Assessment Sensitivity (Critical Notice)

Assessment–Sensitivity FILIPPO FERRARI [FORTHCOMING IN ANALYSIS] —I— Relativism has long been at the centre of traditional philosophical interests. As long as it remains so, John MacFarlane’s exceptionally ingenious book Assessment Sensitivity: Relative Truth and its Applications1 is surely destined to be essential reading for philosophers working on truth and, more generally, in philosophy of language and formal semantics. It is fair to say that Assessment Sensitivity is the most powerful and comprehensive defence of relativism about truth to date. The ideas elaborated in the book have circulated widely among philosophers since at least early 2009, generating an impressive amount of discussion and reinforcing a new wave of interest in relativism, started a few years earlier with some of the previous works by MacFarlane himself and the publication of Max Kölbel’s Truth without Objectivity in 2002. MacFarlane’s overall objective is to provide an explication of philosophical talk of “relative truth.”2 This is to be achieved by means of two sub-projects: (i) a purely theoretical one consisting in showing that relativism about truth is a conceptually coherent position by arguing that languages containing assessment sensitive expressions are conceptually possible; (ii) an empirical project, aimed at showing that English contains some assessment-sensitive expressions and thus that ‘true’ in English should, at least locally, be treated relativistically. To these two sub-projects are dedicated, respectively the two main parts of the book: “Foundations” (chapters 1-6) and “Applications” (chapters 7-11). The book ends with a chapter defending the rationality of believing and asserting assessment-sensitive propositions (chapter 12). My aim here is to provide three constructive criticisms targeting MacFarlane’s discussion of the norms governing the making and retracting of assertion (as outlined in chapter 5) in connection with his discussion of disagreement in the domain of basic taste (chapters 6 and 7). Given the limited space at my disposal here and the extraordinary width in the range of topics covered by MacFarlane, I am afraid I will have to remain silent on many other important and potentially controversial issues discussed in his fascinating book. The plan is thus as follows: in sections II and III I briefly introduce those core elements of MacFarlane’s assessment-sensitive relativism (which I will call simply ‘relativism’) that are needed for the three the critical point I intend to raise in sections IV-VI. —II— The thesis that truth is relative to context is compatible with a variety of semantic pictures. In particular it is compatible with both indexical and non-indexical forms of contextualism according to which the truth of an assertion somehow depends on features of the context in which it is made. What is distinctive of MacFarlane’s truth relativism is the idea that the truth of what is asserted—the proposition—by an assertoric use of a sentence in the context of assertion depends not only on features of that context, but also on features of another context, that of assessment. As an example, consider an assertoric use of a sentence about basic taste: “sushi is tasty.” According to relativism, this assertion expresses the same proposition—i.e. that sushi is tasty—relative to every context, but the truth-value of this proposition depends not only on 1

Assessment Sensitivity: Relative Truth and its Applications by John MacFarlane. Oxford: Oxford University Press, 2014. xvi+p344 pp. £30.00. 2 See Assessment Sensitivity p. 44. Unless otherwise stated, all references below are to MacFarlane’s book. 1

features of the context of assertion—e.g. the world and time at which the assertion is made— but also on features of the context of assessment—the gustatory standard of the assessor. The same proposition that sushi is tasty might be true as assessed by Yum but false as assessed by Yuk, not because of a change in the proposition but because of differences in the gustatory standard of Yum and Yuk. It is precisely in this sense that certain expressions—e.g. ‘is tasty’— are said to be assessment-sensitive. More formally, we have the following characterisation of relative truth: REL

A proposition p is true as used at the context of use (c1) and assessed from a context of assessment (c2) if and only if p is true at , where wc1 is the world of c1 and gc2 is the taste of the agent of c2 (the assessor).3

This is sufficient to highlight the formal difference between relativism and either forms of contextualism. With respect to indexical contextualism the contrast is clear since the latter predicts that the proposition expressed by an assertoric use of “sushi is tasty” will vary from one context of use to another, in tandem with variation in the gustatory standards of the respective users. This is because, to simplify, “is tasty” works similarly to indexical expressions of English—expressions like ‘that’ or ‘I’—and thus expresses a different property in different contexts. The formal contrast between relativism and non-indexical contextualism (NIC) is less immediate. NIC agrees with relativism that an assertoric use of “sushi is tasty” invariantly expresses a single proposition—that sushi is tasty. What they disagree about is what gustatory standard is relevant for the assessment of the truth of the proposition expressed. According to NIC the relevant standard is always provided by the context of assertion whereas according to relativism such a standard is always provided by the context of assessment. —III— However, if to such a formal difference corresponds no practical difference—i.e. a difference in the kind of predictions that these semantic theories make concerning assertoric practice— there would be little reason for preferring MacFarlane’s framework to the more conservative (with respect to the Kaplanian orthodoxy)4 NIC. This is why MacFarlane dedicates a lot of effort to explaining the difference his theory makes with respect to core normative aspects of enquiry, such as norms governing the making and retracting of assertion and the peculiar normative impact that a certain kind of disagreement exhibits. Starting with the idea that truth is the core normative notion, MacFarlane subscribes to the following norm of assertion (Reflexive Truth Norm): RTN

An agent is permitted to assert that p at context c1 only if p is true as used at c1 and assessed from c1.5

According to RTN the truth of an assertion made at c1 and assessed from c1 is a necessary condition for the permissibility of the asserting at c1. However, as the reader can no doubt see, if RTN were the only normative principle there would be no appreciable practical difference between relativism and NIC. In fact, although the assessment context is formally mentioned in the principle, operationally it is an idle wheel. This is because, whenever we make an assertion in a context we are also assessing it from that same context. In this respect, NIC and relativism turn out to be normatively equivalent theories. 3

p.105. NIC adds no structural complexity to Kaplan’s original semantic machinery—it only adds one (or more) parameter(s) to the set of information (circumstance of evaluation) relevant for determining the truth-value of a sentence in context. Relativism, on the contrary, by adding a context of assessment increases the structural complexity of the semantic machinery. This can be appreciated by the array of new post-semantic notions envisaged by relativism, beside that of relative truth—see pp.68-70. 5 p.103. 4

2

According to MacFarlane, the practical difference between the two views can be fully appreciated only by considering norms governing retraction. By ‘retraction’ MacFarlane means “the speech act one performs in saying ‘I take that back’ or ‘I retract that’.”6 Thus retraction is a speech act targeting another speech act—a past, unretracted assertion, made by the same subject—and it is constrained by the following norm (Retraction Norm): RN

An agent in context c2 is required to retract an (unretracted) assertion of p made at c1 if p is not true as used at c1 and assessed from c2.7

is a prescriptive norm—it says that the untruth of a proposition as used at c1 and assessed from c2 is sufficient to require retracting a previous assertion of it. The effect of retracting is that of rendering the previous assertion null and void—i.e. to disavow the assertoric commitments undertaken in the original assertion.8 With both RN and RTN on board, we can fully appreciate the practical difference between NIC and relativism. Because NIC does not distinguish operationally between a context of assertion and a context of assessment, when it is equipped with a retraction norm it predicts that a subject ought to retract a previously unretracted assertion just in case that assertion was impermissible—i.e. expressing a false proposition in the context in which it was performed. By contrast, MacFarlane’s relativism in trading on the distinction between context of assertion and that of assessment, leaves open the following possibility: that a subject is required to retract an assertion of p, because p is false from her current context of assessment, even though such an assertion was permissible in her previous context of assertion according to RTN. To illustrate, if in c1 I enjoy sushi and I assert “Sushi is tasty” but in c2 I change my tastes and I no longer enjoy sushi, then relativism, but not NIC, would predict that at c2 I am obliged to retract my previous assertion of “Sushi is tasty” since the proposition it expressed is assessed as false in c2, even though, let’s assume, that assertion was permissible in c1 according to RTN. The last distinctive feature of MacFarlane’s relativist framework I would like to introduce has to do with the normative significance of disagreement. Among the various major contributions that MacFarlane has made to the recent debate in philosophy of language, his idea—he shares credits with Torfinn Huvenes (2011)—that disagreement comes in several varieties has been particularly influential. Rather than trying to capture the essence of ‘real disagreement’ we should be pluralist about disagreement and welcome the thought that there is more than one way in which two subjects can disagree about some subject matter. This acquires a crucial significance when discussing what is perhaps taken to be the main advantage of endorsing relativism about truth, namely that of offering a consistent account of the possibility of faultless disagreement—i.e. a disagreement where neither party need be at fault in violating RTN. Among the various legitimate notions of disagreement, one in particular is, according to MacFarlane, appropriate for assessment-sensitive discourse. MacFarlane labels this notion of disagreement “preclusion of joint accuracy”: RN

PJA

The accuracy of my attitudes (as assessed from any context) precludes the accuracy of your attitude or speech act (as assessed from that same context).9

The notion of accuracy used in PJA is a technical one, intended to capture the idea of an assertion being true relative to the contextual information that matters for its evaluation (e.g., in the case of assertions about taste, the world and time of the context of assertion and the taste

6

p.108. Ibid. 8 Ibid. 9 p.129. 7

3

of the assessor). Taking into account the possibility of assessment-sensitive expressions, MacFarlane characterises accuracy as follows: ACC

An attitude or speech act occurring at c1 is accurate, as assessed from a context c2, just in case its content is true as used at c1 and assessed from c2.10

Thus, to illustrate what the notion of PJA amounts to, suppose Yum asserts “Sushi is tasty” and Yuk asserts “Sushi is not tasty”. PJA predicts that there is no single context of assessment relative to which both assertions are accurate. However, given RTN and RN it may be that both assertions are permissible in their respective contexts of assertion, and thus that neither Yum nor Yuk is required to retract from within their original context of assertion. In other words, although Yum’s assertion precludes the accuracy of Yuk’s assertion (and vice versa) it may be that both Yum and Yuk succeed in complying with RTN and RN. —IV— In this and the following sections, I would like to raise three critical points. The first concerns the notion of retraction. MacFarlane takes retraction to be a speech act targeting other speech acts—in particular assertions. In considering whether we could have retraction at the mental level, MacFarlane seems quite negative: “Why is there nothing like retraction in the case of belief? Can’t one give up a belief, just as one can retract an assertion? Here it is important to keep in mind a metaphysical difference between assertions and beliefs. An assertion is an action, and hence also an event, while a belief is not an action or event, but a state that an agent can be in over a period of time. The inception of a belief may be an event, but the belief itself is not.”11 What about the inception of a belief, then? If the problem of understanding the cogency of retraction as applied to belief is that belief is a state while retraction properly targets actions, why not consider acts of judging? It is common to understand a judgement as a cognitive mental act of affirming a proposition.12 MacFarlane considers this option in a footnote, where he says: “If we countenance mental acts of judgement, then we might try talking of retractions of such acts. This would open up the possibility of making sense of assessment sensitivity without assertions. However, it is not clear to me either that there are mental acts of judgement, or that there is anything corresponding to retraction of such acts. We should beware of carrying the analogy between thought and discourse too far.”13 This is about all MacFarlane says on the issue of retracting judgements. And it sounds disappointing especially given the fact that, as he points out, there is no special problem in understanding assessment-sensitivity at the mental level.14 So it is not clear why MacFarlane is taking this line. In fact, I believe that there are at least two reasons for arguing that he should not take this line. First, nothing prevents us from making sense of the notion of retracting judgements—if retraction is indeed a genuine normative epistemic phenomenon. Second, making sense of the notion of retraction at the mental level is required in order to make sense of retraction at the level of speech act. I will illustrate both points, starting with the first. 10

p.127. p.115. 12 See, Shah & Velleman (2005): 503. 13 Footnote 23, p.115. 14 As MacFarlane says: “Once we understand what it is to assert an assessment-sensitive proposition, there is no obstacle to countenancing beliefs with these propositions as their contents.” p.114. 11

4

Regardless of whether a structural analogy between the normative architecture of speech acts and that of mental acts can be drawn, the point is that if a normative phenomenon grounded on epistemic/alethic factors takes place at the level of speech acts, one might expect to see it reflected at the mental level as well. To deny this connection requires some argument. We certainly do have alethic and epistemic norms governing the formation, maintenance and relinquishment of cognitive mental attitudes. In fact, there is a well-established philosophical tradition (that was christened by William Clifford’s 1877 essay “The Ethics of Belief”) that locates the primary source of epistemic normativity at the mental level. What is so special about retraction that would make it inapt to occur at the mental level? According to MacFarlane the key point is that there seems to be nothing more at the mental level that a subject can do once she gets evidence of the falsity of a certain proposition p than ceasing to believe p, or withholding the judgement that p. Nothing comparable to the disavowing effect of the kind of taking back associated with retraction of speech acts is taking place in the mental realm. Is this right? There is a clear sense in which an act of retraction targeting a cognitive mental act would differ from retracting a speech act—it would have no, immediate, public effect. Unless I make my mental retraction explicit by means of an appropriate speech act, I would still be held accountable for any assertoric commitment I might have undertaken by asserting the proposition in question. This might be right, but it has nothing to do directly with epistemic or alethic normativity. The kind of responsibility in question and the possible reproach that might follow from neglecting it clearly aren’t epistemic, but moral (latu sensu). Moreover, there are reasons in favour of acknowledging a notion of retraction that targets cognitive mental acts. Let’s understand retraction as an act aiming at disavowing one’s epistemic commitments incurred by a previous a belief. One option we have at the level of belief-maintenance would be that of retracting a previous act of judgment with the effect of disavowing the epistemic commitments associated with the holding of the proposition in question. In this respect—and contrary to what MacFarlane says in the passage above—a mental act of retraction wouldn’t be just a punctual act of ceasing to believe or withholding a judgement, but it would partly consists in many other mental actions aimed at preserving the consistency within the subject’s system of beliefs. For instance, there are many other judgements in the vicinity of the retracted one that require some epistemic intervention aimed at preserving consistency. Such cognitive activities are not necessarily involved in the mere acts of ceasing to believe a particular proposition or withholding a previous judgment. In this way, retraction at the mental level would present some structural analogies with the way in which MacFarlane conceives of retraction of speech acts. The second point I would like to make is that we actually need retraction occurring at the cognitive mental level if we want to make sense of retraction of speech acts. We cannot actually retract a speech act—it’s out there, it’s done, it’s an event that took place in the past and thus isn’t available to the subject anymore. In retracting a previous speech act I am not trying to make the case that I did not perform it. So what am I doing? What is left in my context of retraction are two things: my endorsement of the asserted proposition—i.e. the belief—and the assertoric commitments, if any, engendered by my previous assertoric speech act. Thus, when I am required to retract in accordance with RN I am required to do two things: to withhold my belief and to disavow any assertoric commitment associated with my previous assertion. However, both things are clearly grounded in something attitudinal that can be thus expressed: I don’t stand by what I have said—so don’t take me to be committed to that. This standing by is clearly attitudinal and what it targets is our endorsement of the proposition in question as a belief. The withdrawal of a commitment marks an attitudinal change. Thus the point is that MacFarlane is already committed to describing the phenomenon at the level of mental attitudes. There has to be an attitudinal description of what is going on, of what the

5

import of retraction is at the mental level, if we are to understand what the speech act of retraction does.15 —V— Let us now turn to a discussion of the notion of preclusion of joint accuracy. My first concern is about whether PJA is the right model for understanding disagreement in the domain of taste. According to MacFarlane things seem to tip in favour of a positive answer: “[T]he parties to taste disagreement think of themselves not just as trying to change the other party’s attitudes, but as trying to refute them—where the sign of successful refutation is not just that the other party now holds the content of her original claim to be false, but that she retracts her original assertion as inaccurate. Disputes of taste do seem to have this flavor. If Yuk eventually gets Yum to dislike the taste of licorice, Yum will feel pressure to withdraw her earlier assertion that it is tasty. In this respect, disputes of taste are like disputes about any objective matter—for example, the age of the earth.”16 That things are this way is crucial to defend the thesis that assessment sensitivity offers a more adequate framework for the domain of taste than NIC. What NIC can explain is a situation of disagreement in which Yuk and Yum put forward doxastically non-cotenable attitudes which are nevertheless both accurate—this is because there is no operational distinction between the context of use and that of assessment. The assessment-sensitive model predicts something more—namely that the disagreement between Yuk and Yum is one in which there is PJA. But is the evidence adduced by MacFarlane sufficiently strong to drive the point home? There are good reasons to be sceptic. The crucial element here is that the disagreement is teleologically constrained by an attempt by each party to refute their respective opponent’s assertion—where a successful refutation leads the refuted party to retract. Evidence for this, according to MacFarlane, is given by the fact that, once refuted, Yum will feel pressure to withdraw her earlier assertion. There are two sets of reasons to be sceptic about whether this should be taken as evidence for relativism. On the one hand, to say simply that Yum would feel pressure to withdraw her earlier assertion once refuted is clearly not to show unambiguously that the domain of taste exhibits assessment-sensitivity. In fact such an expression underdetermines the distinction between merely regarding the content of her original claim as false and regarding the original assertoric speech act as inaccurate (as characterised by ACC—i.e. as expressing a proposition which is untrue as used at Yum’s previous context of assertion and assessed from the context in which the disagreement takes place). In other words, the kind of evidence adduced by MacFarlane underdetermines the distinction between Yum’s disposition in the context in which the disagreement takes place to distance herself from the proposition expressed by her assertion on the ground of its falsity, and her disposition, in that same context, to retract—in MacFarlane’s technical sense—her previous assertoric speech act on the ground of its inaccuracy. Unambiguous evidence for assessment-sensitivity should be evidence that unambiguously supports both things—but the kind of evidence cited by MacFarlane fails to do that (at least in the basic taste case). One might think that this is just a particular exemplification of a more general point—namely that the kind of linguistic data discussed by MacFarlane in support of the cogency of the phenomenon of retraction underdetermines whether what is withheld is the assertoric speech-act the production of which antecedes its withholding or, merely the

15 16

On this, see also Greenough, P. (2011): 214-17. p.132. 6

proposition expressed by such a speech act.17 For all that has been said so far we cannot exclude that expressions of retraction, as presented by MacFarlane, are nothing more than an emphatic way of voicing one’s change of mind with respect to the truth-value of the proposition in question. On the other hand, even granting MacFarlane that the kind of considerations he discusses could be used to unambiguously support his relativism, there is the further question of whether folk’s intuitions concerning disagreement about basic taste point in the direction he suggested. Is it really the case that people engaging in disagreement about basic taste aim at controverting their opponents’ judgments? Is an attribution of inaccuracy to a contrary view about basic taste something we would accept? The following is what might be thought as a plausible reaction to a disagreement about basic taste: “Despite our disagreement, your contrary opinion is just as good as mine.” This is in effect the thought that a certain kind of faultlessness is indeed endorsable from within a committed perspective—something along the lines of what Wright has called parity18. In typical cases—i.e. where it is clear that both judgments involved in the disagreement have been formed in accordance with respectively the tastes of the two subjects—we would expect not only that no criticism is forthcoming after the disagreement between Yuk and Yum comes to light, but also that the divergence in opinions is serenely accepted without further argument aimed at refuting the opposite party. Now, is attribution of inaccuracy to a contrary view in line with this thought? An answer to this question ultimately depends on what MacFarlane has to say about the comparative evaluation of standards of taste—whether, e.g., in disagreeing with Yum, Yuk is licensed to evaluate her own tastes as better than Yum’s. This, I conjecture, depends on an examination of what kind of value folk would typically attribute to basic taste. MacFarlane seems to suggest that there are certain senses of ‘better taste’ according to which it would be consistent with the assessment-sensitive framework for a subject to evaluate certain tastes as better than others. If that’s right, then certain disagreements about basic taste—those in which one party has the right to claim that her tastes are better than those of her opponents—would elicit a rather substantive negative assessment of someone’s holding a contrary opinion. This seems to be enough evidence that in MacFarlane’s framework an attribution of inaccuracy to a contrary view would be sufficiently negative to justify Yuk’s attempt to persuade Yum to change her mind and thus to withdraw her judgment. And this would be incompatible with the parity thesis. Of course, MacFarlane has the option of denying that parity is part of folk’s conception of the normative significance of disagreement in basic taste. But it would be helpful to have some argument in support of this denial. As a side issue, if we engage in the project of a comparative analysis of disagreement as it occurs in various domains where a relativistic thesis about truth looks prima facie appealing, we might be persuaded by the idea that the normative significance of disagreement varies in relation to the subject matters of the dispute. For instance, disagreements about basic taste might be perceived as unimportant, thus encouraging an attitude of live and let live. By contrast, disagreements about more refined aesthetic matters—such as a disagreement about whether Michelangelo’s La Pietà is among the finest achievement of renaissance’s plastic arts—might be perceived as involving a much greater importance and thus as justifying an attitude of rational criticism and attribution of (some kind of) fault to a contrary view. This is just one example, and many other cases should be considered in order to have a full appreciation of the kind of variability in the normative reaction that people might have when presented with disagreements about different subject matters. I suspect that the normative

17

See Wright, C. (2007): 272-275, for a critical discussion of the assessment-sensitivity about epistemic modals. See Marques, T. (2015) for a criticism of the obligatoriness of retractions in some of the areas discussed by MacFarlane. 18 See Wright, C. (2012). 7

architecture endorsed by MacFarlane is not flexible enough, as it stands, to account for the kind of variability just discussed. Something needs to be added.19 One last point about the notion of PJA is in order. One might wonder whether the distinction between PJA and a notion in the vicinity—what MacFarlane labels “preclusion of joint reflective accuracy” (PJRA)—is really stable. MacFarlane characterises PJRA as follows: PJRA

The accuracy of my attitudes (as assessed from my context) precludes the accuracy of your attitudes (as assessed from your context).20

According to MacFarlane the distinction between PJA and PJRA captures the contrast between an objectivist treatment of a subject matter and an assessment-sensitive one. PJRA is supposed to cover cases of disagreement about purely objective matters—i.e. assertions that are neither use- nor assessment-sensitive. If Yum and Yuk disagree about whether the Sun is bigger than the Earth, then it cannot be case that Yuk’s assertion is true as used and assessed from Yuk’s context and Yum’s assertion is true as used and assessed from Yum’s context. This is because the accuracy of assertions concerning objective matters is absolute. At most one among Yum’s and Yuk’s claims can be accurate. However, one might doubt whether the distinction between PJA and PJRA is indeed that clear. We can easily construct a case about scientific disagreement in which we would have PJA but not PJRA. Let the assessment point be determined by a state of information. Yum says that p while Yuk says not-p, where p is fully objective. Since neither standpoint in particular signs up for both p and not-p, we have PJA. But they are jointly reflexively accurate because each subject is assessing her own judgment in the light of the state of information available at her assessment context. So how is the contrast to be drawn between the objective matter and the relativistic matter on that understanding? MacFarlane would no doubt reply that the notion of accuracy he is using is alethic whereas in the case described above the intended notion of accuracy is epistemic. That might be right. However, if we say that the notion of joint reflexive accuracy is truth-entailing then with respect to those cases in which joint reflexive accuracy is not precluded—e.g. the basic taste case—we should be able to say that both views are true. But, from which perspective are we saying that? Surely the participants in the dispute cannot say that. In fact, each participant is committed to assess the opponent’s assertion as false and inaccurate. But a neutral referee cannot say that either. Although she has no view on the subject matter of the dispute, she is not in a position to coherently say that both Yum’s claim and Yuk’s claim are true. Thus, if the notion of joint reflexive accuracy is truth-entailing, it seems that there is no single standpoint from which two contradictory assertions can be said to be jointly reflexively accurate. As a result, it seems that MacFarlane is facing the following dilemma. If the notion of joint reflexive accuracy is non-truth-entailing then the objective claims match it and we do not have the disanalogy that MacFarlane is pressing between relativistic subject matters and purely objective ones. If, on the contrary, it is truth-entailing, then it looks as if the claim of simultaneous truth of both p and not-p is one that no one in particular can endorse.21 This means that the distinction between PJA and PJRA is not one that can be coherently expressed within the object language—and thus not a distinction that can be appreciated by ordinary speakers. However, this seems problematic since we do seem to clearly appreciate the difference in normative significance between disagreements about basic taste and those about scientific matters. That we do appreciate it is evident from the difference in the normative assessment of a contrary view that we typically make in the two cases. Thus some other explanation is called for. 19

I deal with this in Ferrari, F. (2016). See also Ferrari and Zeman (2014) for a comparative analysis of retraction in the taste and moral domains. 20 p.130. 21 This point emerged in conversation with Patrick Greenough, Giacomo Melis and Crispin Wright. See also Wright, C. (2016): pp.6-7. 8

—VI— Let us now turn to my last critical point. If my discussion of PJA is on the right track, it points to some difficulties with that notion. This leaves us with the question of which notion of disagreement we should take on board. Among the other options discussed by MacFarlane, what he calls ‘doxastic noncotenability’ appears as the most appropriate one for truth-apt discourse: DNC

To disagree with someone’s doxastic attitude towards the proposition p is to have doxastic attitudes whose contents are jointly incompatible with p.

However, if we apply DNC across the board we seem to get implausible results in certain cases. MacFarlane takes this fact to be evidence for the need of the distinction between DNC and 22 PJA. The starting point of MacFarlane’s discussion is an argument given by Cappelen and Hawthorne23 against the coherence of temporally neutral propositions—e.g. propositions of the form that it is raining in Boston. Here is MacFarlane’s slightly modified version of the argument: (a) (b) (c) (d) (e)

Two parties disagree if there is a proposition that one believes and the other disbelieves. Suppose that tensed propositions [what I call ‘temporally neutral propositions’] can be the contents of beliefs. Then it should follow that if Bill believed, two days ago, the tensed proposition—It is raining in Boston—and Janet disbelieved the same tensed proposition two weeks ago, they disagreed. But this pattern of attitudes does not constitute disagreement. So, by reductio, tensed propositions cannot be the contents of beliefs.

What the argument shows is that if we assume temporalism—i.e. the view that there are temporally neutral propositions—and we understand ‘disagreement’ as DNC then we end up saying that Sebastiano and Filippo disagree if Filippo believed on Tuesday the proposition that it is raining in Bologna while Sebastiano believed on Wednesday the proposition that it is not raining in Bologna. However, intuitively, that between Filippo and Sebastiano does not seem a disagreement at all. Thus, either temporalism is flawed in claiming that temporally neutral propositions can be the object of belief or the DNC model of disagreement is inadequate for this case. MacFarlane rejects the argument as invalid. He argues that once we take the distinction between DNC and PJA on board we can appreciate that the argument equivocates between these two senses of ‘disagreement’—‘disagreement’ in (a) being intended as DNC while ‘disagreement’ in (d) as PJA. If this is right, the argument cannot then be taken as a refutation of temporalism. If we rewrite the argument by disambiguating ‘disagreement’ as DNC, then we would have that Filippo and Sebastiano do really disagree—which seems quite an implausible prediction. If, on the other hand, we intend disagreement as PJA we would have that Filippo and Sebastiano do not disagree—which seems the right prediction. This move preserves temporalism alongside our intuition that there is no intuitive sense in which Filippo and Sebastiano disagree. However, we don’t need the notion of PJA—and thus the relativised notion of accuracy—in order to make temporalism compatible with our intuition that Filippo and Sebastiano do not disagree. We could instead refine the notion of DNC, focusing on the notion of judgement, in the following way: 22 23

p.127. See Cappelen & Hawthorne (2009): 96-8. 9

DNC*

To disagree with someone’s judging that p is to issue judgments whose contents are jointly incompatible with p—where the content p of a judgment J1 issued by subject S1 is incompatible with the content q of a judgment J2 issued by subject S2 if and only if p, or any paraphrase of p which is truth-conditionally equivalent to it in the context in which S1 judges that p, cannot be imported into the context in which J2 is issued by S2 without generating (any additional) inconsistency within S2’s system of belief.24

The additional complexity here is meant to exclude from the set of candidate contents generating disagreement temporally/locationally neutral propositions—e.g. or —whose truth values varies as a function of the time and location of the context of use. Since in the context of Filippo’s judging that it is raining in Bologna there is an available truth-conditionally equivalent content—i.e. that it is raining in Bologna on Tuesday—which can be imported into the context of Sebastiano’s judging, on Wednesday, that it is not raining in Bologna without generating any inconsistency in Sebastiano’s doxastic attitudes towards the weather—the outcome is that Filippo and Sebastiano do not disagree. By contrast if Filippo judges that sushi is tasty while Sebastiano judges that it is not, then they would disagree because, assuming content-invariantism in the taste domain as relativists do, there’s no truth-conditionally equivalent content to available in Filippo’s context that can be imported into Sebastiano’s context without generating inconsistency in Sebastiano’s system of beliefs. Therefore, we do not need the notion or PJA, nor the relativized notion of accuracy, in order to explain why Filippo and Sebastiano do not disagree about the weather but they do disagree about sushi. The upshot of these brief considerations—if on the right track—is not only that the dialectic between NIC and relativism need be reassessed but also that other competitors—i.e. minimalist conceptions of truth à la Horwich or Wright equipped with the kind of flexible normative architecture I have defended elsewhere25—should be payed due attention. Let me conclude by emphasising that Assessment Sensitivity is an immensely important book, with many groundbreaking features that make a very significant contribution to contemporary philosophy of language and metaphysics. I cannot but urge whole-heartedly that philosophers of language and semanticists should engage with the project of Assessment Sensitivity as a matter of priority.26 Institute of Philosophy, University of Bonn, Am Hof 1, 53113, Bonn, Germany Department of Philosophy & Communication, University of Bologna, 40122, Bologna, Italy

24

A similar refinement of DNC is discussed in Wright, C. (unpublished manuscript): pp.11-12. See Ferrari, F. (2016). 26 I would like to thank Elke Brendel, Patrick Greenough, Giacomo Melis, Sebastiano Moruzzi, Eva Picardi and Crispin Wright for their valuable comments. The research for this paper was carried out during my tenure of a postdoctoral fellowship in the project “Disagreement in Philosophy”, sponsored by the German Research Foundation (DFG – Deutsche Forschungsgemeinschaft), at the University of Bonn. I gratefully acknowledge the support of the Trust. 25

10

References Cappelen, H. and J. Hawthorne. 2009. Relativism and Monadic Truth. Oxford: Oxford University Press. Ferrari, F. 2016. Disagreement about Taste and Alethic Subrogation. Philosophical Quarterly (doi: 10.1093/pq/pqv116). Ferrari, F. and D. Zeman. 2014. Radical Relativism and Retraction. In New Frontiers in Truth, eds. F. Bacchini, S. Caputo, M. Dell'Ultri, 80-102. Cambridge: Scholar Publishing. Greenough, P. 2011. Truth-Relativism, Norm-Relativism, and Assertion. In Assertion: New Philosophical Essays, eds. J. Brown and H. Cappelen. Oxford: Oxford University Press. pp.197-232. Kölbel, M. 2002. Truth without Objectivity. London: Routledge. Huvenes, T. 2012. Varieties of Disagreement and Predicates of Taste. Australasian Journal of Philosophy 90(1): 167-181. MacFarlane, J. 2005. Making Sense of Relative Truth. Proceedings of the Aristotelian Society, 105: 321-339. Marques, T. 2015. Retractions. Synthese (doi: 10.1007/s11229-015-0852-8). Shah, N. and D. Velleman. 2005. Doxastic Deliberation. Philosophical Review 114(4): 497534. Wright, C. 2012. Replies Part III: Truth, Objectivity, Realism and Relativism. In Mind, Meaning and Knowledge: Themes from the Philosophy of Crispin Wright, ed. A. Coliva, 418450. Oxford: Oxford University Press. Wright, C. 2016. Assessment-Sensitivity: the Manifestation Challenge. Philosophy and Phenomenological Research (doi: 10.1111/phpr.12262). Wright, C. (unpublished manuscripts) Can Sense be Made of Faultless Disagreement?

11