I. INTRODUCTION
According to the legal rationalist, the law claims to give its subjects reasons for action. The leading legal rationalist, Joseph Raz, says, “the law claims that the existence of legal rules is a reason for conforming behaviour.”Footnote 1 Putting the same point more casually, he writes:
The law sets things straight: telling people “this is what you should do and whether you agree that this is so or not, now that it is the law that you should you have the law as a new, special kind of reason to do so.”Footnote 2
Jules Coleman, who also at times plays the part of the legal rationalist, agrees:
The prevalent view among legal positivists today is that law purports to govern conduct as a practical authority. The distinctive feature of law's governance on this view is that it purports to govern by creating reasons for action. Footnote 3
Or more succinctly, “Law claims to create reasons for acting.”Footnote 4
The legal rationalist is concerned to argue that the law claims to give its subjects reasons for action because he is concerned to establish the intuitively plausible thought that the law is normative, that the law makes a normative difference in the lives of its subjects. Coleman has put the same point in terms of a practical difference, “a difference, that is, in the structure or content of deliberation and action.”Footnote 5 The idea comes originally from H.L.A. Hart, who famously argued against Austin's sanction-based theory of law in part on the grounds that it could not account for the fact that, at least sometimes, people take their legal obligations as reasons for performance.Footnote 6 Reasons for action being “the key normative phenomenon,”Footnote 7 the most obvious way to establish that the law makes this normative difference in people's lives is by arguing that the law claims to give reasons for action to its subjects. Call this thought the normative reasons intuition.
In addition to the intuitively plausible thought about the normativity of law, the legal rationalist, in arguing that the law claims to give reasons for action, can rely on a second intuition, that that people in general seem to take the law as a reason. People sometimes say things like “I don't download songs from BitTorrent because it's a violation of the Copyright Act” or “Sorry I took so long to get here, but I was driving under the speed limit.” When they do so, they seem to be explaining their actions by reference to the law's directives and so treating the law as their motivating or explanatory reason. Although there are important distinctions between explanatory and normative reasons, it is plausible to suppose that an agent who ϕs because of some consideration p—that is, an agent who is motivated by p to ϕ and guided by p in her ϕing—takes p to be a normative reason.Footnote 8 And while it is possible to be confused about our normative reasons, it seems unlikely that everyone is confused all the time; so the fact that people consistently take the law as their motivating reason and guide their conduct by the law seems to suggest that the law does provide normative reasons for its subjects.Footnote 9 Call this thought the explanatory reasons intuition.
Based on these two intuitions, the legal rationalist argues that the law claims to give reasons for action to its subjects. We can make the legal rationalist's claim more precise by noting how it follows from the more familiar idea that the law claims to be a legitimate authority.Footnote 10 It follows in two steps. First, “It is widely assumed in the literature that a legitimate directive of a practical authority imposes an obligation on the relevant subjects to comply.”Footnote 11 And since the law claims to be an authority, the law claims that when it directs an agent to ϕ, the agent thereby has an obligation to ϕ. Second “the fact that one has an obligation is a reason—a reason of special force—for doing whatever one has an obligation to do.”Footnote 12 Putting the two points together, we get:
To claim legitimate authority is to claim that one's directives provide reasons for acting that apply to those over whom one claims authority.Footnote 13
And if it's a necessary feature of law that it claims authority, then the argument seems to establish that law claims to provide reasons for its subjects.
Schematically, the argument goes like this:
(1) The law claims to be a legitimate authority (premise).
(2) If a legitimate authority directs a subject to ϕ, the subject is obligated to ϕ (definition of legitimate authority).
(3) The fact that an agent is obligated to ϕ is a reason for her to ϕ (premise).
(4) So the law claims that its directives give reasons to its subjects.
Note the two distinct premises. First there is the Razian premise that the law claims to be a legitimate authority. And second there is the premise, which I illustrated above with a quote from John Gardner, that I call obligations as reason-giving (or OAR for short):
Obligations as Reason-Giving (OAR): the fact that A is obligated to ϕ is itself a reason for A to ϕ.
Unfortunately for the legal rationalist, OAR is false. Legal obligation has certain features that mean that the fact of a certain action's being legally obligatory cannot itself constitute a reason to perform that action.
As I argue below, legal obligatoriness is what is sometimes called verdictive or conclusory, as opposed to contributory, and the verdictive does not provide reasons (see Section II); moroever, an action's being legally obligatory is something that obtains in virtue of a process of actual human deliberation in a way that means that it cannot itself be a reason for action (see Section III). So the legal rationalist's argument that the law claims to give reasons for action fails. To vindicate the thought that the law does not claim to give reasons for action, I need to provide some explanation—or explaining away—of the two intuitions that the legal rationalist claims in support of his argument (Section IV). So I argue (Section V) that we can explain the explanatory reasons intuition away by noticing that much of the time, when people refer to the law in explaining their actions, they are referring not to the sort of reasons contemplated by OAR, but rather to other sorts of reasons triggered by the law in a way that does not help the legal rationalist.
And (see Section VI) the same goes for the normative reasons intuition. I show how we can establish that the law makes a normative difference in the lives of its subjects in a way that is consistent with the falsity of OAR. To say that the law does not claim to give its subjects reasons for action is not to deny that the law claims that its directives have some special practical force on its subjects’ deliberations; it is not to deny that, as Coleman recently put it, “Law impacts what we have reason to do.”Footnote 14 What I deny, rather, is that the law's effects on our practical deliberation or on what we have reason to do must include reason-giving. In a way, this is just the logical next step to take after accepting the Razian idea of second-order reasons: Raz teaches us that the law's impact on our reasons is not limited to adding weighty reasons in its favor but that it might also include second-order exclusionary reasons. Once we make that step, we can abandon the idea that the law creates first-order reasons and still capture our intuitions about how the law affects or claims to affect what we ought to do. To close my argument, I consider one objection to the effect that the law might interestingly claim to give reasons in a way not captured by OAR (Section VII), and finally (Section VIII) I suggest some implications of my argument.
Before getting started, I need to make four sets of remarks by way of clarifying what I take to be some important but peripheral issues. First, on the idea that the law claims to give reasons to its subjects:Footnote 15 saying that the law makes normative claims might seem mysterious, since the law is not an agent, and only agents can make claims. But it isn't. The claims that the law makes are just those claims made by some set of legal officials when they are acting in their capacity as legal officials.Footnote 16 I do not say anything here about the conditions under which the law's claim might be true, or genuine, or what have you. That's because, as important an issue as understanding those conditions is, it is not the issue that is my concern here: my inquiry is about how to understand the kind of normativity that the law claims for itself, about what we can coherently understand the law to be claiming.Footnote 17
Second, on the legal point of view: the idea of the legal point of view is important here because in what follows I need to be careful of a pretty serious terminological difficulty. The phrase “legal obligation” is, as Scott Shapiro notes, ambiguous between two interpretations.Footnote 18 On the one hand, “legal” might be read “adjectivally,” according to which a legal obligation is a kind of obligation, like a promissory obligation or a professional obligation. On the other hand, “legal” might be read “perspectivally” according to which “legal obligation” refers to an obligation from the legal point of view. On this perspectival interpretation, when I am under a legal obligation to ϕ, I am under a genuine (perhaps moral) obligation to ϕ from the legal point of view. The easiest way to understand that idea is to say that when I am under a legal obligation to ϕ, the law claims that I am under a genuine (perhaps moral) obligation to ϕ.Footnote 19 My argument in what follows needs to tread this line carefully. On the one hand, OAR relies on the fact that law claims to be an authority and thus that it claims to create genuine obligations. My argument against OAR is based, in effect, on accepting that these claims might sometimes be true—which I think I can do on the (roughly Razian) grounds that while the law's claim to authority might be false, it cannot necessarily be false; this seems to me to suggest that we can fruitfully assume that it might sometimes be true in order to see just what such a claim would entail—and on challenging the conclusion that the law could create reasons even if they were. As such the entire argument accepts the law's claim or takes place, as it were, from or within the legal point of view.
On the other hand, even from the legal point of view, according to which legal obligations are genuine obligations, it still seems plausible to say that there is something to the adjectival reading of “legal obligation.” The idea is that obligations created by law have certain characteristics that perhaps not all obligations have. In particular, legal obligations are created by legal institutions, in contradistinction to, for example, the paradigmatic moral obligations not to lie, murder, and so forth. And I rely on these special features from time to time in order to demonstrate that even from the legal point of view, the fact that one is under a legal obligation to ϕ is not itself a reason for one to ϕ.
Third, on some features of obligation in general (and legal obligation in particular): I suggest above that we can make progress in understanding legal normativity by granting the law's claim to legitimate authority in order to see what would follow. So suppose we do so, and in so doing, grant that legal obligations are genuine obligations. Then any account of legal obligation (or its features) needs to be consistent with some set of intuitions or platitudes about obligation in general.Footnote 20 My aim is not to provide a full account of obligation, legal or otherwise. But it is part of an account of one aspect of obligation—its relation to reasons—and so it's worthwhile at least to mention a few such platitudes.
For one, it is very plausible to suppose that when an agent is obligated to ϕ, she has some reason to ϕ.Footnote 21 As we see below, this plausible but less precise idea might be behind some of the intuitive force of OAR. But being obligated to ϕ means more than just having a reason to ϕ in at least two distinct ways. First, obligation is linked in some way to wrongness: generally speaking, it is wrong to fail to do what one is obligated to do. Second, being obligated to ϕ seems to remove the question of whether or not to ϕ from the obligated agent's proper scope of deliberation: obligation is peremptory. Each of these two features of obligation are important in Section II, and I discuss them there. Any general theory of obligation needs to explain these features (or explain them away).Footnote 22 What about legal obligation?Footnote 23 I note above that even on the perspectival reading of “legal obligation,” one might plausibly think that legal obligation is an importantly distinct subset of obligation more generally, because legal obligations are created by legal institutions. In Section III, I rely on an idea in this neighborhood as part of my argument against OAR.Footnote 24
Fourth, on obligations and reasons and facts: You might wonder why OAR is so awkwardly phrased, with that reference to the fact that A is legally obligated to ϕ, given how often people say things like “A's legal obligation to ϕ is a reason for A to ϕ”? It is phrased that way because I take the view that obligations are not the right sort of things to be reasons:Footnote 25 reasons are most plausibly understood to be facts,Footnote 26 and obligations are not facts.Footnote 27 In discussing a related point about desires and reasons, Mark Schroeder writes, “if we say that someone's reason is her desire, then we can be plausibly reinterpreted as meaning that her reason is the fact that she has that desire.”Footnote 28 What Schroeder says goes for desires, I say goes for obligations. Thus the claim “A's legal obligation to ϕ is a reason for A to ϕ” can plausibly be reinterpreted as the claim “the fact that A has a legal obligation to ϕ is the reason for A to ϕ,” which is to say, as the claim “the fact that A is legally obligated to ϕ is a reason for A to ϕ,” which is to say, as OAR. With those four clarifications in hand, we can move on to the argument.
II. LEGAL OBLIGATION AND VERDICTIVENESS
One well-established idea in jurisprudence is that the normative effect of a legal obligation to ϕ on a subject cannot simply be to provide the subject with one additional reason to ϕ. The legal rationalist surely agrees with Hart that the law, in directing an agent to ϕ, claims to “cut off or exclude” deliberation on the merits of ϕing—that the legal obligation is meant to be a “peremptory” reason to ϕ.Footnote 29 Part of what Hart is after here is the idea that deliberation on the matter of whether or not to ϕ is closed or settled by the obligation. But deliberation is cut off because the law has reached some sort of conclusion or answer or verdict on the question of whether or not to ϕ. The legal obligation not to ϕ contains a practical verdict about ϕing: it is not to be done.
This idea—that the legal obligation not to ϕ provides or contains a verdict that ϕing is not to be done—is the idea that legal obligation is verdictive. For something to be verdictive is for it to express or embody a verdict or conclusion as to what ought to be done.Footnote 30 It is plausible to say that wrongness is verdictive. So to say that such-and-such an action would be wrong is to express a verdict on that action—it is one that should not be done. To get a handle on the idea of the verdictive, it helps to contrast it with the contributory. A contributory consideration is one that contributes to the verdict, by increasing the weight (as it were) on the verdict's side. So that ϕing would harm someone, or would cause extreme suffering, or would have costs that vastly outweighed any benefits, say, would be contributory considerations on the side of the verdict that ϕing would be wrong.Footnote 31
The verdictive cannot provide reasons. The basic thought is inherent in the idea of a verdict: the verdict is based on the underlying contributory considerations and so cannot itself then count as an additional contribution in favor of the same conclusion. If some action is (morally) wrong, that's because some set of reasons not to perform it makes it wrong. To take some examples from Phillipa Foot, that an action would hurt someone or that it would humiliate him are reasons not to do it. The action is wrong when those reasons not to do it defeat the reasons to do it.Footnote 32 That the action is wrong precisely is just the verdict on the reasons pro and contra. But if the wrongness is verdictive—if that an action is wrong is a verdict on the reasons for and against that action—then it cannot be the case that an action's wrongness is also itself a reason not to do it. As Jonathan Dancy says, “It is incoherent, in this light, to suppose that the [wrongness] can add to the reasons on which judgment is passed, thus, as one might say, increasing the sense in which, or the degree to which, it is true.”Footnote 33
The idea is this: wrongness itself, as verdictive, attaches to an action in virtue of the reasons pro and contra that action. It is, as Dancy says, “incoherent” to think that it can thereby also be a reason for that same verdict—because a verdict cannot be a reason for itself.
Now shift from moral wrongness to legal obligation. We can start by recalling that wrongness and obligation seem to be very closely linked indeed. In fact, it is plausible to suppose that for A to be legally obligated not to ϕ just is for it to be legally wrong for A to ϕ. I think this is the underlying idea behind judicial statements to the effect that, for example, “A trespass is a legal wrong,”Footnote 34 or “A breach of contract is in itself a legal wrong,”Footnote 35 since we are legally obligated not to trespass and to perform our contracts. The thought that wrongness is verdictive is equivalent to the thought that:
Wrongness as Reason-Giving (WAR): the fact that it is wrong for A not to ϕ is itself a reason for A not to ϕ
is false. And legally (i.e., from the legal point of view), there seems to be a direct equivalence between “A is legally obligated not to ϕ” and “it is legally wrong for A not to ϕ,” that is, they are expressions of the same fact. So we can simply substitute the one for the other, and, at least when it comes to legal obligations, if WAR is false, we can conclude straightaway that OAR is false, too.Footnote 36 The verdictiveness of legal wrongness, then, just is the verdictiveness of legal obligation.
We can see the same point—that legal obligation is verdictive—in a different way if we return to the notion of peremptoriness. The idea that a legal directive not to ϕ is, in Hart's words, “not intended to function within the hearer's deliberations as a reason for doing the act,” has, I think, got to be the idea that legal directives are not contributory (or at least are not meant to be contributory by the issuing authority). They are not considerations that are supposed to be taken into account in determining whether or not it would be wrong to ϕ; rather they are—to repeat myself—meant to settle the matter. They are overall verdicts of the form “ϕing is not to be done.”Footnote 37 Legal obligation is verdictive, and therefore the fact that A is legally obligated to ϕ cannot be a reason to ϕ.
Now here one might object that nobody is claiming that a legal obligation is a reason for itself. Which of course is correct. But the considerations suggesting that a legal obligation to ϕ is meant to settle the question of whether or not to ϕ and not to “function within the hearer's deliberation as a reason for” ϕing suggest that that legal obligation to ϕ, as verdictive, is based on some underlying considerations counting for and against ϕing. (A Razian might call these the dependent reasons.)Footnote 38 Thus the legal obligation is a conclusion or verdict on the question of whether or not one ought to ϕ. And the natural thought is that that verdict, based as it is on the underlying considerations, cannot then itself count as a new underlying consideration on the same issue. To put it differently: the legal obligation to ϕ obtains because ofFootnote 39 the reasons to ϕ and not to ϕ and thus cannot itself count as a new reason to ϕ.Footnote 40 So OAR is false.Footnote 41
But we need to be careful here, because it is easy to confuse two different ideas. One—which is the target of this section—is that it is in the nature of the verdictive that verdicts cannot be reasons for the actions to which they attach. Verdicts are based on some set of underlying contributory considerations, where the sense of “based on” that is appropriate here is metaphysical rather than epistemological. That is, these underlying considerations make it the case that the verdict applies, and this is true independent of any agent's going through the process of weighing the underlying contributory reasons. The second idea—which I discuss in the next section—turns away from this metaphysical claim to the thought that verdicts, and especially the verdicts embodied in legal directives and obligations, are reached through a process of deliberation. There I suggest that the nature of deliberation is such that once we reach a decision or verdict about some action, we cannot then use that decision as a further reason for that action.
III. LEGAL OBLIGATION AND DELIBERATION
In the section above, I argue that OAR is false because legal obligation is verdictive, and that therefore the argument that the law claims to give reasons for action fails. Here I take a different tack. The beginning of the bootstrapping argument is the observation that the law is a human creation: legal obligations are created by humans to accomplish purposes through some process of reasoning and deliberation.Footnote 42 But as Michael Bratman shows, the result of such a process of practical deliberation—a decision, or an intention, or a plan, or a policy—is not a reason to carry itself out. That is, that A has decided to ϕ is not a reason for A to ϕ; that B has planned for C to ψ is not a reason for C to ψ; that D and E have a policy of ξing is not a reason for them to ξ. To hold otherwise, says Bratman, is to sanction an impermissible form of bootstrapping.Footnote 43
To put some meat on these bones, suppose I am trying to decide whether to roast a chicken for dinner tonight. I weigh the reasons pro and contra, and let us say I decide to roast the chicken. My decision will be characteristically associated with a set of reasons for me then to carry it out:Footnote 44 perhaps I buy the chicken, or turn the oven on, or get the butter out of the refrigerator to soften, or open a bottle of chardonnay. Each of these actions will have consequences in the world that will then ground reasons for me to carry out the plan to roast the chicken. But it is a much more difficult and subtle question whether the fact of settling on that intention—just the fact of making the decision to roast—itself creates a reason for me to do so.
Michael Bratman's theory of intention gives us reason to doubt that it does. Bratman argues that to see the intention to ϕ as creating a new reason to ϕ would “sanction unacceptable forms of bootstrapping.”Footnote 45 We can see Bratman's argument by considering my chicken in more detail. Assume for a second that the reasons to roast it are actually outweighed by the reasons not to do so and to have salad instead (the amount of butter I put on the chicken before I roast it, let's say, is really just more than I ought to be eating). Nevertheless, I choose to have chicken and I form the intention to roast it. Some time later, I am faced with the question of whether to buy a chicken. If my intention to roast a chicken is a reason to roast one, then it must also be a reason to buy one (since chicken buying is, let's assume, a necessary means to chicken roasting). The reasons for and against buying the chicken should therefore now take into account the reasons to roast the chicken, the reasons to have salad instead, and, now, the reason given by my intention to roast the chicken.
It might be the case that these reasons balance out in favor of my buying the chicken. But buying the chicken is a means to roasting the chicken, and we know by hypothesis that I do not in fact have sufficient reason to do that—the reasons to skip chicken and eat salad outweigh the reasons to roast. But by making the (wrong) decision to roast a chicken, it seems, it is possible that I have made chicken buying the all-things-considered right option, the option supported by the most reasons. As Bratman says, we can see this even more clearly if we imagine that I decide to buy the chicken the moment I decide to roast it. Here it is clear that I have made the wrong choice, but if we see the intention as a reason, then, by breaking the process down into two steps, it seems I can make the wrong option into the right one.
I have more to say about the following point below but I do want to flag it right now: notice that in following Bratman and committing myself to the view that my decision is not a reason to roast the chicken, I have not ruled out the possibility that the decision has some other kind of practical significance. That is, Bratman's view is perfectly consistent with the thought that the decision should affect how I deliberate about what to do in the future (perhaps it even has some sort of peremptory status). There is of course much more to say here, but this gives us the basic idea of Bratman's claim.Footnote 46
Now the analogyFootnote 47 with legal obligation should be clear: the fact that an action is legally obligatory is a fact that, like intentions, arises based on deliberation about the merits of actions.Footnote 48 Raz's view, which I mention above, suggests this. But so does a plausible intuitive idea about how legal directives come into being (or at least how the law claims that they do so). Whether by legislation or by judicial creation, it seems, a legal obligation not to ϕ is created based on consideration of or deliberation about the reasons pro and contra ϕing. Judges, in handing down their decisions, and legislators, in passing laws, are supposed to weigh the merits of what they are doing and (judges at least) are, I think, plausibly seen to be doing just that. If that is the case, though, to say that the decision made to create the obligation—the legal directive—itself creates a new reason for performance, just in virtue of its being issued and not because of any effects triggered by the creation (on which more below), seems to license the same sort of impermissible bootstrapping that Bratman argues against in the personal decision case. So the fact that A is under a legal obligation to ϕ cannot itself be a reason to ϕ, and OAR is false.
Now I need to consider three objections. First, it might be objected that there is a major disanalogy between the decision case, on the one hand, and the case of legal obligations, on the other. The difference is this: a decision or intention is something I form on my own, whereas a legal directive is created by legal officials and meant to apply to citizens at large. These citizens had (in general) no part in making that decision. So much is true, but why think that the practical significance of a decision is any different when one person makes it for another? What if I am making the chicken not for myself but for my wife? Again, by hypothesis imagine that the reasons not to have the chicken defeat the reasons to have it. If my decision were a reason, then it could have the power to make the chicken the all-things-considered better option for her. But that does not make any sense, and I cannot see why things would be different in this case from what they were when I was just making the chicken for myself. Perhaps in this case there are reasons for my wife to guide her conduct according to my decision (maybe—just maybe—I am the better cook?). If so this can be accomplished by seeing my decision as peremptory.
The second objection is related because it focuses on the legal case especially. Here the objection accepts my argument and replies that bootstrapping new reasons into place is, effectively, the whole point of legal directives. That is, the law changes what we have reason to do just by its say-so,Footnote 49 and so to claim that the law can create new reasons just by its say-so is not a bug but a feature. The problem with this objection is that it ignores the thought—shared by Hart, Raz, Coleman, Shapiro, and many others—that legal normativity is in some sense continuous with other forms of normativity. By this I do not mean that legal normativity is a subset of, say, morality (as Hart explicity denies by suggesting that legal obligations are somehow sui generis). Instead I mean that if the law claims to impose obligations on us as a legitimate authority, then its claims must be held up against the nonlegal facts of the matter as to what obligations are and how they work. As a parallel, take Hart, who thinks that legal obligations arise out of rules. On this view, I think, Hart's account must be responsive to facts about rules that we might establish outside of the legal context.Footnote 50 That is, there are features that a rule must have to make it a rule. So if legal obligations are rules, they, too, must have such features. The argument I am trying to make in this section claims that rules, to stick with Hart, are not themselves reasons for their being carried out. So while the bare assertion that the law bootstraps reasons might not seem problematic, the more plausible and helpful assertion that the law could bootstrap reasons through rules must be wrong, since its invocation of rules makes it responsive to a wider set of considerations about the way that normative phenomena work.Footnote 51
The third objection asks if the bootstrapping argument proves too much. The essence of the bootstrapping argument is that we cannot, simply by making a decision to ϕ (say), give ourselves a reason to ϕ. Its sting can be lessened by the thought that decisions can have practical significance without being reasons themselves, perhaps in some second-order fashion.Footnote 52 But why? Why is it not the case that the bootstrapping argument also defeats the very idea that our decisions (or laws) can have any effect on our practical deliberations at all? The answer to this question is a long story that we cannot get into in full here. But the basics are easy: there are good reasons for us to make decisions and to stick to them. These are largely pragmatic considerations to the effect that by treating our decisions as having some genuine practical effect (that is, as changing how we ought to deliberate), we will tend to do better in achieving our goals and ends and, more generally, in doing what we have reason to do.Footnote 53
The echoes with the Razian justification of authority are obvious, and, consistent with that, we can see that what is at play here is a distinction between reasons to deliberate in various ways and reasons to act in particular ways in particular situations. The former are second-order reasons that apply to us in general regardless of our particular goals but rather because they help us achieve goals in general; the former depend on the merits of the action in question and therefore a simple act of decision-making (which cannot, after all, affect those merits)Footnote 54 cannot create them on the spot. So there is a genuine distinction between reasons to carry out a given decision (which the decision itself cannot create) and reasons to give decisions special power in our practical deliberation (which arise in general because of the way that decisions work and because of facts about our psychology); the bootstrapping argument applies only to the former and not to latter.
IV. EXPLAINING THE INTUITIONS AWAY
Let us take stock. So far, I am offering two arguments that OAR,
the fact that A is obligated to ϕ is itself a reason for A to ϕ,
is false. If either (or both) of these two arguments is sound, and OAR is in fact false, then the legal rationalist is wrong to argue that the law claims to give reasons for action.Footnote 55 But I have a problem. Or two, really: the explanatory reasons intuition and the normative reasons intuition. A really solid refutation of the legal rationalist's argument would provide something of an explanation for these intuitions—I state above that I would explain them away in a manner that is consistent with a denial of OAR. The rest of the paper attempts to do so.
The normative reasons intuition, roughly, says that the law is normative and so must provide normative reasons. Since Hart pointed out the problem with Austin's view, legal philosophers have been at pains to avoid a view about law according to which the law makes no normative difference to how we ought to act.Footnote 56 The natural way to make that argument is to rely on the notion of a reason, the basic normative notion. And the notion does seem to fit here: a reason to ϕ is a consideration that counts in favor of ϕing and contributes to the case for ϕ's being permissible or justifiable or rational. And the law does certainly seem to claim that its obligations work in just that way. It seems normal to think that the law against trespass is a reason for me not to walk on your lawn, a consideration that counts against my doing so and renders it somehow less permissible or justifiable if I do.
Related but distinct is the explanatory reasons intuition. People sometimes say things like the following: “Why are you driving so slowly?” “I'm obeying the limit.” Or “Why did you pay him that money?” “I was performing my contract.” Here, the law seems to be playing the role not of favoring actions but of explaining them. That is, the existence of a legal obligation is taken to be an explanatory reason rather than a normative reason. Explanatory reasons and normative reasons are distinct notions. As it is usually put, normative reasons count in favor of actions, and explanatory reasons explain why agents acted as they did. To illustrate, let us suppose that the fact that I have this excellent bottle of chardonnay already chilled is a (normative) reason for me to roast a chicken for dinner (since the chardonnay is excellent with roast chicken). If I roast the chicken in recognition of that fact—motivated by my knowledge of it and guided by it my acting—then it is the explanatory reason which explains my roasting the chicken (in addition to being, as we supposed, a normative reason for me to do so). But suppose I forget about the chardonnay or do not roast a chicken at all: in those cases, the fact that I have it chilled is still a normative reason but it is not an explanatory reason since it does nothing to explain my action. Conversely, suppose that I do not have the chardonnay chilled but that I think I do, and that based on that erroneous belief I roast the chicken. The fact that it is chilled cannot count in favor of my action, since it is not a fact at all; but it does seem to explain my action in some way.
These types of considerations have lead philosophers to say a great deal about the relationship between explanatory and normative reasons; too much, really, to get into the details here.Footnote 57 Instead, I just highlight one widely accepted connection between normative and explanatory reasons, as it is the one that is most important for our purposes here: for it to be the case that A ϕs because of p, A must take it to be true that p is actually a (normative) reason for her to ϕ.Footnote 58 How does this matter to the argument? Like this: if we tend much of the time to offer our legal obligations as explanations of why we acted the way that we did—that is, if we tend to offer our legal obligations as our explanatory reasons—then it seems that we take our legal obligations to be normative reasons. Our routine reference to the law in explaining our action suggests that we routinely take law to justify our actions. To put it differently, people take their legal obligations as reasons to perform. Hence the explanatory reasons intuition. In the remainder of the paper I provide explanations for these two intuitions that allow us to deny OAR. I take them in reverse order and start with the explanatory reasons intuition.
V. AN ERROR THEORY OF LEGAL EXPLANATORY REASONS?
According to the explanatory reasons intuition, people routinely refer to the law in explaining their actions; and since an agent who ϕs because of some consideration p must take p to be a reason for her to ϕ, this routine explanation of action by reference to law suggests a widely held view that the law provides normative reasons.
Now, of course it is not the case that just because an agent refers to some consideration as a reason, it is actually a reason for her. That is because we can be wrong about our reasons: if I think that the door is locked then if you ask me why I'm getting my keys out, I'd say, “because the door is locked and I need to unlock it to go inside.” But if the door is unlocked, I have no reason to get my keys out. Footnote 59 So logically, the fact that we refer to legal obligations as reasons does not entail that they are in fact reasons—we could just be wrong about this. I say “logically,” but I doubt that helps much: while in any given case we could be confused or wrong about our reasons, the fact that people almost universally refer to legal obligations as reasons seems to call for a more robust explanation. We might call it an error theory. The explanation has three parts.
The first part of the explanation requires us to note a distinction between two different ways that the law could give reasons to its subjects. On the one hand, the law could give us reasons, as I say, derivatively: the existence of a legal obligation to ϕ is characteristically associated with a set of reasons to ϕ grounded in the downstream effects or consequences of the creation of the obligation. The examples are familiar: often there are sanctions for not ϕing, sometimes ϕ is a salient solution to a coordination problem, perhaps ϕ is required by some fair system of cooperation.Footnote 60 Reasons created in these ways are only derivatively given by the law because all the law does is, as David Enoch puts it, trigger some reason for action that could be triggered just as easily by a nonlegal act. Take the sanction case: the reason given by the sanction—that I want to avoid the pain, say, that would be caused by not-ϕing—could just as easily by triggered by a gunman as by the law. There is nothing distinctly legal about such a reason. That fact—that the reason triggered by the law could just as easily be triggered by something else—is the idea that the reason is given only derivatively.Footnote 61
Derivative reason-giving is contrasted with direct reason-giving. Here the idea is that the law gives reasons to its subjects not through any downstream effects of the directive but rather in some special way that is distinctive of the directive's imposition of a legal obligation. Where derivative reason-giving involves the law's triggering a normative consideration that could be triggered by something else, direct reason-giving does not. Instead it is the very act of the issuance of the directive itself that is supposed to ground the reason.Footnote 62 I think it is clear that the legal rationalist argues that the law claims to give reasons directly rather than derivatively.Footnote 63 The legal rationalist thinks that the fact that A is under an obligation to ϕ itself constitutes a reason to ϕ in a way that no other kind of fact could.
One might ask here what the distinction between direct and derivative reason-giving has to do with the more well-known distinction between content-independent and content-dependent reasons.Footnote 64 There are several different attempts in the literature to try to explain just what Hart had in mind with this latter distinction,Footnote 65 but as content-independence is not my interest here, I do not choose one or offer my own. Instead I just note that the basic idea behind Hart's distinction does seem to be the basic idea behind the one I am drawing here: on the one hand are reasons that have to do in some way with the merits of the action in question or its consequences (the content-dependent reasons or those given derivatively) and on the other are reasons that have to do instead with the bare fact that a legal directive was issued (the content-independent reasons or those given directly). I leave a full consideration of this question to another dayFootnote 66 and end the aside here.
Returning to the explanation of the explanatory reasons intution, then, we can ask what work this distinction between derivative and direct reason-giving is doing. The explanatory reasons intuition relies on the fact that people tend to refer to the legal obligations all the time in the explanation of their actions. The thought here is that much of the time, when people refer to their legal obligations as reasons, they are not referring to any reasons created directly by law but instead are referring elliptically to derivative reasons. Sanctions are again illustrative (and in fact play the role I have in mind most frequently). When I say, “I don't BitTorrent because it's a violation of the Copyright Act,” I am elliptically saying “I don't BitTorrent because I don't want to have to pay a fine in the event that I am caught violating the Copyright Act.” In my own casual conversational experience, for what it is worth, most people will give this second answer when pressed about the first.Footnote 67 Others, perhaps more civic-minded, will give an explanation that refers to some other of the reasons I mentioned above that are characteristically associated with legal obligations: they are worried about fairness or about upsetting the salience reasons that solve a coordination problem. (These people say something like, “yeah, but what if everyone did that?”)
Each of the various kinds of derivative reason-giving I mention above has in the past been considered and rejected as a complete account of the reasons given by the law; and surely none of them is sufficient on its own. But taken together, these explanations are much more powerful, because in many (or even most) cases, the existence of a legal obligation will trigger a reason in one of those derivative ways. Elliptical reference to derivative reasons created by the law (in one or another of the ways mentioned) can explain the lion's share of the phenomenon that gives rise to the explanatory reasons intuition.Footnote 68
Beyond these reasons that characteristically tend to be triggered by the imposition of a legal obligation, of course, there are often very good reasons to perform any given legal obligation in any particular case. Some trespasses are harmless, but many are not, and we have good reason not to harm others’ property. Some breaches of contract are efficient, but most are not, and we have good reason not to create unnecessary economic costs. Violation of many criminal prohibitions would be harmful to particular individuals. More broadly, in many cases legal obligations are imposed to engender support for some valuable regime, and some nonzero number of violations might lead to the erosion of that regime. All of these concerns are contingent, but in many cases they are the very reasons that were relied upon in the creation and imposition of the legal obligation. They are, in Razian terms, the dependent reasons. And when people refer to the law as a reason, they might, again, be elliptically referring to these dependent reasons in explaining or justifiying their actions.
That is, when I say, “I am ϕing because I am legally obligated to ϕ,” it seems plausible that in fact I am giving a shorthand for: “my legal obligation to ϕ was imposed for and is based in part on some set of valid dependent reasons to ϕ which justify my ϕing in this case.” While this might be rarer in normal, everyday conversation, I think that this is a very plausible explanation for pronouncements by judges and other legal officials to the effect that we ought to perform our legal obligations. These officials, remember, are elaborating the legal point of view as to our actions and so are espousing the view that our obligations are actually based on some valid dependent reasons and that these reasons justify our actions.Footnote 69 That is the second part of the explanation.Footnote 70
The final part of the explanation is related to the parallels between legal obligation and legal wrongness that I draw in Section II above. T.M. Scanlon's account of the relationship between wrongness and reasons suggests that wrongness can relate to reasons in two ways. First, an action is sometimes wrong in virtue of the fact that there are conclusive reasons not to perform it (consistent with the account of wrongness as verdictive that I discuss above). Scanlon also thinks that wrongness plays a second role, which “lies in shaping the way I should think about the decision I face, and in determining which other considerations I should take to be reasons.”Footnote 71 Here he explains in more detail:
In the most common kind of case in which wrongness plays a shaping role, two kinds of reasons are in play, corresponding to two kinds of “why?” questions. There are first-order reasons such as “it would hurt someone” or “you promised,” which explain why a certain action would be wrong. In addition, there are higher-order reasons, which might be offered in response to the question, “Why care about wrongness?” or “Why accept that as they way to think about what to do?” The conclusion that an act would be wrong claims that considerations of the first of these kinds count decisively against it. So, considered in this role, wrongness is not itself a reason-providing property.Footnote 72
Extending Scanlon's claim to legal obligation and then drawing on the connection between legal wrongness and legal obligation, we get something like this: the fact that I am under a legal obligation to ϕ is not itself a reason to ϕ but is instead a reason that explains why certain other considerations count decisively in favor of ϕing. To legal philosophers, the next step here will be obvious: the fact of the legal obligation is not, as Scanlon says, a “first-order reason.” Instead it is a second-order reason. Second-order reasons are reasons to deliberate about other reasons in particular ways. Raz's use of the notion (in particular under the guise of exclusionary reasons) is the most prominent here, but there are other examples that attempt to generalize Raz's view toward the idea of shaping that Scanlon seems to have in mind here.Footnote 73
This shaping provides an explanation (or an explaining away) of the fact that people often seem to treat legal obligations as reasons that fits well with the first two explanations I offered. Those first two explanations each claimed that people use “I had a legal obligation to ϕ” as a shorthand or elliptical formulation of something else—sometimes for “I don't want to be sanctioned for not ϕing” or some other characteristic reason triggered by the law's manipulation of the nonnormative environment, other times for “the legal obligation is based on valid reasons that justify my action.” Here, again, the idea is that “I had a legal obligation to ϕ” is elliptical for something more complicated. That more complicated thing this time is something like this: “there are good second-order reasons to deliberate according to the legal point of view according to which the reasons that made up the first two parts of the explanation count conclusively in favor of ϕ.” Between these three phenomena I think we can explain all the situations in which people seem to refer to their legal obligations as reasons and defuse the strength of the explanatory reasons intuition as a reason to accept the legal rationalist's argument about law and reasons.
VI. LEGAL NORMATIVITY WITHOUT LEGAL REASONS
Let us turn, then, to the normative reasons intuition. For a consideration to be a (normative) reason for some action is for that consideration to count in favor of the action and perhaps contribute to the case for the action's permissibility or justifiability or rationality. And the legal rationalist can surely say that the law plays this kind of role in our lives (or at least claims to do so): according to the law, that we are legally obligated to ϕ surely counts in favor of ϕing and improves the case for ϕing (or at least worsens the case for not ϕing). That is, the law surely claims to have a normative effect on our lives, and, reasons being the basic normative phenomenon, the most obvious explanation of that effect is in terms of the law claiming to give us reasons.
But we do not need to go this far. To say that legal obligations (or, perhaps more broadly, the law) is normative, we need not commit ourselves to the idea that each legal obligation is a reason for its performance. Instead, all we need to do is provide the resources to deny what we could call the no-normative-difference thesis,Footnote 74 according to which the fact that A has a legal obligation to ϕ makes no normative difference as to whether or not A ought to ϕ. As I note above, Coleman provides a helpful statement of what we need to explain here, under the guise of the practical difference thesis: “authoritative pronouncements must in principle be capable of making a practical difference: a difference, that is, in the structure or content of deliberation and action.”Footnote 75
The statement of the practical difference thesis presents the solution to our problem.Footnote 76 In order for our legal obligations (“authoritative pronouncements” in the official version) to make a practical difference—that is, to be normative—all we need is for them to make a difference in the structure or content of our deliberations or to make a difference with respect to what we have reason to do. But, and again this is a familiar point for legal philosophers, there are considerations that make a practical difference in our deliberations without themselves being reasons for action: second-order reasons. It seems that we have failed to appreciate the power of Raz's insight into the notion of a second-order reason. Raz's breakthrough was to see that a consideration can affect what an agent has reason to do in ways beyond being a reason for a particular action. A second-order reason can change what an agent to whom it applies has reason to do. But once we see that a consideration can affect what an agent has reason to do without being a reason for a particular action, we can abandon the idea that an obligation, which certainly affects what an agent has reason to do, must itself be a reason for that agent.
Presenting a full account here is outside my scope (and it is related to the points I take from Scanlon above, where I say the same thing), but the thought can be captured by modifying Raz's account. According to Raz, the fact that A is under an obligation to ϕ is a reason for A to ϕ as well as a second-order reason for A not to act on (at least) some other reasons not to ϕ. If my arguments above are sound, though, we need to drop from Raz's account the idea that the obligation is itself a reason to ϕ. And so we are left with the claim that it is a second-order (exclusionary) reason for A not to act on some other reasons not to ϕ.Footnote 77 Here is the crucial point: on its own, this is enough to rebut the no-normative-difference thesis. Suppose that, legal obligation aside, the reasons not to ϕ outweigh the reasons to ϕ. If we put the obligation into the picture, some of those reasons not to ϕ are excluded, and now the nonexcluded reasons to ϕ outweigh the reasons not to ϕ. So the fact that A is under the obligation made a normative difference as to what A ought to do.
A full account is required here, and it is outside my scope to provide it in this paper.Footnote 78 The important point is this: the desire to explain how law can make a practical difference can be satisfied without resorting to OAR's commitment to the idea that the fact that an agent is obligated to ϕ is a reason for her to ϕ. For my own part, I do not think that the exclusionary-reason account is the way to go, but I use it because it is familiar here.Footnote 79 The point is just that we can see that obligations can have the right kind of practical effect without having to think that obligations themselves provide reasons, as OAR requires. We do not need to resort to OAR to deny the no-normative-difference thesis and to account for the normative reasons intuition. And according to the arguments in Section V, we also do not need to resort to OAR to account for the explanatory reasons intuition. The legal rationalist's argument from OAR to the thought that the law claims to give reasons to its subjects fails.
VII. LEGAL REASONS WITHOUT LEGAL OBLIGATION?
Perhaps, however, the legal rationalist's thought that the law claims to give reasons for action does not rely on OAR. I can think of one alternative route that the legal rationalist might take. The thought would be that the law's claim is that a legal directive, “do not ϕ,” in effect does two separate things. Namely, it (1) creates an obligation not to ϕ and (2) creates a reason not to ϕ. I have to confess that while this seems like a logical possibility, I am not at all sure how it would work.
Consider that Jules Coleman's analysis of the moral semantics of law, which he takes to be a spelling-out of the notion of the legal point of view,Footnote 80 suggests that the law's claim is just that its directive, “do not ϕ,” can be redescribed as a (moral) obligation not to ϕ. According to the suggestion on the table, however, the law's claim to give reasons for actions to its subjects must be understood as being the claim that it is the fact that the law imposed the directive that is the reason not to ϕ. If the arguments I am proposing against OAR are sound, though, it is impossible for the fact that an agent is under an obligation to ϕ to be a reason for that agent to ϕ. So the proposal is that the law claims that one single fact, the fact that it issued a directive “do not ϕ,” is both a reason not to ϕ and an obligation not to ϕ (which cannot be a reason not to ϕ). So to say that the law claims both things is necessarily to say that the law is making inconsistent claims. This is not a plausible suggestion.
What if we take a step back and suggest that the thought that the law gives reasons is distinct from the law's claim to impose obligations, such that the law could succeed in giving a reason even without imposing an obligation? This proposal fares no better, because setting aside the law's claim to impose obligations requires setting aside the law's claim to be a legitimate authority and reduces its directives to the orders of a mere de facto authority or a gunman. And such orders can create reasons only derivatively, through triggering. While it is clear that the law does in fact create reasons in this way, it should be equally clear that this is not the sort of reason-giving that interests legal philosophers.Footnote 81
VIII. WHERE TO NOW?
I want to close by making some remarks on the implications of my argument, because while I do maintain that the legal rationalist is wrong to say that the law claims to give reasons, in some important respects, my conclusions here are modest: I maintain that our legal obligations do make a practical difference in how we ought to act and that the law is normative and part of the task of explaining the nature of law is explaining that normativity. So are we really making any progress? Sure. Three points: first, and most generally, if, as I say above, a part of the task of explaining the nature of law is explaining the law's normativity, then we had better be sure that we are talking about the right thing under the heading of “law's normativity.” So it is good to know that explaining law's normativity means not explaining how the law creates reasons but rather explaining how the law shapes the reasons that we have or perhaps creates second-order reasons. We can put the point differently by returning to the idea that the law claims to be a legitimate authority: my conclusion here tells us that understanding and evaluating the law's claim to authority is not a matter of understanding or evaluating any claim of the law to give reasons to its subjects, since without OAR the argument for the law claiming to give reasons fails.
Second, and more specifically, Scott Shapiro's planning theory of law, which has, I think, a solid claim on being the most sophisticated and best-worked-out postivist theory of the nature of law out there, is committed to my view. According to Shapiro, the fundamental normative structure of the legal system rests on plans. And according to Bratman, plans are not reasons. Indeed, Bratman's theory of planning developed in part as a way to explain how it is that intentions and decisions can make a practical difference in our lives while maintaining that they are not themselves reasons for carrying themselves out. So if I am wrong, OAR is correct, and legal obligations are reasons for performance, that is a reason to doubt Shapiro's theory.
Third, and even more specifically, recently John Gardner and Ernest Weinrib, among others, have engaged in a debate about how to understand the relationship between A's tort law obligation to, say, take due care, and the remedial obligation that arises when A fails to take due care and injures B.Footnote 82 If we want to understand this relationship between these two related obligations in terms of reasons, as both Gardner and Weinrib sometimes suggest, we had better be sure that we have a clear idea of the relationship between reasons and obligations in the first place.
