Legal philosophers often make bold, contentious claims about the nature of law. Some claim that law necessarily involves coercion, while others disagree. Some claim that all law enjoys presumptive moral validity, while others disagree. Whenever philosophers make these, and other, bold claims about law’s nature, I always wonder how anyone is in a position to know which one of these claims is true.
Now this expression of doubt serves as an important premise in a larger argument. The contention here is that we should see our bold claims about law’s nature as practical claims, I will say ‘pragmatic claims’, and not attempts at description or personal expression at all. My contention, written schematically, is the following.
[1] Bold claims, that is, controversial claims legal philosophers make about law’s nature, must be interpreted as descriptions, as expressions of a speaker’s non-descriptive attitudes, or as claims to be assessed on the basis of by practical reasons.
[2] In interpreting bold claims, we should choose the interpretation that fares best along two dimensions: semantic fit and epistemic propriety.
[3] A view that fares well along semantic fit will not attribute to a philosopher a judgment that the philosophy cannot plausibly be understood to make.
[4] A view that fares well along epistemic propriety will not attribute to a philosopher a claim that it would be epistemically improper for one to make.
[5] Descriptivism,Footnote 1 that is, interpreting bold claims as descriptions, fares well along semantic fit but poorly along epistemic propriety.
[6] Expressivism, that is, interpreting bold claims as expressing a person’s non-descriptive attitudes, fares poorly along semantic fit but well along epistemic propriety.
[7] Pragmatism, that is, interpreting bold claims as claims to be assessed on the basis of practical reasons, fares well along semantic fit and well along epistemic propriety.
[8] Therefore, if we are interpreting bold claims, we should choose pragmatism.
About this schematic argument, a few notes are in order. Much of my paper concentrates on vindicating [5], but first I will write a few words about the rest. I do not defend [1] here, and in fact, it would be likely be impossible to show that these three interpretive options, viz., descriptivism, expressionism, pragmatism, exhaust logical space. So, I rely on a trio that captures live options in the literature. If one finds more, I will be content to have shown that pragmatism is the best of the three.
[2] is concerned about semantic fit and epistemic propriety and is, I admit, somewhat controversial as written. It might be supposed to serve as a kind of principle of charity along the lines of Davidson,Footnote 2 but one need not to see it only that way. Perhaps [2] does seem addressed to interpreting other people’s behavior. However, we might instead see it as the best interpretative standard telling us how to characterize our own legal philosophical thought. In fact, one can understand [2] as serving to personalize the argument and perhaps make it more plausible. If, for instance, we think of our own philosophical reflection in the descriptivist way, we should be worried about epistemic impropriety and thus, we might, because of [2], want to recast our own projects.
Now [3] and [4] are just definitions. My paper concentrates on vindicating [5], particularly the claim about epistemic propriety. After that hurdle is surmounted, [6] and [7] are proved rather quickly. [8] then follows deductively.
The paper proceeds as follows. In §1 and §2, I offer a skeptical argument. There, I maintain that, in order to know that a description of law is right, we have to be able to answer a preliminary question that we cannot answer. In §3, I remark upon the fact that there is an epistemic condition on making descriptions. These three sections, then, amount to the claim that descriptivism is a poor interpretive strategy because it fails on the epistemic propriety dimension. In §4, I compare expressivism and pragmatism in hopes of demonstrating that pragmatism is, by far, the better interpretation of the two.
1. The What-Kind Question
In purporting to describe law, a theorist must decide early on what kind of thing law is. Call this the “what-kind question.”Footnote 3 Specifically, a legal theorist must decide whether law is a social kind or an abstract kind. In what follows, I explain this distinction and show why deciding the what-kind question must be done and has to precede most other investigation. In the following section, I offer reasons to think that everyone lacks the epistemic resources to decide the what-kind question.
It is a familiar thought that objects can be classified into three major categories, viz., natural kinds, social kinds and a third category, which one might call abstracta, non-natural or immaterial kinds, or Forms. This tri-partite division seems eminently sensible: a given object either owes its existence to a special relation to human attitudes or does not; if it does not, it is either physical or not. Thus, we have the three categories, social, natural, and abstract, respectively. To make this familiar thought more perspicuous and thus theoretically useful, I stipulate a brief description of those objects that fall under the extension of each broad category.
Social Kinds: that which is dependent upon human attitudesFootnote 4
Natural Kinds: that which is physical and not dependent upon human attitudes
Abstract Kinds: that which is neither physical nor dependent upon human attitudes
This ontological framework will be useful for our purposes if it is exhaustive of the kinds of objects that there are and has the epistemological consequences which I elaborate below. Because I developed this framework with only these goals in mind, the framework may be inappropriate for other theoretical purposes one might have. For instance, my gloss on natural kinds might be thought ‘too weak’ in that it does not specify what it takes for a collection to be a proper kind. One might complain that, on my framework, a collection of the planet Jupiter, all the binary star systems in the Andromeda Galaxy, and all the individual snowflakes atop Mont Blanc right now fall into the natural kind category.Footnote 5 The sense that my natural kind category is too permissive stems from the fact that the term natural kind is often reserved for talk of those collections that play an essential role in our best scientific theories, participate in natural laws, and so on. Developing a framework so that natural kinds can play those sorts of roles is useful for some theoretical purposes, such as developing a theory of reference, but this is not my purpose. My purpose is twofold: to suggest categories into which all particulars must fall and to suggest that it matters epistemically the category into which we place all the law particulars. My aim is not to police the boundaries of a proper kind.
Let us now turn to the epistemic consequences of the framework. Natural kinds, of which water is a paradigm example, have particulars that are physical and whose existence does not depend upon human attitudes. It would seem that these particulars are best investigated empirically. This is not just because the empirical sciences have “delivered the goods,”Footnote 6 as it were, with respect to investigating natural kinds, but also because it is hard to see how non-empirical methods could be expected to produce knowledge about these sorts of objects. Social kinds, of which money is a paradigm example, have particulars whose existence does depend in part upon human attitudes. It would seem that these are best investigated, in part, by probing the relevant human attitudes, which is also an empirical venture. Abstract kinds, of which propositions are a paradigm example, have particulars that are immaterial and whose characteristics, if discoverable at all, are discovered by a priori reflection.
In short, given what they are, we must investigate social kinds and abstract kinds differently. Knowing about those particulars under the social kind heading requires empirical investigation; whereas, knowing about those particulars that fall under the abstract kind heading requires non-empirical methods. Of course, this is a little too strong, for I could use an empirical method to find out about an abstract kind. Mathematical objects, we can assume, are abstract kinds, and I might interview some mathematicians to learn more about these objects. If I do that, I will have learned something about an abstract kind through an empirical method of investigation because interviewing people is a standard empirical technique. For this reason, it is not the case that every person must use non-empirical methods to learn about abstract kinds (or even that every person must use empirical methods to learn about social kinds).Footnote 7 Rather, we, as an inquiring community, must use empirical methods to learn about social kinds and must use non-empirical methods to learn about abstract kinds.Footnote 8 In my case about math, while I can defer to the mathematical experts to learn something, this method cannot be the sole or even primary means of learning about math, for someone must have begun the a priori investigation which is the precondition for someone having mathematical expertise.
In this qualified way, we should accept the difference in the method of investigation for social kinds and abstract kinds, and if we do, something else follows. The difference in the method of investigation suggests that, in order to know much about something, one needs first to discover whether it is a social kind or an abstract kind (or a natural kind); that is, one must answer the what-kind question. An answer to the what-kind question then dictates the direction of further research. Without answering the what-kind question, it is hard to see how further investigation is possible.
Putting it this way is a little too strong, for there are instances when we can know about something, even if we are agnostic about the kind of thing that it is. For instance, in moral theory, there is a dispute regarding the what-kind question with respect to moral properties. Some claim that moral stuffs are natural kindsFootnote 9 while others think they are abstract kinds.Footnote 10 This disagreement, however, does not undermine our (putative) knowledge that it is morally wrong to torture puppies for sport. In epistemology, we find a similar state of affairs. Some hold that knowledge is a natural kind,Footnote 11 while most contend that it is an abstract kind, but still, nearlyFootnote 12 everyone agrees and (arguably) knows that Charles cannot know that he has marijuana in his car, if he does not believe he has marijuana in his car.Footnote 13 These, then, are cases where one can have knowledge about something, even when no one may know the answer to the related what-kind question.
These kinds of case call for me to moderate the point about the urgency of the what-kind question, but, for three reasons, they do not call for us to give up the point entirely. First, our most well-reputed and successful fields of inquiry have settled the what-kind question for their respective objects of inquiry, and indeed, this enabled their success. To see that, imagine if one of our successful fields did not settle the question. Imagine the state of particle physics if some physicists were wedded to the idea that their objects of inquiry were somehow mind-dependent or thought they were abstract objects. When the natural-kinders propose building particle accelerators to try to detect subatomic particles, the abstract-kinders might just run a Moorean open-question argument. Knowledge would not advance in that setting. Second, we should still insist upon the need to answer the what-kind question because the ‘counterexamples’ are limited in number. How much can we learn about magnetism, Moldova, or modus ponens if we do not implicitly answer the what-kind question with respect to each? Third, the counterexamples can be explained away. In those cases where we can have knowledge about x without answering the what-kind question with respect to x, this owes in part to the vast agreement about the features of x. Normally, we do need to settle the what-kind question before any further inquiry can occur, but sometimes, we can bypass the question because widespread agreement provides evidence about the object of inquiry and may even provide some resources to carry the inquiry forward. Of course, when the situation is otherwise, when there is no widespread agreement to which one can appeal, moving forward with an inquiry requires settling the what-kind question.
In the case of law, answering the what-kind question looks pressing. When we focus on the kinds of bold claims with which I began, claims about whether law requires coercion or whether law requires secondary rules, we find little agreement. It follows that we do need to settle the what-kind question to be in a position to know whether these claims accurately describe law’s nature. In the following section, I look at various ways that one might settle the what-kind question with respect to law.
2. Methods of Deciding
Here I consider three methods to decide the what-kind question. For each, I suggest that the method does not tell in favor of any particular answer to the question. If this is right and if there are no alternative methods on offer, we really cannot make progress on downstream questions about law’s nature, and thus, we really do not know whether some of the ambitious claims about the nature of law are accurate descriptions.
2.1 It’s Obvious
“Isn’t it just obvious what-kind of thing law is?” This question encapsulates the first method I consider. When we consider most objects, we find the answer to the what-kind question completely obvious. For instance, I never find myself puzzling over whether mitochondria are abstract kinds or social kinds. It is obvious to me that, if they exist, they are natural kinds. A similar story might be told about arugula, black holes, and cerium. Is an answer to the what-kind question with respect to law just as obvious?
Some, like Brian Leiter, seem to think so. Leiter writes, “The concept of law is the concept of an artefact.”Footnote 14 If we can assume that all artifacts fall into the social kind category, then Leiter has proposed an answer to the what-kind question: law is a social kind. This answer is not merely the truth, contends Leiter; it is an obvious and uncontroversial truth. In fact, those who disagree require “psychological… investigation,”Footnote 15 or so he jests. Of course, I doubt the situation is so obvious.
To be as obvious as he says, there would have to be a striking lack of controversy, and pace Leiter, I do not think this exists. Leiter is surely correct that few openly deny that law is a social kind.Footnote 16 Nonetheless, many of the jurisprudents that Leiter cites as on his side (e.g. “Kelsen, Hart, Raz, Dickson and Shapiro”Footnote 17) performatively disagree with him. Instead of focusing on any one theorist in particular, I note a common jurisprudential methodology employed by nearly all of them, a priori analysis.Footnote 18 If law is a social kind, what use have we of that? Leiter’s answer is both simple and correct: none.Footnote 19 Of course, if this is right, we can run a modus tollens. That is, if we do need a priori analysis, law cannot be a social kind. Social kinds, just to repeat the remarks from above, do not require a priori analysis. In fact, it seems bizarre to attempt to learn about a social kind primarily through a priori reflection. Consider something that is unambiguously a social kind such as the game of basketball. Nobody thinks that we learn much about basketball through a priori analysis. This is so even though one can dream up ‘philosophical’ questions about basketball; for instance, one can ask whether we could eliminate the prohibition on traveling and still have basketball.Footnote 20 Despite such possibilities, basketball is a social kind that we should investigate primarily with empirical methods. The same should be true of law, were it a social kind too. The jurisprudents who forswear empirical methods and primarily employ aprioristic methods are implicitly committed to seeing law as an abstract kind that is epistemically accessible via their aprioristic methods. This is why I do not take it as obvious that law is a social kind.
Below, I consider two attempts to save the “it’s obvious” method of deciding the what-kind question. The first I call the Only Apparent Reply. According to this reply, jurisprudents’ methods are only apparently aprioristic. If the methods are not aprioristic, one cannot infer that jurisprudents are committed to seeing law as an abstract kind.Footnote 21 Why might one doubt that the methods are aprioristic? Well, so the reply continues, jurisprudents’ reliance on intuition-pumping thought experiments might be thought to be an instance of conceptual analysis, or the analysis of some linguistic community’s understanding of a given concept. The jurisprudents would, on this picture, be making an empirical judgment, based on their own experiences.Footnote 22
The Only Apparent Reply, though ingenious, raises a question: if jurisprudents are trying to figure out what our shared concepts are, why should they perform conceptual analysis from the armchair? Going out to ask people what they believe or using whatever empirical methods garner esteem from the best social scientists—these are vastly more reliable methods. If this observation about reliable methods is correct, it reveals a deep problem about the Only Apparent Reply. Recall the dialectic. My central claim in this section is that legal philosophers are not in an epistemic position to describe law’s nature. Specifically, I am arguing that legal philosophers find themselves in this unfortunate epistemic position because they cannot decide the what-kind question. However, it is open for me to claim, as well, that legal philosophers find themselves in this unfortunate epistemic position because they are using unreliable methods. This charge would be true if philosophers were purporting to discover a community’s beliefs by making educated guesses from their offices. In other words, the thought that jurisprudents ‘obviously’ see themselves as analyzing a social kind while using obviously unreliable methods for the investigation of social kinds makes it all the more plausible that jurisprudents are not in a position to accurately describe law’s nature.
The second way of rescuing the “it’s obvious” method might contend that my remarks about the ubiquity of a priori analysis work all too well. Given the ubiquity of a priori analysis in contemporary jurisprudence, maybe it is obvious that law is an abstract kind, or so this suggestion goes.
This response also fails. If many legal philosophers are committed to seeing law as an abstract kind while openly claiming that it is a social kind, this suggests that deep confusion reigns in thinking about law’s nature. Philosophers’ deep confusion in categorizing law as a social or abstract kind belies any claim that the answer to the what-kind question is obvious.
2.2 Truisms & Desiderata
Since the answer to the what-kind question is not obvious, we have to dig deeper, bring more theoretical resources to bear. We might try to answer the what-kind question by appealing to various uncontroversial claims about law. Call the general method of selecting a hypothesis about x by appeal to uncontroversial claims about x the Truisms and Desiderata method, or just T&D.Footnote 23 There are various ways to carry out T&D. Variation comes in determining how we come up with the uncontroversial claims, whether we think the correct hypothesis about law has to be consistent with all or most of the claims, and, if one answers ‘most’ to the previous question, in determining whether some subset of the claims has more weight than others such that it is more important for a hypothesis to accord with members of that subset of claims than with other members.
T&D is widely used in philosophy.Footnote 24 However, because philosophical methodology is an underdeveloped area, I have yet to see a sustained conversation about which variant of T&D is best. Here is not the venue to begin this kind of conversation either. In lieu of this, I offer a few reasons for proceeding as I do. To find the uncontroversial claims, I asked people. I take it that philosophers often consult their own intuitions, but I worry that doing that can unconsciously stack the deck in favor of whatever hypothesis one wants. In simply asking other people what they think is an uncontroversial claim about law, I did nothing approaching the rigors of good experimental philosophy, but I hope to have improved upon whatever might have resulted from consulting my own idiosyncratic intuitions. As to whether the best hypothesis must conform to all or most of the uncontroversial claims, I say ‘most.’ Like any reasonable person, I think that the nature of reality might be opaque to us, maybe opaque to a great many of us.Footnote 25 For that reason, we should remain open to the idea that a hypothesis about law can be correct even if it does not conform to everything we antecedently thought about law. Finally, while I do think that some of the uncontroversial statements about law are more core than others, we will not need decide which ones, because, as I argue below, the uncontroversial statements are consistent with both hypotheses about law, that law is a social kind, and that law is an abstract kind. If that is right, it shows that T&D is inadequate to solve the what-kind question.
Consider the following list of truisms about law.
(A) Law is a norm
(B) Law comes from humans
(C) Law creates some property regime
(D) Law prohibits some killing
(E) Law protects people
(F) Law reflects a society’s values
(G) Laws can be unjust
(H) Laws can change
(I) Laws regulate (external) behavior
(J) Legal systems enjoy authority
(K) Legal systems have punishment and coercion
(L) Legal systems have some adjudication function
(M) Legal systems must have some legitimating narrative
I contend that each of these is consistent with thinking of law as a social kind or as an abstract kind. Let us consider the first few. There seems to be no clear winner, social kind or abstract kind, when considering (A). That law (or anything) is a norm does not favor either answer as some norms like etiquette are social kinds, while other norms, like moral norms, are arguably abstract kinds. One might worry that (B) cannot be squared with the thought that law is an abstract kind; however, consider the following about numbers. Surely, the Arabic base-10 number system came from humans, but that does not mean that arithmetic depends, in any interesting way, on humans. A similar story might be offered for the case of law. For (C), (D), (F), (G), and (I) these are just claims about the content of law, and it would seem that these sorts of content claims do not favor a particular answer to the what-kind question.
I skipped over (E), which initially seems hard to square with the idea that law is an abstract kind. (E) looks hard to square because it attributes causal power to law, but abstract kinds, qua non-natural stuffs, are not thought to have causal efficacy at all. One option for the abstract kinder is to go full-blown Platonist and to insist that abstract kinds, despite being non-natural, do have causal efficacy.Footnote 26 A more promising option for the abstract kind advocate is to note that it is debatable whether law itself causes anything. It is, therefore, far less contentious to maintain that certain widespread attitudes toward law cause (or constitute) safety. Putting it this way enables the abstract kinder to capture the flavor of (E) in a different way. Instead of claiming that law itself protects people, maybe one could claim (E*).
(E*) A condition of general obedience to law is (oftenFootnote 27) safer than its opposite.
(E*) avoids the strange causal attribution and is thereby more plausible than (E). I take it that (E*) captures the spirit of (E) because it still countenances the familiar notion, traceable to Hobbes,Footnote 28 that living under a legal system promises greater physical safety to average persons than they would enjoy, absent a legal system. (E*) should, thus, replace (E) on our list. When it does, we see that there is no problem for the abstract kind advocate because (E*) is equally compatible with law as an abstract kind or a social kind.Footnote 29
(H) is another statement that looks like a trouble spot for the abstract kind advocate because abstract kinds are classically conceived as changeless.Footnote 30 One could abandon the thought that all abstract kinds are changeless. If, however, the abstract kind advocate wanted to pursue a less controversial strategy, one might understand legal change as a change in the applicability of a given legal norm. The abstract kind advocate can thus account for legal change while maintaining that law, in a certain sense, is changeless. The remaining question is whether something is lost in speaking this way as opposed to the way from (H). I doubt this, and if I am right, (H) is another truism that is equally compatible with thinking of law as a social kind or an abstract kind.
I leave the task of determining how the rest of the truisms fare to the reader. The foregoing has made it all but undeniable that the above set of truisms is consistent with both answers to the what-kind question. One more thing should be said before concluding this section. It might be held that, while one cannot solve the what-kind question by appealing to the truisms, one can settle other questions about law’s nature by reference to these same truisms. In other words, this challenge asks whether T&D itself is a way to circumvent the what-kind question and to directly answer the skeptical challenge I raise. The short answer: probably not. If there were a straightforward set of inferences from truisms to controversial conclusions about law’s nature, the debates mentioned at the outset of the paper would not persist.Footnote 31
2.3 Appeals to Simplicity and Naturalism
It might seem that one could easily answer the what-kind question by just denying that there are abstract kinds in the first place. If there are no abstract kinds, a fortiori, law is not one; if law is not an abstract kind, it is a social kind—or so an argument might go.Footnote 32 A concern for ontological simplicity and a thoroughgoing naturalism would motivate the first antecedent and thus carry us to the claim that law had better be a social kind. Is there anything to be said to block this strategy?
I have nothing general to say about the viability of thoroughgoing naturalism of this sort, so my response will not take a position on the existence of abstract kinds. Instead, I question the inference from “there are no abstract kinds” to “law is a social kind.” To see the problem with this inference pattern, consider a different case. Consider what thoroughgoing physicalists might say about numbers.Footnote 33 They would deny that numbers are abstract kinds, but they might insist that numbers, were they to exist, would be abstract kinds. Accordingly, they might strongly deny that numbers are social or natural kinds; instead, they would contend that numbers do not exist at all. The numbers case shows that mere skepticism about abstract kinds is not enough to establish that something, thought to be an abstract kind, must fall into some other category of kinds. Eliminativism is a viable strategy too.Footnote 34
To return to thinking about law, naturalism is not enough to show that law must be a social kind, for the very same naturalism licenses us to be eliminativists about law. To round out the argument, then, more would have to be done to explain why we should think of law as a social kind as opposed to advocating for eliminativism about law. Without this work done, there is no easy way to resolve the what-kind question and thus we cannot know much about law’s nature.
Of course, it might be doubted whether eliminativism really is a reasonable option. Let me elucidate this doubt before explaining where it goes wrong. The person who worries about eliminativism might begin by noting that it surely seems as if law structures a good bit of our reality. Therefore, it seems like, if we are sure of anything, we are sure that there is law. Any argument purporting that there is no law is on shakier ground than one contending that there is law.
I can begin to diffuse this sort of worry by mentioning an analogous case in the philosophy of race. This is a debate about whether race is a natural kind or social kind.Footnote 35 Some people in this debate think that race is the kind of thing that, if it existed, it would be a natural kind, just given how we think of it. These same people, however, doubt that race exists; they are eliminativists about race. Thus, racial eliminativists can wholly agree that “Race has profoundly shaped life in the Americas and elsewhere from 1600 onward,” but they think such sentences are actually loose talk. Technically speaking, race has not shaped anything because it does not exist, but false beliefs—that there are races and that members of these races have particular characteristics—those have profoundly shaped life.
This example about race goes to show that eliminativism even about something that appears to have left an indelible mark upon the world can be made plausible. Perhaps the same might be true of eliminativism about law. Here is not the place to substantiate the claim. All I contend here is that eliminativism is a viable strategy, and if it is viable, naturalism alone is not sufficient grounds for concluding that law is a social kind. There needs to be an independent argument for explaining why naturalism should not lead us to thinking that there is no such thing as law, after all. Until such an argument is furnished, we cannot know much about law’s nature because we cannot answer the what-kind question.
Before concluding this section, I raise and respond to one last attempt to show that eliminativism about law is non-viable. I consider this attempt simply because it is suggested by our conversation thus far. One might think the list of truisms itself provides ample evidence that eliminativism cannot be true. How could law have any of the characteristics in (A)-(M), if it does not exist? Here is the very beginning of a response. Perhaps, the eliminativist can assent to all of these truisms by replacing law itself with false beliefs people have about law. For instance, consider (E*), which claims that a condition of general obedience to law is often safer than its opposite. The eliminativist can claim,
(E**) A condition of general obedience to what people believe to be law is often safer than its opposite.
(E**) preserves the general notion underlying both (E) and (E*), and (E**) might be thought to be more explanatory than either of these since it tries to explain intentional actions via people’s mental states, not through external stimuli alone. This response, to be clear, is not a vindication of eliminativism, but it gestures at the sort of strategy eliminativists can employ. If that type of strategy succeeds, it proves that naturalism and simplicity concerns do not show that law must be a social kind. Law could be the kind of thing that, if it existed, it would be an abstract kind, but as it happens, law does not exist since there are no abstract kinds.
Finally, the eliminativist strategy that I have outlined not only shows that the naturalist cannot quickly conclude that law is a social kind; it also presents another reason to worry about our ability to describe law’s nature. One does not only need to settle the what-kind question, one also needs to show that law exists in the first place.
3. Descriptivism Rebutted
Having shown the great difficulty jurisprudents face in answering the what-kind question and demonstrated the skeptical upshot of this for the project of describing law’s nature, now, we can move to the next phase of the argument. Here, I explain why the skeptical upshot has repercussions for the descriptivist interpretation of jurisprudential discourse.
Descriptivism, as I understand the term here, is an interpretive strategy for understanding a particular domain of discourse, and it has two components, a semantic component and a normative component. According to the semantic component, to be a descriptivist about a discourse D means to think that sentences in D Footnote 36 are propositions, propositions that refer to entities and properties that the surface grammar of the sentence indicates. In short, to be a descriptivist about D is to be a cognitivist about the semantics of D-sentences. According to the normative component, to be a descriptivist about D means to think that those who assert D-sentences err insofar as they would be epistemically unjustified in holding that those D-sentences name states of affairs which obtain. In short, to be a descriptivist is to think that a “robustly epistemic”Footnote 37 norm governs D-sentences. I put this vaguely because there is no need to precisify the specific epistemic norm which descriptivism requires. It could be the popular Knowledge Norm of Assertion, which claims that one should assert p only if one knows that p.Footnote 38 Alternatively, it could be something less demanding like Jennifer Lackey’s Reasonable to Believe Norm of Assertion, which claims that one “should assert that p only if (i) it is reasonable for one to believe that p, and (ii) if one asserted that p, one would assert that p at least in part because it is reasonable for one to believe that p.”Footnote 39
This stipulated definition of descriptivism as an interpretive strategy aims to capture the intuitive notion of what it is like to understand someone as trying to describe something. When we understand someone as trying to describe something, we see them as saying something about the world, something that could be accurate of the world or not. Also, when reading someone as trying to describe something, we think the person is criticizable if that person has been epistemically irresponsible in offering something as a description.
Having explained descriptivism generally, it should be plain what a descriptivist interpretation of jurisprudential conversations involves. It should also be plain that a descriptivist interpretation is, by far, the most natural reading of such discourse. And it should also be plain that descriptivism suffers as an interpretation of jurisprudential discourse. While descriptivism does well in terms of glossing the semantics, it fails on the normative side because of the epistemic problems that attend trying to describe law’s nature, as I demonstrated in the previous section. Why, one might ask, does a good interpretation have to minimize normative errors, as I have suggested?
In choosing an interpretation of jurisprudential discourse (or any other practice), we should be wary of those interpretations that suggest participants are routinely not following the rules, especially if, on another interpretation, participants are great at following rules, just different rules for a slightly different practice. This interpretive principle is a version of Donald Davidson’s principle of charity. For Davidson, employing the principle of charity is a precondition for interpreting anyone at all. One might embrace the principle on that basis; the thought would be that the principle of charity is non-optional. There is another, less ambitious reason for adhering to the principle of charity. Unless we adopt this principle, that is, unless we see participants’ behavior as largely correct, we cannot see the practice (as practiced) as worth salvaging. In the abstract, this consideration may not seem important, but when the practice is one to which we, ourselves, are deeply committed, it is obviously important to see our own favored practices as worthwhile. Thus, if we can help it, we should not see ourselves as engaged in a practice for which we all regularly flout the internal norms.
4. Two Non-Descriptive Reads
At this point, I have shown that, if jurisprudents continue making bold claims about law’s nature, we should not see them as making descriptive claims, lest we attribute to them epistemic error. Another way of putting the point: if we, ourselves, have been making bold claims about law’s nature, we must re-cast these claims as something other than description.
In this final substantive section of the paper, I turn to discussing two alternative ways we might understand claims about law’s nature. Though there are other logically possible options, there are really two main contenders. The most familiar non-descriptive way to understand claims is in an expressivist manner, that is, to see the relevant claims as expressions of conative states such as desiresFootnote 40 or commitment to plans.Footnote 41 Expressivism has a long and illustrious history in moral theory,Footnote 42 and it has begun to surface in jurisprudential discussion as well.Footnote 43 For reasons I discuss below, an expressivist take on debates in jurisprudence presents difficulties. Given those difficulties, I advocate for a pragmatist understanding of claims about law’s nature. On a pragmatist read, jurisprudential claims are suggestions about how to view law, suggestions that are to be assessed on the basis of practical reasons.
4.1 Expressivism—Boo!
Many views go under the heading of expressivism,Footnote 44 and many figures call themselves expressivist. Consequently, it is hard to say much about expressivism without becoming mired in exegesis and making qualifications. Here is not the place for that work, so instead, I stipulate a definition of expressivism below and try to show that it does not name a promising alternative to thinking of jurisprudential claims as descriptions.
To give an expressivist understanding of a person’s judgment is to see the judgment as primarily functioning to express a non-descriptive attitude of the personFootnote 45 and not to describe the state of affairs indicated by the surface grammar of the judgment. Consider a simple example. Suppose I were to judge that Riesling is good. On an expressivist understanding, that judgment primarily functions to express one of my non-descriptive attitudes, perhaps my warm feelings toward Riesling, my plan to drink Riesling in such-and-so circumstances, or my desire for others to drink Riesling. On any expressivist understanding, the judgment certainly does not primarily function to describe Riesling or to link it with the property of goodness (or the properties that constitute goodness-for-Riesling).
Expressivism is not a theory or interpretative strategy for one-off judgments. Instead, it is used for entire domains of discourse. Though it is most common to be expressivist about moral discourse, one might employ that strategy for theological discourse,Footnote 46 logic,Footnote 47 or even globally. I mention expressivism’s generality because sometimes the right way to understand a speaker’s statement, despite the surface grammar, is as an expression of something about that person. For instance, if someone pours a glass of Riesling in my presence but does not offer me any, I might say, “Riesling sure is delicious.” In this case, a hearer might rightly understand my statement to express my desire to drink a glass of Riesling, but this alone does not render the hearer an expressivist. If the hearer held that all aesthetic language were expressive and not descriptive, this would amount to expressivism. Expressivism does not concern what is implicated Footnote 48 by a given statement; indeed, it is semantic theory about certain kinds of discourse.
Having offered a characterization of expressivism, now let us turn to explaining how an expressivist read of jurisprudential language avoids the problems of descriptivism with respect to law’s nature. If in judging that, for instance, coercion is central to law’s nature, theorists are not trying to describe law but rather trying to express a desire or some other non-descriptive attitude of theirs, it does not matter that a theorist does not know much about law’s nature. The epistemic failing that one would make, were one engaged in an act of description, is simply not attributable to her. Thus, expressivism is a semantic escape route if we want to continue making claims about law’s nature.
As a semantic theory, expressivism has been prey to much criticism. Here, I mention just three worries. None of these criticisms is new, and none of them may ultimately prove devastating, but these are reasons to prefer an interpretative strategy that, all else equal, does not incur these problems. As I argue in the next section, pragmatism is such a theory. For now, I turn to the criticisms of expressivism.
The first worry, a problem for any kind of expressivism, is how to make the relevant judgments, when construed expressivistically, act as truthbearers.Footnote 49 This worry is best seen by considering a simple example. Suppose the judgment “The norm L is a law” is construed expressivistically such that the judgment expresses the judger’s plan to follow L. If this is how the judgment is to be construed, it seems rather obvious that the judgment is neither true nor false, since plans are neither true nor false. Plans might be well- or ill-conceived; they may rely on falsehood, which could make them ill-conceived, but, strictly speaking, they cannot be true or false. Plans are not truthbearers and do not act as such. This is not yet a problem. The problem emerges when one notes that some of the judgments that expressivists have read expressivistically do act as truthbearers, such that one might infer that the expressivists are wrong in interpreting the judgments as they have. The most famous instance of this kind of problem is the Frege-Geach Problem.Footnote 50 Here is an instance of a Frege-Geach Problem, using our simple example.
Premise 1: L is a law or 2+2=5
Premise 2: It is not the case that 2+2=5
Conclusion: L is a law
This looks like a good inference, but it is not clear how it could be so, if we understand “L is a law” in the expressivist way. The argument above looks like an instance of the following schema.
Premise 1`: p or q
Premise 2`: Not-q
Conclusion`: p
But it is not. “L is a law” is not a truthbearer; it is a plan. As a plan, it cannot be plugged into a slot where only a truthbearer can go. This particular version of the Frege-Geach problem is known in the literature as an embedding problem. There are other problems associated with making expressivistically-understood judgments act as truthbearers. These include problems about negation and contradiction. There are proposed solutions for the different facets of this general problem (and rebuttals), but here is not the place for probing. My aim is not to disprove expressivism but rather to saddle it with problems that my preferred theory does not face.
The second problem with expressivism is what I call the Massive Error Theory Problem.Footnote 51 Again, I refer to our simple example to illustrate this. If the average legal theorist, much less the average person, says, “L is a law,” one likely takes oneself to say something about law, not to express one’s own plans. To interpret this judgment expressivistically is to claim that the average person who makes such judgments is in deep error about one’s own thoughts. It is a general principle within philosophy to regard massive error theories with suspicion because our consensus in favor of the relevant beliefs, our successful navigation of the world using those beliefs, and other things besides are all evidence counting in favor of the beliefs. Of course, error theories are not always wrong, so it is not exactly a criticism to mention that expressivism involves an error theory. The criticism comes in thinking about the scale of this particular error theory versus its payoff. If the expressivist suggestion is to view ordinary claims about law’s nature as expressing a theorist’s conative states, we get a situation where nearly every jurisprudent in the world is deeply mistaken about thoughts in that person’s own head. That is the scale. What is the payoff? The payoff is that now we do not view every jurisprudent as making a different systematic epistemic error. This payoff just swaps one large error for another. If a theory could have this same payoff with a smaller error theory, it would be an advantage.Footnote 52
The third and final problem I address focuses specifically on expressivism with respect to claims about law’s nature. This problem has analogues elsewhere but is particularly pressing in this domain. What is the specific attitude that claims about law’s nature should be thought to express? The suggestions of early ethical expressivists are clearly inadequate to the task. Claims about the centrality of coercion to law do not seem like saying “Coercion, boo!” or “Coercion, rah!” Something more sophisticated is being said, but what? In ethical discourse, connecting ethical judgment to a certain set of conative states (e.g. desiring, planning to live a life in a certain way, or commanding others to behave in a certain manner) has a kind of intuitive appeal because doing so can explain ethical motivation. No analogue is present in the case of jurisprudential language; thus, it is hard to know where to start. I leave it as a standing challenge for the expressivist to explain which attitudes are being expressed by jurisprudential language, given that really simple answers look implausible and that motivation provides no helpful theoretical starting point.
4.2 In Praise of Pragmatism
An inability to answer the what-kind question led us away from seeing claims about law’s nature as descriptions; the three worries raised in the last section led us from seeing claims about law’s nature as non-cognitive expressions of the person making the claim. Here I offer a new way to see claims about law’s nature, a pragmatist way.
It is an old saw to note that pragmatism can be understood in various ways.Footnote 53 At the outset of this article, I offered a specific stipulation of the term. A pragmatist interpretation of jurisprudential discourse contends that jurisprudential discourse, including the bold claims, are claims to be adjudicated by practical reasons. Before explaining why this interpretation is to be preferred to the descriptivist and expressivist interpretations, it may help to compare my stipulative understanding of pragmatism to other, perhaps more familiar, understandings. This comparative, expository work will facilitate the persuasive work ahead.
What I mean by pragmatism bears little relation to the pragmatist theory of truth, most often associated with the term. The question here has all along been about interpretation—how we ought to interpret claims about the nature of law. The pragmatist theory of truth offers conditions under which a truthbearer, such as a proposition, is true. A well-known slogan provides those conditions: “truth is what works.”Footnote 54 If we wish to make that idea more specific and manageable, one might claim that, on this theory, p is true if and only if believing that p is optimific or resolves doubt. However, I reject the pragmatist theory of truth, which is in keeping with several notable pragmatists. The present effort simply does not concern the metaphysics of truth.
Historically, pragmatism also has been thought “a doctrine concerning the meaning of propositions.”Footnote 55 This is closer to my project. Pragmatism as a theory of meaning has a variety of iterations. On one iteration, the meaning of p is determined by the set of consequences an agent expects upon believing that p. On another iteration, the meaning of p is determined by the ways that an agent uses p. There are, no doubt, various critiques one might raise to both of these, but since my goal is to distinguish my version of pragmatism, not to disparage others, I forgo any reproof. How similar is my project to either version of meaning-pragmatism? While my pragmatism concerns interpreting jurisprudential discourse, it is not a theory of meaning. I do not claim that jurisprudential judgments mean this or that; instead, I offer a norm by which to assess them. (Of course, the judgments have to mean something so that they can be assessed, but the pragmatism on offer here does not provide that. More on this in a moment!)
Some pragmatists have been very clear that they are not offering theories about the metaphysics of truth or semantic theories; instead, they claim to offer a metaphilosophical view about how inquiry is to be conducted. Such “inquiry-pragmatists” argue that we ought to assess claims by the practical consequences of believing them or acceptingFootnote 56 them. To fix ideas, consider the famous passage in Pascal’s Pensées where he argues for the claim “God exists” on the basis of an argument about the expected utility of believing that God exists.Footnote 57 Pascal does not claim that God’s existence depends on the expected utility calculation; Pascal does not even say (à la the pragmatist theory of truth) that the truth of “God’s exists” depends on the expected utility calculation. Instead, the claim is that the norm for settling the question of whether God exists is expected utility. What I am calling an inquiry-pragmatist approves of this way of assessing claims. Of course, individual pragmatists within this type might disagree with utility as the proper criterion of assessment, but all such pragmatists will urge using practical norms to decide the given questions.
Pragmatism as a norm of inquiry is identical to the view I advocate here. Admittedly, my advocacy has been oblique. The question I have asked is how jurisprudential discourse is to be interpreted, and I am in the course of arguing that we should select the pragmatist interpretation. By the pragmatist interpretation, I have meant that we should see jurisprudential claims as claims to be adjudicated by practical reasons. This should already make the connection between my view and inquiry-pragmatism close, but consider another fact. Jurisprudence is a field of inquiry. Thus, I am advocating that, for a specific field of inquiry, claims made within that field should be assessed by practical reasons.
As the previous point makes clear, inquiry-pragmatism can be limited to a specific domain of discourse. One need not adopt a global version of it. One could, as I have, advocate for pragmatism for reasons that are specific to a domain.
I make one last expository point. I am far from the first person to argue very explicitly that jurisprudential claims should be assessed by practical criteria. One of the first was Liam Murphy.Footnote 58 Other advocates include Natalie StoljarFootnote 59 and Juan Carlos Bayón.Footnote 60 Ronald Dworkin’s views in jurisprudence bear some resemblance to my own, especially insofar as we both take issue with jurisprudence as a quest to describe the nature of law.Footnote 61 However, assessing the closeness of the positions is a tough exegetical task best left for another day.Footnote 62
Having offered a characterization of pragmatism, now let us turn to explaining how a pragmatist reading of jurisprudential language avoids the problems of descriptivism with respect to law’s nature. For the pragmatist, a theorist is not answerable to the same norms as one would be when engaged in description. Thus, for the pragmatist, the mere fact that one does not meet the epistemic norms internal to description is no problem. It is as if one’s journey is guided by a different treasure map, so the fact that one cannot reach the descriptivist’s destination is irrelevant. It is important to stress that, unlike the expressivist, the pragmatist has no metaphysical story about why norms of description do not apply to jurisprudential language. The expressivist has a semantic theory about particular judgments: such judgments have different contents than what the descriptivist claims, and that is why the norms of description fail to apply. The pragmatist is quietist about the semantics.Footnote 63 The pragmatist can say that the judgments are not descriptions, but all that amounts to is a normative claim, namely a denial that the norms of description apply. Thus, pragmatism is a normative escape route if we want to continue making claims about law’s nature.
Between pragmatism and expressivism, pragmatism is the preferable alternative to descriptivism about jurisprudential language because the three main problems for expressivism are largely absent. The first problem, the problem of getting the content to behave as truthbearers despite not being truthbearers, does not arise for pragmatism. Since pragmatism offers no semantic claim, there is no worry about getting it wrong. Pragmatists can help themselves to whatever turns out to be the best gloss of the semantics.Footnote 64 The second problem, the Massive Error Theory Problem, remains but is much diminished. Pragmatists need not to insist that jurisprudents are wrong about their own judgments, as the expressivist must say. Instead, the pragmatist claims that some jurisprudents are wrong about the norms applicable to their judgments. Of course, to be wrong about the standards of correctness within a domain of discourse is a great failing, but it is clearer how one might become confused about this as opposed to confused about what judgments one is, in fact, making. The third problem, that of specifying the conative attitudes that jurisprudential judgments express, is totally absent for the pragmatist. Again, one can help oneself to whatever semantic theory offers the best theoretical benefits.
5. Conclusion
In sum, we should view bold claims about law’s nature as practical claims, that is, as claims to be assessed on the basis of practical reasons. Another way to say the same thing: we should understand these claims as ones about how to view law, irrespective of its true nature. This is what I have called a pragmatist understanding of jurisprudential language. This understanding is superior to the more familiar descriptive strategy because, were we to see the claims that way, we would understand jurisprudents as falling into epistemic error, specifically, we would see them as violating the norms internal to any act of description. The pragmatist understanding is also superior to a rival non-descriptive strategy, namely expressivism. I have claimed that expressivism faces grave theoretical problems that either fail to arise or loom less largely for pragmatism.
