to go to the cinema and in deciding not to go the cinema but to get medicine instead (if you justifiably believe that you can’t do both). In Case 2, you would not be reasoning correctly if you moved from the justified belief that going to the cinema will lead you to miss an important deadline to deciding not to go, given your justified belief that you ought to go to the cinema.

(8) is highly plausible. To see this, consider the contrapositive: if you are justified in deciding not to φ, you can have some justified attitudes from which you could correctly reason to making that decision. To deny this is to allow for baseless justified decisions.

(9) follows from (4), (7) and (8).

Is (10) true? Even if you cannot make a justified decision without deciding to φ, one might think, it does not follow that you ought to decide to φ. Perhaps you may simply make no decision at all as to whether to φ.

However, if you are justified in making no decision regarding φing, it is hard to see how you could also be justified in believing that you ought to φ. Whatever considerations justify you in thinking that you ought to go the cinema—for example, that you promised to meet your friend there—would presumably defeat any considerations which might otherwise justify making no decision as to whether to do so—for example, that it does not matter whether you go to the cinema. Since we are concerned only with cases in which you justifiably believe that you ought to φ, we can set aside the above doubts about (10).¹³

(11) follows from (4), (9), and (10).

Once again setting aside cases involving WKRs (cf. Sect. 3.1), (12) follows from (11).

So, from (1), the assumption that you justifiably believe that you ought to φ, plus additional premises which we have defended, we arrive at (12), you ought to φ. This establishes Ought Infallibilism.

4 Interval

Ought Infallibilism might seem incredible (in a bad way). Consider:

Justified false beliefs are possible. This claim commands almost universal agreement among contemporary epistemologists. (Greco 2014: 203).

By common consent, justification is distinct from truth […] False beliefs can be justified. (Alston 1993: 535)

The logical independence of justification and truth is a staple of contemporary epistemology. (Vahid 2006: 305).

There are, as always, exceptions. Some claim that all justified beliefs are true (e.g. Littlejohn 2012; Steglich-Petersen 2013; Sutton 2007). But this is very much a minority position. In any event, as stressed above, Ought Infallibilism claims only that justified beliefs about a particular subject matter, namely, what you ought to do, must be true. This is consistent with the denial that justification is in general factive.

If it is typically possible to have false justified beliefs, why should beliefs about what you ought to do be any different? We now turn to objections to Ought Infallibilism. Addressing some of them will allow us to answer this question, and provide a further argument for the principle.

In doing so, we make a further assumption: that what you should and may do have implications for what you have reason to do (and vice versa). According to one view of this sort:

You ought to φ if and only if you have most reason to φ; you may φ if you have sufficient reason to φ.¹⁴

It is important to note that this assumption is not needed for our argument for Ought Infallibilism. Its role is just to make it easier to formulate and discuss certain objections to it; that is to say, it is for the benefit of our opponent.

5 The hybrid view

As mentioned above (Sect. 2), there is a disagreement about how to understand the deliberative ‘ought’. According to objectivism, what you ought to do is determined by all the facts of your situation. According to perspectivism, what you ought to do is determined by your perspective on the facts, that is, by your epistemic position. What constitutes your epistemic position? Different answers to this question result in different versions of perspectivism. Perhaps your perspective is determined by the known facts, the knowable facts, how the facts appear to you to be, the evidence you have, and so on. The differences between these views do not matter for present purposes.

There is a corresponding debate about reasons (which is unsurprising, if what reasons you have determines what you ought to do). For objectivists, a fact can be a reason for you irrespective of your epistemic position. For perspectivists, only considerations within your perspective can be reasons.¹⁵

The debate between objectivists and perspectivists arises in both the practical and epistemic domains. Many philosophers implicitly or explicitly are objectivists about what we ought to do and perspectivists about what we ought to believe (hence, what we are justified in believing).¹⁶ Call this the hybrid view.

If the hybrid view is true, Ought Infallibilism is false. There might be situations in which, given your perspective, you ought to φ but, given the facts, you ought not to φ (cf. Titelbaum 2015: 266).

In response, we argue that the hybrid view is unmotivated. The arguments for objectivism about reasons for action, if successful, support objectivism about reasons for belief, while the arguments for perspectivism about reasons for belief, if successful, support perspectivism about reasons for action. The same is true of arguments for objectivism and perspectivism about the deliberative ‘ought’. Needless to say, we cannot explore all the existing arguments for or against objectivism or perspectivism. Instead, we will focus on the most prominent.¹⁷

5.1 For perspectivism

One consideration which might seem to support perspectivism, and count against objectivism, is that a normative reason for φing must be capable of motivating a person to φ.¹⁸ That is, if that p is a reason someone has to φ, she must be capable of φing for that reason. But only what falls within a person’s perspective is capable of motivating her. So, normative reasons must fall within a person’s perspective. This means, in turn, that what a person ought to do depends on her perspective.

If this argument is successful, it succeeds in the case of belief too (cf. Gibbons 2013; Pollock 1986: 126ff). If a person has a reason to believe something, she must be capable of believing for that reason. But only what falls within a person’s perspective can be a reason for which she believes. So, normative reasons for believing must fall within a person’s perspective. This means, in turn, that what a person ought to believe, and so what she is justified in believing, depends on her perspective.

Another line of thought which might point to perspectivism is that objectivism is too harsh—if a person φs when φing is not the best option, she is not criticisable or blameworthy for doing so if, say, all the information she possesses suggests that φing would be best. Suppose that all the evidence available to Doctor suggests that giving Patient a pill will relieve Patient’s headache, but it in fact causes Patient terrible suffering. Given objectivism, Doctor acted wrongly—she ought not to have given the pill. But Doctor is not to blame for the suffering and seems beyond reproach. If a failure to do what you ought to do makes you liable for blame or criticism, objectivism must be false.¹⁹

If this line of thought is persuasive, it is as persuasive in the epistemic domain (cf. Steup 1999). Since all the evidence suggests that the pill will relieve the headache, Doctor is not blameworthy or criticisable for believing this. So, whether a person ought to believe something, or is justified in doing so, is not determined by the facts but by her perspective on the facts.

Perhaps the most influential argument for perspectivism about the practical ought appeals to cases like the following:²⁰

Doctor is deciding whether to give drug A, drug B, or drug C to treat Patient’s painful but nonfatal disease. Doctor knows that B will partially cure Patient and that one of A and C will completely cure Patient and the other will kill her. However, she is not in a position to know which of A and C will kill Patient and which will completely cure her. In fact, A is the complete cure.

According to objectivism, Doctor ought to give A. But that seems wrong. If Doctor gives A there’s a 50/50 chance that Patient will die, whereas Patient will certainly be partially cured if she gives B. According to perspectivism, the doctor should give B, which seems right.

Parallel lines of thought have been offered in support of perspectivism about the epistemic ought. Suppose that a fair coin has been tossed but you have not seen how it landed. If the coin landed heads, objectivism implies that you may believe it landed heads. But this seems wrong—from your perspective, there’s a 50/50 chance that it landed tails instead. Given the information available, it seems that you should suspend judgment. This is the verdict perspectivism delivers (cf. Feldman 1988a: 245).

5.2 For objectivism

An influential line of thought in support of objectivism, and against perspectivism, about the practical ought begins with reflection on contexts of advice (see Graham 2010; Thomson 2008: 187–191). When advising a person as to whether she ought to φ, the advisor takes into account (what she takes to be) the facts, not the advisee’s epistemic situation. Suppose that an advisee asks, ‘Should I go to the cinema?’ The advisee has no knowledge or views which would support going over not going. In contrast, the advisor knows that a new Coen brothers film is showing. In light of this information, the advisor says, ‘You should go to the cinema’. Given perspectivism, this advice is false, at least prior to being given. Worse still, the advice changes the advisee’s epistemic position. In that case, it is hard to see how it could count as informing the advisee of what she ought to do, rather than altering this. These are not problems for objectivism.

We find the same when we turn to advice about what to believe. Suppose that an advisee asks, ‘Should I believe that Smith killed Jones?’ The advisee has no knowledge or views which would support believing or disbelieving this. In contrast, the advisor knows that Smith’s fingerprints were on the gun. In light of this information, the advisor says, ‘You should believe that Smith killed Jones’. Given perspectivism, this advice is false, at least prior to being given. Worse still, the advice changes the advisee’s epistemic position. In that case, it is hard to see how it could count as informing the advisee of what she ought to think, rather than altering this. These are not problems for objectivism.

A related line of thought for objectivism and against perspectivism in the practical domain appeals to judgments of hindsight (cf. Ross 2002 [1930]: 32). To return to an earlier example, on learning that, contrary to the evidence, the pill caused Patient terrible suffering, Doctor might say to herself, ‘I shouldn’t have given Patient the pill!’ or, ‘I was wrong to give the pill!’ If perspectivism is true, these thoughts would be mistaken. Objectivism, in contrast, vindicates them.

If this line of thought succeeds, it applies in the case of belief too. On learning that the pill causes terrible suffering, Doctor might think to herself, ‘I should never have believed that it would cure her!’ or ‘I was wrong to think the pill would cure her!’ If perspectivism is true, these thoughts would be mistaken. Objectivism, in contrast, vindicates them.

The phenomenology of first-person deliberation seems to point to objectivism. When deliberating as to whether to believe that Smith killed Jones, the focus is not on one’s epistemic situation but on, say, whether Smith had a motive—that is, on the facts. If the aim of theoretical deliberation is to answer the question, ‘What should I believe?’, the most natural explanation for this is that the answer to that question is determined by the facts (cf. Shah 2003).

Likewise, when deliberating as to whether to go the cinema, the focus is not on one’s epistemic situation but on, say, whether the Coen brothers film is showing—that is, on the facts. If the aim of practical deliberation is to answer the question, ‘What should I do?’, the most natural explanation for this is that the answer to that question is determined by the facts (cf. Shah 2008).

5.3 Interim conclusion

As this survey suggests, if an argument supports objectivism in the practical domain, it supports objectivism in the epistemic domain, and vice versa. Likewise, if an argument supports perspectivism in the practical domain, it supports perspectivism in the epistemic domain, and vice versa. So, the hybrid view is unmotivated.²¹ We should either be objectivists about both what we ought to do and what we ought to believe or be perspectivists about both. Call this the uniformity view.

Note that the above suggests, not just that practical and epistemic oughts are both epistemically constrained or both epistemically unconstrained, but that, if the relevant oughts are epistemically constrained, the constraints are the same in each case.

It is important to stress that we are not endorsing the arguments for objectivism or perspectivism. The claim is rather that objectivism or perspectivism in the epistemic domain stands or falls with its counterpart in the practical domain. This paper is neutral as to which stands and which falls.

With the hybrid view rejected, Ought Infallibilism should seem considerably more plausible. Indeed, some might think that the uniformity view implies Ought Infallibilism.

However, this is a mistake. Even granting the uniformity view, Ought Infallibilism will be false if (A) there are reasons for believing you ought to φ which are not reasons to φ or (B) reasons not to φ which are not reasons not to believe that you ought to φ. Reasons of either sort might allow you to justifiably believe that you ought to φ when it is not the case that you ought to φ.

In the following sections, we consider whether there are any cases of type (A). (What we say carries over to putative cases of type (B).) We will suggest that whether or not there are such reasons, Ought Infallibilism holds. Together with our rejection of the hybrid view, this will amount to a further argument for Ought Infallibilism.

5.4 WKRs

One kind of putative counter-example to Ought Infallibilism involves WKRs. Consider:

An eccentric billionaire offers you a reward to believe that you ought to go the cinema, which will be withdrawn if you go. The offer is not a reason to go. In this case, you have most reason to believe that you ought to go to the cinema, hence, you are justified in so believing, but it is not the case that you ought to go.

In passing, note that cases like this challenge Ought Infallibilism, even when justification is understood in objectivist terms.

The most straightforward response to such cases is to deny that there are WKRs, more specifically, that there are pragmatic reasons for believing.²² It is beyond the scope of this paper to defend this response so we will offer some alternatives.

We could reformulate Ought Infallibilism as follows: If you are epistemically justified in believing that you ought to φ, you ought to φ.²³ Since the billionaire’s offer is not an epistemic justification for believing, it is not a counterexample to this principle, which remains interesting and contentious.

Even if there are WKRs for believing, we cannot believe on the basis of them. This is a point that those who deny that there are WKRs often stress but, importantly, it is typically granted by those who claim that there are (e.g. Foley 1987: 215–16; Reisner 2009). Ought Infallibilism is not a claim about when there is justification for believing but about when a subject believes on that basis. So, even if there are WKRs, they do not provide counterexamples to Ought Infallibilism.

5.5 Deontic testimony

Testimonial cases seem to provide further counterexamples to Ought Infallibilism. Consider:

A reliable friend tells you that you ought to go to the cinema. This testimony provides a reason for believing that you ought to go to the cinema – other things being equal, you might justifiably believe that you ought to go on the basis of this testimony. But that your friend has told you that you ought to go to the cinema is not a reason for going. So, it cannot make it the case that you ought to go. So, Ought Infallibilism is false.²⁴

This objection gets much of its force from the hybrid view. Suppose that, in the example above, you act on your friend’s advice and go to the cinema. It turns out that you hate the film. This might seem like a case in which you justifiably but mistakenly thought that you ought to go to the cinema. But this reaction presupposes that the fact that you would hate the film—a fact that was outside your perspective—was a reason not to see the film, but not a reason not to believe that you ought to see it. This requires the hybrid view.²⁵

Nonetheless, the objection does not get all of its force from the hybrid view. There is also the worry that the fact that your friend said that you ought to φ is not a reason to φ because it is not a ‘right-maker’—it is not the kind of thing that could explain why you ought to φ.

Kearns and Star argue that testimony is a right-maker (2009). So one option is to accept their defence of the claim that the fact that your friend told you to do something is a reason to do it.

However, there are other options. Ought Infallibilism could be true even if testimony that you ought to φ is not a reason to φ. Ought Infallibilism requires only that whenever deontic testimony puts you in a position to justifiably believe that you ought to φ, there are reasons to φ strong enough to ensure that you ought to φ. It does not require that the testimony is itself such a reason.

In many cases, deontic testimony will either be accompanied by or implicate further claims. For instance, if testimony that you ought to eat cabbage puts you in a position to justifiably believe that you ought to eat cabbage, you will typically also be put in a position—given ordinary background knowledge—to justifiably believe that cabbage is, or is likely to be, healthy. And that cabbage is, or is likely to be, healthy is a good reason to eat cabbage (cf. Kiesewetter 2013: 264–5).

Deontic testimony may not always do this. To adapt an example from Foot (2002: 107), suppose a reliable source tells you that you ought not to look at hedgehogs in the light of the moon. In this case, you may have no idea why you ought not do this. So, if testimony is not itself a reason, you may learn of no reason not to look at hedgehogs in the light of the moon. However, it is far from clear that this testimony justifies you in believing that you ought not to look at hedgehogs in the light of the moon. More generally, it is very unclear whether bare testimony justifies beliefs about what you ought to do.²⁶

So it is plausible that in cases in which deontic testimony justifies you in believing that you ought to φ, you also ought to φ. Testimonial cases are not compelling counterexamples to Ought Infallibilism.

5.6 Another route to Ought Infallibilism?

The defence of the uniformity view, together with the discussion in the preceding sections, delivers an answer to the question, broached above, of why justified normative beliefs must be true. In doing so, it also provides the further argument for Ought Infallibilism, promised above.

With the possible exception of cases involving WKRs and deontic testimony, the considerations which determine what you ought to believe, hence, what you are justified in believing, are the very same as the considerations which determine what you ought to do. If the facts determine that you are justified in believing that you ought to φ, those facts also determine that you ought to φ. If your perspective determines that you are justified in believing that you ought to φ, your perspective also determines that you ought to φ.

For the perspectivist, this is not always the case. The considerations which justify you in believing that it is raining need not determine that it is raining. In contrast, the objectivist is committed to the view that, WKRs aside, justified beliefs are always true, whether or not they concern normative matters.

As for WKRs, we doubt that there are any but, if we are wrong about that, any case in which you justifiably believe that you ought to φ will not involve WKRs. So, WKRs do not undermine Ought Infallibilism.

As for cases involving deontic testimony, either such testimony makes it right to φ and to believe that you ought to φ, or in any case in which testimony makes it right to believe that you ought to φ there will be considerations which make it right to φ. Either way, Ought Infallibilism stands.

So, at this point we have not only addressed an objection to Ought Infallibilism but provided another route to it via the rejection of the hybrid view.

6 Ought implies can

If certain normative claims imply non-normative claims, this might seem to pose a problem for Ought Infallibilism. Consider, for example, the principle that ought implies can (OIC).²⁷ It might seem obvious that you can be justified in believing that you ought to φ even though you cannot. Suppose that you know that there is a Coen brothers film showing this evening. On the basis of this, you justifiably believe that you ought to go the cinema. However, though you are in no position to know this, your house has been sealed and you cannot leave. Since you cannot go to the cinema, and given OIC, your justified belief that you ought to do so is false. So, given OIC, Ought Infallibilism is false.

It is important to stress that this is only a problem for the perspectivist. The objectivist will claim that, given OIC, you cannot be justified in believing that you ought to φ when you cannot φ.

For the purposes of this paper, we do not take a stand in the debate between perspectivists and objectivists. On behalf of perspectivism, we will instead make a case for thinking that the perspectivist should reject OIC.²⁸ If it fails to persuade, a fallback option is to qualify Ought Infallibilism so that it applies only to justified beliefs about actions that you can perform (cf. Gibbons 2013: 188; Kiesewetter 2013: §7.8).

The first point to consider is that OIC does not sit happily with some motivations for perspectivism. Consider:

Doctor is deciding whether to give drug A or drug B to Patient. She knows that A will relieve 100 % of Patient’s suffering, while B will relieve it by 80 %. Due to regulations imposed by the pharmaceutical industry, if Doctor tries to give one of the drugs, the other will be unavailable. And unbeknownst to Doctor, she cannot give A. The information Doctor possesses suggests she ought to give A. But, given OIC, that cannot be what she ought to do. So, perspectivism and OIC together entail that she ought to give B.

If she tries to give A, Doctor would surely not be criticisable or blameworthy for failing to do what she ought to do. Since one of the arguments for rejecting objectivism proceeds via reflection on what you are criticisable or blameworthy for doing, parity of reasoning suggests that the perspectivist should deny OIC.

The next point serves to explain why perspectivism supports rejecting or revising OIC.

It is widely recognised that there are two kinds of considerations which determine whether you ought to φ—reasons for/against φing, which can be weighed against one another, and enabling/disabling conditions (see e.g. Dancy 2004; Horty 2012; Pollock 1986; Schroeder 2007). That you cannot φ is presumably not a reason not to φ; instead, it seems to be a disabling condition. For example, that you are unable to go to the cinema is not a reason against going; rather, it disables what would otherwise be a reason to go. If you hold that reasons—one kind of consideration which determines what you ought to do—are epistemically constrained, it would be natural also to hold that disabling conditions—another kind of consideration which determines what you ought to do—are epistemically constrained too.

An attractive proposal which bolsters this point is that disabling conditions are themselves reasons, namely, reasons not to place weight on a consideration in reasoning (cf. Horty 2012; Schroeder 2007). That you cannot go to the cinema is not a reason against going, but it might be a reason to place no weight on the fact that a Coen brothers film is showing when deliberating about what to do this evening.

If disabling conditions are themselves reasons, the perspectivist must hold that they are epistemically constrained. So, if considerations about what you cannot do are disabling conditions, the perspectivist must admit that they make a difference to what you ought to do only when they satisfy the relevant epistemic constraint. So, the perspectivist must reject OIC.

Needless to say, these are large issues and it is beyond the scope of this paper to resolve them. But we hope to have made something of a case for thinking that, while the objectivist should deny that OIC undermines Ought Infallibilism, the perspectivist should deny OIC. Either way, there is no objection to Ought Infallibilism from OIC.

7 No justified consequentialists?

You might think that it is simply implausible to deny that one can have justified false beliefs about some subject matter. When the subject matter is normative, you might think it especially implausible. Suppose that consequentialism is false. Is no-one justified in believing consequentialism? Surely not.

Objectivists and perspectivists will say different things in response to this concern. We will outline their responses in turn.

7.1 No objectively justified consequentialists?

Objectivists will argue that, while consequentialists do not justifiably believe that they ought to maximize value, they may be rational or, alternatively, blameless in so believing (cf. Littlejohn 2012; Steglich-Petersen 2013). They may argue that this explains why we may have thought consequentialists were justified.

It is worth noting that for reasons that have nothing to do with Ought Infallibilism objectivists will need to draw some kind of distinction between what you ought to do—hence, what you are justified in doing—and what you are rational or blameless in doing. Given objectivism, there will be cases in which you ought to φ even though you could not reasonably be expected to realise this, or even though the considerations which determine that you ought to φ are beyond your ken. In such cases, a failure to φ will be rational or blameless (cf. Graham 2010).

In making this response on behalf of the objectivist, one might think that we are re-introducing the hybrid view. The response grants that what it is rational to believe depends on your perspective. Thus, you can rationally believe that you ought (given all the facts) to φ, even when that is false. But it might be thought that what it is rational for you to believe is just what, according to the perspectivist, you are justified in believing. So, our suggestion is in effect that you can be justified (given your perspective) in believing that you ought (given all the facts) to φ, even when that is false. That might sound like the hybrid view.

Our proposal is not that the objectivist retreat to the hybrid view. First, it is a controversial issue whether justified belief, for the perspectivist, is rational belief. For example, some perspectivists hold that what you are justified in believing depends on your factive mental states, but that what it is rational for you to believe depends on your non-factive mental states.²⁹

Second, the mark of the hybrid view is that it treats normative statuses such as justification differently as they apply to belief and action—for instance, it takes justified belief, but not justified action, to depend on one’s perspective. The response we are suggesting does not accept this. Instead, it distinguishes a normative status which depends on one’s perspective—rationality—from a normative status which does not—justification.³⁰

Third, even if the objectivist concedes that ‘justification’ can be used to express the notion of rationality, and thus to express a notion which depends on one’s perspective, she does not thereby concede the hybrid view unless she also holds that it is this notion of justification which connects to the deliberative ‘ought’, that is, that it follows from there being this kind of justification to φ that it is not the case that you ought not φ, in the sense of ‘ought’ which structures deliberation. Our objectivist does not concede this further claim.

7.2 No perspectivally justified consequentialists?

In response to the above concern, perspectivists can go one of two ways.³¹ According to some perspectivists, what you ought to do is determined by your perspective with respect to the non-normative facts, together with the true normative principles. This kind of perspectivist is in the same boat as the objectivist—she will have to say that consequentialists, for example, may be rational or blameless without being justified.

According to other perspectivists, what you ought to do is determined by your epistemic perspective with respect to both non-normative facts and normative principles. This kind of perspectivist can allow that consequentialists can indeed be justified (in which case, their beliefs about what they ought to do are true).

But even this perspectivist, like the objectivist, has independent reasons to draw some distinction between what one ought to do and what one is rational or blameless in doing. Unless you are always in a position to tell what your perspective involves—unless luminosity holds of the items in your perspective—there will be cases in which you will be blameless or rational despite failing to do what you ought to do (cf. Hawthorne and Srinivasan 2013). Following Williamson (2000), many would deny that luminosity holds.

8 Is Ought Infallibilism trivial?

In the previous section, and throughout, we allowed that there are different senses of ‘ought’ and ‘justification’. In doing so, we allowed that are different readings of Ought Infallibilism, depending on how these terms are to be understood. One might object that, once the intended reading is made clear, specifically, once it is stipulated that the relevant terms are being used in their deliberative senses in each of their occurrences, the thesis is trivially true, hence, uninteresting. It was only ever the availability of alternative readings that made it seem like something in need of argument.

There are a number of things to say in response to this. First, the proponent of the hybrid view understands ‘ought’ and ‘justified’ as they occur in Ought Infallibilism to express deliberative notions and nonetheless maintains that it is false. This is due to her substantive commitments—she takes what you ought in the deliberative sense to do to depend on the facts, and what you are justified in the deliberative sense in believing to depend on your epistemic situation. Since Ought Infallibilism is true only if the hybrid view is false, and since the hybrid view is a significant and widely-held position, Ought Infallibilism is a nontrivial thesis.

Second, even with the hybrid view out of the picture, that is, even with the uniformity view in place, Ought Infallibilism does not trivially follow. If objectivism is true, and what you ought to do in the deliberative sense is determined by the facts, Ought Infallibilism is still threatened by pragmatic reasons for belief. If perspectivism is true, and what you ought to do in the deliberative sense is determined by your epistemic situation, Ought Infallibilism is still threatened by pragmatic reasons for belief, deontic testimony, and the principle that ought implies can. Responding to those threats is a nontrivial matter.³² Moreover, whether perspectivism or objectivism is true, Ought Infallibilism will strike many as counterintuitive. Accommodating the relevant intuitions by, say, distinguishing being justified (in the deliberative sense) from being rational or blameless is, again, a nontrivial matter.

Third, recall Greco’s observation, quoted above, that the claim that there are false justified beliefs ‘commands almost universal agreement among contemporary epistemologists’. He continues:

Most would go further, and claim that the possibility of justified false belief isn’t restricted to special subject matters. Rather, in no domain is justification an infallible guide to truth. At the very least, if there are domains in which justification entails truth, this requires some special explanation. (2014: 203)

Greco does not distinguish different notions of justification. Thus, we take his claims here to be intended to cover all significant notions of justification, or perhaps the notion of justification which is of central interest to epistemology, which we take to be the one in play in deliberative contexts. So, if Greco is right about orthodoxy, then Ought Infallibilism, read in the way we intend, runs counter to it.

Finally, Ought Infallibilism has nontrivial consequences, some of which we outline in closing.

9 Closing remarks

We argued for Ought Infallibilism by considering the role of normative beliefs in reasoning. Having done so, we addressed objections to Ought Infallibilism and thereby provided another route to it.

Ought Infallibilism is an independently interesting thesis. It also raises further questions, which we will have to leave for another occasion. For instance, does Ought Infallibilism generalize to justified beliefs about what one ought to believe,³³ or to justified beliefs about what others ought to do, or to justified beliefs involving other normative notions such as reasons or rational requirements?³⁴ Ought Infallibilism might also have some important implications. While exploring and assessing those implications in full is another task for future work, we conclude by sketching two.

First, Ought Infallibilism has implications for the debate about what is wrong with akrasia—that is, with believing that you ought to φ without intending to φ. Some philosophers (e.g. Broome 2013) suggest that to explain what is wrong with akrasia we need to posit a ‘wide-scope’ requirement against akrasia—a requirement to (intend to φ, if you believe that you ought to φ). Others (e.g. Kolodny 2005) have argued that we should seek to explain what is wrong with akrasia without positing such requirements. Ought Infallibilism offers a way to do this. Given Ought Infallibilism, the akratic agent always has either a belief that she is not justified in having or fails to intend to do something that she should do. The akratic agent is thus guaranteed to be going wrong in either her belief or lack of intention.³⁵

Second, Ought Infallibilism might provide a new route to the idea that there is ‘pragmatic encroachment’ on justified belief, and thereby on knowledge.³⁶ Consider:³⁷

Low :

On Friday, you drive past the bank. You have a cheque to deposit though there is no urgent need to do so. Since there are long queues, you think to yourself, ‘if the bank is open tomorrow, I should wait until tomorrow’. You recall that on previous Saturdays the bank was open and conclude that you should wait until tomorrow.

High :

Like Low except that there is an urgent need to deposit the cheque. You have a mortgage payment to make by Sunday, when the bank is closed, and your house will be repossessed if you fail to make it.

Intuitively, in High, you should not wait until Saturday to deposit the cheque—after all, the bank might have changed its hours. It follows from Ought Infallibilism that you are not justified in believing that you should wait until Saturday. And given closure of justification under modus ponens, it follows that you are not justified in believing that the bank is open on Saturday. In Low, in contrast, you might be justified in believing that the bank is open on Saturday, and that you ought to wait until then. Since the only difference between the two cases is what is at stake, it seems that pragmatic considerations bear on whether your beliefs are justified. If there is pragmatic encroachment on justified belief, it follows that there is pragmatic encroachment on knowledge, on the assumption that you know a proposition only if you are justified in believing it.³⁸

Again, these remarks are not intended to establish pragmatic encroachment on justification or knowledge but only to point toward what appears to be a consequence of Ought Infallibilism. Whether that really is a consequence, it remains the case that, if you justifiably believe you ought to do something, you ought to do it.