/ Assertoric Force Perspectivalism: Relativism Without Relative Truth


According to relativist accounts of discourse about, e.g., epistemic possibility and matters of taste, the truth of propositions must be relativized to nonstandard parameters. This paper argues that the central thrust of such accounts should be understood independently of relative truth, in terms of a perspectival account of assertoric force. My point of departure is a stripped-down version of Brandom’s analysis of the normative structure of discursive practice. By generalizing that structure, I make room for an analogue of the “assessment sensitivity” MacFarlane characterizes in terms of relative truth. I argue that my reformulation supplies a stronger rationale for the most distinctive feature of MacFarlane’s brand of relativism, its account of when speakers ought to retract assertions. Furthermore, I show that the view usually regarded as a “moderate” alternative to MacFarlane’s “radical” relativism requires the more radical deviation from an absolutist account of assertoric force.

1. Introduction

The last decade has seen an explosion of interest in relativism in the philosophy of language. Relativist accounts have been proposed to explain discourse about knowledge, epistemic possibility, matters of taste, and contingent future events. In this context, relativism is usually taken to be, or to presuppose, a semantic thesis. According to relativists, understanding how some discourses function requires recognizing that speakers express propositions whose truth or falsity must be evaluated relative to parameters in addition to a possible world (see especially Kölbel 2002; MacFarlane 2003; 2005a; 2011a; 2014; Richard 2004; Egan et al. 2005; Lasersohn 2005; Stephenson 2007).

In this paper, I propose a different way to think about how the discourses in question motivate relativism in the philosophy of language. The central thrust of relativism, I argue, can and should be understood independently of any semantic framework of relativized truth. Instead, relativism should be understood in pragmatic terms, as corresponding to a particular understanding of assertoric force. The idea starts with Robert Brandom’s analysis of “fact-stating discourse” (1994: 607) as a “game of giving and asking for reasons” whose basic move is asserting. Brandom’s account of assertion makes no appeal to truth.[1] My proposed revision of his analysis makes room for a broader conception of fact-stating discourse, by allowing assertoric force to depend on speakers’ perspectives. What is distinctive and plausible about relativism, I will argue, is best captured by the resulting liberalized version of Brandom’s game of giving and asking for reasons.

This way of reconceiving relativism has consequences for the debate between proponents of two versions of relativism: so-called “moderate” and “radical” relativism.[2] Understood in terms of relative truth, the difference is a matter of how contextual parameters get supplied when an assertion is evaluated. So-called moderate relativists look to a context of use to supply these parameters: they attribute what John MacFarlane calls “use sensitivity” to the discourses at issue. By contrast, so-called radical relativists, including MacFarlane, argue that we also need to look to a context of assessment: they attribute what he calls “assessment sensitivity.” All parties have agreed that attributing assessment sensitivity is the more radical departure from absolutist orthodoxy.[3] On the received view, then, there are three degrees of increasing relativism:

  1. Absolutism: propositions have absolute truth conditions. (By a proposition’s “truth condition” I mean a condition whose obtaining in a world is necessary and sufficient for the proposition to be true at that world.)
  2. Use sensitivity: propositions have relative truth conditions, and how an assertion should be evaluated depends on the asserted proposition’s truth relative to parameters supplied by the context of assertion.
  3. Assessment sensitivity: propositions have relative truth conditions, and how an assertion should be evaluated depends on the asserted proposition’s truth relative to parameters supplied by the context of assertion and a context of assessment.

My proposed reformulation of relativism will yield a pragmatic analogue of MacFarlane-style assessment sensitivity. Indeed, I will argue that the reformulation points to a better explanation of why our discourse would display assessment sensitivity. But I will also argue that assessment sensitivity is the less radical of the two relativist deviations from absolutism. In short, I will propose replacing the above three degrees of relativism with a pragmatic alternative, one that reverses the order between the two relativistic options:

  1. Absolutism: assertions have absolute assertoric force.
  2. Assessment sensitivity: assertions have hearer-relative assertoric force, in a manner that is determined by the context of assertion and a context of assessment.
  3. Use sensitivity: assertions have hearer-relative assertoric force, in a manner that is determined by the context of assertion alone.

The paper’s structure is as follows. Sections 2-3 explain how I understand the relativist’s explanatory task and strategy. Section 4 introduces assessment sensitivity, as presented by MacFarlane in terms of relative truth. The main argument starts in Section 5, where I explain why an initially promising motivation for attributing assessment sensitivity yields a dilemma. My response to the dilemma, which consists in reconstruing assessment sensitivity in terms of perspective-dependent assertoric force, is laid out in Sections 6-7. Reconstrued as proposed, I argue in Section 8, assessment sensitivity is less radical than use sensitivity. Section 9 replies to objections.

2. Evaporative disputes and frictionless denial

The goal of advocates of relativist semantics has been to explain the apparently distinctive functioning of a wide-ranging class of expressions. As my examples, I will take epistemic modals such as ‘it’s possible that...’ and so-called predicates of personal taste such as ‘tasty’ and ‘fun’. At issue is the way their use appears to have a subjective or perspectival aspect. This section explains how I understand the phenomena relativists seek to account for, and one respect in which my presentation differs from some in the literature.

The key phenomenon is pointed to in Huw Price’s groundbreaking book Facts and the Function of Truth (1988), which deserves to be recognized as a close precursor to the current discussion of relativism.[4] This is the seeming possibility of what Price calls “evaporative disputes,” which he says “populate the margins of factual discourse” (1988: 159ff ). Price himself focuses on disputes over whether something is probable, but his observations apply equally to disputes involving epistemic possibility and predicates of personal taste.[5] Here are two examples:[6]

  • Two physicians, Dr. Adams and Dr. Brown, are discussing a mutual patient. Adams says: ‘It’s possible the patient has Lyme disease’. Brown protests: ‘No, that’s not possible! You must be misinterpreting the test results.’
  • Two friends, Alice and Ben, are discussing which cheeses to buy for a party. Alice says: ‘Limburger is a tasty cheese. My sister told me so.’ Ben replies: ‘Did she really? I’ve tried it, and it’s not tasty at all!’

To all appearances, both examples satisfy the following schematic description:

  • (i) one party has affirmed, and the other has denied, that P,[7]
  • (ii) both parties recognize (i), and
  • (iii) each takes (i) to mean that there is something mistaken about the other’s speech act.

Suppose the appearances are correct and (i)-(iii) are satisfied. In a familiar sense, this suffices for the parties to be engaged in a dispute about whether P. Thus Adams and Brown are engaged in a dispute about whether it’s possible the patient has Lyme disease, and Alice and Ben are engaged in a dispute about whether Limburger cheese is tasty.

Imagine, now, that Adams and Brown discover that only Brown has seen the most recent test results ruling out their patient’s having Lyme disease. According to Price, (ii) will continue to obtain. In particular, Brown will continue to regard Adams as having affirmed that it is possible that the patient has Lyme disease. Yet (iii) now seems to fail: each will cease to regard the other’s assertion about the patient as mistaken in any way. In particular, Brown will not merely excuse Adams’s mistake as blameless given Adams’s uninformed perspective (Price 1988: 162-163). Rather, in view of that perspective, Brown will cease taking Adams to have made any mistake, even though Brown takes Adams to have described something as possible that is not in fact possible. What was formerly a dispute about whether it’s possible that the patient has Lyme disease has evaporated —or so it has seemed to Price and his relativist heirs. Similarly, suppose that in the course of their subsequent shopping, Alice and Ben learn that Alice delights in pungent cheeses, while they have no appeal to Ben. One possible outcome is that each will cease to regard the other’s assertion as mistaken in any way, even though Ben continues to maintain that Alice described as tasty a cheese which is not in fact tasty. What was formerly a dispute about whether Limburger cheese is tasty has now evaporated—or so it has seemed to relativists.

So far, I have characterized evaporative disputes without using any notion of truth or of a proposition. But in order to generalize over the instances of our schema, we can introduce a (non-relativized) truth predicate and describe one party as denying the truth of the very proposition she recognizes the other party to have asserted. Taking the appearance of evaporative disputes seriously, as relativists do, thus involves taking seriously the idea that one may regard another person as having asserted a false proposition while insisting that there’s nothing mistaken about that person’s assertion. Relativism about the target discourses may be understood as an attempt to explain how this is possible.[8]

Evaporative disputes are usually regarded as a special case of a more general phenomenon, “faultless disagreement” (Kölbel 2002; 2003). Price himself speaks of “no-fault disagreement” (1983: 403; 1988: 170, 173), but he appears to use ‘disagreement’ as a synonym for ‘dispute’. In the relative-truth literature, by contrast, two people who aren’t engaged in any dispute are often said to stand in a relation of disagreement (see Cappelen and Hawthorne 2009: 60). I will avoid talk of disagreement, mainly because it’s seldom clear what relation is intended. Is there any familiar (or theoretically useful) sense of ‘disagree’ in which Alice and Ben continue to disagree after their dispute about Limburger cheese evaporates? For that matter, should they even be said to disagree before their dispute evaporates, or does its evaporation show that they merely take themselves to disagree?[9] I have no reason to address these questions here. In order for evaporative disputes to motivate relativism, we need no more “disagreement” than is captured by the claim that (i) and (ii) are satisfied.

Although I won’t be treating evaporative disputes as instances of “faultless disagreement,” I agree that they are instances of a phenomenon that also seems to exist in cases where there are no disputants. Such are the much-discussed eavesdropping cases (Egan et al. 2005; Egan 2007; 2011; MacFarlane 2011a). Suppose Dr. Evans overhears Dr. Adams asserting that it’s possible the patient has Lyme disease. Having seen the negative test results he knows Adams hasn’t seen, Evans then protests (to himself, or to a third party) that this isn’t possible. Here, again, it appears that Evans has denied what Adams has asserted, and recognizes this, yet doesn’t take there to be anything mistaken about Adams’s assertion. Borrowing more imagery from Price, we can say that Evans’s denial of what he recognizes Adams to have asserted is normatively “frictionless” (Price 1998; 2003). The apparent possibility of such frictionless denial is the general phenomenon of which evaporative disputes are a special case. All I say below about evaporative disputes applies equally to eavesdropping cases: when I speak of an asserter’s “audience,” I don’t assume that asserter and audience are engaged in conversation.[10]

3. Relativism as a pragmatist strategy

Recent relativists have followed Price in rejecting two familiar explanations according to which the appearance of evaporative disputes (and, more generally, frictionless denial) is an illusion. According to these explanations, apparent evaporative disputes are cases where condition (i) isn’t actually satisfied. Contrary to appearances, the parties don’t take the relevant facts to be different: they are not, respectively, affirming and denying the same proposition. Both explanations accomplish this by denying that the apparent subject-matter of the target discourses should be taken at face value. They may therefore be classified as metaphysically eliminativist regarding this apparent subject-matter.

One such option is a nonfactualist expressivism. On this view, as applied to predicates of personal taste and epistemic modals, we must reject the appearance that ‘tasty’ is used to ascribe any property, and that ‘it’s possible that. . . ’ is used to assert the obtaining of any states of affairs. Instead, we use these words to express various states of mind. In the case of ‘tasty’, this is the state of finding something’s taste appealing. In the case of ‘It’s possible that. . . ’, it is the state of leaving open some possibility.[11]

A second metaphysically eliminativist option is a traditional (“indexical”) contextualism. On that view, we use ‘tasty’ to attribute various perfectly “objective” properties, such as that of tasting good to a particular person, or group of people, at a particular time. But there’s no such property as that of being tasty, simpliciter. Likewise, according to contextualism about epistemic modal expressions, we use these to affirm various perfectly objective states of affairs concerning what’s compatible with a given person’s, or group’s, body of information (DeRose 1991). But there’s no such property as that of possibly having Lyme disease, simpliciter.[12]

By contrast, the relativist response to apparent evaporative disputes isn’t metaphysically eliminativist. Indeed, the relativist’s pronouncements in a metaphysical key are often indistinguishable from those of the “absolutist.” For example, MacFarlane affirms that there is such a property as being tasty, simpliciter. The predicate ‘tasty’, he says, “invariantly expresses a single property, the property of being tasty” (2014: 152, 105).[13] As he recounts, he has come to believe that this property is not possessed by fish sticks (2007: 20; 2014: 13-14). Relativists about epistemic modality should likewise agree that there’s a property a species has when it’s possible that it isn’t a rodent. Studies in molecular phylogenetics, I’m told, have revealed that the guinea pig has this property.[14]

How, then, does relativism differ from absolutism about tastiness and epistemic possibility?[15] What might it mean to say that a property or state of affairs is “subjective” rather than “objective”? Addressing these questions by appealing to relative truth needn’t amount to offering a metaphysical answer. Instead, MacFarlane’s answer is a pragmatist one: once again in line with Price, he focuses on how the vocabularies at issue fit into the normative structure of the practice of assertion.[16]

The next section explains MacFarlane’s own pragmatics of assessment sensitivity, which invokes relative truth. The rest of the paper will argue that once we see how assessment sensitivity is best motivated, we can see that recourse to relative truth is unnecessary. My aim here isn’t to establish that relativism about epistemic possibility or matters of taste is superior to its metaphysically eliminativist rivals, nonfactualism and contextualism. Rather, I’ll be defending a new way of understanding and elaborating the thought, shared by Price and MacFarlane, that “fact-stating” discourse admits of degrees of objectivity.[17]

4. What is assessment sensitivity?

MacFarlane argues that the practice of assertion may be understood as constitutively governed by two rules (2014: ch. 5). In this section, I will first present the rules in a simplified form, and then explain why my purposes justify the divergence from MacFarlane’s own versions. Both rules will be stated in terms of a proposition’s truth relative to a context, where it is understood that this notion can in turn be defined in terms of truth relative to some sequence of parameters supplied by the context.

The first rule is a necessary condition for being permitted to make an assertion:

(A) An agent in context c is permitted to assert p only if p is true relative to c.

In my present context, I’m not permitted to assert that it’s possible Canberra is a coastal city. That’s because, in view of the information I possess, that modal proposition is not true with respect to this context. Yet someone who lacks the relevant geographic information may not be forbidden by (A) from asserting that it’s possible Canberra is a coastal city. In making this assertion, then, such a person has not violated our constitutive rule of assertion. To that extent, at least, she need have made no mistake.

Even if (A) were the only rule constitutive of assertoric practice, we might thus already have a way to make sense of evaporative disputes. But MacFarlane argues that the practice is constitutively governed by a second rule, a sufficient condition for being obligated to retract a previous assertion:

(RAS) An agent in context c′ is required to retract an assertion of p she made, in a prior context c, if p is untrue relative to c′.

Assuming I asserted in the past that it’s possible Canberra is a coastal city, I am now required to retract that assertion, to render it “null and void” (MacFarlane 2011b: 83; 2011a: 147-148). It’s not enough that I now deny the same proposition—in addition, I’m not permitted to stand by my prior assertion. The speech act of retraction can be performed by saying ‘I take that back’. In some contexts, it may also be the act performed by saying ‘I was wrong’. Notice, however, that I may be required to retract an assertion even though I was permitted to make it in the first place.

We can now state what assessment sensitivity amounts to, when it is understood in terms of relative truth. The following general formulation aims to set the stage for my later discussion while remaining faithful to MacFarlane’s basic intentions.

(AS) The practice of assertion exhibits assessment sensitivity provided: whether an agent in context c′ is obligated or permitted to respond to an assertion with some speech act (by retracting it, rejecting it, conceding it, etc.) can depend on whether the proposition asserted is true relative to c′.

A practice governed by the retraction rule (RAS) will exhibit assessment sensitivity according to (AS). That’s because whether or not an asserter must retract her assertion at some subsequent context depends on the asserted proposition’s truth relative to that context. By contrast, consider a different retraction rule:

(RUS) An agent in context c′ is required to retract an assertion of p she made, in a prior context c, if p is untrue relative to c.

A practice of assertion with (A) and (RUS) as its sole constitutive rules would exhibit only use sensitivity, not assessment sensitivity.

It’s crucial not to conflate assessment sensitivity, as defined by (AS), with something far less interesting. Take our scenario in which Adams asserts it’s possible the patient has Lyme disease. Consider now the proposition that Adams asserted a true proposition (in the ordinary, unrelativized, sense of ‘true’). Call this the assessment proposition. And suppose that whether an assertion of the assessment proposition is correctly made depends on its truth relative to the assessor’s context (cf. Cappelen and Hawthorne 2009: 17). Would this suffice for assessment sensitivity? Not at all—it would simply show that there is use sensitivity in the assertion of assessment propositions. Some commentators hold that what’s essential to assessment sensitivity is that “we can get the same utterance correctly assessed as true by one assessor, and correctly assessed as false by another” (Egan 2011: 228; cf. López de Sa 2011: 113). Yet an utterance’s truth is most naturally understood as the truth of the proposition it expresses. In that case, the cited condition requires that the correctness of asserting that Adams’s utterance expresses a true proposition varies depending on the contexts in which that assessment proposition is used. Once again, this doesn’t amount to assessment sensitivity in the sense of (AS). Use sensitivity in the practice of assessment is one thing; assessment sensitivity is another.

I stress again that (AS) is my formulation, as are the rules (A) and (RAS). The remainder of this section explains one respect in which I am simplifying MacFarlane’s characterization of assessment sensitivity, and why the simplification doesn’t matter for present purposes. The reason his account needs additional complexity is that he allows that there may be propositions such that whether a speaker is required to retract a prior assertion of the proposition depends on features of the context of use that can differ from those of the context of assessment.[18] For example, suppose there is a time-neutral proposition that it is raining in Boston, which is true relative to some times and false relative to others. Suppose I assert this proposition during a Boston downpour. Then (RAS) would, absurdly, require me to retract the assertion once the rain stops.

Yet MacFarlane doesn’t want to give up assessment sensitivity by replacing (RAS) with (RUS). Instead, he formulates the retraction rule in terms of a proposition’s truth relative to an ordered pair of contexts:

(RUAS) An agent in context c′ is required to retract an assertion of p she made in a prior context c if p is untrue relative to 〈c, c′〉.[19]

Consider the time-neutral proposition that Alice is eating a tasty cheese. Suppose I’m assessing whether I must retract an assertion of this proposition I made a year ago. The answer will depend on my taste standard now, but on what Alice was eating a year ago. So whether I need to retract my assertion will depend on whether the proposition is true relative to a pair of contexts: last year’s context of use (which supplies the time parameter) and my present context of assessment (which supplies the taste parameter).

In Section 7, I will propose replacing (AS) with a very different explanation of what it means for a practice of assertion to exhibit assessment sensitivity. One upshot will be that it’s more revisionary to claim that retraction obligations are sensitive to features of the context of use. For this reason I will ignore MacFarlane’s (RUAS) and focus instead on the simpler (RAS). The important point is that (RAS), though formulated in terms of singly-relativized propositional truth, counts as positing assessment sensitivity according to (AS).[20]

5. Motivating the retraction norm: a dilemma

How might one motivate the claim that being subject to rule (RAS) is essential to assertion? What’s needed is an explanation of why there might be a practice that essentially involves this particular rule, rather than no retraction rule, or (RUS) instead. What difference does a retraction rule that mentions a context of assessment make to the practice governed by it? The most promising strategy for addressing this question, I believe, starts with the following basic idea about the communicative function of the speech act of assertion:

(E1) An assertion is a kind of endorsement, by which the asserter authorizes others to employ the asserted proposition as a premise in their theoretical and practical reasoning. (Here “others” include not just those the asserter believes to be privy to the assertion, but also any eavesdroppers as well as anyone who subsequently learns of the assertion.)

MacFarlane (2005b; 2007) argues that this view of assertion is compatible with his relativism. Elsewhere, he invokes assertion’s endorsing role when explaining the speech act of retraction: “ [I]n retracting an assertion, one disavows the assertoric commitment undertaken in the original assertion. This means, among other things, . . . that others are no longer entitled to rely on one’s authority. . . ” (2014: 108).[21] I will now argue that (E1), by itself, fails to motivate a retraction norm that refers to a context of assessment. My ultimate goal, however, is to show how (E1), once supplemented with a second principle, can be used to motivate assessment sensitivity.

Let’s continue to assume that propositions have context-relative truth conditions. In that case, using a proposition as a premise in one’s theoretical or practical reasoning presumably involves relying on its being true relative to the context in which one is doing the reasoning. We might then understand (E1) to mean that by asserting a proposition at a given context, a speaker endorses the proposition’s truth relative to any subsequent context in which the proposition has become available to be used, on the asserter’s authority, in some other agent’s reasoning or acting.

(E1RT) In asserting a proposition, an agent endorses the proposition’s truth relative to any subsequent context in which the assertion is registered by another agent.

Suppose now that the asserted proposition is untrue relative to some such subsequent context, perhaps due to the taste standards or information possessed by the agent of that context. Then it would seem that the asserter’s continuing endorsement becomes inappropriate. Hence it will be in order to withdraw the endorsement by means of a retraction.

Unfortunately, this defense of a retraction norm faces an obvious problem. We have been assuming that there are propositions, e.g. the proposition that Limburger cheese is tasty, that are true relative to some contexts and false relative to others (in the same world).[22] Yet, as Wright (2008: 173n19) in effect notes, there would surely be little point to a practice that involved endorsing propositions such as this one as true relative to the perspective of each potential audience member. If that were our practice, it would follow that for most assertions that something is tasty, immediate retraction would be in order. If the guiding thought (E1) is unpacked as (E1RT), it motivates an unreasonable retraction norm far more demanding than (RAS). Perhaps, then, the endorsing effect of an assertion should be captured, not by (E1RT), but rather by

(E1RT′) In asserting a proposition, an agent A endorses the proposition’s truth relative to any subsequent context whose agent is also A.

There is reason to think MacFarlane would accept this suggestion. In several papers, he describes assertion as involving commitment to defend the asserted proposition’s truth relative to a context of assessment, where the “relevant context of assessment is the context in which the asserter is evaluating the putative refutation” that has been offered by another agent as a challenge to the assertion (2005b: 336, emphasis added; also 2005a: 229 and 2007: 29). In other words:

(E1RT″) In asserting a proposition, an agent A undertakes to provide grounds for the proposition’s truth relative to any subsequent context whose agent is also A, and in which the assertion is properly challenged by another agent.

However, neither (E1RT′) nor (E1RT″) can provide the motivation for the assessment-sensitive retraction rule we’ve been looking for. That is because, unlike (E1RT), they no longer draw support from (E1), the natural thought that an asserter is endorsing a proposition for use by others. To see the problem, consider MacFarlane’s childhood assertion that fish sticks are tasty. Let’s assume the proposition is true relative to his context of utterance, but false relative to the context he occupies today. The defender of (RAS) now faces a dilemma. In asserting that fish sticks are tasty, did young MacFarlane put the proposition forward for others to use today? Suppose he did. This promises to account for his present obligation to retract, assuming that the proposition in question is false relative to the contexts of some of those at whom his endorsement remains directed. But it doesn’t explain why his obligation to retract should depend on his own present dislike of fish sticks. Suppose, then, that he didn’t put the proposition forward for anyone besides himself to use today. In that case, it’s unclear why resolving a clash between his earlier endorsement and his present dislike of fish sticks should obligate him to perform a public speech act of retraction.[23]

We can pose the same dilemma in terms of an asserter’s commitment to respond to challenges. Learning of MacFarlane’s childhood assertion today, I challenge him by denying that fish sticks are tasty. Once again, either young MacFarlane put forward the proposition for me to use today, or he didn’t. Suppose he did. In that case, why would successfully vindicating his assertion in response to my challenge require showing that the proposition is true relative to the gustatory perspective he now occupies, rather than the one I do? Suppose, then, that he didn’t put the proposition forward for me to use. Then why should my denial that fish sticks are tasty count as a challenge to his earlier assertion, one he is obligated to respond to on pain of retracting that assertion?

In the face of this dilemma, can we still motivate assessment sensitivity in terms of the idea that assertion is an endorsement for others? In the next two sections, I will show that the answer is “yes.” The key to escaping the dilemma will be the following claim:

(E2) Whether an asserter counts as having issued an endorsement for a given audience member to rely on can depend on the asserter’s own circumstances at the time when the audience member registers the assertion.

For example, it is only because MacFarlane presently shares my attitude toward the taste of fish sticks that his unretracted childhood assertion counts as an endorsement for me to rely on today. If this is right, the first horn of the dilemma ends up being acceptable after all.

Obviously, much more needs to be said to fill in this sketch. In the remainder of the paper, I will formulate a pragmatic analogue of assessment sensitivity based on principles (E1) and (E2). As I will explain, this will not require describing the normative structure of assertoric practice in terms of relative truth. Still, I won’t argue that my reply to the dilemma is incompatible with principles and rules stated in terms of relative truth.[24] Rather, my claim is that once we see how to avoid the dilemma, we can see that the real work in motivating a role for contexts of assessment in connection with retraction is done by (E1) when supplemented by (E2) in the way I will propose. Neither of these two principles is formulated using a notion of truth, absolute or relative. Accordingly, Section 7 will offer a characterization of assessment sensitivity that employs no notion of truth.

First, however, let me briefly contrast my proposal with a very different rationale MacFarlane has recently offered for a retraction rule that involves assessment sensitivity (2014: 310-316). He argues that conforming to the rule (RAS) imposes a lesser “cognitive burden” on agents than does conforming to (RUS). That’s because in the former case, an agent’s ability to determine when to retract her prior assertion doesn’t require that she have retained information about the perspective she occupied at the time of assertion. Instead, all that matters is the perspective she herself occupies when determining whether she should retract. From the “engineering point of view,” such an agent is easier to implement.

I have two main reservations about this proposal. First, it addresses at most the question whether retraction should be governed by (RAS) or instead by (RUS). By contrast, I have sought to explain why there should be retraction obligations in the first place, regardless what engineering constraints such obligations may impose. Second, as far as MacFarlane’s stated rationale in terms of cognitive burden goes, it looks entirely accidental that assertions can have an audience, i.e. that assertoric practice has a communicative dimension. (This is related to the fact that on MacFarlane’s implementation of assessment sensitivity, whether what is asserted is true or false relative to the context of some agent other than the asserter plays no role whatsoever.[25]) By contrast, in appealing to (E1) I aim to make sense of the retraction norm by understanding why retraction can be an appropriate response to a challenge by one’s audience. In this respect, my proposal accords better with MacFarlane’s own explanation of retraction as the disavowal of assertoric commitment, as a result of which “others are no longer entitled to rely on one’s authority.”[26]

6. Assertoric force perspectivalism introduced

I start with a brief sketch of Brandom’s account of assertion as “the fundamental sort of move in the game of giving and asking for reasons” (Brandom 1994: 141). According to Brandom, asserting is a speech act by which “one not only licenses further assertions on the part of others, but commits oneself to justifying the original claim” when challenged (Brandom 1983: 641; 1994: 171-172). Asserting thus involves “coordinate dimensions of authority and responsibility” (1994: 238). The responsibility undertaken in asserting a proposition is to “vindicate” one’s entitlement to assert the proposition when one’s assertion is appropriately challenged. Entitlement is vindicated in various ways: by inference, by deferral to another’s authority, and by invoking observational authority (1994: 174). The authority claimed in asserting a proposition is to license others to assert the same proposition and use it as a premise in their theoretical and practical reasoning.

When an assertion succeeds in such licensing, the hearer is then in a position to vindicate her entitlement to assert the proposition through a deferral of responsibility to the original asserter. Whether an assertion succeeds in such licensing depends on two factors: whether the asserter possesses the entitlement in question, and whether the hearer has some status that precludes her from inheriting that entitlement. Unlike Brandom, I will have little to say about what kind of status might serve as such a defeasor.[27] But one defeasor will be crucial to my argument. According to Brandom, someone’s having issued an unretracted denial of a proposition counts as precluding her from inheriting entitlement as a result of another person’s assertion of that same proposition.

A consequence of Brandom’s analysis is that there can be no frictionless denial. Suppose B denies a proposition she recognizes A to have asserted. In denying the proposition, B will take herself to “void . . . the communicative authority,” implicitly claimed in A’s assertion, to entitle potential audience members (including B) to assert the same proposition (Brandom 1994: 178). And when B is denying the proposition, she can’t take the blame for this failure of communicative authority to lie with her own denial, rather than with A’s assertion. Hence Brandom’s analysis neatly accounts for the following feature of fully “objective” discourse stressed by Price:

(D) If B takes herself to be denying the proposition A asserted, B thereby takes there to be something defective about A’s assertion.

As Price puts it: “We are prepared to make the judgment that a speaker is incorrect, or mistaken, . . . simply on the basis that we are prepared to make a contrary assertion” (Price 2003: 176; Price 1998: 246). But in allowing that “factual discourse” admits of degrees of objectivity, Price also suggests that there is room for a practice characterized by only a qualified version of feature (D). To understand such a practice, I will now argue, we need to make a modification to Brandom’s analysis of the “social articulation of the inferential practice of giving and asking for reasons” (1994: 54). In terms of a distinction drawn by Kukla and Lance (2009: 16-22), we need to allow that the entitlement-transmitting function of assertion is in general “agent-relative” (rather than “agent-neutral,” as they, following Brandom, hold it to be).

Consider, once again, Alice’s assertion that Limburger cheese is tasty. On the view I propose, Alice no longer claims authority to license all who become aware of her assertion to assert this proposition. Rather, she only licenses such assertion by those potential audience members who share a relevantly similar standard of taste. Given Ben’s aversion to pungent cheeses, his denial that Limburger is tasty doesn’t count as a challenge Alice is obligated to respond to. Stated generally, the idea is that assertoric force has a perspectival aspect.[28] On this view, an assertion may carry its licensing potential not vis-à-vis its entire potential audience, but rather vis-à-vis a target audience restricted to those who share a particular perspective. Here is a first attempt at implementing the idea:

(AFP) In asserting p, an agent A claims authority to license the assertion of p by any audience member who occupies a context relevantly similar to the context occupied by A herself at the time of assertion.[29]

I have proposed that in the case of a proposition about tastiness, the target audience is picked out by possession of a relevantly similar standard of taste. In the case of propositions about epistemic possibility, the target audience is picked out by possession of a relevantly similar body of information.[30]

Assertoric force perspectivalism promises to explain the possibility of frictionless denial. Once Ben learns that Alice delights in pungent cheeses, he may cease to regard his own denial that Limburger is tasty as a challenge to her assertion. Once Dr. Brown learns that Dr. Adams doesn’t know the results of the relevant medical tests, she may cease to regard her own denial that the patient might have Lyme disease as a challenge to Adams’s assertion that the patient might have Lyme disease. In these circumstances, principle (D) breaks down, just as Price claims it can at the “margins of factual discourse.” We are no longer prepared to attribute a defect merely on the basis of our own denial of the proposition the speaker asserted.

7. Assessment sensitivity without relative truth

As presented so far, assertoric force perspectivalism explains frictionless denial but fails to make sense of the retraction obligations characteristic of assessment sensitivity. Consider again MacFarlane’s childhood assertion that fish sticks are tasty. Given how we’ve been applying (AFP) to claims about tastiness, it would seem that young MacFarlane wasn’t endorsing the proposition for assertion by those with adult MacFarlane’s standard of taste. Accordingly, if challenged by such a person, he would seem subject to no requirement to retract his childhood assertion. In short, I have yet to show how assertoric force perspectivalism underwrites anything like assessment sensitivity.

To capture the retraction obligations characteristic of assessment sensitivity, we need to revise the proposal in a way that implements claim (E2) of Section 5:

(AFP′) In asserting p, an agent A claims authority to license the assertion of p by any audience member who occupies a context relevantly similar to a context occupied by A herself either at the time of assertion, or at the time that audience member registers the assertion.

As long as MacFarlane hasn’t retracted his childhood assertion that fish sticks are tasty, he still counts as endorsing this claim for assertion, on his authority, by anyone who (in the relevant respects) shares his present standard of taste.[31]

(AFP′) gives us an analogue of the rule (RAS), stated without any truth predicate:

(R1) Suppose A’s assertion of p at time t is challenged at t′ by someone who would (at t′) belong to the target audience of an assertion at t′ of p by A. Then, unless A can vindicate her entitlement (at t) to assert p, she should retract the assertion.

For example, since MacFarlane can’t vindicate his present entitlement to assert the proposition, he should retract his childhood assertion when I challenge him. That is because, according to (AFP′), the endorsement effected through his childhood assertion extends to the present, vis-à-vis any audience whose standard of taste is relevantly similar to his present standard. Principle (R1) attributes assessment sensitivity in a sense resembling MacFarlane’s: an agent’s obligation to retract an assertion she made in one context may depend on her perspective at a subsequent context.

Assertoric force perspectivalism also yields an explanation of the conditions in which an audience member will take it that an assertion is defective as an endorsement. This is a qualified version of Price’s principle (D):

(D′) If B takes herself to be denying a proposition A asserted, and takes herself to belong to the potential audience vis-à-vis which A thereby claims entitlement-transmitting authority, then B thereby takes there to something defective about A’s assertion.

And (D′) has consequences for retraction, provided we understand assertoric force perspectivalism in terms of (AFP′). If MacFarlane presently denies that fish sticks are tasty, he thereby takes there to be something defective about his childhood assertion, a defect that would be remedied by retracting that assertion. In general:

(R2) If an agent takes herself to be denying a proposition she previously asserted, she thereby takes it that there is something defective about her earlier assertion—a defect that would be remedied by a retraction.

Taking someone’s assertion to be defective as an endorsement doesn’t mean recognizing any sense in which she ought not have made the assertion in the first place (cf. MacFarlane 2014: 110). In particular, one may still take the assertion to have succeeded in transmitting entitlement to hearers who at the time of the assertion belonged to its restricted target audience.[32]

I have now argued that when assertoric force perspectivalism is understood in terms of (AFP′), it underwrites versions of MacFarlane’s claims about the appropriateness of retraction, the phenomenon that leads him to attribute assessment sensitivity. As advertised, I have made no use of relative truth. Next, I will explain why allowing this pragmatic analogue of assessment sensitivity is a less radical move than allowing a pragmatic analogue of use sensitivity.

Figure 1: Which speakers does a past assertion presently target? (The shaded areas of perspective space contain perspectives of members of the assertion’s target audience.)
Figure 1: Which speakers does a past assertion presently target? (The shaded areas of perspective space contain perspectives of members of the assertion’s target audience.)

8. Degrees of pragmatic relativism

Recall that on Brandom’s own account of assertion, asserting a proposition is endorsing it for use by any potential audience member. This is the pragmatic counterpart of the absolutist position in semantics which attributes to every proposition a condition for being absolutely true. I have proposed liberalizing Brandom’s account by allowing for the endorsement of certain propositions to be directed at restricted target audiences, audiences restricted to those who occupy a certain perspective. But I’ve pointed out that there are two very different versions of this liberalization.

Consider first the version I’m recommending, which allows an analogue of MacFarlane’s assessment sensitivity. I call this “mitigated relativism” because it imposes a limit on how an assertion’s target audience may be restricted (see Figure 1). According to (AFP′), an assertion always involves sticking one’s neck out with respect to one’s own future contexts. To determine which potential assessors belong to an assertion’s target audience at a given time, we need to consider not only features of the asserter’s context at the time of assertion, but also features of the asserter’s context at the time of assessment. In this sense, assertoric force is assessment-sensitive. As we’ve seen, this underwrites principles (R1) and (R2) concerning retraction.

The second way of liberalizing Brandom’s account, which I call “excessive relativism,” allows an analogue of MacFarlane’s use sensitivity. On this view, captured by (AFP), asserting a proposition counts as endorsing it for use by those who share a certain perspective occupied by the asserter at the time of the assertion. Assertion thus need not involve sticking one’s neck out with respect to one’s own future contexts. Take the case of epistemic possibility. According to excessive relativism, in asserting that Canberra might be a coastal city an asserter claims entitlement-transmitting authority only vis-à-vis those whose state of information is relevantly similar to the asserter’s own at the context of assertion. In that case, there would be no reason to expect (R1) and (R2) to hold. There would be nothing wrong with someone’s asserting that Canberra might be a coastal city and refusing to retract the assertion when challenged in a subsequent more informed context in which she can’t vindicate her entitlement. Nor need she see any normative friction between her original assertion and a subsequent denial that Canberra might be a coastal city. She could say: ‘Yes, what I asserted then is false, but why is there any problem with that?’[33]

Like the views that attribute what MacFarlane calls “use sensitivity,” excessive relativism is committed to the following claim:

(ER) There is a proposition p such that the same person, in the same world, may assert p at one time and deny p at a later time (and retract neither speech act) without incurring any normative deficiency.

I hope this brings out how use-sensitivity, allegedly the more “moderate” of our two brands of relativism, actually presents the more radical challenge to a traditional conception of how propositions are individuated. One of my aims has been to show how attributing assessment sensitivity allows a theorist of subjective discourse to retain an account of the practice of assertion in terms of “propositions” for which (ER) is ruled out.[34] It’s essential to such a practice that a practitioner counts, in virtue of her unretracted assertion, as endorsing some proposition for use by all those for whom she would be endorsing that proposition if she were to presently assert it. Another way to make this point is to say, with Brandom (2002: 53-54), that being a linguistic agent involves being a diachronic “locus of conceptual commitment and responsibility.”[35]

A putative practice described by excessive relativism can be viewed as a less extreme version of a fictional practice described by Price (1998; 2003). He contemplates a community of “merely opinionated asserters” who never attribute any normative failing when “denying” the claim another has “asserted.” Price holds that this isn’t a coherent description of a practice in which propositions get asserted. I have been suggesting that the same is true of a description according to which (ER) obtains. The possibility of assessment-sensitive assertoric force gives substance to Price’s thesis that broadly “fact-stating discourse” admits of different grades of objectivity. But use-sensitive assertoric force may not be a feature any fact-stating discourse can display.[36]

Suppose I’m right that fact-stating discourse has room for assessment-sensitive assertoric force, even if it has no room for use-sensitive assertoric force. An important question remains. Why would our language be such as to allow fact-stating discourse about tastiness and epistemic possibility at all? Why don’t we instead merely engage in non-propositional expressions of our states of mind, of the sort envisioned by expressivists: for example the state of finding something’s taste appealing, or of leaving a certain possibility open?

Here I’m sympathetic to an idea voiced by both Price and MacFarlane. This is that incorporating the expression of personal perspectives into assertoric practice results in an opportunity for controversy, which in turn encourages the interpersonal coordination of perspectives. According to MacFarlane (2007: 29-30),[37]

Assessment-sensitive expressions are designed, it seems, to foster controversy. . . . But what is the point of fostering controversy in “subjective” domains, if there is no (nonrelative) truth on which both parties can converge? . . . Perhaps the point is to bring about agreement by leading our interlocutors into relevantly different contexts of assessment. If you say “skiing is fun” and I contradict you, it is not because I think that the proposition you asserted is false as assessed by you in your current situation, with the affective attitudes you now have, but because I hope to change these attitudes. Perhaps, then, the point of using controversy-inducing assessment-sensitive vocabulary is to foster coordination of contexts.

The suggestion that the underlying purpose of assessment-sensitive assertoric force is to encourage coordination by fostering dispute is compatible with the recognition that some of the disputes will end up evaporating, and thus failing to promote that purpose.

In MacFarlane’s example of ‘fun’, the perspectives whose interpersonal coordination is encouraged by allowing disputes about whether something is fun are standards of fun. In the case of epistemic modals, the point of fostering controversy about what is possibly the case may lie in promoting coordination among speakers with regard their attitudes of leaving certain possibilities open. Yalcin (2011: 310-312) identifies such coordination as the aim speakers pursue in making epistemic possibility claims. Though he defends a nonfactualist expressivism about these claims, the proponent of assessment-sensitive assertoric force could argue that coordination is better promoted by a practice in which epistemic possibility claims express propositions and figure in at least a liberalized version of Brandom’s game of giving and asking for reasons.

I have now tentatively endorsed an explanation of why our language might allow the expression of propositions whose assertion involves perspectival assertoric force. Such an explanation would be an important component of a defense of relativism against rival accounts of the target discourses, such as expressivist or contextualist ones. But my purpose in this paper hasn’t been to defend relativism. Rather, it has been twofold. First, I’ve argued that the relativism MacFarlane describes in terms of relative truth is more revealingly described in terms of perspectival assertoric force. Second, I’ve argued that such a relativism, which posits assessment sensitivity, amounts to a mitigated form compared with alternatives that posit use sensitivity.

9. Objections and replies

I close by addressing four objections to my proposal to reconceive assessment sensitivity in terms of assertoric force perspectivalism. (I will have to disregard objections that apply equally to MacFarlane’s relativism formulated in terms of relative truth, or that apply equally to Brandom’s account of assertion in its original absolutist guise.)

Objection 1 Accounts of an expression’s behavior in terms of force often encounter difficulty dealing with the behavior of sentences where the expression appears in an embedded context, e.g. in the antecedent of a conditional. The problem is that the force in question (e.g. imperatival force) typically doesn’t get exercised in the use of such compound sentences. Since I’m explaining the subjectivity of epistemic modals and predicates of personal taste in terms of their effect on the force of an utterance, doesn’t my proposal face this “Frege-Geach problem” (Geach 1960)?

Reply According to assertoric force perspectivalism, the perspectival aspect of an assertion’s force is independent of whether the relevant expression appears in an embedded context. Just as an assertion that Limburger cheese is tasty counts as an endorsement vis-à-vis a perspectivally restricted audience, so does an assertion that if Limburger cheese isn’t tasty, the deli doesn’t sell it. The perspective-based restriction on the target audience depends on the concepts involved in the proposition, not on how they are embedded. Thus, the proposal faces no obvious Frege-Geach problem. (However, Objections 2 and 3 also concern embeddings.)

Objection 2 Suppose Alex asserts the conjunctive proposition that Canberra has over a million inhabitants and it’s possible that it’s a coastal city. On the view I’ve suggested, Alex doesn’t thereby claim entitlement-transmitting authority vis-à-vis someone, such as Beverly, who is better informed about political geography. Hence, were Beverly to respond by denying that Canberra has over a million inhabitants (thus denying a logical consequence of the proposition Alex asserted), she wouldn’t count as challenging Alex’s assertion. Yet this clearly should count as a challenge.

Reply Here the solution will require integrating what Brandom calls the “intrapersonal” and the “intra-content” dimensions of the “game of giving and asking for reasons” (1994: 175). According to Brandom, there are content-based restrictions on attributions of entitlement. For instance, being entitled to assert a conjunction entails being entitled to assert each conjunct. Hence, we might say that in asserting a conjunction, one claims entitlement-transmitting authority not only for the conjunction, but also for each conjunct. (How far this could be extended is a difficult topic I won’t discuss.) In our example, then, Alex has claimed entitlement-transmitting authority for the proposition that Canberra has over a million inhabitants, vis-à-vis those potential audience members who share a perspective relevantly similar with regard to that proposition (as determined by the concepts it involves). Unless the concept of an inhabitant has a perspectival aspect, this will be all potential audience members, including Beverly. Hence, in denying that Canberra has over a million inhabitants, Beverly does place herself in normative tension with Alex’s conjunctive assertion.

Objection 3 There are features of epistemic modal discourse which my proposal can’t explain. I have proposed that assertion of a proposition involving the constituent that it’s possible that φ carries entitlement-conveying force vis-à-vis those who possess no more information relevant to whether or not to leave open that φ than does the asserter. Recall our two physicians: Dr. Adams, who hasn’t seen the patient’s test results for Lyme disease, and Dr. Brown, who has seen negative test results. Now consider the following two propositions

p = the proposition that it must be the case that the patient has Lyme disease (i.e., the proposition that it’s not possible the patient doesn’t have Lyme disease)

q = the proposition that if it’s possible the patient doesn’t have Lyme disease, then further tests for Lyme disease are needed.

On the proposed account, the potential audience Adams would target in asserting either of these propositions excludes Brown, who has seen a test result ruling out Lyme disease. Yet it seems that Brown’s denial of p should count as challenging Adams’s assertion of p.[38] Likewise, suppose Brown denies q, knowing the patient doesn’t have Lyme disease, and therefore thinking further tests unnecessary. Then her denial should count as challenging Adams’s assertion of q.

Reply These phenomena can be explained without appealing to the entitlement-conveying aspect of assertoric force. On the view I’ve proposed, the reason we should take Brown to be disputing Adams’s assertion of p isn’t that Brown is in the target audience for this assertion. Rather, the reason is that Adams’s entitlement to assert p wouldn’t be defeasible by the new information available to Brown. If Adams were entitled to assert that the patient must have Lyme disease, and the only relevant difference in their perspectives is that Brown has seen an additional test result, then Brown too would be entitled to assert this. Since Brown isn’t entitled to assert that the patient must have Lyme disease, then neither is Adams. In short, though Adams’s assertion lacks licensing potential relative to Brown, it remains true that if Adams is entitled to assert p, then so is Brown. Hence, if Brown takes herself to lack entitlement to assert p, she is committed to Adams’s lacking entitlement as well. The same style of explanation can account for the observation concerning q. In the case of a proposition that if it’s possible that φ, then ψ, entitlement to assert the proposition can’t be defeated merely by possession of additional information to the effect that φ. Though Adams’s assertion of the conditional q lacks entitlement-conveying force relative to Brown, it’s nonetheless true that if Adams is entitled to assert q, then so is Brown. Hence, if Brown take herself to lack entitlement to assert q, she is implicitly calling into question Adams’s entitlement to assert q.

In sum, the objection is met by distinguishing two different ways one can count as challenging another person’s assertion. First, provided one belongs to the targeted potential audience, one may defeat the asserter’s entitlement-transmitting authority by denying the asserted proposition. Second, even if one doesn’t belong to the targeted audience, one may challenge the asserter’s entitlement to make the assertion in the first place.

Objection 4 Those who have invoked relative truth in theorizing about epistemic modals and predicates of taste have done so in connection with a compositional semantics for languages with these expressions. In such a semantics, the “semantic value” assigned to a sentence at a context is typically a function that yields a truth value given an “index” that includes nonstandard parameters such as information state and standard of taste. In proposing that relativism about the target discourses doesn’t require relativized truth, am I rejecting such compositional semantics?

Reply It’s important not to exaggerate the connection between a compositional semantics with nonstandard indices and an appeal to relativized propositional truth in accounting for the target discourses’ distinctive functioning. Recent work by relativists as well as their critics has stressed a gap between the concepts used in giving a compositional semantics and the concepts used in describing or explaining a language’s communicative functioning.[39] In light of this gap, I will argue, attribution of distinctively relativistic normative structure to our target discourses needn’t be based on a compositional semantics, even if the latter is required for other theoretical purposes.[40]

Let me first explain the relation MacFarlane sees between relativism and compositional semantics. Compositional semantics recursively characterizes an expression’s “extension” with respect to a context and an index of parameters, written \(⟦e⟧^{c}_{i}\). For example, supposing an index is a world/time pair, \(⟦{I\ am\ U.S.\ Secretary\ of\ State} ⟧^{c}_{i}\) = Truth when c is any context whose agent is Hillary Clinton, and i has as its world parameter the actual world and as its time parameter a time in 2012. But this notion is not the one MacFarlane appeals to in characterizing the assertoric use of sentences. That role is played instead by the following relation between a sentence and a pair of contexts:

S expresses a proposition at c that is true relative to c′.[41]

We saw how this notion can be used to formulate norms governing the assertion of S and the subsequent retraction of such an assertion, for example the twin rules (A) and (RAS).

MacFarlane calls the notion just mentioned a “postsemantic” one. As this label suggests, he takes there to be a close connection between it and compositional semantics. If we ignore complications including the need to handle assignments of values to variables, the connection he affirms amounts to the following biconditional (2014: 151-152, 67):

(PS) S expresses a proposition at c that is true relative to c′ iff the extension of S with respect to c and the index determined by c and c′ is Truth, i.e. \(⟦ S ⟧^{c}_{i(c,c')}\) = Truth.

If the only expression responsible for assessment sensitivity is ‘tasty’, the index i(c, c′) is said to be the ordered triple 〈wc, tc, gc〉 consisting of the world and time of c together with the gustatory standard of c′. In view of (PS), the objector is right that MacFarlane relates the central notion of his relativistic account of the assertoric use of sentences to his compositional semantics.

However, as MacFarlane himself stresses, the relativistic flavor of his account doesn’t result from the presence of nonstandard index parameters in the compositional semantics.[42] The very same compositional semantics is compatible with both relativistic and non-relativistic accounts of assertoric practice, including “indexical” contextualist theories on which the sentences in question express propositions with absolute truth conditions (Ninan 2010), and nonfactualist theories on which the sentences in question express no propositions at all (Yalcin 2011). In view of what is thus already agreed to be a loose connection between (a) giving a compositional semantics and (b) explaining how our communicative use of the target discourses is sensitive to perspectives, I see no obvious obstacle to undertaking (b) in a way that doesn’t depend on a compositional semantics, or even on the “postsemantic” notion of a proposition’s context-relative truth.[43] An account that takes this route, such as the assertoric force perspectivalism proposed in this paper, may still concede the importance of a compositional semantics with nonstandard index parameters. Such a semantics would have its proper role in a different explanatory project: the project of explaining or (at least) modeling aspects of our semantic competence with compound sentences, in particular our judgments concerning entailment relations.[44]


This paper originated in a talk given in 2011 at the conference EPR3: Expressivism, Projection and Rules at the University of Sydney, where I benefited greatly from audience comments, especially by John MacFarlane, Huw Price and Robert Brandom. I presented successive drafts during 2011-12 at Arché (University of St Andrews), the University of Connecticut, the University of Nevada, Las Vegas, and at a conference on inferentialism hosted by UNED Madrid. Many thanks to these audiences, and to the audience of a recent Discourse in Philosophy Colloquium talk at ILLC (University of Amsterdam). Thoughtful comments from anonymous referees are responsible for a number of improvements and corrections.


  • Brandom, Robert (1983). Asserting. Noûs, 17 (4), 637–650. http://dx.doi.org/10.2307/2215086
  • Brandom, Robert (1994). Making It Explicit: Reasoning, Representing and Discursive Commitment. Harvard University Press.
  • Brandom, Robert (2002). Tales of the Mighty Dead. Harvard University Press.
  • Cappelen, Herman, and John Hawthorne (2009). Relativism and Monadic Truth. Oxford University Press. http://dx.doi.org/10.1093/acprof:oso/9780199560554.001.0001
  • Cappelen, Hermann, and John Hawthorne (2011). Reply to Lasersohn, MacFarlane, and Richard. Philosophical Studies, 156 (3), 449–466. http://dx.doi.org/10.1007/s11098-010-9627-8
  • DeRose, Keith (1991). Epistemic Possibilities. Philosophical Review, 100 (4), 581-605.
  • Egan, Andy (2007). Epistemic Modals, Relativism and Assertion. Philosophical Studies, 133 (1), 1–22.
  • Egan, Andy (2010). Disputing about Taste. In Richard Feldman and Ted Warfield (Eds.), Disagreement (p. 247-286). Oxford University Press. http://dx.doi.org/10.1093/acprof:oso/9780199226078.003.0011
  • Egan, Andy (2011). Relativism about Epistemic Modals. In Steven Hales (Ed.), A Companion to Relativism (pp. 219–241). Wiley-Blackwell.
  • Egan, Andy, John Hawthorne, and Brian Weatherson (2005). Epistemic Modals in Context. In Gerhard Preyer and Georg Peter (Eds.), Contextualism in Philosophy (pp. 131–168). Oxford University Press.
  • Geach, Peter (1960). Ascriptivism. Philosophical Review, 69 (2), 221–225.
  • Kölbel, Max (2002). Truth without Objectivity. Routledge. http://dx.doi.org/10.2307/2183506
  • Kölbel, Max (2003). Faultless Disagreement. Proceedings of the Aristotelian Society, 104 (1), 53–73.
  • Kölbel, Max (2008). Introduction: Motivations for Relativism. In Manuel García-Carpintero and Max Kölbel (Eds.), Relative Truth (pp. 1–38). Oxford University Press. http://dx.doi.org/10.1093/acprof:oso/9780199234950.003.0001
  • Kukla, Rebecca, and Mark Lance (2009). Yo! and Lo!: The Pragmatic Topography of the Space of Reasons. Harvard University Press.
  • Lasersohn, Peter (2005). Context Dependence, Disagreement, and Predicates of Personal Taste. Linguistics and Philosophy, 28 (6), 643–686. http://dx.doi.org/10.1007/s10988-005-0596-x
  • Lewis, David (1980). Index, Context and Content. In Papers in Philosophical Logic (pp. 21–44). Cambridge University Press, 1998.
  • López de Sa, Dan (2011). The Many Relativisms: Index, Context, and Beyond. In Steven Hales (Ed.), A Companion to Relativism (pp. 102–117). Wiley-Blackwell.
  • MacFarlane, John (2003). Future Contingents and Relative Truth. Philosophical Quarterly, 53 (212), 321–336. http://dx.doi.org/10.1111/1467-9213.00315
  • MacFarlane, John (2005a). The Assessment Sensitivity of Knowledge Attributions. Oxford Studies in Epistemology, 1, 197–234.
  • MacFarlane, John (2005b). Making Sense of Relative Truth. Proceedings of the Aristotelian Society, 105 (1), 321–339.
  • MacFarlane, John (2007). Relativism and Disagreement. Philosophical Studies, 132 (1), 17–31. http://dx.doi.org/10.1007/s11098-006-9049-9
  • MacFarlane, John (2009). Nonindexical Contextualism. Synthese, 166 (2), 231–250. http://dx.doi.org/10.1007/s11229-007-9286-2
  • MacFarlane, John (2011). Simplicity Made Difficult. Philosophical Studies, 156 (3), 441-448. http://dx.doi.org/10.1007/s11098-010-9626-9
  • MacFarlane, John (2011a). Epistemic Modals Are Assessment-Sensitive. In Andy Egan and Brian Weatherson (Eds.), Epistemic Modality (pp. 144–178). Oxford University Press. http://dx.doi.org/10.1093/acprof:oso/9780199591596.003.0006
  • MacFarlane, John (2011b). What is Assertion? In Jessica Brown and Herman Cappelen (Eds.), Assertion: New Philosophical Essays (pp. 79–96). Oxford University Press. http://dx.doi.org/10.1093/acprof:oso/9780199573004.003.0004
  • MacFarlane, John (2014). Assessment Sensitivity: Relative Truth and its Applications. Oxford University Press. http://dx.doi.org/10.1093/acprof:oso/9780199682751.001.0001
  • Ninan, Dilip (2010). Semantics and the Objects of Assertion. Linguistics and Philosophy, 116 (5), 355–380. http://dx.doi.org/10.1007/s10988-011-9084-7
  • Price, Huw (1983). Does ‘Probably’ Modify Sense? Australasian Journal of Philosophy, 61 (4), 396–408. http://dx.doi.org/10.1080/00048408312341151
  • Price, Huw (1988). Facts and the Function of Truth. Basil Blackwell.
  • Price, Huw (1998). Three Norms of Assertibility. Philosophical Perspectives, 12, 41–54.
  • Price, Huw (2003). Truth as Convenient Friction. Journal of Philosophy, 100 (4), 167–190.
  • Price, Huw (2011). Naturalism without Mirrors. Oxford University Press.
  • Rabern, Brian (2012). Against the Identification of Assertoric Content with Compositional Value. Synthese, 189 (1), 75–96. http://dx.doi.org/10.1007/s11229-012-0096-9
  • Recanati, François (2007). Perspectival Thought: A Plea for (Moderate) Relativism. Oxford University Press. http://dx.doi.org/10.1093/acprof:oso/9780199230532.001.0001
  • Recanati, François (2008). Moderate Relativism. In Manuel García-Carpintero and Max Kölbel (Eds.), Relative Truth (pp. 41–62). Oxford University Press. http://dx.doi.org/10.1093/acprof:oso/9780199234950.003.0002
  • Richard, Mark (1981). Temporalism and Eternalism. Philosophical Studies, 39 (1), 1–13. http://dx.doi.org/10.1007/BF00354808
  • Richard, Mark (2004). Contextualism and Relativism. Philosophical Studies, 119 (1/2), 215–242. http://dx.doi.org/10.1023/B:PHIL.0000029358.77417.df
  • Rovane, Carol (2011). Relativism Requires Alternatives, not Disagreement or Relative Truth. In Steven Hales (Ed.), A Companion to Relativism (pp. 31–52). Wiley-Blackwell.
  • Schaffer, Jonathan (2011). Perspective in Taste Predicates and Epistemic Modals. In Andy Egan and Brian Weatherson (Eds.), Epistemic Modality (pp. 179–226). Oxford University Press. http://dx.doi.org/10.1093/acprof:oso/9780199591596.003.0007
  • Stephenson, Tamina (2007). Judge Dependence, Epistemic Modals, and Predicates of Personal Taste. Linguistics and Philosophy, 30 (4), 487-525. http://dx.doi.org/10.1007/s10988-008-9023-4
  • Stojanovic, Isidora (2012). On Value-Attributions: Semantics and Beyond. Southern Journal of Philosophy, 50 (4), 621-638. http://dx.doi.org/10.1111/j.2041-6962.2012.00137.x
  • Strawson, Peter (1950). Truth. Aristotelian Society Supplementary Volume, 24, 129–156.
  • Willer, Malte (2013). Dynamics of Epistemic Modality. Philosophical Review, 122 (1), 45-92.
  • Wright, Crispin (2008). Relativism about Truth Itself: Haphazard Thoughts about the Very Idea. In Manuel García-Carpintero and Max Kölbel (Eds.), Relative Truth (pp. 157–185). Oxford University Press. http://dx.doi.org/10.1093/acprof:oso/9780199234950.003.0007
  • Yalcin, Seth (2007). Epistemic Modals. Mind, 33 (464), 983–1026. http://dx.doi.org/10.1093/mind/fzm983
  • Yalcin, Seth (2011). Nonfactualism about Epistemic Modality. In Andy Egan and Brian Weatherson (Eds.), Epistemic Modality (pp. 295–332). Oxford University Press. http://dx.doi.org/10.1093/acprof:oso/9780199591596.003.0011


    1. In this respect he follows Strawson (1950: 143-144), who opposes attempts at “elucidating the fact-stating type of discourse” in terms of truth.return to text

    2. The terminology derives from Recanati (2007; 2008); cf. López de Sa (2011), Kölbel (2008: 18-20) and Schaffer (2011: 186).return to text

    3. MacFarlane argues that use sensitivity remains “on the safe side of the really interesting line” (2014: 89; cf. 191). By MacFarlane’s lights, we will see, some attempted characterizations of assessment sensitivity really capture only use sensitivity.return to text

    4. Price reserves the label ‘relativism’ for one application of his general account how discourses can be perspectival, namely to cases where the relevant differences in perspective aren’t just a matter of the different evidence available to speakers (1988: 177; cf. 195).return to text

    5. Price also argues that there may be room for evaporative disputes concerning indicative and subjunctive conditionals, aesthetic and moral matters, modality, causation and secondary qualities (1988: ch. 8).return to text

    6. The first is a variant of one used for a different purpose by DeRose (1991), though see also Price (1988: 163). On disputes involving ‘tasty’, see Kölbel (2003), Lasersohn (2005) and Egan (2010).return to text

    7. Here, and in the rest of the paper, I follow Price in regarding the assertion of a proposition’s negation as at least one way of denying that proposition.return to text

    8. Price describes the situation differently, as one where one disputant ceases to regard the other as having spoken falsely, even though she denies the very claim she takes the other to have asserted. Since “ ‘true’ and ‘false’ are evaluative terms,” he writes, we “should expect that speakers who discover that a [dispute has evaporated] would be reluctant to say that the other party’s initial claim was false. And this does seem to be what happens” (1988: 160, 162-163). In other words, Price holds that ‘A’s claim that S is false’ is taken by ordinary speakers to involve an evaluation that is missing from ‘A claimed that S, but not-S’. Questions about the ordinary use of ‘false’ will not matter for my purposes.return to text

    9. MacFarlane seeks to explain a sense of ‘disagreement’ in which parties to evaporative disputes do disagree, both before and after the dispute evaporates (2007; 2014: ch. 6).return to text

    10. In the case of future contingents, it takes very artificial scenarios to yield cases of evaporative dispute. By contrast, it’s easier to describe something like frictionless denial. In 2000, Alan believes it isn’t settled whether an African-American will be elected U.S. President before 2020. For this reason, he rejects, though he doesn’t deny, the proposition that there is an African-American U.S. President prior to 2020 (he would also reject its negation). In 2010, Beth learns of Alan’s speech act. She responds by asserting the very proposition she takes him to have rejected, yet insists that there was nothing mistaken about his rejecting it.return to text

    11. For Price’s rejection of nonfactualist expressivism about his target discourses, see Price (1988: 166); that approach is discussed at length in MacFarlane (2014). For a defense of nonfactualist expressivism about epistemic modal discourse, see Yalcin (2011).return to text

    12. For Price’s rejection of an “indexical treatment” of his target discourses, see Price (1988: 167-169); that approach is discussed at length in MacFarlane (2014). For defenses of indexical contextualism about epistemic modal discourse and discourse about matters of taste, see e.g. Schaffer (2011) and Cappelen and Hawthorne (2009). MacFarlane uses ‘contextualism’ more broadly, to encompass also theories that invoke context-relative truth but fail to attribute assessment sensitivity. His term for such theories is ‘nonindexical contextualism’ (2009).return to text

    13. His relativism thus contrasts with the view Wright (2008: 158) calls “proto- or ground-level relativism,” whose “characteristic expression in ordinary discourse” is “There is no such thing as simply being Φ.”return to text

    14. When MacFarlane invokes possible worlds, he describes them using non-perspectival vocabulary. Yet there seems no reason why he couldn’t allow it to be a feature of some worlds that, in those worlds, fish sticks are tasty. (He could add that these are worlds in which the “objective” features of fish sticks differ from those in the actual world.) Likewise, he could allow that it’s a feature of the actual world that, in this world, the guinea pig might not be a rodent. Since he affirms the “(alethic) modal rigidity of epistemic modals” (2011a: 169-170; 2014: 271n27), he would add that this state of affairs obtains in all possible worlds.return to text

    15. Cappelen and Hawthorne (2011: 460-461) see this question as pointing to an “internal tension” or “instability” within MacFarlane’s relativism.return to text

    16. Cf. Kukla and Lance (2009: 82): “We propose to try to understand objectivity . . . by beginning with a pragmatic story about the nature of objective and subjective claims rather than with a metaphysical story about the nature of the referents of those claims.” This theme also runs through Price (1988) and Brandom (1994).return to text

    17. Price concludes that there is “no sharp distinction between fact-stating and non-fact-stating uses of language” (1988: 201; cf. 2011: 18). Unlike Price, I won’t explain the different degrees of “factuality” in terms of differences in how the truth predicate is employed.return to text

    18. If the only relevant parameters are invariant between the context of use and contexts in which the assertion is assessed for retraction, (RAS) and (RUS) become equivalent and the distinction between use and assessment sensitivity vanishes. Typically, the world of use would be such an invariant. (However, MacFarlane’s analysis of talk about a contingent future rests on denying that there need be a unique world of use. See 2014: 208.)return to text

    19. More precisely, as a referee stressed, MacFarlane writes ‘true as used at c and assessed from c′’ in place of ‘true relative to 〈c, c′〉’ (2014: 108). This makes no substantive difference, since he only appeals to doubly-relativized propositional truth in principles, such as (RUAS), where c and c′ are already specified as contexts of use and assessment. In my own formulation, by contrast, the basic notion is truth relative to a context c, where that context figures in (A) as a context of assertion and in (RAS) as a context of assessment.return to text

    20. The fact that (RUAS) uses a doubly relativized propositional truth predicate shouldn’t be confused with the fact that MacFarlane uses a doubly-relativized sentential truth-predicate. That’s consistent with singly-relativized propositional truth: sentence S is true with respect to 〈c, c′〉 provided the proposition expressed by S at c is true relative to c′.return to text

    21. While the notion of assertoric commitment, central to MacFarlane (2005b; 2007), isn’t officially employed in MacFarlane (2014), it figures here at a crucial point.return to text

    22. As explained in note 19, the basic notion of relative truth I am using is just that of a proposition’s being true relative to a context. It’s only when we state principles concerning the practice of assertion that we must specify whether the context in question is a “context of use,” as in (A) and (RUS), or a subsequent “context of assessment,” as in (RAS) or (E1RT).return to text

    23. Presumably, the point of his retraction is not just to prevent others from mistakenly inferring (on the basis of his past endorsement) that he presently takes fish sticks to be tasty. That could be accomplished by a simple denial that fish sticks are tasty, with no retraction required.return to text

    24. Indeed, the dilemma can be avoided by replacing ‘whose agent is also A’ with ‘relevantly similar to one which A occupies at the time of that context’ in both (E1RT′) and (E1RT″).return to text

    25. He considers, but does not in the end recognize, a “rejection rule” that would give truth relative to contexts of interpersonal assessment a role in the account of assertoric practice (2014: 111-112).return to text

    26. Taking the interpersonal dimension into account also diminishes the difference in the cognitive burden imposed by the two options MacFarlane is comparing. Suppose MacFarlane is right that one needn’t keep track of one’s own prior perspectives in determining whether or not to retract one’s assertions. Still, there remains a further cognitive task. This is to recognize whether or not, in denying some proposition, one counts as challenging a past assertion by another speaker. On the view I will be proposing, being in a position to make such determinations will require keeping track of others’ prior perspectives.return to text

    27. In particular, I won’t mention a status of being committed to a proposition —explaining what Brandom means by this requires laying out more of his theory than I wish to endorse.return to text

    28. Schaffer (2011) contrasts two varieties of “perspectivalism” in semantics: “meaning perspectivalism” (here which proposition a sentence expresses depends on the speaker’s perspective) and “truth perspectivalism” (here perspective figures in how the truth of the expressed proposition is evaluated). As Schaffer notes, MacFarlane’s relativism is a version of truth perspectivalism. My suggestion is that MacFarlane’s purposes can be served more directly by a third—now pragmatic rather than semantic—brand of perspectivalism.return to text

    29. Egan (2007: 20) defends the different but compatible view that propositions with context-relative truth conditions are “felicitously asserted” only in conversations where there is a presupposition in effect that asserter and audience share a relevantly similar context. (He qualifies this claim in Egan 2010: 273-275.) Huw Price has urged in discussion that the default status of such a presupposition is important for explaining how evaporative disputes can arise.return to text

    30. Kukla and Lance (2009: 201; also 21, 83) argue that “all declarative truth-claims must take the universal community of all rational agents as their domain, for otherwise they cannot be hooking on to an objective world.” On the present proposal, some declarative truth-claims count as “subjective” precisely insofar as they violate this condition. In this respect, the proposal resembles that of Rovane (2011). She construes relativism as a view on which “there may be persons whom we cannot teach and from whom we cannot learn—not on all matters, but on those matters concerning which our beliefs are normatively insulated from theirs” (2011: 38). Where Kukla and Lance would object that such persons “cannot be hooking on to an objective world,” Rovane would agree: according to the relativism she discusses, “there are many worlds rather than one” (2011: 36). The present proposal involves no talk of worlds.return to text

    31. Whether an assertion succeeds in transmitting entitlement to an audience member at a later time depends on whether the entitlement the asserter has at the time of assertion still counts as entitlement at the later time, notwithstanding any change of perspective. In order for entitlement to have been transmitted, it’s not enough that the asserter would be entitled to assert the proposition at the later context. Had young MacFarlane asserted, with deceptive intent, that fish sticks aren’t tasty, he wouldn’t count as having thereby entitled those sharing his present tastes to assert that proposition.return to text

    32. Thanks to Theodora Achourioti for pressing me to clarify this.return to text

    33. As MacFarlane notes, “ [i]t is characteristic of [views that attribute use sensitivity] that monadic truth ascriptions can come apart from . . . retraction obligations” (2014: 224, cf. 191).return to text

    34. Recanati, appealing to Egan (2007), proposes in effect that it’s enough that (ER) is ruled out in “shared-situation cases” (2007: 88; 2008: 57).return to text

    35. The idea of a locus of responsibility encompassing both the past context of assertion and the present context of possible vindication-or-retraction also helps explain a difference between mitigated relativism and MacFarlane’s version of assessment sensitivity. Suppose one’s assertion of a proposition is challenged by a hearer whose perspective is relevantly similar to one’s own perspective at the time of one’s assertion, but one can’t give reasons that would vindicate entitlement to assert the proposition on the part of an occupant of that perspective (who lacks a defeasor status). According to mitigated relativism, retraction is in order, even if one can vindicate one’s own entitlement to assert the proposition at the time of the challenge. In asserting, one makes oneself accountable to challengers whose perspective resembles one’s own at the time of the challenge, but remains accountable to challengers whose perspective resembles one’s own at the time of assertion. (Thanks to Johannes Marti for calling my attention to this.)return to text

    36. Still, I wish to allow that it might be useful for other purposes to recognize a sense of ‘proposition’ in which there can be, say, a time-neutral proposition which the same speaker may accept and reject at different time without normative deficiency. Such “propositions” might be required for certain kinds of psychological explanation. But this won’t be the sense of ‘proposition’ that will figure in the pragmatics of assertion. Cf. Richard (1981) and Recanati (2007; 2008).return to text

    37. For similar suggestions, see Price (1988: chs. 6.4 and 7; 1998; 2003). In the displayed passage, MacFarlane is contrasting expressions like ‘tasty’ and ‘fun’ with indexical expressions like ‘pleasing to my taste buds’. He is not here addressing the expressivist alternative, nor explaining why the use of perspective-relevant propositions displays assessment sensitivity rather than use sensitivity.return to text

    38. I owe this observation to John MacFarlane.return to text

    39. MacFarlane already calls attention to the gap in MacFarlane (2003) and elaborates on it in MacFarlane (2014: chs. 3-4); for non-relativists who do the same, see esp. Ninan (2010), Rabern (2012), Stojanovic (2012) and Yalcin (2011). These discussions draw heavily on Lewis (1980).return to text

    40. Though I focus on truth-conditional semantics here, my reply also pertains to a parallel objection on behalf of those who approach some of our target vocabularies using dynamic semantics (e.g. Willer 2013). Such analyses, which associate sentences with “context change potentials,” might appear closer in spirit to the pragmatist approach pursued here. On the present approach, however, understanding the subjective nature of our target vocabularies doesn’t require any compositional account of what effect uptake of an assertion of the sentence will have on a hearer.return to text

    41. As explained in Section 4, this is a slight simplification: MacFarlane actually appeals to S’s expressing a proposition at c that is true relative to the pair of contexts 〈c, c′〉.return to text

    42. This is a running theme of MacFarlane (2014); see also MacFarlane (2005b; 2007; 2009; 2011).return to text

    43. Significantly, MacFarlane himself doesn’t use his compositional semantics to explain what it is for a sentence to express a given proposition at a context. Instead, he merely formulates constraints on the assignment of propositions by defining the intension of the proposition expressed by S at c in terms of the compositional semantics. See e.g. MacFarlane (2014: 152).return to text

    44. In the case of epistemic possibility, for example, MacFarlane explains how his compositional semantics yields the result that ‘For all I know, it is possible that they won’ is logically equivalent to ‘For all I know, they won’ (2014: 266). Building on Yalcin (2007), MacFarlane also uses his semantics to define a notion of consequence on which ‘If they didn’t score, then they didn’t win’ implies ‘If they won, they scored’, yet ‘If they didn’t score, then it’s not possible that they won’ fails to imply ‘If it’s possible that they won, then they scored’ (2014: 252-253, 271).return to text