I. PLANS

To repeat, for Shapiro, laws are plans. Plans exercise an important normative force upon those adopting them. Having made the plan to cook dinner, the plan “generates rational pressure to act accordingly.”Footnote ¹³ I ought not to reconsider my choice unless there is a compelling reason.Footnote ¹⁴ If I reconsider in the absence of a compelling reason, I am irrational.

Notice that this normative force is not moral. Assume that I have adopted a plan to torture a kitten. As I am about to begin torturing, I suddenly find myself reconsidering my project, not because I have recognized its immorality (which is a compelling reason) but for no real reason at all. This is irrational, even though my irrationality happens to have good moral consequences.

It is worth putting this normative constraint in more intuitive (if more vague) terms. There is a certain mystery in how, once one has decided to do something, action follows. It is tempting to view performance of an intended action as irrational or compulsive—a robotic doing that simply happens after I have made the decision to act. But the fact that action follows upon decision should instead be understood in terms of responsiveness to rational constraints upon one's agency. Rationality can recommend that one not consider one's reasons for action.

Personal plans exercise other normative constraints on us: having made the plan to cook dinner, I am rationally obligated to generate subplans at the appropriate time concerning how my plan is to be effectuated. If I adopt the plan without creating subplans, I am once again irrational.Footnote ¹⁵ As Shapiro puts it, plans are partial (they leave open details, such as the means to bring about the plan),Footnote ¹⁶composite (they can consist of parts—subplans—that are themselves plans),Footnote ¹⁷ and nested (the creation of subplans is in the context of and rationally constrained by previous plans).Footnote ¹⁸

Following Michael Bratman, Shapiro argues that a shared plan can put rational constraints on a group that adopts it that are similar to those constraints imposed by individual plans.Footnote ¹⁹ Rationality constrains the reconsideration of the shared plan by adoptees and recommends the creation of nested subplans. What is more, a shared plan can rationally recommend cooperation among adoptees, including the acceptance of subplans created by other members of the group.Footnote ²⁰

It is this notion of a shared plan that stands at the heart of Shapiro's theory of law. The basis of a legal system is a master plan adopted by the members of a planning organization composed of officials and perhaps of other members of the community. This master plan is intended to overcome the fact that the community faces numerous and serious moral problems whose solutions are complex, contentious, or arbitrary (which is not to say that the master plan succeeds in this purpose or that all those who accept it do so to effectuate that purpose). Once this master plan is adopted, it asserts a rational force upon its adoptees.

II. NORMS

Let us now consider plans’ ontological status a little more. Shapiro describes plans as norms. The following is Shapiro's definition of the term “norm”:

Norms do not include merely directives, which tell agents to perform or not to perform certain acts,Footnote ²² but also permissions and authorizations.Footnote ²³ Norms are not the same as the sentences that describe them: “‘No vehicles in the park’ is a sentence that describes the no-vehicles-in-the-park norm and hence cannot be the norm itself.”Footnote ²⁴ Shapiro defines a “rule” as a type of norm, namely one that offers a generalized—as opposed to an individualized or particular—standard of conduct.Footnote ²⁵

Shapiro insists that norms, including rules and plans, are “abstract propositional entities”Footnote ²⁶ that do not exist within space and time.Footnote ²⁷ One reason he comes to this conclusion is that they “have a potentially infinite domain” that cannot be satisfied by the physical, psychological, or social.Footnote ²⁸ An unbounded number of actions violate the Securities Exchange Act—more than all the would-be defrauders could ever generate or even imagine. It is for similar reasons that many are inclined to think that games such as chess are abstract objects.Footnote ²⁹ An unbounded number of matches fall within the scope of the game of chess, including matches longer than a human life.Footnote ³⁰

Another reason to treat norms as abstract objects is that they have no particular location in space. Just as it makes no sense to ask where chess is, so there is no place to which we can assign a norm. The Securities Exchange Act is not located at the piece of paper Roosevelt signed, nor is it located in the heads or minds of the members of Congress who passed it or of the officials who enforce it. Indeed, if norms were somehow in people, there would be as many plans as there are people who have them, making it impossible for people to share plans. Just as two people can believe the same thing—a fact that motivates us to assume the existence of a descriptive proposition that is the object of their belief—so two people can adopt the same plan. The psychological act of adopting a plan, Shapiro therefore argues, must be different from the plan itself.Footnote ³¹ Intentions “take plans as their object.”Footnote ³² By adopting a plan, a person “places herself under the governance of a norm,”Footnote ³³ and two people can place themselves under the governance of the same norm.

Shapiro speaks of norms as existing not merely outside space but also outside time. The notion that abstract objects have no temporal parameters has sometimes been challenged. Although, as we have seen, chess can be considered an abstract object, it sounds odd to some to say that it exists outside time—that it was present when there were dinosaurs, or at least no less present than it is now. Chess, some argue, came into being when it was first thought up and will end when it is forgotten.Footnote ³⁴ It would no doubt sound odd to these people that norms existed before there were human beings.

I think Shapiro is right, however, to insist that norms are timeless. The view that abstract objects come into being with our thinking of them confuses the psychological and the propositional. When I think that grass is green, there is something that I think and that others can think as well, namely, that-grass-is-green. If the content of my thought were created in my act of thinking, not only would others be unable to think what I do, my thought would be unconstrained. I could not aim at anything in my thought, because the goal of my thought would be created, and so reached, simply in thinking. The same point is true of norms. When I think that no vehicles should be in the park, there is the object of that thought, the no-vehicles-in-the-park norm, to which my thought is directed and to which others can also direct their thoughts.

Shapiro speaks of norms as propositional entities that “purport to tell you what you ought to do and what is desirable, good, or acceptable, but whether they actually succeed at this task is another matter entirely.”Footnote ³⁵ “A norm that tells you to do something that you shouldn't do is an invalid norm. It is a bad norm, not a nonnorm.”Footnote ³⁶ Although Shapiro usually uses the word “valid” to refer to legal validity, as one would expect in a book on the philosophy of law, here he is using the term in a general sense that can apply to any norm. A norm is valid if it actually succeeds in what it purports to do. It need not succeed to be a norm. Just because there is a no-vehicles-in-the-park norm does not mean that, in fact, vehicles shouldn't be in the park, any more than the existence of the proposition that-there-are-no-vehicles-in-the-park means that the proposition is true.

Shapiro does not offer a general account of what it takes for a norm to be valid in this sense. But we know that there are such norms. Correct logical and moral norms, for example, are valid. What they say we ought to do, we really ought to do. Indeed, Shapiro describes correct logical and moral norms as having “intrinsic validity,” meaning, I take it, that they are valid no matter what we do. In contrast, plans become valid by being adopted.Footnote ³⁷ Only when I adopt a plan does the plan succeed in what it purports to do, because only then is it the case that I ought to do what the plan says I ought to do.

III. CAN NORMS BE CREATED?

Although plans become valid upon their adoption, the fact remains that for Shapiro, plans are abstract propositional objects and so must exist prior to their adoption. For this reason, one would expect Shapiro to say not that we create plans by adopting them but that we make them valid by doing so. At times he appears to put things this way. He says, for example, that by adopting a plan, someone “places herself under the governance of a norm,”Footnote ³⁸ suggesting that the norm was there already. But far more often he speaks of plans as generated,Footnote ³⁹ formed,Footnote ⁴⁰ created,Footnote ⁴¹ or made.Footnote ⁴² Plans, as he puts it, are positive norms—they “are created through adoption and sustained through acceptance.”Footnote ⁴³

Shapiro also speaks of social rules as created. A social rule, he argues, is a rule that “exists and has the content that it does because, and only because, of certain social facts.”Footnote ⁴⁴ In particular, a social rule “is brought into existence simply by virtue of being accepted and practiced by members of a group.”Footnote ⁴⁵ Indeed, under Shapiro's planning theory of law, laws are social rules: the master plan and subplans of a legal system are generated by social facts, namely, their acceptance by officials.

But how can this be? Things outside space and time cannot be created, for creation is an event in time. Furthermore, abstract objects cannot enter into causal relations with things in the material and psychological worlds, and creation is usually understood as causal in nature.

It appears that when Shapiro says that someone creates a plan, what he really means is that she makes a plan valid, in the sense that she now ought to do what the plan says she ought to do. Likewise, what he calls a “social rule” is not a rule created by social facts but a rule that is made valid by the existence of such facts. In speaking of plans and social rules as created in those circumstances in which they are made valid, Shapiro appears to ignore his own admonition not to conflate the existence of norms with their validity.

Such statements are not a serious problem for his position, however. One solution is to simply replace passages in which Shapiro speaks of plans and social rules being created with language about their being made valid. To be sure, we have to conclude that the Securities Exchange Act as a plan stands outside space and time—that it was no less existent when dinosaurs roamed the earth than when Roosevelt signed a piece of paper in 1934. Nevertheless, the Securities Exchange Act became valid in 1934, because that is when it was adopted by American officials.

An alternative solution—the one I prefer—is to reject the view that a plan is a norm simpliciter and understand it as a norm plus the psychological fact that it has been adopted. Until adoption, it is not a plan. It is just a norm. Under this reading, when Shapiro says that a plan is created, what he means is that we transform norms into plans by adopting them. In this way, we “create” plans, even though we do not create the norms out of which they are constituted. Under this reading, the Securities Exchange Act as a norm would exist when dinosaurs roamed the earth but as a plan would exist only in 1934. The same points would be true of social rules. A social rule is not just a rule (that is, a norm with a generalized standard of conduct) but is instead a rule combined with the social facts that make the rule valid.

An advantage of the second reading is that it can explain how norms can perform dual purposes. Consider the norm described by the sentence, “Do not intentionally cause unjustified harm.” By itself, this can be considered a moral norm that is valid no matter what I do. But I can also turn the norm into a plan by adopting it. Having adopted it, I have added to the preexisting reasons to follow the norm new Bratmanian reasons: I should not reconsider conformity to the norm in the absence of compelling reasons, should adopt appropriate subplans, and so on.

IV. DO NORMS EXIST ONLY IF THEY ARE VALID?

A similar puzzle about the relationship between the existence of norms and their validity comes up when Shapiro speaks of logical and moral norms. Consider, for example, norms 1 to 4 described by the following sentences:

1. 1) From the proposition A or B, infer the proposition A.

2. 2) From the proposition A and B, infer the proposition A.

3. 3) Kill everyone you see.

4. 4) Do not intentionally cause unjustified harm.

Norms 2 and 4 are valid. (Norm 2 is a valid logical norm and norm 4 is a valid moral norm.) Norms 1 and 3 are not. When Shapiro speaks of norms as propositional objects that merely “purport to tell you what you ought to do and what is desirable, good, or acceptable,”Footnote ⁴⁶ he means that norms 1 to 4 all exist. Norms 1 and 3 are no less existent by virtue of being invalid.

At times, however, Shapiro says that logical and moral norms “exist simply by virtue of their ultimate validity.”Footnote ⁴⁷ That suggests that invalid norms (like 1 and 3) do not exist. If the absence of a reason for acting in accordance with a norm meant that the norm as a propositional object did not exist, we would apparently not be able to have immoral or illogical attitudes, because there would be no entities with which we could establish a relationship in order to bring those attitudes into being.

Again, I do not think that this puzzle is a serious problem for Shapiro's argument. By “logical and moral norms,” Shapiro clearly wants to refer only to norms, like 2 and 4, that are in fact valid. That they are valid is in part what defines them as logical and moral norms in the relevant sense. Thus, the terms “logical norm” and “moral norm” are, for Shapiro, like “plan” or “social rule.” Each term takes into account something more than the bare existence of the norm as a propositional object, something that makes the norm valid. In the case of plans, this something more (the validity-maker of the norm) is adoption. In the case of social rules, it is a set of social facts. In the case of logical and moral norms, however, the validity-makers are not facts about human beings, which is why Shapiro describes their validity as intrinsic.Footnote ⁴⁸

V. CAN REASONS FOR ACTION BE CREATED?

To repeat, I do not think that the infelicities identified above are a genuine problem for Shapiro's position. But there is one more issue to consider. When I adopt a plan, is it really true that I create reasons for action (that is, reasons not to reconsider the plan, to adopt appropriate subplans, and so on)? Are there new reasons that I lacked previously? Relying upon Mark Schroeder's work,Footnote ⁴⁹ David Enoch draws a helpful distinction between robust reason-giving and triggering reason-giving.Footnote ⁵⁰ In triggering reason-giving, there is a change in the nonnormative circumstances, which triggers a dormant reason for action.Footnote ⁵¹ The dormant reason is often conditional. Creating the event E makes it such that one ought to X, because E triggers a preexisting reason of the form if E then X. In robust reason-giving, in contrast, there is no preexisting reason for action that is triggered by E. Instead, E itself constitutes or generates a new reason to X. E has changed the fundamental normative framework rather than simply triggering a reason for action that was there all along.

A possible example of robust reason-giving is divine-command theory. For the divine-command theorist, God is responsible for morality, in the sense that his commands create (or are) moral reasons for actions. In contrast, if God's commands are only triggering reasons, God is not responsible for morality after all, for his commands have moral consequences only in the light of preexisting moral reasons for action (which God did not create), for example, a reason that takes the form if God says to X, then X.Footnote ⁵²

In the end, Enoch doubts that robust reason-giving in the above sense is possible. The idea that E robustly gives one a reason to X leaves the relationship between E and the reason to X utterly mysterious. If E makes it such that someone has a reason to X, there must be some explanation of why that is so, and the explanation will be in reference to a preexisting reason for action. Because he finds it so mysterious, Enoch offers an alternative, nonmysterious account of robust reason-giving, under which it is a form of triggering reason-giving.Footnote ⁵³ But I am interested here in the distinction between triggering reason-giving and robust reason-giving in the mysterious sense.

Consider promising. It is natural to say that in making a promise to take you to the zoo, I have created a moral reason, if only a prima facie reason, to take you to the zoo. One way to read this statement is that I have created a triggering reason. I have brought into being an event (my psychosocial act of promising) that triggers a dormant reason for action that preexisted my promise.

There are many accounts of the nature of this preexisting reason for action. Some argue that it is based in the promisor's grant of authority over his actions to the promisee.Footnote ⁵⁴ For others, the antecedent of the reason must include more than a bare promise. One ought to X having promised to X only if other conditions are met, such as the promisee's expectation of performance by the promisor. It is the duty not to frustrate the expectation of performance that stands behind promissory obligations.Footnote ⁵⁵ These debates are not our concern. Under all of these accounts, the reason-giving of promising (or, if one prefers, of promising with an expectation of performance by the promisee) is not robust in the relevant sense. The connection between a promise to X and a reason to X is explained in the light of moral reasons for action that preceded the promise.

But is it possible that by promising to take you to the zoo, I have robustly created a reason to take you to the zoo? To do so I would have to be, in some sense, like the God of the divine-command theorist. I would have to have the power to create and extinguish fundamental moral reasons for action at will. And such a theory would deny us an explanation of why, having promised to take you to the zoo, I have a moral reason to take you to the zoo. It would leave the relationship between promising and reasons for action mysterious.

With this distinction between robust and triggering reason-giving in mind, what is Shapiro's position on planning? Does the acceptance of a plan robustly create a reason for action or merely trigger a preexisting reason? I am confident that Shapiro would choose the triggering approach. After all, he does not leave the relationship between accepting a plan and having Bratmanian reasons for action unexplained. Having accepted a plan to X, I have reasons not to reconsider the plan, reasons to create subplans, and so on, because of the “norms of instrumental rationality.”Footnote ⁵⁶ Our power to bind ourselves by adopting plans exists only in the light of these background reasons for action. Planning triggers these reasons for action; it does not robustly create reasons for action.

VI. SHAPIRO'S CATEGORY-MISTAKE ARGUMENT AGAINST HART

With Shapiro's ontology clarified, we can now consider his criticism that Hart's theory of law suffers from a “category mistake.” The foundation of Hart's theory of law is his practice theory of social rules. For a social rule to exist, there must be more than patterns of conduct among participants in the practice. They must also take what Hart calls the “internal point of view”: participants must be motivated by a commitment (although, for Hart, not necessarily a moral commitment) to conformity, such that deviation will be met with resistance and criticism.Footnote ⁵⁷ This criticism will generally manifest itself in what Hart calls “internal statements”Footnote ⁵⁸ framed in normative language like “You ought to do this” or “That is wrong.”Footnote ⁵⁹ Hart himself accepts the description of his account of social rules as “the practice theory”:

As Shapiro puts it, for Hart, social rules are social practices.Footnote ⁶¹

Relying on this theory of social rules, Hart argues that a legal system comes into being due to the existence of “a rule of recognition,” that is, a social rule constituted by a practice among officials of enforcing norms only if they have satisfied certain ultimate criteria.Footnote ⁶² The existence of the rule of recognition, like the existence of any other social rule, is a matter of social fact.Footnote ⁶³ For example, the legal system of the ancien régime existed by virtue of a practice in which French officials would enforce norms only if they were identified as enforceable by the king of France. Taking the internal point of view, French officials would resist deviation from that pattern and would criticize the deviant with normative statements, such as “You ought to do this” or “That is wrong.”

But Hart's theory of rules, Shapiro argues, confuses orders of being:

Indeed, the rule of recognition to which officials appeal cannot be understood as a practice, for the rule has a potentially infinite domain. A boundless number of norms fall under the rule, more than could ever be instantiated in an actual practice.

Every philosopher of law I am aware of who has discussed this argument has dismissed it as uncharitable to Hart.Footnote ⁶⁵ It is true that Hart speaks of the existence of a social rule as a matter of social fact. But that is from the external point of view. The internal point of view, in contrast, “is that of a participant in such practice who accepts the rules as guides to conduct and as standards of criticism.”Footnote ⁶⁶ When adopting the internal point of view, an abstract object is employed.Footnote ⁶⁷ As Stefan Sciaraffa puts it:

For example, in the ancien régime, French officials followed an abstract object according to which, roughly, they ought to enforce only norms identified by the king of France.

But does Shapiro really ignore the role of the internal point of view in Hart's theory of law? Consider a very simple scenario. A judge in the ancien régime is deciding whether to enforce a norm against a defendant. Assume the judge employs the abstract object enforce-only-norms-identified-by-the-king-of-France to guide his action. The question remains why he is conforming his behavior to that abstract object. After all, an infinite number of such objects exist. Why not conform his behavior to an object that identifies ancient Roman law or the law of a fictional country like the kingdom of Gondor in Tolkien's Lord of the Rings?

The question, in short, is not whether the official uses an abstract object to guide his action but whether he can justify his action by reference to the object. It is clear that Hart thinks officials use legal rules not merely “as guides to the conduct of social life” but also as justifications: “as the basis for claims, demands, admissions, criticism, or punishment.”Footnote ⁶⁹ A judge takes “the breach of the rule as his reason and justification for punishing the offender.”Footnote ⁷⁰ For officials, “the violation of a rule is not merely a basis for the prediction that a hostile reaction will follow but a reason for hostility.”Footnote ⁷¹

Indeed, one of the main goals of The Concept of Law is to explain this justifying aspect of rules in the face of skeptical arguments that it assumes “there is something external, some invisible part of the fabric of the universe guiding and controlling us.”Footnote ⁷² Hart hopes to make this talk of rules as justifications for action metaphysically respectable. If all he were trying to explain was an abstract object that guides us, an object that does not give us any reason to be so guided, it is hard to see why philosophers of law thought they were so mysterious. To question their existence would, it seems, mean questioning our very ability to think about standards for our behavior. Although some philosophers worry about the existence of rules in this sense,Footnote ⁷³ such worries are not central to the philosophy of law. The mysterious phenomenon that Hart wanted to explain was not—or not only—the guiding function of a rule but also “our own powerful ‘feelings’ of compulsion to behave in accordance with the rule and to act against those who do not.”Footnote ⁷⁴

Hart's problem, therefore, is this: the French judge justifies his action by reference to a rule. What can that rule be? In particular, can it be an abstract object? If the rule of recognition were like the norms of logic or morality, he would be able to justify his actions by reference to it. But that is because the norms of logic and morality are, as Shapiro puts it, intrinsically valid—we have a reason to conform to such objects no matter what any of us may do. And it is not true that everyone always has a reason to abide by the rule of recognition. The rule of recognition is a social rule: it is binding only upon participants in the practice.

If the rule of recognition as an abstract object cannot do the justification, what does? As Hart sometimes describes it, officials consider the practice as a social fact in the light of background reasons for actions (or purported reasons) that exist independently of the practice. Or, to use Enoch's language, they point to the practice as a trigger. Some officials treat the practice as triggering nonmoral reasons for action, such as “calculations of long-term interest; . . . an unreflecting inherited or traditional attitude; or the mere wish to do as others do.”Footnote ⁷⁵ Other officials conform because they “conceive of themselves as morally bound to do so.”Footnote ⁷⁶ For example, they might think that by willingly joining the practice, they implicitly promised to conform, and that promises ought, at least prima facie, to be abided by. Or they might conform because deviance would upset others’ settled expectations, and settled expectations should not be upset absent a compelling reason. But whatever the justification, its form is the same: officials have a reason to conform because the practice as a social fact (plus other contingent psychological and social facts) triggers reasons for action that exist independently of the practice.

We can, therefore, interpret Shapiro's category-mistake argument against Hart as follows. When officials take the rule of recognition as their reason for action, what is that rule? Not the propositional object guiding their action, for that cannot justify conformity to the object. Not the preexisting reasons for action triggered by the rule of recognition as a social fact, for these do not apply solely to the officials. Morality and prudence bind everyone, everywhere. The rule of recognition is a social rule; it is justificatorily relevant only to participants in the legal system. The justifying “rule” must therefore be the social practice itself.

It is true that Hart generally says officials taking the internal point of view presuppose rather than describe the existence of the rule of recognition as a social fact.Footnote ⁷⁷ But on one reading of the internal point of view, it is not the case that this social fact must be presupposed rather than explicitly referred to. Hart himself admitted that “[i]f the truth of this presupposition were doubted, it could be established by reference to actual practice: to the way in which courts identify what is to count as law, and to the general acceptance of or acquiescence in these identifications.”Footnote ⁷⁸ On this reading of the internal point of view, although our French judge would commonly stop his justification with the rule that he should enforce only norms identified by the king of France, thereby presupposing the practice and the background reasons for action that justify his being guided by the rule, he would still be inhabiting the internal point of view if he made the role of the social practice and the background reasons it triggers explicit. Indeed, a complete justification of his actions would have to appeal to these reasons for action.

Justification of official action is very different in Shapiro's planning theory. In a well-functioning legal system, officials do not—indeed should not—refer to their social practice and the background reasons for action it triggers. Instrumental rationality recommends that an adopted plan not be reconsidered except for compelling reasons. Thus, having adopted a master plan, legal officials show their obedience to instrumental rationality by not reconsidering the plan—by stopping justification of their actions with the plan.

Seen in this light, Shapiro's category-mistake argument turns out to be a powerful objection to Hart's theory of law. In Hart's theory, official justification of action ultimately refers to the wrong sorts of entities: social practices and background reasons for action, rather than social rules. Why does Hart make this mistake? Because he fails to recognize that law is a planning activity. He assumes that legal officials justify their actions the way that people do in the absence of plans. Without a plan, nothing is settled, and one must therefore consider background reasons for action and the contingent facts that trigger them.Footnote ⁷⁹

So understood, Shapiro's category-mistake argument identifies an important element of an adequate theory of law: it must account for the phenomenology of legal justification, in which officials do not offer social practices as justifications of their actions but instead terminate their justification with legal rules. To the extent that they justify their actions through social practices and background reasons for action, legal rules drop out of consideration.

For example, imagine that our ancien régime judge, in justifying why he ordered someone to prison, said the following:

Furthermore, assume that upon encountering another judge who has decided to enforce norms on the basis of whether they satisfy the standards of Plato's Republic, our judge said something like, “That is wrong, because it frustrates people's settled expectations. You ought to enforce only norms identified by the king of France.” Despite the fact that our judge has apparently adopted the internal point of view, he sounds like someone who denies the existence of legal rules. If a legal system made up of such judges satisfies Hart's theory of law, Hart's theory is inadequate, because it fails to account for the phenomenology of legal justification.

Because officials’ justification of their action terminates in rules rather than practices and background reasons for action, revolutions will look like changes in the normative order. In 1788, officials in France ended justification of their actions with the rule that only norms identified by the king of France should be enforced. In 1791, they ended with a very different and apparently incommensurable rule: that only norms identified by the National Assembly (or the people of France it represents) should be enforced. Under Hart's theory, French officials in 1788 and in 1791 would experience no normative discontinuity. After all, the norms they used to justify their actions would remain the same. Only the triggering social facts would have changed. For example, a French judge who enforced the king of France's commands because he thought people's settled expectations should not be upset would find himself appealing to the same norm in 1791. He would still be trying not to upset people's settled expectations, although what their expectations were would, of course, have changed dramatically.

Notice that one way of explaining the incommensurability of legal justification between legal systems is to insist that French officials in 1791 simply had fundamental reasons for action that were different from those they had in 1788. French officials would not merely appear to have shifted to a different and incommensurable normative order; they really would have shifted to such an order, in much the same way that God's commands change fundamental reasons for action for the divine-command theorist.

But like robust reason-giving, the view that French officials inhabited a different normative order after the French Revolution leaves us without an explanation of why their fundamental reasons for action were different. A virtue of Shapiro's planning theory of law—one not emphasized even by Shapiro himself—is that it is able to explain the incommensurability of legal justification in different legal systems in a manner that is compatible with our all actually inhabiting the same normative order. Because different legal systems involve the adoption of different master plans, officials in different systems will stop justification at different plans.

Indeed, Shapiro's planning theory appears to account precisely for the phenomenology of legal justification during transitions between legal systems, for it also explains the point at which officials will give up looking to plans and revert to all-things-considered reasoning on the basis of background reasons for action and the social facts that trigger them. At some point in 1789 (around the time of the Tennis Court Oath), French officials began to find compelling reasons to abandon the master plan that looked to the king of France. They no longer stopped justification of their actions at the rule that his word should be obeyed. Situated in the middle of a revolution, they had to justify their actions through background reasons for action and triggering social facts. They began to ask themselves: “Can a stable practice be found? Do morality or prudence recommend conforming to it? Do morality or prudence recommend the creation of a different practice?” By 1791, however, French officials had adopted a new master plan of looking to the National Assembly. At that point, their justifications ended with the new plan, making it appear as if they were in a different and incommensurable normative order from the one they inhabited in 1788.

VII. COORDINATION CONVENTIONS

As I interpret Shapiro's category-mistake argument, the problem with Hart's theory of law is that officials would justify their conformity by reference to the wrong sorts of entities. Instead of terminating their justification with an abstract object, they would ultimately appeal to a social practice that triggered background reasons for action.

But the category-mistake argument is not the only criticism of Hart's theory that can be found in Legality. Shapiro offers a second criticism directed at a different version of Hart's theory that, Shapiro claims, does not suffer from a category mistake. Under this reformulated theory, social rules are not equated with social practices, but “social practices necessarily generate social rules in their wake.”Footnote ⁸⁰ Although Shapiro is not entirely clear why the reformulated theory does not suffer from a category mistake, it must be that under the reformulated theory, officials justify their conformity by appeal to a social rule—understood as an abstract object—rather than to their practice and background reasons for action.

The problem with this reformulated theory, Shapiro argues, is that Hartian social practices do not necessarily generate social rules. That participants in a practice feel that conformity is justified and criticize deviation does not mean that a social rule is created:

Let me reformulate this criticism in the light of my clarification of Shapiro's legal ontology. By saying that Hartian social practices do not necessarily create social rules, Shapiro means that the existence of a Hartian social practice is not necessarily the validity-maker of the norms to which participants in the practice appeal. Why? The norms might have no triggering conditions or might be triggered by something other than the existence of the practice. People have reasons not to smoke, after all, whether or not they are part of a social practice.Footnote ⁸²

A theory of law that Shapiro thinks does not suffer from this second criticism is one that bases legal systems in coordination conventions. Unlike a Hartian social practice, the existence of a coordination convention necessarily creates a social rule. As Shapiro puts it:

The convention of driving on the right necessarily creates a social rule, in the sense that it necessarily makes valid the drive-on-the-right norm to which participants in the convention appeal when justifying conformity.

Of course, Shapiro thinks that the coordination theory of law is defective for other reasons.Footnote ⁸⁴ For example, the solution to a coordination convention is arbitrary, and officials are usually not indifferent to the content of the rule of recognition. Few American officials would be willing to go along with a constitution provided that other American officials did the same.

The coordination theory of law has another problem, however, one not noticed by Shapiro. It is vulnerable to the category-mistake argument. Shapiro apparently thinks the coordination theory is not susceptible to the category-mistake argument because officials would justify their conformity by appeal to an abstract object rather than the convention as a social fact. But the coordination theory lacks any account of why officials would stop their justification at an abstract object rather than looking to the convention and background reasons for action. Consider, for example, the coordination convention of driving on the right. Why should participants stop their justification at a drive-on-the-right norm? Why not argue that the reason one should drive on the right is that the social fact that everyone else drives on the right, combined with background reasons for action such as morality and prudence, recommends that one do the same?

Because the coordination theory of law provides officials with no reason to stop justification at an abstract object, it is contrary to the phenomenology of legal justification. Under a coordination theory, officials would experience no normative incommensurability as they shifted from one legal system to another. French officials in 1788 and in 1791 would justify their actions in the same way: by reference to social facts (about coordination conventions) triggering background reasons for action. The norms of morality and prudence to which they appealed would be the same before and after the revolution.

To be sure, a coordination theory of law could accommodate the phenomenology of legal justification if it insisted that when officials recognized they were in a coordination convention, they adopted a plan of conformity. Having adopted a plan, they would terminate their justification with the plan, instead of moving on to the convention and background reasons for action. But with such an addition, we would have returned to Shapiro's theory of law.

For the same reason that the coordination theory of law is vulnerable to the category-mistake argument, Shapiro should have denied that Hart's theory can be reformulated to make it immune to the argument. Hart's theory, as Shapiro has described it, lacks an explanation of why officials would terminate their justification with abstract objects rather than practices and background reasons for action.

VIII. HART'S EXPRESSIVE THEORY OF LEGAL RULES

Up to this point, I have defended Shapiro's category-mistake argument. Although widely dismissed, I think it identifies an important requirement for an adequate theory of law. Such a theory must account for the fact that officials justify their actions by reference to legal rules rather than pointing to social practices triggering background reasons for action. For Shapiro, the rules to which officials appeal are plans. Because plans, as abstract propositional objects, are not intrinsically valid, in ending their justification with them, officials would appear to have left their action unjustified. But that is precisely the point. Rationality dictates that plans, when adopted, should not be reconsidered in the absence of a compelling reason.

Although I am sympathetic to Shapiro's category-mistake argument against Hart, I would like to end with a suggestion that the argument, as it stands, is incomplete. Careful attention to Hart's idea of the internal point of view shows that he, like Shapiro, tried to account for the phenomenology of legal justification.

Previously, I assumed that it is not essential to the internal point of view as Hart understands it that officials presuppose rather than explicitly appeal to the existence of the rule of recognition as a social fact when justifying their actions and criticizing deviants. An official could adopt the internal point of view even if he points to official practice and the background reasons for action that the practice triggers. But Hart occasionally argues that officials must presuppose the existence of the rule of recognition as a social fact for the internal point of view to arise.Footnote ⁸⁵ To the extent that officials’ justification refers to the practice as a social fact, they will have abandoned the internal point of view within which legal rules are perceived to exist. In other words, Hart also hopes to explain the phenomenology of legal justification. He, too, wishes to explain why legal justification ends with legal rules.

But how can Hart accomplish this feat? Shapiro uses planning to explain why official justification would end with an abstract object. But for Hart, law is not essentially a planning activity. The question remains, therefore, why Hart thinks official justification can end with an abstract object. In ending justification with an abstract object, officials appear to be treating the object as intrinsically valid, as if appeal to the object is sufficient to justify conformity. But this seems false. Conformity is actually justified because the practice as a social fact triggers reasons for action that exist independently of the practice.

It what follows I would like to offer a very brief account of what I take to be Hart's explanation of the phenomenology of legal justification, an explanation that is very different from Shapiro's. Shapiro explanation seeks to show why the fact that officials terminate their justification in an abstract object is itself justified by instrumental rationality. He does so by arguing that the abstract object is a plan. Hart, in contrast, does not treat the phenomenology of legal justification as rationally justified.

As Shapiro recognizes, there is a connection between Hart's conception of the internal point of view (and the “internal legal statements” that express that point of view), on the one hand, and noncognitivist or expressivist theories of evaluative language and judgment, on the other.Footnote ⁸⁶ I think Hart relied upon expressivism to explain why legal officials justify their actions by reference to legal rules rather than social practices. (I assume here that Hart held an emotivist account of evaluative sentences and judgments, rather than another form of expressivism, because of its simplicity and because it is the position with which Hart would most likely have been familiar.)

Consider the evaluative sentence “One ought to keep one's promises.” For the emotivist, this sentence does not describe a state of affairs but instead expresses the speaker's desire or disposition of the will concerning promise-keeping. Notice that emotivists insist that evaluative sentences are the expression, not the description, of affective states. “One ought to keep one's promises” is not semantically equivalent to the sentence, “I have a positive attitude toward promise-keeping.”Footnote ⁸⁷

Emotivism, so described, is a semantic theory. But it suggests a psychological theory of evaluative judgment as well, in which such judgment involves the expression of desires and other affective states. And here, too, emotivists insist that the evaluative component in such judgment is not descriptive of the expressed states. When I judge that one ought to keep one's promises, I am not judging that I have a positive attitude toward promise-keeping.

The emotivist will also deny that an evaluative judgment is justified by reference to affective states. To be sure, when giving a descriptive account of the judgment, one should say that it occurred because of such states. But a justification of an evaluative judgment will itself be evaluative and so will involve the expression of the same or allied affective states as the judgment itself. As C.L. Stevenson puts it:

When asked why the judgment that one ought to keep one's promises is justified, we appeal not to our desires to keep promises but to a reason for action that does not depend upon our desires. As a result, emotivism can accommodate our commitment to the existence of norms whose validity is independent of our judgments and attitudes. This position, now commonly described as quasi-realism,Footnote ⁸⁹ was a part of emotivism even in the early 1960s.Footnote ⁹⁰

I believe that Hart relies on emotivism to explain the phenomenology of legal justification. But in Hart's theory, what is expressed in an internal legal statement is an affective state within the context of a certain practice. To the extent that one justifies one's action by reference to the practice, one will have abandoned the perspective within which legal rules are perceived, although one will not necessarily have abandoned a normative perspective.

To see how this approach might work, consider a simpler case. Imagine that I sit down with a companion at a chess table. I try to move my queen like a knight, and my companion says, “You can't do that. The queen ought to be moved like this.” One possible rendering of my companion's statement is that social facts have triggered background reasons for action. For example, by deviating from the practice of chess, I have frustrated her expectations, and settled expectations ought not be frustrated without a compelling reason. But assume my companion was not saying that I was unjustifiably frustrating her expectations; she was talking about a rule of chess, not the morality of playing games. Furthermore, in speaking of this rule, she was not, or not merely, speaking of an abstract propositional object, for such an object would not give me a reason to move the queen one way rather than another. The rule she was referring to was a justification for action.

By terminating her justification in a rule, my companion appears to be treating the rule as if it were intrinsically valid—as if appeal to the rule is sufficient to explain what I ought to do. And such a rule is mysterious. An accurate explanation of why I ought not move my queen like a knight would appeal to social facts triggering background reasons for action. For example, I ought not move my queen like a knight because it would frustrate my companion's expectations, and expectations ought not be frustrated absent a compelling reason, or because in sitting down to play the game, I implicitly promised to my companion to move my pieces in certain ways, and promises ought, prima facie, to be abided by. Nevertheless my companion's statement and her commitment to the rules of chess as justifications can be explained in terms of the expression of her attitudes in the context of the practice. Taking the internal point of view, she will use language that speaks of such rules.

Hart, I would argue, treats legal rules similarly. From the internal point of view, an official will speak of legal rules as justifications for conformity. If the justificatory role of official practice is made explicit, legal rules will vanish. Conformity will then be justified because the practice has triggering background reasons for action. But when expressing his attitude toward actions within the context of the practice, the official will speak of legal rules.

Notice that Hart's theory, like Shapiro's, can explain why revolutions appear to involve changes in the normative order rather than simply changes in social facts. In 1788, officials in France, taking the internal point of view, would terminate justification of their action in the legal rule that only norms identified by the king of France should be enforced. In 1791, in contrast, they would appeal to a very different and apparently incommensurable legal rule that only norms identified by the National Assembly (or the people of France that it represented) should be enforced. Thus, the revolution appears to have shifted normative orders in the same way the divine-command theorist thinks that changes in God's commands puts us into a new normative order. But this appearance of normative incommensurability will be only from the internal point of view. From the external point of view, the revolution will reveal itself to be a social fact that triggered background reasons for action that applied with equal force before and after the revolution.

To be sure, Hart probably was an expressivist concerning judgments about background reasons for action as well. But to the extent that he was, he allowed for two expressive stages. In the first, one expresses one's attitudes by speaking of background reasons for action such as morality and prudence. What is unique about Hart's account of social rules, as I describe it here, is that social rules are the result of a second stage in which one expresses attitudes in the context of a certain practice. In that second stage, the rules one offers as one's reasons for action cannot be reduced to the justifications provided in the first stage. If one justifies one's actions in the first stage, one has abandoned the internal point of view within which social rules are perceived to exist. Thus, when Hart says that officials taking the internal point of view presuppose rather than describe the existence of the rule of recognition as a social fact, he means that they must presuppose it for the internal point of view to arise.Footnote ⁹¹

My goal in this final section is to describe briefly how an expressive theory of legal rules might explain the phenomenology of legal justification. Because Hart, like Shapiro, believes that legal justification terminates in legal rules, not social practices and background reasons for action, Shapiro's category-mistake argument against Hart is, as it stands, incomplete. By this, I do not mean that Shapiro would have no response. My guess is that he would argue that a theory of law must do more than simply explain why officials terminate justification of their action in legal rules. It must also explain why they are rational in so doing. Subject to this constraint, Hart's theory must be judged a failure.

But whether the phenomenology of legal justification is the result of a rational process is an open question. What at first appears to be a deficiency in Hart's approach may, in the end, be its strength. In any event, I believe—and try to argue here—that in their views about what would count as an adequate theory of law, Shapiro and Hart are remarkably similar, for both believe that such a theory must explain how officials justify their action through legal rules rather than social practices.