Hybrid Dispositionalism and the Law

Among the many things that humans disagree about, disputes over socalled matters of fact are often described as an idealized paradigm— there’s an objective fact of the matter. Either we get the facts right or we don’t. And when we disagree, we can’t both be right. At least one of us is wrong when I believe that p and you believe that notp. In turn, this requires the existence of a common content of belief over which we can disagree. This idealization is simplistic and unrealistic. First, it is not clear if it applies to all “matters of fact.”1 Second, other concerns arise once we focus on presumably more anthropocentric issues. For example, in spite of the saying de gustibus non est disputandum, disagreements


I. Introduction
Among the many things that humans disagree about, disputes over so-called matters of fact are often described as an idealized paradigm-there's an objective fact of the matter. Either we get the facts right or we don't. And when we disagree, we can't both be right. At least one of us is wrong when I believe that p and you believe that not-p. In turn, this requires the existence of a common content of belief over which we can disagree.
This idealization is simplistic and unrealistic. First, it is not clear if it applies to all "matters of fact." 1 Second, other concerns arise once we focus on presumably more anthropocentric issues. For example, in spite of the saying de gustibus non est disputandum, disagreements over matters of taste are pervasive in our social life. If we have different standards of taste, it is pretentious to insist that one of us is wrong, or that there is a single right answer about what is of good taste. Similar pervasive and persistent disagreements occur in other domains, from aesthetics to knowledge attributions. The simple model described above does not appear to be well-suited to all such disputes. It is questionable if in each of those domains there is one single correct answer to the question of what is the case. It is not clear that in all cases either one gets the facts right or one doesn't. In the last decade, there has been an intense debate about whether disagreement without fault is possible. 2 These concerns intersect various kinds of questions. First, there's the metaphysical question of what, if anything, distinguishes matters of fact from matters of value, and of matters of personal taste. Second, there is the epistemic question of whether it is the case that we can disagree without either of us being at fault. Also, there is the semantic question of what the content of our mental states and speech acts must be for us to disagree or have a conflict.
One possible answer to the metaphysical question makes these facts dependent on our perspectives. For instance, one could say after Hume that beauty is "no quality in things themselves, " or that moral values are projections we make onto the world. To capture the perspectival nature of taste, beauty, or moral value, various authors argue for dispositional or response-dependent theories of value. 3 Different answers to the metaphysical question recommend different answers to the semantic question. It is possible that in some domains there is near universal convergence on responses that have an appearance of objectivity. But it is hard to make the case that, across the board, from personal taste to morality, there's one right answer for each evaluative or normative question. Insofar as answers to the metaphysical question recommend the dependence of normative and value facts on our perspectives, affective responses, or dispositions, the answer to the semantic question about the content of normative discourse and thought may likewise require semantic dependence on perspectives or standards.
However, semantic solutions that incorporate ontological perspective-dependence face substantial objections. One central objection concerns the explanation of speakers' intuitions about disagreement: if two people do not refer to the same standards, how can they intelligibly disagree? A theory that makes norms and values dependent on the perspectives from which they are valid puts forward contents that can be accepted by anyone. If we can all accept the same contents, we apparently don't disagree. Another objection concerns the motivational force of evaluative and normative discourse and thought. It seems that it is possible to accept that something is a value, or a norm, given certain standards or sets of norms, and not be motivated to act by that value or norm. Yet, evaluative and normative thought is motivational in a way that merely descriptive thought is not.
In Section II, I offer a brief summary of the problem of legal disagreements for social or collective accounts of the Law. In Section III, I present hybrid dispositionalism, and show how it offers an account of three core features of evaluative discourse, and solutions for resilient conflicts and disagreements. My view combines a contextualist semantic account In Law's Empire, Dworkin illustrates the argument with the famous example of Riggs v. Palmer. In this New York state civil court case, Mrs Riggs and Mrs Preston tried to invalidate their father's will. The defendant of the case, Elmer E. Palmer, had killed his grandfather to inherit the estate. At the time, the New York statute of wills said nothing explicitly about whether someone named in a will could inherit if he had murdered the testator. Palmer's aunts, Riggs and Preston, sued the administrator of the will, demanding that they, and not Palmer, should inherit the property. Judge Gray, who wrote the dissenting opinion in the trial, argued for a literal interpretation of the law, which made no exception for murderers, and argued that Palmer was already being punished for the murder. Judge Earl, who wrote for the majority, argued that the legislators couldn't have intended murderers to benefit from their crimes. The statute, on this interpretation, would not be merely the written text, but rather that text interpreted according to the intentions of the legislators, which would exclude the interpretation on which a murderer could benefit from the crime. Judges Gray and Earl hence had different interpretations of what the law fundamentally is: If two lawyers are actually following different rules in using the word "law, " using different factual criteria to decide when a proposition of law is true or false, then each must mean something different from the other when he says what the law is. Earl and Gray must mean different things when they claim or deny that the law permits murderers to inherit: Earl means that his grounds for law are or are not satisfied, Gray has in mind his own grounds, not Earl's. So the two judges are not really disagreeing about anything when one denies and the other asserts this proposition. They are only talking past one another. Their arguments are pointless in the most trivial and irritating way, like an argument about banks when one person has in mind savings banks and the other riverbanks. (Dworkin, 1986, 43-44) Theoretical disputes raise a challenge to legal positivism-legal officials often and fundamentally disagree about what norms they have a duty to apply. Since there is no single set of obligations that is accepted by all legal officials, there should be no law. But there is. Hence, the law cannot be merely what is accepted as such by legal officials. Dworkin's argument fundamentally rests on the premise that the best way to explain how an exchange between two speakers expresses a disagreement is to suppose that speakers mean the same thingthat is, express the same concepts-with the words they use in that exchange.
In the alternative, Dworkin argues for interpretivism. Dworkin claimed that the concept Law, like other concepts (Democracy, Equality, Freedom), is an interpretative concept, and that in theoretical disputes participants disagree about the concept's correct application. Interpretivism is the thesis that the law includes not only the rules accepted by a community, but also "the principles that provide the best moral justification for those enacted rules" (Dworkin 2011, 402). Interpretivism contradicts legal positivism because the content and existence of the law is to be determined by things other than social facts about what legal officials are disposed to accept.

III. Hybrid Dispositionalism
In recent work (Marques 2016a), I argue for a hybrid form of contextualism about aesthetic predicates. My account is committed to a dispositional metaphysical account of aesthetic properties, which is close to Lewis's 1989 conditionally relative dispositional theory. Lewis's theory allows for value pluralism, but does not require it. It also allows us to be mistaken about what is a value. Both are desirable features in a theory of value. I suggest that a semantic contextualist semantic implementation of Lewisian dispositionalism about values has the resources for a comprehensive explanation of value discourse: it can explain its cognitive role, its expressive role, and its connection-building role. Moreover, a contextualist implementation of dispositionalism allows people that prima facie disagree and are in conflict to denote different value properties. People may simply not share standards. This gives a reply to objections to contextualism similar to Dworkin's objection to legal positivism: that in situations like this, people "are only talking past one another." In that chapter, I argued that the action-guiding nature of the dispositional value properties denoted in discourse, when supplemented by expressive conversational implicatures, explains unifying features of evaluative and normative discourse. Moreover, the additional expressive dimension can be the focus of resilient normative or evaluative disagreements.

A. Dispositionalism and Contextualism
Many philosophers have thought that facts about values are tied to facts about us, i.e., to the responses elicited from us under certain conditions. David Lewis's (2000) dispositional account of value offers one way to cash out this idea. For Lewis, "something of the appropriate category is a value if and only if we would be disposed to value it under ideal conditions" (Lewis, 2000, 103), and a value property is something of the following kind: x is a value iff x is disposed to elicit response R in group of agents G in circumstances C.
Or, to the same effect, x is a value iff we are disposed to give response R to x in circumstances C.
For Lewis, the elicited response R is a certain kind of motivational, action guiding, state, desiring to desire. We are the relevant group of agents G, and C are conditions of full imaginative acquaintance. Although Lewis's account has been disputed, it has virtues. The theory is cognitivist, naturalist, internalist about motivation, and it is subjectivist, in that it is tied to facts about us (Lewis, 2000, 68-69). The view is also conditionally relative: I say X is a value; I mean that all mankind are disposed to value X; or anyway all nowadays are; or anyway all nowadays are except maybe some peculiar people on distant islands . . . or anyway you and I, talking here and now are; or anyway I am . . . What I mean to commit myself to is conditionally relative: relative if need be, but absolute otherwise. (Lewis, 2000, 129) Which semantic implementation fits conditionally relative dispositionalism? Semantic invariantism could be adopted under the empirical precondition that there is universal (or near universal) convergence in the relevant responses under ideal conditions (of imaginative  4 But there is plenty of evidence against uniformity in evaluations. Taste and humor are clear cases; and there is cross-cultural variation in specific moral codes. 5 If this is right, it puts pressure on any account that postulates a uniform right answer to all matters of value. Semantic contextualism can accommodate this possible variability in value standards. 6 Contextualism about evaluative predicates holds, first, that they have denotation relative to an evaluative standard. A value standard may be absolute or not. If it is absolute, it does not vary with context. If it is not absolute, some feature of the context of the use of the word must contribute to determine the relevant standard of the context. In spite of contextual variation, a dispositional contextualist theory should accommodate the core features of value terms that justify the membership of a word in the category evaluative term. The core features are, in summary, that value terms communicate cognitive descriptive contents, they play an expressive and motivational role, and finally that uses of evaluative discourse contribute to build connections among discourse participants. Arguments for contextualism about evaluative discourse are often supported by semantic considerations. Recently, McNally andStojanovic (2014 andLiao et al. (2016), for instance, have investigated common features of value adjectives that support a context-dependent treatment, specifically of aesthetic and personal taste adjectives. Drawing on work on gradable adjectives, McNally and Stojanovic (2014) offer a taxonomy of aesthetic adjectives. They argue that some of the features that are typical of relative gradable adjectives are also present in aesthetic adjectives. 7 Relative gradable adjectives, such as "big" or "long, " are multidimensional. The same object can be big in some respect, dimension, or with respect to some comparison class, but not on other dimensions or comparison classes. Relative gradable adjectives are also measurable: they have degrees and thresholds of application that vary with contexts. The context of use of a relative gradable adjective requires that relevant parameters be supplied by the local conversational context. We can say that something is big for an X, we can use modifiers such as very big, comparatives such as X is bigger than Y, etc. Relative gradable adjectives also allow certain inferential patters. For instance, "Mary is richer than Betty" entails neither "Mary is rich" nor "Betty is not rich" (Kennedy and McNally (2005). Now, value terms such as "good" or "beautiful" also admit modifications with "very, " "more . . . than . . . , " ". . . in some respect, " and "for an X." "Good" is to "rich" like "better" is to "richer." The same patterns of inference of relative gradable adjectives are also manifest with aesthetic and taste predicates. For instance, "Bob is prettier than Meg" similarly entails neither "Bob is pretty" nor "Meg is not pretty." 4 For a defense of such a view for aesthetic and taste predicates, see Schafer (2011). 5 See for instance recent results in Sarkissian et al. 2011, and(Khoo andKnobe 2016). 6 (Marques 2016a); (Marques and García-Carpintero 2014); (Brogaard 2008); and (Schaffer 2009) argue for contextualism about value predicates. Contextualism about the semantics of evaluative and normative language is sometimes classified as relativism (cf. Harman 1975;and Dreier 1999). 7 In recent presentations, Stojanovic has proposed to generalize the taxonomy to different evaluatives. Liao et al. (2016) make the case that aesthetic adjectives don't pair neatly with relative or absolute gradable adjectives. Aesthetic adjectives are like relative gradable adjectives in the sense that their standards of application derive from aesthetic comparison classes. But they are unlike relative gradable adjectives because the relevant comparison classes are not contingent on the immediate situational context of their use. One of the main objections to semantic contextualism about evaluative and normative discourse appeals to intuitions about disagreement that competent speakers allegedly share. 8 The examples typically report an interchange that prima facie presents a real disagreement. The contextualist analysis would arguably deprive interlocutors of a common subject matter over which to disagree.
Also, Andy Egan (2012) challenges the contextualist semantic implementation of a dispositional theory of value. His objection rests on the possibility of non-converging dispositions. Egan rightly argues that it is not clear that everybody would desire to desire alike under the same conditions of full imaginative acquaintance. If "we" refers to everybody, then there is a risk that there will not be any values, since it may be that there is nothing that elicits the relevant response in the right circumstances from absolutely everybody (Egan 2012, 558). On the other hand, if what "we" refers to is conditionally relative, then there will be a relativity of values: values humans , values martians , values them , values us , etc. The same value predicate may express different values in different contexts. There will be no common-subject matter for evaluative discourse and thought when evaluative standards diverge. But, Egan says, when I say x is a value and you say it is not, our assertions and thoughts should be in conflict, "our difference should count as a disagreement" (Egan 2012, 568).

B. The Role of Evaluative Discourse: Cognitive, Expressive, and Connective
Evaluative and normative discourse performs three very important roles: (i) it expresses people's evaluative and normative beliefs. Let us call this its cognitive role. (ii) It also (normally) expresses speakers' action-guiding conative attitudes, which have motivational effects; we can call this its expressive role. And finally, (iii), value talk normally establishes commonalities that, quoting Egan, are a "substantial part of the process of building and maintaining interpersonal relationships" (Egan 2010, 260). We can call this the connective role of evaluative discourse.
A dispositional contextualist theory has the resources to give a straightforward explanation of evaluative discourse. It claims that evaluative terms denote value properties, that those properties are response-dependent, and that we can believe correctly or incorrectly that those properties apply to their objects. The dispositional theory hence captures the cognitive role of evaluative talk. The theory does not face the Frege-Geach problem, which purely expressivist theories arguably face. 9 Additionally, the theory must explain how the denoted properties relate to the expressive and connective roles of evaluative talk. Finally, it must also explain the impression of persistent evaluative and normative disagreements.
If we accept that something has the property what we desire to desire (in way w, under conditions C . . .), we accept that we share desire-like attitudes. We hence implicitly accept that we have common responses under similar conditions. Attributing value to something, by the nature of value properties themselves, is essentially a way of establishing commonalities. Yet, although we may be similarly disposed, we are fallible in our value judgments, and our 8 See, e.g., (Egan et al 2005, Egan 2010; (Kölbel 2004b(Kölbel , 2004a  I say that to be valued by us means to be that which we desire to desire. Then to be a value-to be good, near enough-is that which we are disposed, under ideal conditions, to desire to desire. . . . It allows, as it should, that under less-than-ideal conditions we may wrongly value what is not really good. (Lewis 1989(Lewis / 2000 However, we may not be disposed alike, and in such cases we will not have the same standards. In the previous section, I mentioned the problem of non-converging dispositions raised by Egan. One of the possible consequences of non-convergent dispositions is that there may be no values at all, or that we lose a common subject matter. Egan suggests that a solution for the Lewisian dispositional account of value is to modify the nature of the dispositional value properties. He argues that if dispositional properties are de se properties that we self-ascribe, then we guarantee common content. The de se dispositional theory's central claim is that the content of a belief that x is a value is the property of the form being disposed to have response R to x in C. As Egan claims, this version retains the good features of dispositionalism (Egan 2012, 571), and it avoids the problems of non-convergence, because it avoids both an error theory and guarantees a common de se subject matter.
Egan defends an independently motivated Stalnakerian account of assertion (Stalnaker 1978). On Stalnaker's account, assertions add the content they communicate to the conversational common ground. In felicitous contexts, interlocutors accept the asserted content as part of the common ground. And thus, On a de se dispositionalist theory of value, the role of evaluative discourse is going to be to get the participants in the conversation on the same page with respect to how they think they would respond to the objects of evaluation under the appropriate conditions. (For example, on the de se version of Lewis's view, the role of evaluative assertion will be to get the parties to the conversation aligned with respect to what they think they would desire to desire under conditions of full imaginative acquaintance.) (Egan 2012, 574) De se evaluative thought allegedly does not face the challenges of lost disagreement that the contextualist dispositional theory faces. 10 De se thoughts are incompatible when it is subjectively irrational to self-ascribe the two centered properties that are the content of those thoughts. The subjective incompatibility of the thoughts grounds intersubjective incompatibility: if a de se property p is asserted in a given context, it cannot be coherently 10 The difference between believing classical propositions and believing centered, or de se propositions (on the Lewisian understanding of de se propositions that Egan relies on) comes down to this. To believe a classical proposition p is to believe that p is true at the actual world; e.g., that Bea eats the last piece of cake. In contrast, to believe a centered proposition is to believe that a property, q, is true of oneself at the actual world; e.g., eating the last piece of cake. The difference exists also in desires, where to desire a classical proposition p is to desire that p be true at the actual world; e.g., that Bea will eat the last piece of cake. In contrast, to desire a centered proposition is to desire that a property, q, be true of oneself at the actual world; e.g., eating the last piece of cake.
OUP UNCORRECTED PROOF -FIRSTPROOFS, Thu Sep 20 2018, NEWGEN accommodated in the conversational common ground by speakers who already self-ascribe (accept) a de se property not-p. Egan also suggests that disagreement need not always be cognitive-it may involve disagreements (or conflicts) in attitude, and that is what is to be expected in the evaluative case anyway, since value properties are constituted by desire-like states.
Disagreement about values depends thus on three factors: (i) there is disagreement in discourse, on the assumption that Stalnaker's account of assertion as providing uptake conditions on the common ground is correct, coupled with an account of de se content. (ii) The de se contents expressed allow for (a sort of ) disagreement in thought when the desires of the different interlocutors do not converge (it is a sort of disagreement in thought in the sense that it is not rational for an individual to self-ascribe p and not-p). (iii) In cases where people in dispute have non-converging desires, there is also disagreement in attitude, i.e., there are also conflicts of desires. The explanatory priority goes from (iii) to (i): it's because there's a conflict of dispositional attitudes that there is (a sort of ) disagreement in thought, and it's because there's a disagreement in thought that there is a disagreement in discourse (because of incompatible uptake conditions).
I don't have the space in the present chapter for a full discussion of why I think that the de se implementation of dispositionalism does not work. 11 I will just give a summary of why I think the view is problematic. My objection to Egan focuses on the fact that (iii) above is ungrounded. As I argue in Marques (2016b), a purely first-personal subjective rationality constraint does not guarantee that pairs of desire-like states are in conflict. Suppose that I hate liquorice, and have no wish to eat it. You, however, tolerate eating liquorice; in fact you enjoy it. As long as I'm not forced to eat it, our different dispositions do not conflict. Now, the desires that Egan claims are relevant for value properties are purely de se. But this example about liquorice shows that there need not be any intersubjective conflict of de se attitudes when two people's dispositions do not "converge." In short, having different de se preferences is neither necessary nor sufficient for having conflicting attitudes. If so, we cannot explain the presumed intersubjective disagreement in thought that Egan says occurs, and hence we don't motivate the expression of disagreement in discourse.
When I first introduced Lewis's account, I said that a main worry for the theory concerned who we are. Egan's objection targeted this concern. If we allow, as I think we should, for possibly variable standards, we have to explain the apparent persistent disagreements between people who don't share standards. But the objection from possibly non-convergent attitudes leading to "no common subject matter" ignores alternative sources of communicated content and of possible disagreement. I stand by the contextualist implementation of dispositionalism, in spite of Egan's criticism.
In my view, we have action-guiding attitudes that are essentially de nobis, not de se. 12 We may not be entirely aware of the content or the character of these attitudes, but they play an important explanatory role in our lives. The relevant dispositional de nobis attitudes for 11 I give a full discussion in Marques (2017c). 12 Frith and Frith (2012) review work in cognitive science and psychology on various "mechanisms of social cognition." Ongoing research on these mechanisms of social cognition reveals the role they play in learning, cooperation, and language acquisition.

C12.P43n12
OUP UNCORRECTED PROOF -FIRSTPROOFS, Thu Sep 20 2018, NEWGEN value properties are (higher-order) attitudes about ourselves. 13 A value property involves the relevant group to which the speaker belongs, but it does not depend on the speaker's selfidentification. Speakers may be in error about what they value. So, the question is, who are we in a value property, and in the content of the implicatures? There are three alternatives. First, we may be the current interlocutors in a conversation. We can engage in dialogue with people who don't share our standards. In a dispositional property, we cannot depend on who our conversational partners happen to be. Second, then, we must be those of us who share a standard. Finally, we may be all of us who are peers in a more fundamental sense, for instance as moral peers deserving of equal respect. 14 We, qua conversational partners, are also peers qua people equally deserving of respect, although we (in either of these previous senses) may not all be disposed alike. The hypothesis hybrid dispositionalism offers is that, through discourse, the speaker conversationally implicates (through a generalized conversational implicature 15 both that she accepts certain standards, and a desire that we-interlocutors-come to share the same standards.
I think that a semantic contextualist implementation of the Lewisian account is compatible with the expression of additional content. We can call the supplemented account hybrid dispositionalism, HD: HD If a speaker S asserts the sentence 'x is good' , then S denotes a dispositional property what we desire to desire in way w in ideal conditions C by 'good' and: (i) In asserting 'x is good' , S implicates that she desires x (in way w); and (ii) S implicates that she desires that we desire (to desire) x (in way w).
There is some linguistic evidence that these implicatures are associated with evaluative assertions. Imagine that Ana and Bea have the disagreement below: (  (2015), I advanced a conjecture about the evolutionary origin of de nobis attitudes. I suggest that coordination problems are at the root of our having, as humans, evolved to have the de nobis dispositions we have. However, I don't understand de nobis attitudes as "self-ascribable properties." There are different theories of de se contents, and not all regard de se contents as centered propositions. For discussion, see for instance (Recanati 2009). 14 In recent work, Carla Bagnoli (2016) revises Nozick (2001)'s proposal of an ethics of respect, which Nozickhe claimed is rooted in our evolutionary history. Bagnoli favors both the idea that we have irreducibly different moral standards, and that we share a basic commonality-that we are moral peers, not epistemic peers, in that our parity status is rooted fundamentally on the respect owed to each of us (Bagnoli 2016, 17-18). 15 (Strandberg 2012); (Fletcher 2014). 16 In personal conversation, Mike Ridge raised the following objection: Often we only say that we desire something when we don't currently have it. "I want a yacht" implies I will (all else equal) seek a yacht, but it doesn't entail that I have one. In fact, it tends to imply the opposite-I seek it because I don't have it yet. So we get an implicature that the speaker will try to acquire a first-order desire that x, not that she has one already-unless desires are special in being easy to control. But this isn't obvious. When I am on a diet, I wish I desired to If Ana and Bea share value standards, they contradict each other and have a straightforward disagreement. One of them may "wrongly value what is not really good." But they may not share standards, and still appear to disagree. How can the contextualism-dispositionalism combo explain this? How do we justify the claim that the implicatures (i) and (ii) are associated with value statements? Is there any evidence of their existence, and can we explain how they are rationally calculated from the meaning of the used sentences and the conversational contexts where they occur? And do these implicatures play a role in the explanation of persistent disagreement and conflict?
The implicatures (i) and (ii) appear to pass several tests of so-called not-at-issue content. 17 For instance, they appear to be cancelable, and they pass the Hey, wait a minute! test.
First, they are cancelable. There is no contradiction in Ana complementing (3a) by uttering the sentence in (4).
(4) Ana: It's good to respect the sanctity of life, but there are some people who should just die.
It is equally not contradictory for Ana to utter (5), somewhat pedantically: (5) Ana: It's good to respect the sanctity of life, but I don't expect you to respect it.
The implicatures also pass the Hey, wait a minute! test, which is usually regarded as a necessary but not sufficient condition for presuppositionality: [I] f π is presupposed by S, then it makes sense for an audience previously unaware of π to respond to an utterance of S by saying 'Hey, wait a minute, I didn't know that π' . (Yablo, 2006, 165) Thus, implicature (i) passes the test:  (Potts 2007(Potts , 2005, and (Tonhauser et al. 2013) for discussion of projective not-at-issue content. Potts (10) and (11). (10) Ana: It is good to respect the sanctity of life; actually, I hope to always respect it, even when I'm terminally ill. (11) Ana: it is good to respect the sanctity of life, and I hope we all respect it, as we should.
The fact that the speaker's commitment to (i) and (ii) can be reinforced suggests that the contents implicated are not semantically encoded (as a presupposition or a conventional implicature would be). A feature of conversational implicatures is, precisely, that they are reinforceable, as (10) and (11) illustrate (Potts 2007, 669-670).
This supports the idea that value statements pragmatically communicate content that expresses the speaker's conative states (approval or endorsement of a certain standard), and that the speaker desires others to share the same standards.
If (i) and (ii) are conversational implicatures, they should be inferable from the literal content of the sentence used, in agreement with the theory's supposition that what is denoted is a dispositional property. My explanation is similar to Caj Strandberg's 2012 proposal.
First, we can calculate that by saying that it is good to respect the sanctity of life, Ana communicates that she herself desires to respect the sanctity of life. How? Let us assume that Ana is cooperative and rational, and follows the maxim of Relevance: make your contribution so as to be relevant in order to fulfill the purposes of the conversation. Ana utters a sentence to the effect that respecting the sanctity of life is good, thereby saying that we (the set of people that share her standard) are disposed to value respecting the sanctity of life. Since she is one of the people referred to, she implicates that she herself desires to desire respecting life. Normally we seek to satisfy our desires, and hence, normally she would seek desiring to respect the sanctity of life. Hybrid expressivism can explain how the expressive role of evaluative discourse falls off the content of the evaluative statements.
Second, Ana may find herself in conversations where she knows that her interlocutor does not share her standards. Now, Ana cannot be implicating that her interlocutor shares her standard, since she already knows (or may know) that to be false. Nevertheless, she persists and asserts that it is good to respect that sanctity of life, thereby saying that respecting the sanctity of life is something we desire to desire. Since we have no reason to believe that she does not conform to the maxim of Relevance, we can infer that she desires her audience to share the same standards. Thus, even if Ana and Bea don't have the same standards, Ana's assertion plays a connectionbuilding role. Hybrid dispositionalism can account for the cognitive, the expressive, and the connection-building roles of evaluative discourse, even when speakers are not disposed alike.
The theory can also explain the impression of resilient disagreements and conflicts of attitudes, even when interlocutors are not disposed alike. Let us assume that Bea has no desire to share the same standards with Ana. But Ana desires that we come to share the same standards. In that case, Ana and Bea have desires that cannot be jointly satisfied, and hence have conflicting conative attitudes. Hybrid dispositionalism has hence the resources to account for conflicts of attitudes that follow from the very nature of evaluative disputes. A further positive aspect of the account is that the expression of conflicts is the direct consequence of the failure of the connective role of value talk.
We may wonder if a variation of the Frege-Geach problem also arises for the conversational implicature account. 18 We disagree not only when I say "X is good" and you say "X is not good." We also disagree when I say "If X is good then Y is good" and you assert "X is good but Y is not, " etc. But the story about attitude implicature, which is meant to explain the disagreement facts, seems to work only for simple atomic assertions of value. Asserting the conditional appears to be compatible with nihilism.
The problem is merely apparent, however. Conversational implicatures survive under embeddings, unless they're canceled. Sentences such as "It is better to get married and get pregnant than to get pregnant and get married, " or "Bill thinks that there were four children at the party" show how conversational implicatures survive under embeddings. Each of the conjunctions conversationally in the first sentence implicates that one event temporally precedes the other. Those implicatures are preserved under the embedding in "it is better to . . . than to . . . " Likewise, the sentence embedded under "Bill thinks that . . ." conversationally implicates that there were exactly four children.
Moreover, what is evaluative on hybrid dispositionalism is the asserted content itself. Hence, the conditional "If X is good then Y is good" is evaluative in virtue of the meaning of the words uttered. The conversational implicatures are only expressive of attitudes a speaker is presumed to have in virtue of making an evaluative assertion.

IV. Hybrid Dispositionalism and "Shared Acceptance of Norms"
Recent social accounts of the law use resources from social ontology to account for the existence and content of the law as the result of a joint activity of a community of legal officials. Among the current versions that rely on input from theories of collective action, 19 Toh (2005) offered an expressivist interpretation of Hart's view of internal legal statements. In his 2011 paper, Toh's account offers a defense of an expressivist version of legal positivism 18 I'm grateful to Mike Ridge for raising this doubt. 19 For instance, (Shapiro 2002); (Kutz 2001  that relies on the idea of shared norm acceptance, and that has core aspects in common with hybrid dispositionalism. In this section, I suggest how hybrid dispositionalism can complement Toh's proposal. Toh suggests that committed internal legal statements are explanatorily primary, and that external and detached internal legal statements are to be explained derivatively (110). External legal statements would be "attributions" of norm acceptances and of their expressions, and detached internal legal statements would be "expressions of psychological attitudes that simulate norm acceptances" (fn. 5). Although this is not explicitly stated, it seems to entail that legal statements are semantically ambiguous. However, I take it that Grice's modified Occam's razor should be followed whenever possible: don't multiply meanings beyond necessity.
On the noncognitivist interpretation of Hart's account of internal legal statements, where R is the norm that she considers the rule of recognition of the legal system of her community, a speaker that makes a legal statement: (i) Expresses her acceptance of a particular norm that is valid according to some fundamental legal norm R of her community and (ii) Presupposes that R is generally accepted and complied with as the 'fundamental legal norm' by the members of her community.

Yet, Toh voices concerns about how to interpret Hart:
What Hart says is that in making a legal statement, a speaker (i) expresses his acceptance of a particular norm that is valid according to the most fundamental legal norm of his community, or, as Hart calls it, the rule of recognition of his community; and (ii) presupposes that that particular norm is accepted and employed as the rule of recognition by the officials of his community. We can entertain two different interpretations of what Hart means here depending on how we conceive the effect of any possible failure in the factual presupposition of (ii). First, we can think of such a presuppositional failure as rendering the whole legal statement defective. In such a case, a speaker who discovers that the particular rule of recognition that he appeals to is not accepted by the officials of his legal system would be disposed to withdraw his legal statement. Alternatively, we can think of the factual presupposition of (ii) as the usual but not an invariable accompaniment of the more primary normative component of (i). In this latter case, a speaker would not be disposed to withdraw his legal statement upon recognizing a presuppositional failure. (Toh 2011, 116-117) I agree that the presuppositional failure of (ii) is problematic. But I confess I have some trouble seeing how the option of focusing on the "primary normative component of (i) is helpful, once we assume the failure of (ii). So, given the presuppositional 20 failure, a speaker 20 There are accounts that postulate a presupposition of commonality with the content that a rule or standard is commonly accepted and complied with by the members of a community, or the interlocutors in context. See for instance, López de Sa's 2008 account. who makes a legal statement accepts a particular norm that is valid according to a fundamental legal norm R, " since we agree that there is no unique fundamental legal norm of the community. The fact that the speaker accepts a norm that is valid-according-to-a-rule-ofrecognition-R' is insufficient to fully account for the problem of legal disputes. Those who disagree with the speaker can certainly accept that a particular norm N is valid-according-toa-rule-of-recognition-R', and accept that the speaker expresses her acceptance of norm N.
Toh suggests that a further worry concerns how expressivism explains normative and legal disagreement. Dworkin's objection to Hart's legal positivism emphasizes the problem of theoretical disagreements. These legal disputes occur frequently, whenever people do not agree on the fundamental rules of recognition. So the question here is whether (i) can carry the explanatory weight, or whether a modification of (ii) is required.
Toh says that expressivists have the resources, in principle, to explain why people may have conflicts that survive factual agreement. One speaker can accept R and another accept another rule of recognition R'. 21 Toh has in mind pragmatic accounts such as Gibbard's, where it is rational to make conversational demands in contexts where there are no shared norms only if engaging in such interactions can produce benefits that come with the resulting convergence in normative opinions (Gibbard 1990, ch. 12). The problem, however, is that such interactions do not present themselves clearly as normative disagreements (or legal disagreements, in the case that concerns us presently). Gibbard's proposal does not distinguish between negotiations among peers who try to coordinate on which fundamental rule to follow, and the goading of one's interlocutors into accepting the rules we want followed. The concern is that the latter is not a genuine normative or legal dispute. 22 Instead of (i) and (ii), Toh suggests that speakers express a first-personal plural norm acceptance, through which a speaker invites the audience to share the same fundamental norms. He assumes that speaker and audience aim at a joint acceptance of the same set of fundamental norms, adding that joint acceptances are maintained only when there is sufficient uptake: 23 [I] nstead of thinking of the joint acceptance of the fundamental norms as something that is always presupposed, it can be thought of as something that the speaker is sometimes trying to instigate (Toh, 2011, 118-119).
The "content of the norm acceptance" is as follows: Let us φ, on the assumption that: you, of your own accord, think or will come to think likewise, partly as the result of your recognition of this attitude! (Toh 2011, p. 122) I'm sympathetic to what Toh is trying to formulate, and I think that what he spells out as the content of norm acceptance is an instance of an expressive conversational implicature "S implicates that she desires that we desire (to desire) x in way w, " which as I suggested in the previous section accompanies evaluative statements in general, unless they are canceled. In effect, what Toh is proposing is that the speaker expresses a desire that we come to accept the same fundamental norms. However, Toh builds in the content of the norm acceptance the additional features that will guarantee that his account does not suffer from the problem he raises to Gibbard's.
How does the expression of an invitation to share norm acceptances relate to the making of legal statements? Either the joint norm acceptance is semantically encoded in the content of legal statements, or it is expressed pragmatically. Is there any linguistic evidence of which option is the correct one?
Normative claims are often expressed with deontic modals such as "ought, " "might, " "should, " etc. On Kratzer's canonical semantics of deontic modals (1977,1991), modal expressions such as "might, " "may, " "must, " or "ought" function as quantifiers over possibilities, in which the domains of quantification are contextually restricted. Modal sentences contain parameters that require context to determine a circumstantial accessibility relation on a world of evaluation w. This determines a modal base, i.e., a set of worlds accessible from w that are circumstantially like w in relevant ways. Furthermore, context must supply a standard or ordering source as a function of wi.e., a standard that orders the worlds in the modal base as better or worse. Thus, context contributes to determining a proposition by determining both a modal base and an ordering standard.
The standard Kratzerian semantics can apply to normative or prescriptive statements in general, and to the semantics of deontic legal statements in particular. The standard, or ordering source, for particular legal statements would be provided by the rule(s) of recognition of the local jurisdiction. This can be made explicit by preceding the legal sentence with an explicit relativization to the local jurisdiction. For simplicity, roughly, a statement made by a legal official of "According to the law, it must/ may be that φ" assumes that there is a local jurisdiction (a set of existing norms and rule(s) of recognition) and asserts that φ follows from/ is compatible with) the law at the jurisdiction.
Silk (in a chapter in this volume) offers a uniform account of internal and external deontic legal statements based on standard semantics for deontic modal claims, which I'm sympathetic to. 24 As desired here, the semantics is descriptivist-it assigns truth-conditions to deontic claims-the apparent normativity of claims of law follows from a contextualist interpretation of the standard semantics for modals, along with general principles of interpretation and conversation, elucidating the social and interpersonal function of legal discourse.
Normative sentences can be used to make committed statements or uncommitted statements. For instance, (12) seems to be a committed statement. But if we omit the parenthetical contextual information at the end, (12) no longer seems to be committed. A similar 24 Also, in recent work, Chilovi and Moreso (2016) offer an account of sentences of the form "according to the law, φ" where "according to the law" works as an intensional operator that quantifies over a prejacent, φ, and denotes a function that maps any proposition to the truth-value 1 if that proposition is compatible with the content of the law at that jurisdiction  (14) and (15) can have a committed reading or an external descriptive reading.
(14) A leasehold interest is not freely alienable.
(15) The Fourteenth Amendment allows states to regulate bakery employees' work hours.
The parallelism between the legal case and other normative statements with deontic modals, or words such as "rude, " 25 suggests that committed readings can be usefully explained as with conversational maxims. In general, any utterance of these sentences has a descriptive semantic content. The meaning of the words use requires sets of norms or value standards to be provided in context. Additionally, we can infer the committed reading (and we generally do) given sufficient information from context about the identity, location, and background of the speaker and the audience, while assuming that the speaker is being cooperative and rational, and is saying something that is relevant for the discussion In the legal case, if the speaker is a legal official addressing her peers, the fact of peerage is part of the common ground. Legal officials within a jurisdiction form a well-demarcated group. Their status is not just what results from each person's interpretation of her commitments in a jurisdiction. Rather, the status being a legal official is partly constituted by an existing and ongoing practice, and the social roles or positions created by that practice. What is required for someone to be a legal official is for there to exist a (partly) external causal-historical chain of events through which some legal and social conventions have been enacted, and which enable a person to occupy a position as an official within the jurisdiction, qua social structure or matrix. 26 It is of course correct to say that, [I] t is difficult to maintain that people are committed to the most fundamental laws of their legal system as a matter of convention when there is no convention (or any other type of convergent practice) of following those laws because there exists instead persist ent controversies about such laws. (Toh, 2011, 126-127) However, meeting the condition being a legal official does not require that legal officials make the same interpretation of their commitments to the fundamental rule of recognition. It is not required that each official interpret all her duties as an official in the jurisdiction in the same way. Once someone is actually invested as an official, she becomes one. 27 So, the conversational peers in a jurisdiction are the interlocutors who satisfy the legal peerage condition, but not necessarily those people who satisfy the condition people committed to the same fundamental laws. The situations where the implicature is entailed are those that present legal statements as committed.
My suggestion is that it is part of common ground that a speaker addresses her legal peers, but not that she addresses only those who make the same interpretation of their commitments. This understanding of the condition is essential to the very expression of the plural norm acceptances, since "us" must denote the interlocutors who are legal officials at the jurisdiction-we cannot be just those of us who accept the same interpretation of fundamental rules. If that were the case, the content of plural norm acceptance would be infelicitous. So, we need legal peers to engage in debate and to be under the requirement, as Toh suggests quoting Rawls, to be "ready to explain the basis of their actions to one another in terms each could reasonably expect that others might endorse as consistent with their freedom and equality" (Rawls 1993, 218;Toh 2011, 130).
In Dworkin's example, Judge Earl and Judge Gray have a theoretical disagreement concerning how to interpret the statute of wills. Judge Earl persists in asserting, "Elmer should not inherit the estate." If Earl's statement is valid according to the (intentionalist) rule of recognition that he accepts, what he states is correct, and there is no reason for Gray not to accept that statement as true. That statement implicitly relies on the rule of recognition that Earl accepts. Moreover, Earl does not presuppose that Gray accepts and employs the same rule of recognition. Both Earl and Gray know that thisthat they accept and employ the same rule of recognition-is false.
The description of the conversation in the previous paragraph assumes that what Gray (or Earl) expresses is merely his personal acceptance of a fundamental legal norm. This is insufficient to explain their disagreement. I suggest that the plural norm acceptance that Toh proposes is expressed is conversationally implicated. So how is a plural norm acceptance instigated through a conversational implicature? Let's assume that Gray states that according to the law, Elmer may receive the inheritance. The semantics of deontic sentences, we require a contextually provided set of fundamental norms of their jurisdiction to have a complete proposition. Through his assertion, Gray is conveying that this set of fundamental norms is compatible with Elmer inheriting. It is common knowledge that Earl and Gray are peers in the dissent, and that the court in their jurisdiction will enact a single decision. Since they are not merely talking past each other, they must be doing something else. On the assumption 27 As the US government in 2017 and 2018 abundantly illustrates, the behavior of government officials may ignore unwritten norms that were previously assumed to regulate their office. that Judge Gray is rational and cooperative, and follows the maxim of Relevance: make your contribution so as to be relevant in order to fulfill the purposes of the conversation, we can infer that he desires his peers to accept the same rule of recognition. So, we can explain the expression of a desire of a plural norm acceptance. With Toh, by saying that Elmer may inherit the estate, Gray expresses that the rule of recognition R, that justifies his divergent opinion, is the rule that is to be jointly accepted as their shared standard. 28,29 If plural norm acceptances are conversationally implicated, they should pass tests for conversational implicatures, namely the cancelability test and the Hey, wait a minute! test. This would provide additional support for the proposal. The implicature that there is a joint norm acceptance passes the Hey, wait a minute! test. For instance, we may conceive Judge Earl replying to Judge Gray saying, (14) Judge Earl: Hey, wait a minute! I didn't know you were a literalist! (15) Judge Earl: Hey, wait a minute! You can't expect me to endorse a literalist interpretation of the law!
The implicature can also be canceled: (16) Judge Earl: The lawmakers couldn't possibly intend that murderers benefit from their crimes . . . but I no longer expect you to accept an intentionalist interpretation of the law.
There is also support to the claim that the plural norm acceptance is conversationally implicated, and not semantically encoded, in the fact that the implicature can be reinforced: 28 Silk (forthcoming) offers a very similar reconstruction of the underlying reasoning in the dispute between Earl and Gray, based on the semantic properties of deontic modals and on the assumption that the participants are rational and follow the maxim of Relevance (Silk, of the final draft). I also agree with Silk when he says that these disputes are not properly described as metalinguistic or about how to use words Sundell 2013a, 2013b); they are disputes about "what basic legal norms to accept" (Silk,15). Elsewhere, I have expressed my scepticism about the extent and usefulness of metalinguistic negotiations in normative, evaluative, and legal domains (Marques 2017a(Marques , 2017b. 29 Finlay and Plunkett (2017) argue for quasi-expressivism, an alternative descriptivist theory of normative and legal sentences, based on Finlay's (2014) end-relational theory or normative discourse. "Good" or "ought" statements would assert propositions about the statistical relations in which actions stand to ends/ potential states of affairs. On this view, if an agent has a desire toward end e, she will be motivationally disposed toward what she believes stands in an instrumental relation to e. An "ought" statement would express a pragmatic presupposition: that the speaker has a favorable attitude toward end e. I'm skeptical of this proposal, because I don't see how if what "S ought to do A (in order to e)" means is "e is more likely if S does A (than anything else), " it should follow that we can infer in context that the speaker has a favorable attitude/ desire toward end e. Moreover, I don't think that the pragmatic content can be this presupposition, since it can be accepted into the common ground. I can accept what S says and that pragmatic presupposition-that doing A is the most effective way for us to bring about e, and accept that speaker S desires that we bring about e, while disagreeing that we ought to A. C12 The reinforcement, as the previous section indicated, would be infelicitous if plural norm acceptances were part of the semantic content of legal statements, or semantically entailed by them.
The cancelation of committed readings can explain further properties of legal statements. Raz distinguished between committed and detached internal legal statements, where the latter would not display the speaker's commitment or endorsement of a legal system (Raz, e.g., 1975(Raz, e.g., / 1990. The speaker explicitly stating that she doesn't endorse the rule in question can cancel the conversational implicature. Toh gives the example of a libertarian lawyer's tax advice to a client (Toh, 2011, 109). In (18) the libertarian lawyer would be canceling the conversational implicature that he accepts that the government has the right to collect taxes: (18) You must pay your income taxes before the end of this month . . . although, just between us, I actually think the state has no right to collect income taxes. 30 Following my suggestion, detached internal legal statements can be explained as statements that cancel a conversational implicature. This allows us to respect Grice's modified Occam's razor: do not multiply meanings beyond necessity. We do not need legal statements to be semantically ambiguous to explain internal committed, internal detached, and external legal statements.
Besides passing tests for conversational implicated content, plural norm acceptances must contribute to explain resilient theoretical legal disagreements. So, if we say that, in making a legal statement, the speaker conversationally implicates that she desires the joint acceptance of the same fundamental legal norms by her peers-i.e., that she desires us all, "of our own accord, to think or come to think likewise, " we can describe theoretical legal disputes over fundamental legal criteria as involving conflicts of attitudes, or, as Stevenson (1963) would say, "disagreements in attitude"-disagreements where disputants' desire-like states cannot be jointly satisfied. (1962,(1)(2). This offers a reply to Dworkin that does not force the legal positivist to interpret legal officials as making the same interpretation of their commitments, or to accept that moral considerations are part of the background that determines the content of the law.

V. Concluding Remarks
In this chapter, I claimed that hybrid dispositionalism shares some important features with Toh's (2011) account of legal statements as expressions of shared acceptances of norms. 30 Notice that it would be misleading to regard the lawyer's statement in this case as non-normative or merely descriptive. The obligation to pay taxes still exists and is expressed in (18), in spite of the lawyer not accepting all of the background norms. I suggested that Toh's idea of a shared acceptance of a norm is an instance of a de nobis attitude type. Yet, there are some structural differences between my proposal and the expressivist account that Toh offers. Toh claimed that committed internal legal statements are explanatorily primary, and external and detached internal legal statements are to be explained derivatively (Toh 2011, 110). My proposal takes seriously the Gricean recommendation of avoiding semantic ambiguity, if possible. I tried to make the case that hybrid dispositionalism about evaluative discourse offers some theoretical resources that can be applied to legal discourse.
In particular, we can make assumptions about legal peers and about the function of legal practice to explain how we infer conversational implicatures with contents that are essentially plural norm acceptances, i.e., de nobis attitudes. This proposal differs from other accounts that rely on presuppositions of commonality, where these presuppositions have the content that a rule or standard is commonly accepted and complied with by the members of a community. Unlike proposals of this kind, my suggestion distinguishes between the conditions for being a peer, and the conversational implicature that we desire our peers to come to share the same standards or accept the same fundamental rules. I argued that there's good evidence in support of the hypothesis that we communicate de nobis contents with conversational implicatures when we make evaluative statements, and legal statements.
In spite of the structural differences, my positive suggestion for the legal case can supplement Toh's expressivist proposal. In normal contexts, background information from context, the identity of the interlocutors, or other surrounding sentences, will make it clear whether the speaker is just making a description, making a committed normative statement, or making a normative statement in spite of canceling her own endorsement of the background normative standard. My suggestion also helps in defusing the objections from theoretical legal disputes against legal positivism, without compromising positivism