RATIONAL DETERMINATION

In this section, I summarize my argument in HFML and Neta's response. My argument starts from the relatively uncontroversial premise that descriptive facts are among the constitutive determinants of the content of the law. What is controversial is whether descriptive facts can alone determine the content of the law or whether normative facts must also play a part. I argue that a full account of what makes the law have the content that it does will have to appeal to normative facts.

A variety of domains—the aesthetic, the political, the legal—are plausibly not metaphysically or ontologically basic. That is, the facts of those domains are constituted by more ontologically basic facts. In each case, the obtaining of the more basic facts makes it the case that the facts of the target domain obtain. For example, perhaps facts about the arrangement of paint on a surface make it the case that a painting is elegant or clumsy.

I take for granted that legal facts are not a metaphysically basic aspect of the world.⁹

In the legal case, the more basic determining facts include law practices—facts about the attitudes and behavior of various people—and may or may not include various normative facts as well.¹⁰

The nature of the making-it-the-case or constitutive determination relation may vary from domain to domain. To make progress on the central issue of the determinants of legal content, we need to investigate the nature of the determination relation in the legal domain. I claim that this relation is what I call rational determination.

Suppose that we are trying to give a constitutive account of the facts of a target domain. Suppose, further, that we are able to specify constitutive determinants of the target facts that modally entail those target facts. In some domains, this may be all that a constitutive account can do. The constitutive determinants need not provide reasons that explain why the target facts obtain. A reason, in the relevant sense, is a consideration that makes an explanandum intelligible in rational terms (as opposed to, say, emotional or aesthetic ones)—or, as I will sometimes write, that makes an explanandum rationally intelligible.¹¹

By contrast, if the nature of the determination relation in a domain is rational determination, the determining facts must make it rationally intelligible that the facts of the target domain are thus and so. We can call this the rational-relation requirement.¹²

Let us consider an example of a domain in which there need be no reasons of the relevant sort. I mentioned above the view that aesthetic facts obtain in virtue of physical facts. A painting is elegant or clumsy in virtue of the arrangement of paint on the canvas.¹³

But, arguably, the facts about the arrangement of paint need not provide a reason for the painting's elegance or clumsiness (though those facts will no doubt be part of a causal—for example, psychological—explanation of our reactions). For example, a small difference in the arrangement of paint might make a clumsy scene elegant without providing a reason for the difference. (Imagine trying to explain the impact on the painting's beauty to a rational Martian or to a fully rational human being who lacked the relevant sensitivity.)

The claim that the relevant determination relation in law is rational determination is a claim that the relation is different in this respect from the relation between physical and aesthetic facts.¹⁴

There must be reasons that explain why the relevant facts—for example, about decisions of legislatures and courts—have the effects on the law that they do. (This metaphysical truth is reflected in the requirement that judges must articulate reasons for their decisions.) For example, it could not be a brute fact—one that cannot be rendered rationally intelligible—that a particular change in the wording of a statute produces a particular change in the content of the law.¹⁵

Two clarifications will be important for what follows. First, rational intelligibility is a relation between the determinants of legal facts and the legal facts, not a monadic property of the legal facts.¹⁶

The obtaining of particular legal facts, as opposed to others, must be rationally intelligible in light of the determining facts. The rational-relation requirement imposes no requirement on the content of the law independent of the determining facts. For example, it does not demand that the content of the law be reasonable or rational.

Second, the rational-relation requirement does not specify that the reasons it requires must be normative facts. The reasons that the determinants of legal content must provide are considerations that explain in rational terms why the law has the content that it does. As far as the rational-relation requirement is concerned, it is an open question whether there are nonnormative facts that could constitute such reasons—and indeed whether there are normative facts that could do so.¹⁷

(The point is important because otherwise positivists could not accept the rational-relation requirement. The strategy of the argument is to use the rational-relation requirement, which, I claim, most legal theorists implicitly accept, to argue for the controversial conclusion that normative facts must be determinants of legal content.)

The present paper creates a special risk of confusion with respect to this point because I also use the term “reason” in another context. A crucial issue in the debate between Neta and me concerns reasons for actions, and in fact, I distinguish two types of reasons for action.¹⁸

I therefore want to emphasize that the reasons that the rational-relation requirement demands are not reasons for action. Again, they are considerations that explain in rational terms why particular legal facts obtain.¹⁹

In HFML, I sketch some arguments for the proposition that the relation between law practices and legal facts is rational determination, at least in the U.S. and U.K. legal systems and perhaps in all legal systems,²⁰

though I do not attempt anything like a full defense of that proposition. I also give reasons for thinking that the proposition is fundamental to our ordinary understanding of the law and taken for granted by much contemporary legal theory, though seldom articulated.²¹

I then argue that law practices cannot themselves rationally determine legal facts or, indeed, norms of any kind. In brief, the argument is that law practices cannot unilaterally determine their own relevance. Factors independent of law practices must be part of what makes their contribution intelligible in rational terms. In the case of law, the best candidates for such factors are normative facts. (In other domains, there may be other candidates.)²²

For example, facts about democracy and fairness explain the relevance of the publicly promulgated decisions of elected legislators to the content of the law.²³

Neta's leading idea is that descriptive facts²⁴

can alone provide a full account of normativity. He first argues that in all normative domains, descriptive facts in part determine the content of the norms, and the relevant determination relation is rational determination. Next, he tries to show that a full account of moral norms need appeal only to descriptive facts. In particular, he offers an account of the wrongness of promise-breaking entirely in terms of descriptive facts. He concludes that it does not follow from the fact that rational determination is the relevant determination relation that normative facts must play a role. Neta also gives a purported counterexample in the legal domain—a case in which he claims that descriptive facts can themselves rationally determine the content of the law.

I now turn to Neta's attempts to show that descriptive facts can alone determine norms in the moral and legal domains. First (Section II), I discuss how Neta's account of why it is wrong to break promises fails. It begs the question by taking for granted that a person's desires or other motivational states necessarily justify the actions they motivate. Second (Section III), I turn to Neta's alleged legal counterexample. Here I draw on the notion of rational determination to move the argument beyond an appeal to intuitions. I show that the facts to which Neta appeals cannot be reasons of the required sort, for they already take for granted what needs to be explained—the bearing of law practices on the content of the law. Also, Neta fails to consider the possibility that normative facts are playing a silent role in his example. Finally, in Section IV, I turn to the larger issue of how far my argument applies to other normative domains.

NETA'S ACCOUNT OF PROMISING

Neta uses the case of promising to argue that the full metaphysical explanation of normative facts need appeal only to descriptive facts. He offers an example in which “Alice mowed the lawn because John promised to pay her four hundred dollars” and tries to explain “what makes it wrong for John to break his promise.”²⁵

According to Neta, “any plausible answer” involves the claim that “[w]hat makes it wrong for John to break his promise … is that it would be a case of breaking a promise, and it is generally wrong to break promises.”²⁶ This first step appeals directly to a normative fact—that “it is generally wrong to break promises.” The success of Neta's argument depends on whether he can explain, without appealing to any normative (or evaluative) facts, why it is generally wrong to break promises.²⁷ Let us call the moral requirement that one not break promises—the requirement that Neta is trying to explain—“the promising norm.”

Before proceeding, we need to clarify one terminological matter. As I will use the term “norm,” a necessary condition for there to be a norm is that there actually be reasons for agents to act in the way required by the norm. (Occasionally, I will use “genuine norm” rather than merely “norm” as a reminder of my usage.) Other writers sometimes use the term in a purely descriptive way according to which, roughly speaking, for there to be a norm in a particular group requiring that people act in a certain way, it is sufficient that members of the group think and act as if there were a genuine norm requiring that people act in that way. Consider the putative norm of a crime gang that its members must commit an assassination once a year. In my usage, the fact that the gang members accept the putative norm—regularly comply with it, enforce it by punishing those who fail to comply, believe that they are justified in doing so, and so on—does not make it the case that the putative norm is a norm. Neta is very clear that he uses “norm” in my sense.²⁸

I will continue to use “putative norm” to express the purely descriptive sense of “norm.”²⁹

Neta's attempted explanation appeals crucially to the existence of “the practice of promising.” In what follows, I show that Neta's practice-based explanation of the wrongness of breaking promises does not work. I want first to note a problem with the whole approach, however. Neta seems to assume, as many writers do, that promising could not take place in the absence of a particular kind of social practice or convention.³⁰

In my view, this is a mistake. A Robinson Crusoe who has never made a promise and has no practices or conventions with regard to promising can, when he encounters a new castaway for the first time, promise to help the castaway (and it would be wrong for Crusoe to break that promise without good enough reasons). The fact that breaking promises is wrong in the absence of any practice supports the view that the existence of a practice is not what accounts for the wrongness of breaking promises even in cases where there is a practice. On this view, which I do not attempt to defend here, the appeal to practices is misguided from the start. But for present purposes, I consider Neta's practice-based explanation on its own terms.

Neta's attempted explanation begins with the claim that “[t]he practice of promising essentially involves” the promising norm.³¹

This claim might be taken to suggest that he is using the notion of a practice in such a way that the existence of a practice is constituted in part by the obtaining of normative facts. If that were the case, it would be question-begging to appeal to a practice as a necessary part of the explanation, for the goal is to explain normative facts in nonnormative terms.³²

Neta soon makes clear, however, that the claim is that “the institution of promising would quickly cease to exist if people were generally permitted to break their promises.”³³

There is no suggestion here that the existence of the practice is constituted by normative facts. Rather, when Neta claims that the norm is essential to the practice, he means that as a causal or historical matter, if people did not treat promises as creating obligations, the practice would cease to exist.

We need to be clear what exactly is necessary for the practice to exist. Is it that breaking promises be wrong? Or is it that people believe that it is wrong, criticize others for breaking their promises, think it is appropriate to criticize others for this reason, and so on? Neta uses phrases that are ambiguous with respect to this point. That it is “permitted” to break promises could mean either that people permit each other to break promises with impunity (i.e., that they act as if promising were not wrong) or that breaking promises (even without good reason) is in fact morally permissible. What is required in order for people's expectations not to be disappointed and thus for the practice to continue is that many people believe it is wrong to break promises, not that it actually be wrong. In fact, even this formulation seems too strong: what is necessary is not that people believe that it is wrong to violate the norm, but that people act as if they believe it is wrong—or, to oversimplify in order to save words, that people generally comply with the promising norm. Whether in practice people will not comply with the norm unless they believe that it is wrong to violate it (as opposed to believing, for example, that others believe it is wrong and will punish violations) is an empirical question. In sum, all that Neta can take to be necessary for the promising practice is that people often act in accordance with the putative promising norm, not that it be wrong to violate it.

Neta sees that this kind of dependence of a practice on a putative norm does not suffice to show that violating the putative norm is wrong, that is, that it is a genuine norm. But he thinks that all that needs to be added to complete the account is that a person has a reason to participate in the practice:

At this point, Neta takes himself to have established what he set out to show—that it is wrong to break promises.³⁵ In short, Neta's argument is that if a person (believes she) could obtain something she wants³⁶ or enjoys by participating in a practice that would not exist if people did not generally comply with certain putative norms, then those putative norms are genuine norms for her, and it is wrong—even morally wrong—not to comply with them.

Neta's account cannot be right. There are many silly, corrupt, or evil practices that would “cease to exist” if the participants did not generally comply with certain putative norms. The practices of operating fraudulent investment schemes, hazing new recruits, and selling children into sexual slavery are examples. People can obtain things they want by participating in these practices. It hardly follows that it is wrong, even pro tanto, not to comply with the putative norms of the practices.

It is useful to develop one example a little more fully. Consider the protection racket—organized crime's practice of providing “protection” for restaurants, shops, and other businesses in return for regular payments. It is essential to the existence of the practice that many of the lower-level criminals who carry out the work of extorting protection money follow putative norms requiring them to use violence against shop owners who fail to pay protection money and requiring them not to inform on their bosses. Similarly, it is essential to the continued existence of the practice that many of the small business owners follow putative norms requiring them to pay on time and not to complain to the police. Therefore the protection-racket practice “essentially involves” the putative norms that it is wrong to refuse to pay protection money and that it is wrong to report the organizers to the police in exactly the sense in which Neta claims that the promising practice essentially involves the promising norm.³⁷

The criminals who extort the protection fees can obtain things they want by participating in the practice—they earn a living and avoid being punished by their bosses. Even the business owners have self-interested reasons for participating in the practice. They avoid having their shops burned down or worse, and they may in fact receive some protection from petty criminals and other benefits from being on good terms with the organized crime gangs. According to Neta's position, it is enough that they have “some reason to participate,” not that the all-things-considered balance of reasons supports participating (though in some cases it may do so). It hardly follows that the putative norms of the practice are genuine norms, let alone that it would be wrong not to comply, for example, by reporting the organizers to the police or refusing to pay protection money.

Where does Neta's argument go wrong? First, it is not warranted to move from John's wanting to have his lawn mowed to John's having reason to comply with the essential norms of the practice of promising. That one could obtain something one wants by participating in a practice does not show that one has even a narrowly self-interested reason for complying with all the putative norms necessary for the practice to exist. One might get what one wants by making promises without keeping them. Perhaps it is plausible that one must generally comply with the promising norm in order to be able to use promises effectively to get what one wants. It is easy, however, to imagine practices such that one can obtain their benefits through partial or apparent participation, that is, without complying with all of their putative norms. Moreover, at most, John's wanting to have his lawn mowed gives him a reason for keeping his promises when not doing so will hurt his reputation (or cause harm to his reputation greater than the benefit he would receive from breaking them). It is a commonplace that there may be cases in which it is evident that breaking a promise will not harm one's reputation. For instance, John might make an unwitnessed promise to a friend moments before the friend dies.

Neta might respond that John's behavior does not really count as promising unless he intends to keep his promises, believes that it is wrong to break them, and so on. In order to promise, one must be participating in the practice, and the promising norm is “an essential feature” of the practice.³⁸

As emphasized above, however, Neta's argument that the promising norm is essential establishes only that the practice would not exist if many or most people did not comply with (or believe in) the putative norm. It does not follow that John or any particular person must believe it is wrong to break promises or intend to keep promises in order to count as promising. But suppose that we waive this point and grant that one does not count as promising unless one accepts the promising norm. In that case, however, Neta has not shown that John has reason to promise, only that he has reason to make apparent or pseudo-promises. For pseudo-promises may well be enough for John to get what he wants. Hence, even given the assumption that one does not count as promising unless one accepts the promising norm, it does not follow, so far as Neta has argued, that John has any reason to accept the promising norm. Neta here runs into a problem that dogs attempts to show that it is rational to act morally. That one would be worse off if no one complied with a putative norm does not show that one has reason to comply (even if others do comply). Neta is not free to assume away the free-rider problem.

A second problem with Neta's argument is the move from John's having a reason to comply with a putative norm to its being a genuine norm that is applicable to him (“he has reason to comply with the norms of that practice, and so they are norms for him”). This step in the argument seems to depend on an equivocation on the term “reason.”

Michael Smith helpfully distinguishes between “normative reasons,” which are considerations that count in favor of or justify Φing, and “motivating reasons,” which are factors that (potentially) motivate someone to Φ.³⁹

Neta's examples of John's reasons are John's wanting to have his lawn mowed and his enjoyment in participating in the practice. Neta takes it as obvious that these are reasons, and offers no further explanation. Consequently, in pointing out that John has reasons to engage in the practice, Neta must be using the term to mean motivating reason. For the fact that John has a desire that would be satisfied by Φing or enjoys Φing shows immediately that John has a motivating reason to Φ but, without further argument, does not show that he has a normative reason to Φ.

But the plausibility of moving from John's having a reason to comply with a putative norm to its being a genuine norm depends on John's reason being a normative reason. That one believes that one can get what one wants by Φing cannot be sufficient to show that there is a norm requiring one to Φ. Since the issue is whether normative facts can be accounted for entirely in nonnormative terms, it obviously begs the question to move without argument from the existence of a motivating reason—a descriptive fact—to the existence of a normative reason—a normative fact.

It might be objected that the term “reason” is not simply ambiguous. Many motivating reasons are or involve normative reasons. For example, at least one kind of motivating reason to act in a particular way is in part constituted by a belief that there is a normative reason to act in that way.

There are certainly important connections between motivating reasons and normative reasons. But because motivating reasons need not be normative reasons, an explanation of the existence of a normative reason must do more than point to the existence of a motivating reason. The critical question is whether it is necessary to appeal to normative facts as well. Even when a motivating reason constitutes a normative reason, the question remains whether normative facts are part of the explanation of what makes the motivating reason a normative reason. Taking for granted that no explanation is needed to get from the existence of a motivating reason to the existence of a normative reason begs the question of whether the wrongness of promise-breaking can be explained without appeal to normative facts.

It might be wondered whether Neta's position could be defended by appeal to a substantive theory according to which any desire or other state that (potentially) motivates someone's taking a certain action counts in favor of the person's taking that action. In the first place, such a theory is highly implausible. Moreover, the theory makes a normative claim, a claim about what kinds of things provide justifications. So an account of normativity that relies on the theory does not restrict itself to descriptive facts.

At this stage, an obvious suggestion is that only good practices, or practices that it is right to engage in, generate genuine norms. The problem, again, is that the suggestion explains normativity only by appealing to normative facts—facts about which practices are good or which courses of action are right.

There are further problems with Neta's account. That it is wrong to break promises (even prima facie or pro tanto) is a stronger claim than that there is some normative reason for not doing so. This is not the place to attempt a full account of exactly how it is stronger. One type of case will suffice. I have reasons to have annual medical tests. It does not follow that it is wrong for me not to do so, not even pro tanto. Certainly, it does not follow that it is morally wrong.⁴⁰

Furthermore, it would not make a difference if there were a practice to which the putative norm of annual medical tests was essential in Neta's sense. Even if we granted for the sake of argument that all motivating reasons were normative reasons (and assumed that normative facts were not an essential part of the explanation of this proposition), Neta's account would still seem to lack the resources to explain specifically moral wrongness. John's wanting to have his lawn mowed, even if it is a normative reason, is the wrong kind of material with which to explain the moral wrongness of his breaking promises. And since Neta's project is to explain the wrongness of breaking promises without appealing to normative facts, Neta obviously cannot appeal to moral facts.

NETA'S LEGAL COUNTEREXAMPLE

I now turn to Neta's purported counterexample to my claim that normative facts are needed to determine the contribution of law practices to the content of the law—or, as I will say, to determine which model (of the contribution of law practices to the content of the law) is correct.⁴¹

He supposes that the framers of the U.S. Constitution “had included a clause that stated explicitly and precisely how the content of the law was to depend upon the law practices.”⁴²

In the last section, we were able to undermine Neta's claim that nonnormative facts are sufficient to determine normative facts simply by considering the evidence of actual and hypothetical cases. Examples like that of the protection racket are difficult to reconcile with the claim that practices (and motivating reasons to participate in them) necessarily generate normative reasons. (It is for similar reasons that Hart's critics were able so conclusively to dispatch Hart's practice theory of rules as a general theory of moral and other norms.)⁴³

As noted, Neta assumes that legal facts constitute normative reasons for action.⁴⁴

Given that assumption, it would be straightforward to mount an argument against Neta's legal counterexample parallel to my argument against his account of normative facts generally. That a practice involves a document that explicitly and precisely (and without contradiction from other documents) specifies how the practice is to generate rules does not ensure that the practice generates normative reasons for action.

As noted in the introductory paragraphs, however, I want to avoid relying on an assumption that legal facts constitute reasons for action (or, more generally, that legal facts are normative facts), so as not to beg any questions against legal positivists who do not accept the assumption. Consequently, my argument that descriptive facts cannot themselves determine legal facts must take a different form from my argument that descriptive facts do not determine moral and other normative facts.

HFML introduced a new way of answering the question of whether descriptive facts can themselves determine legal facts. This approach appeals to an argument about the properties the determinants of legal facts must have. In particular, I argue in HFML that in law, in contrast to many other domains, the determinants must make the facts of the target domain rationally intelligible.

In Neta's example, it is plausible, though by no means obvious, that the explicit constitutional clause would be foundational—that it would determine the relevance of other law practices to the content of the law and that its contribution to the content of the law would not be affected by other law practices. The relevant question, however, is whether the existence of that clause (and other descriptive facts, such as the absence of competing clauses) could be the full explanation of why that clause was foundational. As we shall see below, an important possibility, which Neta does not consider, is that normative facts are part of the explanation. For example, values of democracy, stability, and fairness support giving great weight to what the ratifiers of the Constitution and the public believed the ratifiers to be agreeing on.

Should we accept Neta's view that the descriptive facts of his hypothetical case are sufficient to determine that the imagined constitutional clause is foundational? Or do normative facts do part of the work? We can adjudicate between these and other competing accounts of legal facts by asking whether their proposed determinants satisfy the rational-relation requirement.

Let us begin by distinguishing two very different approaches that a positivist could take in attempting to show that normative facts are not necessary to satisfy the rational-relation requirement. According to the first approach, the explanation of why descriptive facts are able to satisfy the rational-relation requirement is special to the legal case. Such an explanation appeals to an alleged basic truth about the nature of law (or perhaps a conceptual truth about law). An example of such an alleged truth would be the Hartian thesis that officials' “acceptance”⁴⁵

of a given rule of recognition makes that rule the correct model in the community's legal system. A theorist might argue that understanding this thesis—grasping the explanatory significance of officials' acceptance of a rule—is part of understanding what law is.

In “Hartian Positivism and Normative Facts,”⁴⁶

I consider an attempt to defend a Hartian account of law—the most influential contemporary version of legal positivism—in this way. I argue that the Hartian thesis is not part of our reflective understanding of law.⁴⁷

The second approach makes no appeal to specifically legal truths. Ordinary descriptive facts—for example, about the behavior and attitudes of people—are sufficient to satisfy the rational-relation requirement. Neta's view falls in this second camp, as he makes no appeal to specifically legal truths and in general assimilates the legal domain to other normative domains. In discussing Neta's example here, we should therefore set aside any ideas we might have of basic truths about law. The question is whether the ordinary empirical facts about the sayings and doings of various people are themselves able to satisfy the rational-relation requirement.⁴⁸

In HFML,⁴⁹

I considered the possibility of a “foundational practice”—one that determines the bearing on the content of the law of other practices but whose own bearing does not depend on that of any other practice. I argued that a putative foundational practice could not by itself make rationally intelligible that it was in fact foundational. More fundamentally, it could not make rationally intelligible its own significance to the law.

Neta concedes that if there were “two putatively foundational but inconsistent practices,” some additional factor would be needed to make intelligible why one practice, rather than the other, is foundational. He focuses instead on a case in which “we have only one putatively foundational [practice], the dictates of which are consistent with all of the law practices.” In this case, Neta suggests that the sole putatively foundational practice can provide the reason “why it is itself foundational.”⁵⁰

Neta's description of the case takes three elements as given. First, the practice in question is putatively foundational. Second, no other practice is putatively foundational. Third, the content of the putatively foundational practice (its “dictates”) is consistent with the content of all other law practices.

Stipulating each of these three elements begs the question of whether the descriptive facts can themselves determine which model is correct. The reason is that these three elements are not descriptive facts but facts that are logically posterior to what is at issue—the correct model in the legal system. For example, for a law practice to be the only putatively foundational practice in a legal system, it must be the case that no other law practice has a bearing on what the putatively foundational practice contributes to the content of the law. Whether that condition is satisfied depends, however, on which model is correct.

Similarly, whether the putatively foundational practice is consistent with all other law practices depends on the legal significance of all the law practices. And, again, the legal significance of law practices depends on which model is correct. In sum, the elements on which Neta relies are not the descriptive facts to which he is entitled but facts that take a model for granted.

What has gone wrong with Neta's argument? It is a descriptive fact that only one legal practice in the legal system has associated with it a sentence asserting that the practice has a foundational role. And it is a descriptive fact that no other legal practice has associated with it an inconsistent sentence. Neta seems to equate these descriptive facts with the facts he stipulates—that there is only one putatively foundational practice and that its content is not in conflict with the content of any other practice. Thus he implicitly assumes that the significance of a law practice is no more and no less than the ordinary nonlegal content of the sentences associated with it.

We know that this assumption is false, however. For example, the fact that judicial decisions have interpreted other law practices in a particular way bears on the contribution of law practices to the content of the law, regardless of whether the interpreting judicial decisions explicitly say anything about how law practices are to be interpreted. More generally, as I argue at length in HFML, the contribution of law practices to legal content is not constituted by the conjunction of the contents of sentences associated with those practices.⁵¹

The significance of law practices depends on reasons, and there are reasons for law practices to contribute to the content of the law in ways that do not track the meaning of sentences associated with those practices.

Another problem with Neta's account is that he seems to assume that it is a matter of contingent historical fact whether a constitutional clause of the sort he describes would make rationally intelligible that the clause itself was foundational. He says that the example of the constitutional clause shows that my account is vulnerable to “historical falsification” and “depends on a matter of contingent, and possibly unknown, fact of legal history.” This claim seems to confuse the question of whether the clause would be foundational with the question of whether the descriptive facts (of the existence of the clause and the absence of a conflicting practice) are the full explanation of its being foundational.⁵²

As noted above, Neta neglects the possibility that normative facts play a role in explaining the bearing of his hypothetical constitutional clause. Again, it is plausible that the clause would have foundational significance. But the best explanation may not be simply that the clause says so. There are powerful reasons—for example, of democracy and fairness—for giving great weight to the plain meaning of the constitutional text.

Hence, even if there were a foundational constitutional clause of the sort Neta describes, it would not follow that descriptive facts alone could rationally determine the content of the law. My argument shows that normative facts must play a role in determining the content of the law in any legal system in which the determination relation is rational determination; it does not depend on contingent facts about the history of our own legal system.

In order to see that normative facts may be playing a silent role in Neta's example, consider how much less effective the example would be if the single putatively foundational clause were in a decision of a minor administrative official. Neta's explanation of the foundational nature of his clause may legitimately appeal only to purely descriptive facts. Given the important normative considerations that support the Constitution's special status, it is difficult to be sure that our evaluation of the example is influenced only by the descriptive facts.⁵³

It might be objected that the problem with Neta's example is that constitutional requirements are themselves part of the content of the law and therefore cannot ultimately determine how practices determine the content of the law. Perhaps a version of Neta's example would fare better if it involved not a constitutional provision but the attitudes and dispositions of legal officials that, on a Hartian account, make a rule of recognition the correct model for a legal system.⁵⁴

As noted above, in “Hartian Positivism,” I consider whether a Hartian account can satisfy the rational-relation requirement. I show that the arguments of HFML yield the same conclusion with respect to officials' acceptance of a rule of recognition as with respect to Neta's constitutional clause.

OTHER NORMATIVE DOMAINS

I have criticized Neta's attempts to explain moral and legal facts entirely in terms of nonnormative facts. The relevance of the moral domain to my original paper depends on a more general claim that Neta makes. The claim is that my argument generalizes to all normative domains, or, more precisely, that all normative domains have the structural features of the legal domain that are necessary to generate my original argument. In order to defend the argument of HFML against Neta, I do not need to deny this claim, for I have already argued against Neta's attempt to explain moral norms in nonnormative terms. But the issue of how far my argument generalizes is of independent interest. In this final section, I first consider and reject Neta's argument purporting to show that all normative domains have the relevant features of the legal domain; I then offer a competing picture of some other normative domains.

A crucial premise of my original argument is that descriptive facts are among the determinants of the legal facts, and the determination relation between the determinants and the legal facts is rational determination.⁵⁵

A version of the argument can be developed only in domains in which a corresponding premise is true. I suggest, for example, that the determination relation in the domains of aesthetics, humor, and mental content is not rational determination.⁵⁶ If that suggestion is correct, a version of the argument cannot be developed in those domains.⁵⁷

Neta claims that all normative domains have the structure that I attributed to law. We can formulate this generalized premise as follows:

I am sympathetic to a view that generalizes my conclusion to all normative domains. In particular, I am sympathetic to the view that a full explanation of the existence of any genuine norms cannot be given entirely in terms of descriptive facts. I did not argue for this view in the original article, however, and will not do so here. As I will show, Neta's argument for the generalized premise is not successful. Moreover, the generalized premise is not plausible. Hence the specific argument that I give in HFML does not generalize in the way that Neta claims.

A number of passages in Neta's comments suggest that he is working with a determination notion different from the notion of rational determination. For example, when he explains the determination notion, he writes:

Similarly, he writes: “there must be some reason for the law to require some things and forbid other things—the law's requirements, unlike the boiling point of water, must be at least somewhat reasonable.”⁵⁹

Neta here contrasts domains in which the notion of reasonableness applies to the content of the relevant propositions and domains in which it does not. It is neither reasonable nor unreasonable that water boils at a certain temperature. It is reasonable that the law imposes certain requirements and not others. Moreover, the quoted passages taken together suggest that Neta thinks the content of the law is rationally determined if and only if the content of the law is “somewhat reasonable.”

Neta's notion of the reasonableness of a norm's content is not the same as my notion of rational determination.⁶⁰

As noted above, to say that the A facts rationally determine the B facts is to say (in part) that the obtaining of the B facts must be intelligible in rational terms in light of the obtaining of the A facts. Consider the legal requirement that contracts for the sale of land be in writing. The intelligibility issue is not whether this requirement is a reasonable or wise one, but whether the existence of the requirement is explained by, or intelligible in the light of, for example, statutes and judicial decisions. Rational intelligibility concerns not the content of the higher-level facts—for example, that contracts for the sale of land must be in writing—but the relation between the determining facts and the higher-level facts⁶¹—for example, between statutes and judicial decisions and the fact that contracts for the sale of land must be in writing.

The reasonableness of the content of a norm is consistent with its being opaque why the determinants have the consequences that they do for the content of the norm. For example, one could consistently think that it is a good idea for the Constitution to ban the death penalty yet that the relevant facts—for example, about the text of the Constitution or the intentions of its ratifiers—do not make the existence of such a ban rationally intelligible. Conversely, even if it is a terrible idea to allow a tax deduction for mortgage interest, the existence of that deduction might be perfectly intelligible in light of the Internal Revenue Code. (Indeed, the relation between the arrangement of microphysical particles and macrophysical facts, such as the fact of water's boiling point, might be intelligible even if the notion of reasonableness is not applicable to the macrophysical facts.) In evaluating Neta's argument for the generalized premise, we need to ensure that we evaluate it with respect to the correct notion.

Neta's argument takes the form of a reductio. He begins with the following conditional: “If the [descriptive] facts do not make it reasonable for the norms to be what they are, then either nothing makes it reasonable for the norms to be what they are, or else the reason for the norms to be what they are includes something independent of the [descriptive] facts.” He then argues that both possibilities lead to contradictions, so the antecedent of the conditional must be rejected. The argument is phrased as if the question is the reasonableness of the content of the norms. But as I have just explained, that is not the right question. So it is important to reformulate the argument in terms of rational determination.

We first need to clarify one point about the antecedent of Neta's conditional. Substituting rational determination for reasonableness (and switching to my terminology), the antecedent becomes: “the descriptive facts do not rationally determine the normative facts.” Does this mean that the descriptive facts do not even in part rationally determine the normative facts or that they do not alone do so? Remember that the crucial premise of my argument in the legal case is that the descriptive facts in part rationally determine the normative facts. Since Neta is trying to show that the corresponding premise is true in all normative domains, the antecedent of the conditional (i.e., the premise of his reductio) must be the negation of that premise. Therefore the correct understanding of the antecedent has to be that the descriptive facts do not even in part rationally determine the normative facts.

We can now address Neta's dilemma. As to the first horn, he argues that if “nothing makes it reasonable for the norms to be what they are,” then “there is no reason to follow those norms,” and thus they are not really norms at all.⁶²

The relevant condition is not, however, that nothing makes it reasonable for the norms to be what they are. It is that the determination relation in the domain in question is not rational determination—that is, that it is possible that the obtaining of the norms is not intelligible in light of the determinants of the norms. Why should we assume that in that case there is no reason to follow those norms? From the fact that the determination relation is not rational determination, nothing follows about the content of the norms. Hence one possibility is that the content of a norm can be the source of reasons, quite apart from the relation between the determining facts and the content of the norm. For example, it might be that the reasons to follow moral norms derive from their content. (Indeed, it is common to contrast legal and moral norms with respect to whether the content of a norm is the source of the relevant reasons.)

More generally, it cannot simply be assumed that putative norms have no normative force if the relevance of the determinants of those norms to the norms need not be reason-based. For that is to assume precisely what is at issue—that rational determination is the relevant determination relation.

With respect to the second horn, Neta argues that “[if] the reason for the norms to be what they are is independent of the [descriptive] facts that determine those norms … it is arbitrary that the norms are binding on all and only those creatures of which the determining [descriptive] facts obtain.” In that case, he concludes, “there is nothing that makes the norms in question binding on creatures like us,” which is to say, again, that they are not norms after all.⁶³

Neta seems to think that the relevant condition is that (1) descriptive facts determine the norms; and (2) the relevant determination relation is rational determination; yet (3) something independent of the descriptive facts provides the reasons that explain the existence of the norms. But it is hard to see why one would take this to be the relevant condition. If rational determination is the relevant determination relation (so we are not on the first horn of the dilemma) and the descriptive facts do not, even in part, rationally determine the content of the law (by hypothesis), then descriptive facts are not among the determinants of the normative facts at all. For example, the normative facts might be necessary truths that do not depend on contingent practices. (I sketch a picture below in which fundamental moral and epistemic norms have such a status.) If descriptive facts are not determinants of the normative facts, there is no concern that “it is arbitrary that the norms are binding on all and only those creatures of which the determining [descriptive] facts obtain.”

So far I have criticized Neta's argument that rational determination is the determination relation in all normative domains. I now say something about a few illustrative cases. Consider first the moral domain. According to a plausible and widely held view, fundamental moral truths⁶⁴

are necessary truths that are independent of any descriptive facts. Applied (or “mixed”) moral truths are applications of the fundamental moral truths to particular circumstances. Perhaps it is a fundamental moral truth that it counts in favor of an action, other things being equal, that it would reduce suffering. In combination with contingent facts about the sources of suffering (and perhaps also with other fundamental moral truths), that fundamental truth yields many applied moral truths—for example, that we should provide infants with nurturing environments.

I take this picture for granted here (though Neta would not accept it). On this picture, the interesting metaphysical question is the explanation of the fundamental moral truths. Given that explanation, the explanation of the applied moral truths follows straightforwardly. And it is obvious that normative facts—the fundamental moral truths—play a role in that explanation. I do not, of course, offer an account of fundamental moral truths here. The present point is that the legal case is a nonstarter as a model for such an account. It is uncontroversial that the content of the law obtains in virtue of law-making practices. Fundamental moral truths do not obtain in virtue of practices. That reducing suffering counts in favor of an action, say, is a necessary truth, not something that is contingent on what people do or decide. There is no legal analogue of fundamental moral truths—legal truths that could not be otherwise and that yield all the more specific legal facts when applied to particular circumstances.⁶⁵

The question of the explanation of the fundamental moral truths is not one on which my argument is intended to shed light.

An interesting example is provided by the assumption that all desires or motivating states justify the actions that they (potentially) motivate. This assumption, we saw, was necessary to Neta's account of promising. Notice that it is better understood as a candidate for a necessary truth than for a contingent one. It would be very strange to maintain that some contingent social practice makes it the case that all desires justify actions. The defender of such an assumption would presumably take it to be a necessary truth, which, in combination with facts about what desires people happen to have, yields truths about what justifies particular actions. Hence Neta's account of the normativity of promises, once we make explicit the assumption that all desires justify actions, can be seen to instantiate the picture of the moral domain that I have sketched.⁶⁶

In my view, the epistemic case is similar to the moral one. Fundamental epistemic norms do not depend on contingent social practices. Again, more specific applied epistemic norms are the result of applying the fundamental norms to particular circumstances.

Neta also considers the case of “the normative content of a particular linguistic system.”⁶⁷

I am not sure what he has in mind, but we can consider an example I mention in passing in HFML.⁶⁸ It is plausible that facts about a person's or community's use of words are among the determinants of what those words mean. I do not think that facts about what words mean are normative facts, so this issue may be a point of disagreement between Neta and me. But this case nonetheless provides a helpful comparison with the legal case. As I say,⁶⁹ it may be that the relation between facts about our use of words and facts about meaning is not rational determination. In sharp contrast to the legal case, I see no reason to think that the relation between the use of words and their meaning is necessarily rationally intelligible. It may be that many possible meanings are ruled out arbitrarily (i.e., in ways for which reasons cannot be given) and that we have nonrational mechanisms that exclude these possibilities from consideration.

I can sum up these brief remarks in an intuitive way. At bottom, we do not create epistemic and moral norms. Fundamental moral and epistemic normative truths are necessary and do not depend on the contingencies of our practices. We do create the facts about what words mean, but we do so in a way that need not be rationally accessible. In the legal domain, we create the facts, and we do so in a way that must be rationally accessible.