1.1 ^{Raz (1983, pp. 156-157)} asks us to consider the following case:

Imagine an orthodox but relatively ill-informed Jew who asks the advice of his friend who is Catholic but an expert in Rabbinical law. “What should I do?” he asks, clearly meaning what should I do according to my religion, not yours. The friend tells him that he should do so and so. The point is that both know that this is not what the friend thinks that he really ought to do. The friend is simply stating how things are from the Jewish Orthodox point of view.

Let us suppose the sentence uttered by the Catholic expert in Rabbinical law here is: “You ought to φ” (where “φ” ranges over actions or conduct). So, the substituend for “φ” here will be some (claimed) requirement of Rabbinical law. Such a statement would be informative (descriptively and normatively), as, ex hypothesi, the addressee is not equipped to work out the requirements of Rabbinical law.

One might object to the following conditional analysis of such a statement (considered, but not subscribed to, by ^{Raz (1999)}):

(1) If you intend to act in accordance with the requirements of Orthodox Judaism, you ought to φ.³

Why so? Well, there are many objections one might offer, but two are of central importance here. First, one might object on the basis that this is a narrow-scope normative conditional: the “ought” operator is contained within the conditional’s consequent. I agree with this objection. Such conditionals permit what is called detachment -which is to say, if the antecedent is met (i.e. the addressee intends to act in accordance with the requirements of Orthodox Judaism), it would follow that the addressee ought to φ. But this would be to legitimate an illegitimate form of bootstrapping: no conclusion about what one ought to do can follow from how one intends to act (on this view). (There are clear parallels here with the so-called “is-ought” question: how can a set of purely descriptive facts entail conclusions about what one ought to do?) So, we have good reason to reject the narrow-scope reading.

1.2 The second objection is more general, and is to any form of conditional analysis of such statements. This objection could come in two forms, of differing strengths. More strongly, one might object that it is always impermissible to give a conditional analysis of the logical form of a statement whose surface-syntax is not conditional. I do not consider this strong objection in detail: I simply reject it (and invite the objector to give me reasons supporting this strong claim). Less strongly, one might object that, if one is to give a conditional analysis of the logical form of a statement whose surface-syntax is not conditional, one must overcome a burden of proof against doing so (i.e. default rule: don’t depart from surface-syntax). And, the objection would continue, no such analysis has yet overcome this burden of proof. Now, I agree with this weaker objection. Moreover, no such analysis has yet overcome this burden of proof because discussion so far has (mistakenly) focused on narrow-scope conditionals, à la (1).⁴

1.3 Let me, then, propose my wide-scope reading of detached statements: wherein the “ought” operator governs the conditional as a whole. Still focusing on the Orthodox Jew case, I propose:

(2) You ought: If you intend to act in accordance with the requirements of Orthodox Judaism, to φ.⁵

A few words about (2). First, pretty obviously, there are two ways for the addressee to meet the requirement enshrined in (2): either failing to intend to act in accordance with the requirements of Orthodox Judaism, or φ-ing.⁶ (2) is neutral on which way of conforming the addressee takes. Though (2) is neutral on this, it is commonplace in considering this example to consider the Orthodox Jew (assuming, pro tem, the correctness of (2) as an analysis) φ-ing. But, of course, the addressee here might -say φ-ing is wholly objectionable to him- reconsider his faith, and no longer intend to act in accordance with the requirements of Orthodox Judaism. Second, and more importantly, (2) does not suffer from the chief defect which plagued (1): such conditionals do not permit detachment, and so no illegitimate bootstrapping is legitimated. This is to say, if one intends to act in accordance with the requirements of Orthodox Judaism, it does not follow from (2) that one ought to φ. The burden of proof on going conditional has been overcome (or, at least, one significant obstacle to going conditional has been removed).

By contrast with (2), it is worth considering:

(SK) You ought: If you intend to be a serial killer, to murder several people.

We’ve seen no bootstrapping is licensed by this formulation: it doesn’t follow from (SK) that if one intends to be a serial killer, one ought to murder several people. However, isn’t it odd to have a normative requirement like this neutral on which way it is to be complied with (whether by failing to intend to be a serial killer, or by murdering several people)? No, provided we recognise it is not the only relevant norm. There is a norm against murdering (several) people ((~M)). And (~M) dominates (SK). So, all things considered, one ought to fail to intend to be a serial killer. Finally, what about the psychopath who cannot fail to intend to be a serial killer? He ought, all things considered, to violate (SK), by intending to be a serial killer, while failing to kill several people: (~M) dominates (SK) both in determining how to conform with it, and in determining whether to conform with it at all. Returning to the Orthodox Jew case, how to conform with (2), and whether to conform with it at all, will be determined by the relevant moral norms bearing on φ-ing. (The same goes, mutatis mutandis, for the coming Vegetarian case.) I return to this issue in the coming Legal case, in 3.1.

1.4 It is appropriate, at this point, to consider the truth-conditions of the propositions semantically expressed by utterances of “You ought to φ”, depending on whether such utterances are detached or committed. It is commonplace in philosophy of language that sentences (semantically) express a proposition in a context.⁷ Thus, one and the same sentence -one and the same lexical string- can express different propositions in different contexts. Still focusing on the Orthodox Jew case, I’ve proposed the following as the proposition expressed by the foregoing utterance in a detached context:

(2) You ought: If you intend to act in accordance with the requirements of Orthodox Judaism, to φ.

And, pretty uncontroversially, the following is the proposition expressed by the foregoing utterance in a committed context:

(3) You ought to φ.

Thus, one sentence-type expressing different propositions in different contexts.⁸ No problem.⁹ Both (2) and (3) can aptly be called normative propositions, on account of expressing propositions with (apparent) normative content. Their truth-conditions, however, will not be identical. (3) will be true just in case you really ought¹⁰ to φ. Different metaethical views will give different truth-conditions for that. We need not enter that dispute here, save to say that at least some metaethical views will give (3) exclusively normative truth conditions, by appealing to, for example, normative -here, plausibly, moral- facts (which will determine whether the requirement of Rabbinical law is one with which one ought to conform).

1.5 (2), by contrast, will have, at root, normative, but here, plausibly,¹¹non-moral, truth-conditions. (2) is a normative requirement governing coherence (or consistency) between how one intends to act and how one acts. Why possess such coherence (or consistency)? That is a difficult question. It is best to approach it indirectly. I seek to ground the normativity of detached statements in cognate norms of pure rationality -rational requirements. The cognate such norm of (2) would be:

(2***) You ought: If you intend to act in accordance with the requirements of Orthodox Judaism, to intend to φ.

Why pure rationality? I follow John ^{Broome (2013)} in taking rationality to supervene on the mind. This means rationality (or otherwise) is, fundamentally, determined by the coherence (or consistency) of one’s mental states. Let me explain with an example. Suppose the Orthodox Jew, upon receiving a statement analysable as (2***) from the Catholic, conforms to it by intending to φ. But suppose a militant Catholic is on the scene who physically prevents our Orthodox Jew from φ-ing. On this view, our Orthodox Jew would be perfectly rational -indeed, just as rational as a counterpart who intended to φ and φ-ed. (Mental states determine rationality. Or: no change in rationality without a change in mental state.) And this would be a putative reductio of conceiving of (2) as a (pure) rational requirement: our physically prevented Orthodox Jew would fail to conform with (2) yet would be perfectly rational (in this regard).

Now it is not plausible to analyse the detached statement in question as (2***). The statement (in part) speaks directly to action: if one intends to act in accordance with the requirements of Orthodox Judaism, one conforms with the detached statement’s directive iff one φ-s. However, there is a bridge principle -a putative conceptual truth- which means that any normativity inherent in (2***) is inherited by (2). That principle is: If you have an intention and nothing prevents you, you perform the intended act. (Note this is not a norm of rationality, as it does not refer to coherence (or consistency) of mental states. Indeed, it is not a norm at all as it is putatively inviolable: if nothing prevents you from performing the act in question, and you fail to perform it, this principle, as a putative conceptual truth, mandates that you didn’t have the intention in the first place.) Finally, here: What, if any, is the normativity inherent in (2***)? Initially, note that the general rational requirement bearing on (2***) is, plausibly, something like the following (as (2***) does not involve strictly contradictory intentions -intending at time t that p and intending at t that ~p):

Instrumental Requirement. Rationality requires of [agent A] that, if

[(i)] [A] intends at t that [end] e, and

[(ii)] [A] believes at t that, if [means] m were not so, because of that e would not be so, and

[(iii)] [A] believes at t that, if she herself were not then to intend m, because of that m would not be so, then

[(iv)] [A] intends at t that m. (^{Broome 2013})

Where, transposing things to (2***), “e” = A acts in accordance with the requirements of Orthodox Judaism, and “m” = A φ-s. And note that I can operate with a broad conception of means/end reasoning so as to include, perhaps as here, means which are (part-)constitutive of the end. (For more on the bases for these conditions -particularly (ii) and, the more controversial, (iii)- see ^{Broome 2013}.)

More generally, though: Why be rational? This is an enormous, and controversial, issue in practical reason. We can note a few things, however. First, and uncontroversially, the norms of rationality are straightforwardly that -normative- in my capacious sense (cf. “Introduction” supra) in that they set up a notion of correctness: following the norms is correct according to the norms. But clearly we want to see if we can say more. In particular: Is there a reason to be rational? Niko ^{Kolodny (2005}inter alia) and ^{John Broome (2007} and ²⁰⁰⁸inter alia) have perhaps written most probingly about this question (though see also Kolodny 2005, pp. 510-511 for a helpful survey of other philosophers’ standpoints). ^{Kolodny (2005, p. 509)} concludes that “rationality is only apparently normative, and the normativity that it appears to have is that of reasons [. . . ]. The normative ‘pressure’ that we feel, when normative requirements apply to us, derives from [. . . ] the reasons that, as it seems to us, we have.” Broome, meanwhile (2008, p. 177), concludes that “[i]t [. . . ] seems intuitively plausible to me that rationality is normative.

Moreover, it seems plausible to me that this is so for a broadly instrumental reason. But I am sorry to say that I simply cannot find an argument.” What to conclude in light of all this? I cannot possibly adjudicate on this issue, other than to make three quick observations. First, note that Broome’s inability to find an argument seems to rest on consideration of “quirky worlds [in which] because of the way causal processes work [. . . ] satisfying the requirements of rationality tends to be unsuccessful” (2008, pp. 173-174). Second, and somewhat tritely, an inability to find an argument doesn’t mean either that no such argument exists or that the conclusion to be reached isn’t true. Third, and to repeat -focusing on the Orthodox Jew case- any normativity inherent in (2***) -and plausibly there is some, even if we cannot currently conclusively establish this by argument- is inherited by (2).

1.6 But (2)’s truth-value will also depend on (partly) descriptive facts -here, what Rabbinical law (as a matter of fact) requires of Orthodox Jews. So, the source of this normative requirement is Rabbinical law. Importantly, ascertaining what Rabbinical law requires may well involve difficult matters of interpretation. Such interpretation may -though need not- involve normative -here, moral- judgment. However the interpretation in question goes, though, there will still be descriptive facts playing a key role. I just do not see anything particularly controversial in all of this. And, this last point -cashing out the content of these norms- leads nicely to the Vegetarian Case.

2.1 ^{Raz (1999, pp. 175-176)} asks us to consider the following case:

If I go with a vegetarian friend to a dinner party I may say to him, “You should not eat this dish. It contains meat.” Not being a vegetarian I do not believe that the fact that the dish contains meat is a reason against eating it. I do not, therefore, believe that my friend has a reason to refrain from eating it, nor am I stating that he has. I am merely informing him what ought to be done from the point of view of a vegetarian. Of course the very same sentence can be used by a fellow vegetarian to state what ought to be done. But this is not what I am saying, as my friend who understands the situation will know.

The natural extension of my foregoing treatment of the Orthodox Jew case would be to propose the following analysis of this statement:

(4) You ought: If you intend to act in accordance with the requirements of vegetarianism, not to eat this dish [it contains meat].

A few words about (4). First, most generally, all the foregoing remarks in opposition to (1), and in defense (and cashing out) of (2), can be marshalled, mutatis mutandis, in defense (and cashing out) of (4).¹² Second, the source of this normative requirement is the tenet(s) of vegetarianism. Third, I put the information “it contains meat” in square brackets, as I take it to be, while important, not central to the normative requirement expressed by the utterance in question -it is, as it were, auxiliary information which licenses the consequent.

2.2 We’ve mentioned in passing that it’s a success constraint on detached statements that they be informative -and, more precisely, I take it, normatively informative.¹³ It might be objected that, unlike the Orthodox Jew case, this success constraint is not met in the Vegetarian case. I think this objection would go as follows. The statement in question is addressed to a vegetarian. We might suppose that it is a conceptual prerequisite on being a vegetarian that you believe that you ought not to (indeed: are obligated not to) eat meat. So, the only way this statement could be informative is by alerting the addressee to the purely descriptive fact that a meat dish is before him. But, so the objection goes, this would only ensure descriptive, and not normative, informativeness. Of course, the committed vegetarian would (likely: barring apostasy) use the information to conclude that he ought not to eat the dish in question. But that is a different matter from, and not expressed or conveyed by, the (content of the) detached statement.¹⁴

Before answering this objection, to ensure that the statement in question has a standing chance of being informative, I think it is best to assume that the dish in question is a very unusual meat dish -perhaps one whose meat components are cunningly disguised. Now- to answer the objection -while this objection has force with respect to narrow-scope conditional readings of such detached statements (see note 12 supra), it palpably does not have force with respect to my proposed wide-scope reading. That is, it surely is the case that, if the only new descriptive information one is giving a vegetarian addressee is the fact that a meat dish is before him, this alone cannot result in it following (barring apostasy) that he (morally) ought not to eat the dish in question (as would be the case on the narrow-scope reading). However, as we’ve seen, there’s nothing particularly puzzling in new descriptive information like this triggering a wide-scope conditional like (4), and thus being normatively -though not morally- informative.

In the language of our Instrumental Requirement of §1.5 supra, the triggering descriptive information at stake is, plausibly, information bearing on the truth of the propositional content of the beliefs contained in instances of conditions (ii) (and (iii)). In other words, in this case -focusing on condition (ii)- if you were to eat this dish, because of that you would not be acting in accordance with the requirements of vegetarianism. Why so? Because -the triggering descriptive information (cf. ^{Enoch 2011})- this dish contains meat. (Regarding the more controversial condition (iii), things are more complicated in this case, as the relevant means involves an omission -though the Orthodox Jew and, with a little modification, Legal cases could also involve omissions. That condition, in this case, reads: if you yourself were not then to intend not to eat this dish, because of that you would eat this dish. Any oddity here can be dispelled by considering the dinner party context and assuming that the absence of an intention not to eat any dish will result in you eating that dish: put vernacularly, assume that your default mode is to have a nibble at each dish. Like points apply, mutatis mutandis, in other contexts in which such a detached statement could be meaningfully uttered. Finally, note that one might explore weakening condition (iii) by replacing “would” with “might”.)