I. Introduction: Two Dimensions of Rational Evaluation

It is increasingly common to distinguish two distinct strains in our ordinary thought and talk about rationality. In one sense, rationality is a matter of correctly responding to the reasons one has.¹ For our attitudes -i.e., our beliefs, intentions, preferences, and the like- to be rational in this sense is for them to be justified or reasonable.² Call this substantive rationality. In another sense, however, rationality is a matter of coherence, or having the right structural relations hold between one’s attitudes, independently of whether those attitudes are reasonable or justified. The relevant notion of coherence is a broad one, and a broadly normative one, encompassing a range of different combinations of attitudes that intuitively clash, or fail to properly “fit” together, where the lack of fit needn’t involve any logical inconsistency in contents. Call this structural rationality.³

Suppose, for example, that you meet someone who claims to be Superman. Suppose further that this person is perfectly sincere -he does, in fact, believe that he’s Superman.⁴ It should be obvious that something has gone wrong. Among other things, he has an unjustified belief -one that flies in the face of all the evidence. But suppose you also find out that despite believing that Superman can fly (“It’s one of his greatest powers”, he says), he lacks confidence in his own ability to fly (“I gave up trying after my third broken leg”). Once again it should be obvious that something has gone wrong. Not only does he have an unjustified belief, he’s also incoherent in failing to believe the obvious consequences of other things he believes.

However, the second failing is interestingly different than the first, as evidenced by the seemingly paradoxical way we’re prone to describe what’s going wrong with the subject -call him ‘Tom’.⁵ On the one hand, it seems right to say that you should believe the obvious consequences of other things you believe, and so there’s a clear sense in which, given his other beliefs, Tom should believe that he can fly. On the other hand, one shouldn’t believe something in the face of overwhelming evidence to the contrary, and so there’s also a sense in which Tom should not believe that he can fly. We’re thus faced with the puzzle of wanting to say both of these things -namely, that Tom should believe he can fly, and that he should not.

Similar examples involving strict means-end incoherence arise in the practical realm. Setiya (2007) offers the following story -inspired by ^{Rawls (1971})- to illustrate the “problem of instrumental reason”:

Imagine that I embark upon on a thoroughly irrational project: I intend to count the blades of grass in my garden... Despite my intention, however, I do not take what I know to be the necessary means. Even though I see that I have no chance to complete the enumeration unless I keep track of how many blades of grass I counted [and] where I counted them, I can’t be bothered with bookkeeping. So, every morning, I am forced to start again [and never] complete the count. (650)

As before, there are at least two things going wrong with such a subject -call her ‘Jane’. On the one hand, given her goal it seems clear that Jane should be keeping track of the blades she’s counting. But on the other hand, Jane shouldn’t be counting grass in the first place, and so shouldn’t be keeping track of them. Again we’re faced with the puzzle of wanting to say both of these things -namely, that Jane should take the necessary means to her end, and that she shouldn’t.

Of course, seemingly conflicting ‘should’-judgments aren’t always puzzling. The demands of morality, for instance, regularly conflict with the demands of self-interest, and there’s nothing especially mysterious about the clash. What’s interesting about cases like Tom and Jane is that the ‘should’-judgments are both naturally understood as claims about what the rational response is in a given situation. To not believe the obvious consequences of other things you believe seems to constitute a rational failing, as does believing something in the face of overwhelming evidence to the contrary. The same goes for not intending the means necessary to one’s end, and intending to do something you have no good reason to do.

It should be clear that we’ve hit upon a pattern, and that the foregoing observations generalize beyond the cases of deductive and means-end incoherence, both of which are extreme examples of an otherwise pervasive phenomenon -one arising whenever there is a conflict or lack of fit between one’s mental states or attitudes, where at least one of them is unreasonable or unjustified (I’ll be using ‘attitudes’ in an artificially broad way to cover both the presence and absence of attitudes). This includes not just beliefs and intentions, but also hopes, fears, concerns, suppositions, worries, preferences, regrets, and the like. I might realize it’s more important to get a good night’s sleep than to stay up late and read the news, and yet prefer to continue reading. I might be deeply concerned about the consequences of smoking, and yet still intend to smoke. I might know that spiders are mostly harmless, and yet still be afraid of them. It’s possible for apparently conflicting ‘should’judgments to arise in cases like these, too. But since it’s clearest (and least controversial) in the case of beliefs and intentions, I’ll focus on them in what follows.

Although there are various possible responses to the puzzle above, I think we should take appearances at face value: there appear to be two dimensions of rational evaluation because there are two dimensions of rational evaluation. As noted at the outset, while in one sense being rational is (roughly) a matter of one’s attitudes being justified or reasonable, in another sense being rational is a matter of one’s attitudes being coherent. The former is substantive rationality, the latter is structural rationality.

It’s worth emphasizing up front that the substantive/structural distinction is not intended to be an “objective”/“subjective” or “external”/“internal” distinction. For one thing, if reasons are understood objectively -i.e., as relative to all the facts and not epistemically constrained in any way- then it’s doubtful there’s any sense of ‘rationality’ that requires one to respond correctly to them.

There’s no sense of ‘irrational’, for example, in which it’s irrational for Bernard Williams’ (1981) gin-and-tonic-lover to drink from the glass that appears to contain gin-and-tonic but in fact contains petrol, even though the latter fact is a decisive reason (in the objective sense) to not take a drink.⁶ It’s only when we focus on reasons that are in some way constrained or determined by one’s perspective that the notion of rationality as reasons-responsiveness becomes plausible. What’s more, it’s perfectly possible to take facts about both structural and substantive rationality to supervene on facts about our non-factive mental states, and hence be an internalist about both, while nonetheless insisting that they differ. Experiences and facts about certain mental processes, for example, might make a difference to substantive rationality (as, say, sources of justification) without making a difference to structural rationality, and facts about “mere” attitudes on their own might make a difference to structural rationality without making a difference to substantive rationality. Indeed, this is how I myself conceive of the difference. In general, though, anyone is who not a pure coherentist about reasons and/or justification should be able to recognize the difference between a justified attitude and a merely coherent one, and that ‘rational’ is naturally used to characterize both.⁷

The distinction between substantive rationality and structural rationality is at least latent in the writings of various philosophers, though there’s no consensus on how exactly it is to be drawn. Some bestow the honorific title of ‘rationality’ on just structural rationality, opting for another label to denote substantive rationality, while others prefer the reverse. I prefer to distinguish between two dimensions of broadly rational evaluation. In doing so I don’t intend to be taking a stand on how exactly they’re related, other than being distinct. The main motivation is instead methodological. For given that our pre-theoretic use of ‘rational’ and its cognates fails to reliably discriminate between facts about coherence and facts about reasonableness, and given that there’s a need to distinguish the two -a conspiracy theorist, for instance, might have a set of beliefs that is quite coherent but far from justified- it’s useful to adopt terminology that demands unqualified judgments concerning rationality be disambiguated. Doing so will put us in a better position to accurately handle our (and others’) otherwise slippery judgments concerning what the “rational” response is, or would be, in a given situation, as well as what the ingredients are that determine the answer.

Not everyone agrees, of course. Indeed, many are simply insensitive to the apparent distinction between reasons-responsiveness and coherence, and proceed on the assumption that our use of ‘rationality’ and its cognates is univocal. But even among those who are sensitive to it, not everyone takes it to be theoretically significant. Some argue that, contrary to appearances, only one dimension of rational evaluation is genuine, or genuinely significant. Whereas some deny (or at least doubt) the rational significance of coherence as such, and hence deny structural rationality, others deny the rational significance of reasons as such, and hence deny substantive rationality. Still others offer theories that can be seen as attempting to provide a single, unified account of our judgments of (ir)rationality.⁸

Although I myself take the distinction between substantive and structural rationality to be genuine, and genuinely significant, Broome does not. ⁹ Indeed, he equates structural rationality with rationality full stop.¹⁰ Denying the rational significance of reasons (as opposed to, say, our beliefs about reasons) is nonetheless compatible with granting their normative significance, as Broome does.¹¹ Fortunately, however, I’ll be able to side-step the controversy over substantive rationality since I’ll be focusing exclusively on structural rationality in what follows. The distinction between substantive and structural rationality is nonetheless important insofar as it helps clarify the intended topic and avoid possible misunderstandings, given that the ordinary use of ‘rational’ and its cognates is insufficiently discriminatory.

Again, the focus in what follows will be on structural rationality. The main goal will be to introduce and clarify Broome’s preferred “wide-scope” view and propose a modification of it that avoids recent objections. The modified wide-scope view is one that builds on the insights of Broome’s large body of work -represented most fully in his Rationality Through Reasoning (2013)- while enjoying additional benefits besides, and it is inspired by the analogy with the law that Broome (and others) stress.

The plan is as follows. I begin by introducing the debate over the existence and nature of structural requirements (Section 2). I then turn to the debate over the so-called “scope” of structural requirements and clarify it by distinguishing three separate debates that can be -and have been- confused (Section 3). Next I explain the distinction between the jurisdiction of a given requirement and its conditions of application (Section 4) and use it to construct modified versions of both the “narrow-scope” and “wide-scope” views (Section 5 and Section 6, respectively). I conclude, however, on an uncertain note: though the modified views may represent progress, it becomes unclear where the debate is supposed to continue from here (Section 7).

II. The Existence and Nature of Structural Requirements

The debate over the nature of structural rationality starts with the observation that which attitudes we actually have -whether or not they’re justified- make an intuitive difference concerning what other attitudes it’s structurally rational, or coherent, for us to have. (For ease of expression, I’ll often drop the ‘structural(ly)’ qualifier in what follows, though for the sake of clarity it will occasionally reappear.) More than merely making a difference, however, it seems that you can be rationally committed to having certain attitudes given that you have certain other attitudes, in such a way that you will definitely do something wrong or exhibit a rational failing if you have the latter without having the former. Similarly, it seems that you can be rationally prohibited from having certain attitudes given that you have other attitudes, in such a way that you will definitely do something wrong or exhibit a rational failing if you have the former attitudes while also having the latter.

For this reason, many philosophers find it natural to think of the domain of structural rationality as corresponding -at least in part- to a distinctive set of rules or requirements that mandate or prohibit certain combinations of attitudes.¹² The basic idea is that for each kind of incoherent combination of attitudes there is a corresponding rule or principle prohibiting it, and that what’s wrong with incoherent agents is that they violate these principles, just as for each kind of illegal action there is a law that prohibits it, and in virtue of which actions of that kind are illegal.¹³ Suppose that Jill steals Jack’s bike. What Jill did was illegal. But why exactly? Subtleties aside, the answer is clear: Jill took Jack’s bike without his permission, and there’s a law that prohibits taking others’ property without their permission. If there hadn’t been a law prohibiting theft, then although what Jill did may have been immoral, it wouldn’t have been illegal.

The requirements of structural rationality are supposed to play an analogous role: just as laws explain why particular actions are legal or illegal, so structural requirements are supposed to explain why particular (combinations of) attitudes are rational or irrational. Tom, above, is irrational. But why exactly? According to the present line of thought, it’s because he fails to believe the obvious consequences of other things he believes, and there’s a requirement that prohibits him (and us) from doing so. Standard examples of structural requirements include consistency requirements (in imperatival form: don’t believe contradictions! don’t intend incompatible things!), instrumental or means-end requirements (intend the means you take to be necessary to your ends!), closure requirements (believe the obvious consequences of other things you believe!), and enkratic requirements (intend to do what you believe you ought to do!). As these examples illustrate, candidate structural requirements involve cognitive attitudes like belief and practical attitudes like intentions, as well as combinations of cognitive and practical attitudes.¹⁴ Call this the requirements-based account of structural rationality.

I say ‘requirements-based’ since the focus in what follows will be on structural requirements. But a more apt term would be ‘rule-’ or ‘principle-based’, since requirements are only one type of rule or principle. There may also be principles of permission, for instance.¹⁵ Although requirements and permissions differ in normative strength, they are alike in being essentially “threshold-y” or “all-or-nothing” affairs, and hence importantly unlike “gradable” or “quantitative” normative notions such as value, justification, and reason, all of which come in degrees. That is, whereas it doesn’t make sense to say of some action-type or state of affairs A that it is more required/permitted than B, it does make sense to say that A is more valuable/justified/ well-supported than B, or that you have more reason/justification to A than to B. Of course, some rules may be more important, or ranked higher, than others, and hence take precedent in cases of conflict. But to say that rules admit of hierarchical relationships, such as rankings, is not to say that rules themselves come in degrees. Rank-ability is not gradability.

^{Broome (1999}, ²⁰⁰⁴, ²⁰⁰⁷, ²⁰¹³) is the most prominent and influential proponent of a requirements-based (or, more generally, rules-based) conception of structural rationality, though many others have followed his lead in taking talk of structural requirements seriously. Structural requirements, in the relevant sense, are supposed to be more than mere necessary conditions for being fully rational, and in at least two ways. First, it’s not enough that a subject fails to be fully rational whenever the requirements aren’t met; the requirements, when violated, are supposed to guarantee something more -namely, irrationality. Unlike mere necessary conditions, then, structural requirements state conditions the failure of which to obtain guarantees the having of a negative evaluative property -namely, being irrational- rather than just the lack of a positive one -namely, being rational. And we obviously shouldn’t conflate being irrational with merely not being rational. Rocks and trees fail to be rational, but they’re not thereby irrational. They’re a-rational -they lack the relevant sort of complex capacities needed in order for them or their states to be apt candidates for rational evaluation.¹⁶ Second, the requirements are supposed to state conditions such that agents who fail to meet them are irrational in virtue of doing so -i.e. in virtue of violating the requirement. Violating a requirement doesn’t merely guarantee that you’re irrational: it explains why you’re irrational.¹⁷ The same goes for compliance: just as it is in virtue of violating such requirements that agents are irrational, it is in virtue of complying with them that they are rational.

Different authors express this point in different ways. Broome, for instance, distinguishes between two senses of ‘requires’ and its cognates: a weaker “property” sense and a stronger “source” sense. As he puts it:

The first appears in constructions where its subject denotes a property: ‘Beauty requires hard work’; ‘Staying healthy requires you to eat olives’; ‘Success in battle requires good horses’; ‘Crossing the Rubicon required determination’... [The second appears in] constructions [where] the subject of ‘requires’ denotes a person or thing that has some sort of real or presumed authority: ‘The minister requires the ambassador’s presence’; ‘The law requires you to drive carefully’; ‘The bill requires payment’; ‘Fashion requires knee-length skirts’; ‘My conscience requires me to turn you in’ (2013: 109).

He thinks the most interesting questions concern what rationality requires in the source sense, not the property sense. In a similar vein, Jonathan Way (forthcoming) draws a distinction between stronger and weaker senses in which one might be “rationally required” to do something:

In [a] weak sense, to say that you are rationally required to do A is to say that doing A is a necessary condition of being fully rational. However, [there’s] a stronger sense in which [it might be thought that] rationality requires coherence. What, we might ask, explains why [deductive] incoherence and means-end incoherence are irrational? A natural answer is that there are rules or principles which require you to be closure and means-end coherent... If you have an incoherent combination of attitudes you are irrational because you violate a rational requirement.

^{Schroeder (2013}) agrees, and takes his own talk of the “rules” of rationality to be equivalent to Broome’s notion of a source requirement:

[Y]ou count as (having the property of being) irrational in virtue of breaking one or more of the rules (source requirements) of rationality, and [the debate concerns] which rules (source requirements) you are breaking... when you have inconsistent beliefs, are akratic or means-end incoherent, or fail to draw the obvious consequences from your other beliefs (299).

I’m emphasizing the law-like status of the requirements (or “rules”) of structural rationality for two reasons. The main reason is that the positive proposal in Section 6 crucially depends on taking the analogy with the law seriously. The second reason is to make clear just how substantive it is. Although the assumption that structural rationality bottoms out (at least in part) in strict law-like requirements is widespread, it’s not obligatory. It may be, for example, that the nature of structural rationality is better understood in terms of something more pro tanto and gradable, and hence as more akin to the normative notions standardly associated with substantive rationality (reason, justification) than the law (requirements, permissions). As a matter of fact, that’s where my own sympathies lie.¹⁸ But for the purposes of this paper I’ll be sticking with orthodoxy in assuming a requirements-based view.

Despite the (near-)consensus concerning the explanatory, lawlike status of rational requirements, there is consensus about little else. Besides the question of their explanatory status, other questions include ones concerning their content (what exactly is required?), scope (do the requirements mandate or prohibit particular attitudes, or instead only particular combinations of attitudes?), source (how do such principles arise, and from how do they get their authority?), jurisdiction (who do the requirements apply to, and under what conditions?), extent (do they typically involve a relatively small number of possible attitudes, and hence “local”, or instead larger groups of attitudes?), temporal nature (are they synchronic or diachronic?), and normative status (in what sense, if any, ought we to comply with them?).

Although each of these issues is important, it’s the question of “scope” that has received the most attention.¹⁹ I’ll consider that next, before turning to the question of jurisdiction.

III. The Scope of Structural Requirements

Two importantly different issues have been traditionally conflated in the debate over the “scope” of structural requirements. (We’ll consider a third in due course.) The first issue -as indicated above- concerns whether structural rationality requires one to have particular attitudes (at least under certain conditions), or instead is exclusively concerned with mandating or prohibiting combinations of attitudes. For example, if you intend an end E and believe that in order to achieve E you have to take means M, does it follow that you are rationally required to intend M? Or are you merely required, at all times and irrespective of your other attitudes, to not have the following combinations of attitudes: intending E, believing that M is a necessary means to E, and not intending M? Similarly, if you believe that you ought to φ, does it follow that you are rationally required to intend to φ? Or are you merely required, at all times and irrespective of your other attitudes, to not have the following combinations of attitudes: believing that you ought to φ and not intending to φ? In each case, “narrow-scopers” think that the first claim is the intuitively correct one, whereas “wide-scopers” opt for the second. Call this the scope debate.

The second issue concerns the proper interpretation of certain natural language conditionals containing modal expressions like ‘ought’ and ‘requires’, such as ‘If you believe you ought to φ, then you ought (or are required) to intend to φ’. To distinguish this from the scope debate, I’ll call this the Scope debate. In brief, the Scope debate arises because conditionals containing modals are traditionally assumed to be scope ambiguous, having one interpretation which can be formally represented as:

Wide O(C1 ⟹ C2)

and the other of which can be formally represented as:

Narrow C1 ⟹ O(C2)

where ‘⟹’ is a two-place conditional operator, ‘O’ is a one-place modal operator representing ‘ought’ or ‘requires’, and ‘C1’ and ‘C2’ are the relevant conditions (e.g. an agent’s having certain attitudes).²⁰ The Scope debate thus concerns the interpretation of certain (intuitively true) conditionals: should the modal operator be interpreted as taking “narrow scope” relative to the conditional operator (a lá Narrow) or instead “wide scope” (a lá Wide)?

It’s the (non-linguistic) scope debate that is of central importance, though it has often been wrongly conflated with the (linguistic) Scope debate.²¹ It’s an easy mistake to make. Structural rationality, after all, is fundamentally a matter of how one’s attitudes relate to each other -of how they “fit” or “hang” together- and conditionals give us a natural way of expressing claims about such relations. In particular, conditionals allow us to express claims about which attitudes are rationally required given certain other attitudes, and thereby give voice to our intuitive judgments of proper and improper fit between them -the very judgments that prompt theorizing about structural rationality in the first place.

Nonetheless, it’s increasingly recognized that the Scope debate rests on a dubious assumption. In particular, the Scope debate proceeds on the traditional philosophical assumption that conditionals are to be formally represented using a two-place conditional operator (=>) that takes a pair of propositions and forms a conditional proposition, in much the same way that clauses joined by ‘and’/‘or’ are formally represented using two-place operators (∧/∨) that take a pair of propositions and form a conjunction/disjunction.²² Call this the operator view. It’s because the operator view takes ‘if ’ to denote a two-place conditional operator that question of relative scope arises whenever there’s a co-occurring modal.

Although the operator view is still accepted by many logicians and philosophers, it is widely rejected by linguists. The dominant alternative -commonly known as the restrictor view- involves a fundamental re-thinking of the compositional structure of conditionals: rather than denoting a two-place conditional operator, ‘if’ functions as a device for restricting the domains of nearby operators. The restrictor view was first introduced by Lewis (1975) to handle conditionals containing adverbs of quantification (‘usually’, ‘always’, etc.), and subsequently generalized to other conditionals by Kratzer (1977, 1981, 2012). The basic idea behind the restrictor view is that just as in sentences like

(1) All/Most/Some men smoke

the common noun (‘men’) restricts the domain of the quantifier (‘all’/‘most’/ ‘some’), so that it only ranges over (in this case) men, so in conditionals like

(2) If you believe it’s going to rain, you usually/always/sometimes/may/ must/ought/are required to carry an umbrella.

the antecedent (‘you believe it’s going to rain’) restricts the domain of the co-occurring quantificational operator, which is what adverbs like ‘usually’/‘always’/‘sometimes’ and modals like ‘may’/‘must’/‘ought’/‘required’ are standardly analyzed as. As a result, the consequent clause (‘you carry an umbrella’) is only evaluated with respect to the restricted set of possibilities where the antecedent is true (i.e. you believe it’s going to rain). Thus, a claim of the form ‘If you believe it’s going to rain, you usually carry an umbrella’ will be true (very roughly) just in case most situations in which you believe it’s going to rain are situations in which you carry an umbrella. And a claim of the form ‘If you believe it’s going to rain, you ought to carry an umbrella’ will be true just in case the normatively best, or “highest ranked”, situations in which you believe it’s going to rain are those in which you carry an umbrella.

Accordingly, whereas philosophers have traditionally assumed conditionals containing modals have the following bipartite logical form, where O is an unary operator and conn is a two-place connective (e.g. &, v, ⟹ :

O(R conn P)

Kratzer and company think they are better analyzed as having following tripartite logical form, where O is a binary operator, R is a (possibly tacit) domain restriction, and P is the prejacent (roughly: the consequent minus the modal):

(O: R)(P)

As ^{Kratzer (2012}) famously puts it:

The history of the conditional is the story of a syntactic mistake. There is no two-place if... then connective in the logical forms for natural languages. If-clauses are devices for restricting the domains of operators. (106)

The upshot for the Scope debate is clear: assuming the restrictor view is true, there is no conditional operator concerning which the question of relative scope (Wide vs. Narrow) makes sense.²³

This isn’t the end of the story, however. For there’s another sense in which the restrictor view might be considered a “narrow scope” view, linguistically speaking, rather than being neither narrow nor wide. Consider, for instance, the English sentence ‘If you believe you ought to φ, then you’re required to intend to φ’. Following ^{Worsnip (2015)}, we might say that a semantic interpretation is “wide scope” in this other sense -let’s call it wide-scope*- if, according to it, what the sentence says is required of you is a disjunction of attitudes (e.g. intending to φ or not believing you ought to φ), and that a semantic interpretation is narrow-scope* if, according to it, what the sentence says is required of you is a particular attitude (e.g. intending to φ).²⁴ The restrictor view would then say that the correct interpretation of the sentence above is narrow-scope*: the sentence says you’re required to intend to φ, where the interpretation of ‘required’ is restricted (per the restrictor view) to situations in which you believe you ought to φ.²⁵ This is the sense in which Worsnip takes the restrictor view to be a “narrowscope” view -one that he then seeks to reconcile with the non-linguistic wide-scope view he (along with Broome) favors.

This raises an important, and more general, issue. For once the philosophical and linguistic issues are clearly distinguished, the question arises as to how they interact. It would be rather surprising -and indeed discomfiting- if one’s theory of structural rationality turned out to be utterly disconnected from our ordinary judgments concerning it, many of which come clothed as conditionals. Our ordinary judgments, after all, are what prompt theorizing about structural rationality in the first place, and their truth -or at least apparent truth- is part of the data that ultimately needs to be explained. Accordingly, for any theory of structural rationality to be complete, it must provide a story connecting theory and practice, with the plausibility of the theory depending (in part) on the plausibility of the story told.²⁶ In this way, our ordinary judgments concerning structural (ir)rationality are properly considered as common ground among, as well as data to be explained by, competing theories of structural rationality.

The good news is that for present purposes we can bypass the linguistic Scope (and scope*) debate and focus squarely on the philosophical scope debate by not formulating the relevant structural requirements using natural language conditionals (‘if’s). For example, one standard way of representing the competing views of the requirement concerning means-end (ME) coherence is as follows, where ‘->’ denotes the material conditional (not ‘if... then’):²⁷

(ME-wide) Rationality requires that ((you intend end E ⋀ believe that M is a necessary means to E) -> you intend M).

(ME-narrow) (You intend end E ⋀ you believe that M is a necessary means to E) -> rationality requires that you intend M.

I should note that I -like Broome- intend the requirements above to be understood synchronically. Nonetheless, both arguably have diachronic upshot. For if rationality prohibits you from having certain combinations of attitudes, it seems to follow that in order to be rational (or at least not irrational) you must -going forward- either avoid or else get rid of that combination of attitudes. Similarly, if rationality requires you to have a certain attitude whenever you have certain other attitudes, it seems to follow that to be rational you must -going forward- either acquire that attitude or else avoid having the other ones.²⁸ These derivative diachronic principles needn’t be understood as requirements in the strong sense, however. It suffices for them to be understood as (non-trivial) requirements in the weak sense -i.e., as (non-trivial) necessary conditions of being rational over time.

Recall that although wideand narrow-scope requirements like (ME-wide) and (ME-narrow) make the same predictions concerning when an agent is irrational, they don’t offer the same explanations of why. Is the agent irrational because they have a combination of attitudes rationality requires them not to have (as wide-scopers maintain), or instead because they fail to have a particular attitude that, in their present circumstances, rationality requires them to have (as narrow-scopers maintain)?

Why might one opt for (ME-wide) over (ME-narrow)? The main reason is that the latter seems subject to counterexample: you might intend end E and believe that M is a necessary means to E without it being the case that you are rationally required to intend M. Suppose, for example, that you believe intending M would have terrible consequences for your family. Is it really true that you’re rationally required to intend M anyways? Arguably not -contra what (ME-narrow) seems to say.²⁹ Similarly, you might believe p and that if p then q without it being the case that you are rationally required to believe q. After all, q might be wildly implausible, and you might recognize it as such, in which case it seems wrong to say that you are required to believe q -contra what the following narrow-scope deductive coherence (DC-) requirement says:

(DC-narrow) (You believe p ⋀ you believe if p then q) -> rationality requires that you believe q.

This problem generalizes: for most if not all narrow-scope requirements, there will be cases in which one has the attitudes in the antecedent and yet the normative claim specified in the consequent seems too strong. Call this the too strong problem.

There’s another problem in the vicinity. For suppose you believe p and that if p then q, but also believe r and that if r then not-q. Given (DC-narrow), it follows that you are rationally required to believe q and you are rationally required to believe not-q, and thus required to have inconsistent beliefs. This seems problematic -as ^{Broome (2013}) notes, structural rationality is supposed to prohibit incoherence, not require it. Alternatively, suppose you intend end E1 and believe that M is a necessary means to E1 but also intend some other end E2 and believe that not-M is a necessary means to E2. Does it follow that you are rationally required to intend M and rationally required to intend not-M, and thus have inconsistent intentions, as (ME-narrow) seems to say? Arguably not. This problem generalizes as well. Call it the conflict problem.

What’s the solution to these problems? According to many philosophers -including Broome- the answer is clear: we should accept the wide-scope requirements. Unlike narrow-scope requirements, wide-scope requirements merely prohibit certain combinations of attitudes, remaining silent as to which particular attitudes we should or shouldn’t have. They thus avoid the too strong and conflict problems above, among others.

Nonetheless, the formulations of the wideand narrow-scope requirements above aren’t totally happy. For one thing, there’s a natural tendency -or at least temptation- to interpret ‘->’ as ‘if... then’, which is a mistake. We can avoid this by replacing the material conditional with other, truth-functionally equivalent combinations of connectives (negation + conjunction, negation + disjunction), but then we run into other problems. Suppose, for example, we opt for the following alternatives:

(ME-wide ') Rationality requires that you not: intend end E, believe that M is a necessary means to E, and not intend M.

(ME-narrow ') Either you don’t intend end E or you don’t believe that M is a necessary means to E or rationality requires that you intend M.

The problem is that whereas (ME-wide') correctly captures the core commitment of the wide-scope view -namely, that structural requirements ban incoherent combinations of attitudes, and that’s it³⁰- (ME-narrow') fails miserably in capturing the core commitment of the narrow-scope view -namely, that you’re structurally required to have particular attitudes in virtue of having certain other (combinations of) attitudes. As stated, (ME-narrow') says nothing about there being any explanatory, or otherwise asymmetric, relationship between disjuncts, and in particular is silent about the relationship between the falsity of the first two disjuncts and the truth of the third. (Entailment is not explanation.)

As a result, although there’s no need to appeal to a non-material conditional operator in stating wide-scope requirements, it looks like we do need to appeal to such in stating narrow-scope requirements. Neither material conditionals nor claims involving ‘if... then’ suffice. Whereas representations involving material conditionals fail to do justice to the genuinely and essentially conditional nature of narrowscope requirements, claims involving the ordinary ‘if... then’ fail to be properly distinctive -as noted above, they are claims the widescoper can (and should) agree are true. The widespread tendency to nonetheless make use of the material conditional and/or the ordinary ‘if... then’ in formulating narrow-scope principles is thus regrettable.³¹

This of course just raises the question: how should we formulate wide -and narrow- scope requirements? To make progress on this question, it’s helpful to return to the analogy with the law.

IV. The Jurisdiction of Structural Requirements

Recall that if we take the ideology of law-like structural requirements seriously, besides the question of scope, there’s also the question of jurisdiction, or domain of governance. That is, for each putative requirement R, there’s the question: does R govern all rational agents at all times, or does it only apply under more selective conditions?

As ^{Broome (2013}) and ^{Schroeder (2014}a) note, this question is akin to one that arises in the legal realm. Here’s Schroeder:

One important feature of laws is that they have jurisdictions. For example, in the state of New York, it is illegal to turn right at a red light. The jurisdiction of that law is drivers in New York, and what it prohibits is turning right on red. In general, anyone who is simultaneously a driver in New York and is turning right on red is in violation of this law, but... being a driver in New York and turning right on red make different contributions to this fact. If you are a driver in New York and you don’t turn right on red, then you are complying with the law, whereas if you are a pedestrian in New York or a driver in Cairo, the law simply doesn’t apply to you. The reason why drivers in Cairo who turn right on red aren’t in violation of New York traffic laws is that the [latter don’t] have jurisdiction over drivers in Cairo.

This is suggestive, but incomplete. For there’s a clear sense in which all residents of New York -whether or not they are driving- are ipso facto subject to its laws, including traffic laws, whereas residents of Cairo are not. It remains true, however, that pedestrians in New York bear an importantly different relationship to traffic laws than drivers in New York do. Intuitively, the difference is that whereas the traffic laws don’t apply to pedestrians in New York, they do apply to those who are driving. This difference in application is nonetheless compatible with the thought that everyone residing in New York -whether or not they’re driving- is within the jurisdiction of the relevant law, and hence prohibited from turning right at a red light while driving, whereas the law is simply silent about those residing elsewhere, such as Cairo.

To capture the relevant differences, then, we need to distinguish the jurisdiction of a given law -those which are “subject to” the law or within its domain- from the conditions of application of the law -those conditions that need to be satisfied by those within its domain in order for it to actually apply to a particular case.³²

Violation and compliance are only possible when the conditions of application obtain; merely being within the jurisdiction of a law and not satisfying it’s conditions of application is not enough. Nor, of course, is falling outside of the jurisdiction of the law altogether. As a result, we can (and should) distinguish three types of nonviolation. A given person can fail to violate a given law L by:

The distinction between (b) and (c) is unfortunately blurred by ^{Broome (2013}). He addresses the question of jurisdiction by first distinguishing between two ways in which a requirement can be “conditional”: it might be conditional in its content or conditional in its application. A requirement is conditional in content just in case what it requires is that some (e.g. material) conditional be satisfied, whereas a requirement is conditional in application just in case it requires something of you if some condition is satisfied. Broome writes:

When a requirement is conditional in its content, it is commonly said to have a wide scope, because what is required is the compound proposition that [p -> q]. When the requirement is conditional in application, it is commonly said to have a narrow scope, because what is required is simply q (132).³³

Broome then proceeds to treat a given requirement’s jurisdiction as a condition of application (§8.2), with legal requirements, for example, only applying to residents, and rational requirements only applying to creatures possessing rational capacities. Like ^{Schroeder (2014}a), then, he collapses the distinction between jurisdiction and conditions of application, treating (c) as a special case of (b).³⁴

The upshot is that although Broome and Schroeder are right to emphasize that you can satisfy or comply with a law only if you fall within its jurisdiction, and that leaving the jurisdiction of the law is sufficient to avoid violating it, we shouldn’t take the distinction between compliance and avoidance to track the notion of jurisdiction.

Why is this important? Because, as Broome and Schroeder both stress, the relevant concept of a rational requirement is supposed to be law-like not just with respect to its explanatory status but also insofar as it allows for a non-vacuous distinction between those within its jurisdiction and those who are not -as well as, I’m urging, between those within its jurisdiction to whom it applies and those to whom it doesn’t. This is what allows us to distinguish between different forms of non-violation -compliance and non-application- and this is thought to allow for intuitive “tests” of whether a given principle should be interpreted as wide or narrow (or perhaps intermediate) scope.

Consider, for instance, the following possible regimentations of the relevant New York State traffic law, where ‘[all x: ...]’ specifies the requirement’s jurisdiction and ‘➢’ is a special operator distinguishing conditions of application from what’s required under those conditions:³⁵

According to (NY-widest), the law has jurisdiction over -and applies to- everyone, no matter where they are, and requires of them that they either not be in New York, not be driving, or not turn right on red. According to (NY-wide), the law has jurisdiction over everyone in New York, and requires of them that they either not be driving or not turn right on red. According to (NY-narrow), the law likewise has jurisdiction over everyone in New York, but it only applies to those within its domain who are driving -those to whom it applies are then required to not turn right on red. And according to (NY-narrowest), the law has jurisdiction over everyone in New York who is driving and requires of them that they not turn right on red.

(NY-widest) and (NY-narrowest) are clearly implausible, whereas (NY-wide) and (NY-narrow) fare better. Following Broome and Schroeder, we can support this claim by considering our intuitive judgments concerning compliance, as well as our judgments concerning jurisdiction and application more generally. Intuitively, for example, whereas drivers in New York who don’t turn right at red lights are complying with New York traffic laws, drivers in Cairo -whether or not they turn right at red lights- are not. That’s because they are outside of the jurisdiction of the law altogether. So (NY-widest) is implausible. Pedestrians in New York, by contrast, do seem to be within the law’s jurisdiction -contra (NY-narrowest)- but nonetheless don’t seem to be complying with the traffic law either. Intuitively, that’s because the traffic law doesn’t apply to them in their capacity as pedestrians -contra (NY-wide). (NY-narrow), then, is the most plausible regimentation of the relevant law -it accords well with our intuitive verdicts concerning compliance, jurisdiction, and application.

V. An Improved Narrow-Scope View

We’re now in a position to offer an improved account of narrowscope requirements -although they have jurisdiction over all rational agents, we should think of them as only applying to agents who have the attitudes specified in their antecedents. Their “antecedents” thus specify conditions of application rather than delimit their jurisdiction. For the purposes of illustration, let’s re-consider the means-end (ME) requirement:

(ME-narrow+) Rationality requires of [all x: x is a rational agent] that ((x intends end E ⋀ x believes that M is a necessary means to E) ➢ x intends M)

Unlike (ME-narrow) or (ME-narrow'), (ME-narrow+) directly captures the core narrow-scope thought that having certain attitudes commits you to having certain other attitudes, and that you are required to have the latter in virtue of having the former. On this view, narrow-scope requirements are best thought of as being conditioned, rather than conditional.

This kind of view inherits many of the advantages of the traditional understanding of the narrow-scope view while enjoying additional potential benefits as well. In particular, it may help the narrow-scoper blunt the force of the “too strong” and “conflict” problems above. Recall (ME-narrow). It says that you are rationally required to intend means M whenever you intend end E and believe that M is necessary to E. But aren’t you at least sometimes permitted -if not required- to revise your end or means-end belief rather than intend the means? (ME-narrow) seems to confer a kind of fixed authority to one’s existing attitudes that is problematic; the agent should at least sometimes have the option of giving them up instead. And the same, of course, goes for other narrow-scope requirements.

The narrow-scoper might try to meet the objector halfway. To begin with, we need to distinguish two senses in which one might be “permitted” to do something. Being permitted to do something, φ, in the first sense -call it weak permission- is a matter of not being prohibited from φ-ing. Being permitted to φ in the second sense -call it strong permission- is a matter of φ-ing being positively sanctioned.³⁶ The distinction is both intuitive and important. Suppose, for example, a child finds candy lying around the house. Assuming no one in authority over her has told her to not eat candy (or to ask before eating, etc.), it would seem to be permissible -in the weak sense- for her to eat it: if she does, she won’t be doing anything she isn’t supposed to do. But that may be just because no one has (yet) considered the question of whether or not she should be allowed to eat candy. Now suppose someone in authority over her does consider the question and decides it’s OK. This changes the normative landscape, and the child is now permitted in the stronger sense to eat candy: if she does, she’ll be doing something that enjoys the positive normative status of being sanctioned rather than just lacking the negative normative status of not being prohibited.

With this distinction in hand, the narrow-scoper can grant that narrow-scope requirements like (ME-narrow+), even when they apply, permit you in the weak sense to give up one or more of the antecedent attitudes -i.e., they don’t forbid you from revising (or otherwise ceasing to have) any of your existing attitudes.³⁷ You might satisfy the antecedent conditions of (ME-narrow+), for example, and hence have it apply to you, and yet not do anything wrong when you revise one of the relevant attitudes. That’s because in doing so you’ll no longer satisfy the conditions of application, and the requirement will cease to apply.³⁸ If it doesn’t apply, then you can’t violate it. So in revising one or more of your antecedent attitudes you won’t have done anything wrong by the lights of (ME-narrow+), since you’ll have rendered it non-applicable, and hence silent.

Similar considerations might be marshalled in an effort to blunt the force of the “conflict” problem as well. Suppose, as before, that you intend end E1 and believe that M is a necessary means to E1 while also intending some other end E2 and believing that not-M is a necessary means to E2. Does it follow that you are rationally required to intend M and rationally required to intend not-M, and thus have inconsistent intentions? According to (ME-wide+) the answer remains ‘yes’. But (ME-wide+) nevertheless permits you in the weak sense to revise your antecedent attitudes -i.e., you’re not prohibited from doing so, as far as (ME-wide+) is concerned.³⁹

However, the narrow-scoper shouldn’t say that narrow-scope requirements like (ME-narrow+) permit you to give up one or more of the antecedent attitudes in the stronger sense of actually saying it’s rationally OK to do so. That’s because there may be other requirements that say that it’s not OK, given your other attitudes. As with other narrow-scope requirements, (ME-narrow+) shouldn’t have the final say on whether any particular attitude is structurally rational tout court, as opposed to structurally rational in a particular (“local”) way. In response, the narrow-scoper might try to claim that revising the antecedent attitudes is only strongly permitted by the lights of (ME-narrow+), not tout court. But it’s hard to make sense of an attitude being strongly rational permissible merely “by the lights” some requirement. One possibility, for example, would be to offer a complex principle like the following:

(ME-narrow?) Rationality requires of [all x: x is a rational agent] that ((x intends end E ⋀ x believes that M is a necessary means to E) ➢ x intends M) and rationality permits of [all x: x is a rational agent] that ((x intends end E ⋀ x believes that M is a necessary means to E) ➢ (x not intend M ⋁ x not believe that M is a necessary means to E )).

But (ME-narrow?) seems bizarre. It requires you to intend the means you believe to be necessary to the end you intend while simultaneously strongly permitting you to revise your end as well as your means-end belief -i.e., the very attitudes that require you to intend the means. This makes the general worry about narrow-scope principles even more pressing: why the differential treatment?

Rather than delve deeper into obscurity, the narrow-scoper might change tack. Another way of making the permission to revise the antecedent attitudes more explicit is the following:

(ME-narrow++) Rationality requires of [all x: x is a rational agent] that ((x intends end E ⋀ x believes that M is a necessary means to E) ➢ (x intends M ⋁ x not intend E ⋁ x not believe that M is a necessary to E)).

The problem, however, is that this would effectively turn the principle into a wide-scope requirement -it requires that agents (not) have a certain combination of attitudes, under certain conditions, and that’s it. Indeed, it’s logically equivalent to the wide-scope requirement (ME-wide+) that we’ll consider next. So the narrowscoper is probably best off sticking to the original version (MEwide+) along with the notion of weak permission.

VI. An Improved Wide-Scope View

Wide-scope requirements are standardly assumed to be compliance symmetric: any way of satisfying the complex condition specified by the requirement is as good as any other, as far as the requirement itself is concerned. It’s precisely this feature of wide-scope requirements that enables them to avoid traditional objections to their narrow-scope counterparts -including the too strong and conflict problems above- since the narrow-scope requirements, unlike wide-scope ones, require a particular response.

But it’s a curious fact that the very feature that allows wide scope requirements to avoid such objections -their compliance symmetry- is also what many have found problematic. That’s because not all ways of complying with wide scope requirements are intuitively on a par. Consider again (ME-wide), this time with the jurisdiction marked explicitly:

(ME-wide) Rationality requires of [all x: x is a rational agent] that ((x intends end E ⋀ x believes that M is a necessary means to E) ̶ > x intends M).

As various authors have noted, while intending means M because you believe M to be necessary to achieve your end E seems to be a perfectly rational response, dropping your belief that M is a necessary means to E because you intend E but don’t intend M seems highly irrational.⁴⁰ The same goes for you not intending E just because you don’t intend M. We might expect this asymmetry to be captured by the requirement governing means-end coherence, and yet (ME-wide) fails to do so.

The same possible complaint arises for other wide-scope requirements. So it turns out that the wide scope view’s greatest strength and main attraction -its compliance symmetry- is also its main source of resistance. Wide-scopers, including Broome, are sensitive to this complaint, but don’t view it as an objection since they don’t think it’s the job of the relevant wide-scope requirements to explain everything that might go wrong in such cases.⁴¹ Instead, they take the lesson to be that such principles need supplementation. Broome and Way, for example, appeal to a special class of “basing principles” that prohibit certain basing relations between attitudes and permit others. Although introducing basing principles comes at the cost of complexifying the overall view, ^{Broome (2013: 142)} rightly points out that narrow-scopers also need to appeal to such principles to explain the full range of our verdicts concerning (im)proper basing.

Although this is right as far as it goes, there’s another worry that remains unaddressed.⁴² For according to what Lord calls the “real” symmetry objection, the narrow-scope requirements make intuitively more plausible predictions than wide scope ones concerning what counts as compliance and non-application (Lord 2014a; cf. ^{Schroeder 2014}a). To see why, consider a particular instance of (ME-wide) and compare it to the corresponding instance of (MEnarrow+):

(Grandma-wide) Rationality requires of [all x: x is a rational agent] that (x not intend to visit Grandma ⋁ x not believe that driving to Grandma’s house is necessary to visit her ⋁ x intend to drive to Grandma’s house).

(Grandma-narrow+) Rationality requires of [all x: x is a rational agent] that ((x intends to visit Grandma ⋀ x believes that driving to Grandma’s house is necessary to visit her) ➢ x intends to drive to Grandma’s house).

Both requirements have jurisdiction over all rational agents, but unlike (Grandma-narrow+), (Grandma-wide) also applies to all rational agents. What it requires is that you satisfy the complex condition. Any way of doing so is a way of complying with the requirement. It is compliance symmetric, per above. By not intending to visit Grandma, then, you thereby comply with (Grandma-wide), regardless of what else you believe or intend. You also comply with it whenever you don’t believe that driving to her house is necessary to visiting her, as well as whenever you happen to intend to drive to her house (for whatever reason). As far as compliance with (Grandma-wide) goes, these routes are just as good as intending to visit Grandma and intending to take the means you believe to be necessary -namely, driving to her house.

This generalizes: for every end you don’t intend and every meansend belief that you lack, as well as for every means that you intend, you thereby comply with a corresponding wide-scope means-end requirement. According to ^{Lord (2014}a), verdicts like these seem “a bit far fetched, to say the least” (460).⁴³ In contrast, (Grandma-narrow) only applies to you if you satisfy certain conditions -namely, if you intend to visit Grandma and believe that driving to her house is a necessary means of doing so. To not intend to visit Grandma ensures non-application, not compliance. Compliance requires that one both satisfy the conditions of application and do what’s required under those conditions (i.e. intend to drive).

This sort of dialectic can be rehearsed for each candidate requirement and its corresponding wideand narrow-scope interpretations. Insofar as a given narrow-scope interpretation delivers more intuitively plausible verdicts concerning compliance and non-application than the wide-scope interpretation, this counts in its favor.

Notice, however, that the wide-scoper can help themselves to the distinction between jurisdiction and conditions of application in much the same way that the narrow-scoper can. By doing so she can then offer modified versions of the various requirements that do a better job jiving with judgments concerning compliance and nonapplication. For example:

(ME-wide+) Rationality requires of [all x: x is a rational agent] that ((x intends end E ⋀ x believes that M is a necessary means to E) ➢ (x not intend E ⋁ x not believe that M is a necessary means to E ⋁ x intends M)).

(Grandma-wide+) Rationality requires of [all x: x is a rational agent] that ((x intends to visit Grandma ⋀ x believes that driving to Grandma’s house is necessary to visit her) ➢ (x not intend to visit Grandma ⋁ x not believe that driving to Grandma’s house is necessary to visit her ⋁ x intend to drive to Grandma’s house)).

As before we can interpret (ME-wide+) as having jurisdiction over all rational agents, but unlike (ME-wide) it only applies to you if you satisfy certain conditions. If you do, then rationality requires that you either intend to drive to your grandmother’s house (in which case you’ll be in a state of compliance) or else revise one of your pre-existing attitudes (in which case the requirement will no longer apply). In ^{Broome’s (2013}) terminology, (Grandma-wide+) is conditional in application and in content -a possibility that Broome recognizes but doesn’t pursue.

The upshot is that the wide-scope view can capture intuitions concerning compliance and non-application just as easily as the narrow-scope view can. In particular, the wide-scoper can capture intuitions of “application asymmetry” and “directedness” in precisely the same way as the narrow-scoper-in both cases the relevant requirements only apply under select conditions (and indeed the same conditions), and in both cases it is in virtue of having certain attitudes that one will be required to have certain other (combinations of) attitudes. It’s therefore a mistake to think, as ^{Lord (2014}a) does, that “[t]here is no way for the wide-scoper to escape the commitments that lead to the symmetry objections”, and that it is “an essential feature of the wide-scope requirements that they are application symmetrical” (462; emphasis in original). In the same way, it’s a mistake to think, as ^{Schroeder (2014}a) does, that the analogy with the law affords the narrow-scoper with “richer explanatory resources” than the wide-scoper (225).

Modifying wideand narrow-scope requirements in the ways suggested is not without costs, however. That’s because by doing so we seem to have effectively erased any important difference between them. For the modified wideand narrow-scope requirements end up sharing exactly the same application, violation, and compliance conditions. (ME-wide+), for example, applies just in case (ME-narrow+) does -an agent needs to intend end E and believe that M is a necessary means to E- it is violated just in case (ME-narrow+) is -an agent needs to intend E, believe that M is a necessary means to E, but not intend M- and it is complied with just in case (MEnarrow+) is -an agent needs to intend E, believe that M is a necessary means to E, and intend M.⁴⁴ The only difference is that whereas narrow-scope requirements single out a particular attitude as being called for, wide-scope requirements don’t -they only require that one (not) have certain combinations of attitudes. This difference in content, however, doesn’t make as much of a difference as one might hope or expect.

VII. Conclusion

This leaves us with the question: what’s left to choose between the wideand narrow-scope views? Though the distance between them has diminished considerably, there are various ways for a gap to reemerge. For one thing, wide-scopers might find fault in the modified narrow-scoper’s response to the conflict problem, insofar as it remains true that there are cases in which narrow-scope requirements issue in conflicting verdicts regarding what an agent is structurally required to believe or intend. Unfortunately, however, a standoff is likely to ensue -whereas wide-scopers will continue to insist it’s implausible for structural rationality to issue in conflicting verdicts, narrow-scopers can insist that this is just what we should expect of incoherent agents, since they may very well “paint themselves into a corner” (as ^{Kolodny 2007} puts it). But perhaps that’s to be expected, given how close such judgments are to philosophical bedrock. A similar standoff threatens to arise regarding the too strong problem -wide-scopers may insist that weak permissibility is too weak for the narrow-scoper’s purposes, while narrow-scopers might insist otherwise.

Alternatively, the narrow-scoper might try to turn the tables by complaining that modified wide-scope requirements like (MEwide+) effectively give up the game insofar as they’re no longer compliance symmetric in any meaningful sense. That’s because there’s only one way to actually comply with such requirements, and it’s the same way one complies with narrow-scope requirements. In order to comply with (ME-narrow+), for instance, an agent needs to intend E, believe that M is a necessary means to E, and intend M -and the same goes (ME-wide+). Any other combination of attitudes will either violate the requirements or else render them inapplicable. But insofar as compliance symmetry is supposed to be a defining feature of wide-scope requirements, this may be seen as an undesirable result. If so, the wide-scoper might be wise to return to unmodified versions of wide-scope requirements, such as (ME-wide), and try to explain -or explain away- the seemingly problematic consequences concerning compliance in another way.⁴⁵ At that point, however, it’s not clear there will be many, if any, dialectically kosher moves left to make.