Reply to Lasersohn, MacFarlane, and Richard

1 Précis of Relativism and Monadic Truth

The beginning of the twenty-first century saw something of a comeback for relativism within analytical philosophy. Relativism and Monadic Truth has three main goals. First, we wished to clarify what we take to be the key moving parts in the intellectual machinations of self-described relativists. Second, we aimed to expose fundamental flaws in those argumentative strategies that drive the pro-relativist movement and precursors from which they draw inspiration. Third, we hoped that our polemic would serve as an indirect defence of a traditional and natural picture concerning truth. According to this picture, what we call ‘Simplicity’, the fundamental structure of semantic reality is best revealed by construing truth as a simple monadic property of propositions that in turn serve as the objects of belief, assertion, meaning and agreement.

Our project was not a straightforward one. So-called relativists are not uniform in their key ideology, are often sloppy, casual, obscure or confused in their self-characterization, and differ in their argumentative emphasis among themselves and over time, thereby presenting a target that is both amorphous and shifty. This is an area where parties will frequently claim not to understand each other and where certain parties will sometimes accuse others of failing to make any sense at all. In such a situation any effort to impose order will inevitably strike some parties as tendentious and unfair. That said, we felt that we had enough of a grip on the relativist movement to recognize it as a degenerating research program, and enough of a grip on the resources available to Simplicity to see it as largely unscathed.

Relativism is often depicted as continuous with a larger tradition in semantics that parameterizes the meanings of utterances and the objects of belief. By the lights of such parameterizations, the relevant contents are not true or false simpliciter but instead true or false relative to this or that parameter value. Such parameterizations are encouraged by talk of propositions being true relative to worlds and times. But insofar as these latter ways of talking are accompanied by a rejection of the claim that propositions are straightforwardly true or false, they already constitute a misstep. The ideology of truth at a world or time is best seen as part of a theoretical vocabulary that can sit alongside, and which is indeed less fundamental than, the ideology of monadic truth and falsehood. One sometimes hears relativism presented in a way that makes many of its core commitments seem somewhat innocent: ‘We have already learned that truth is relative to worlds and times. There is nothing very drastic about adding a few more parameters’. Our reply is that we have learned no such thing. These ways of talking should never have been seen as challenges to monadic truth or falsity, and so cannot pave the way for relativism.

As we see things, there are three main ideas that drive relativism—notwithstanding the dizzying array of articulations that we find in the literature.

• First, relativists proliferate parameters. They draw inspiration from those who reject simple truth and falsity in favour of parameterization of truth to worlds and times and then add a range of further putative parameters. (Those that replace simple truth or falsity with truth relative to times and worlds, and hence insist that the content of a given assertion is neutral concerning times and worlds still concede that particular times and worlds might be relevant to the felicity of a particular assertion. Similarly, for these extra parameters.)

• Second, relativists add a disquotational truth predicate to their expressive resources. Roughly, just as ‘S’ will be true relative to certain parameter values and false relative to others, so ‘‘S’ is true’ will be parameterized in exactly the same way.

• Third, they embrace non-parameterization of belief and meaning ascriptions. Roughly speaking, various parameters for a sentence become redundant when that sentence is embedded in a meaning or belief report. Thus, for example, while ‘Backpacks are cool’ may be true or false relative to a standard of taste ‘Certain philosophers believe that backpacks are cool’ will not exhibit such relativity.

In Chapter One we showed how, with these pieces in place, we can illuminate various relativist ideas including ‘no fault disagreement’ and ‘assessment sensitivity’.

Chapter two is devoted to a discussion of various diagnostics for shared content. Assumptions about content sharing are central to relativism and its precursors. For example, relativists about taste often assume that what I believe when I accept ‘Kebabs are tasty’ is what someone disbelieves then they reject ‘Kebabs are tasty’. By way of ground clearing we thought it helpful to evaluate various diagnostics for shared content. Contrary to what many implicitly or explicitly assume, collective ‘say that’ and ‘believe that’ claims are far inferior to collective ‘agree that’ data. Disagreement data, we argue, has even more evidential force.

Chapter three looks at a style of pro-parameterization argument made prominent in Kaplan. Related ideas are promoted by Lewis and are frequently paraded in current relativist literature. We found that the argument was inexplicit and underdeveloped in various spots. We tried to be explicit about the relevant assumptions. We found all of them to be problematic in certain of their applications and were sceptical about whether any compelling operator arguments were left standing.

Chapter four is devoted to a case study: predicates of personal taste. This is one of the best cases for relativism, in part because the various shared content diagnostics tell prima facie in its favour. Our discussion combines a number of themes. Relativists typically fail to pay attention to crucial linguistic details, neglecting noise introduced by generic constructions, neglecting important distinctions among and between various uses of ‘fun’, ‘disgusting’ and other predicates of taste, neglecting cases that are very problematic for the view, notably ones where superficial contradiction is not accompanied by a judgment of disagreement. Meanwhile, relativists typically argue against a contextualist strawman, failing to pay attention to the array of resources available to sophisticated contextualists. Further, relativists typically fail to notice that ordinary people are often inclined to a naïve realism and that disagreement verdicts seem to wax and wane precisely to the extent that such naivety gains and loses its hold. As well as outlining and defending a sophisticated contextualism, we outline a battery of arguments against relativism. One trouble spot worth noting is that the relativist will have far more difficulty than is usually recognized in distinguishing her own position from that of a naïve chauvinistic realist.

Personal taste is one of the favoured hunting grounds for relativism. Even here, the advantages of relativism prove largely chimeral. The pattern of breakdown is likely to be replicated in other areas—epistemic modals, conditionals, future contingents, and so on—where relativism has been promoted as a viable option. We expect to see many more relativist proto-theories advanced in the near future. We doubt that very much will come of them.

2 Reply to Mark Richard

Some of Richard’s initial remarks strike a curiously critical tone. We noted that collective agreement tests provide only limited support to claims of shared content. Acknowledging this, Richard then argues that the relevant limitations do not carry over to collective disagreement reports. We agree. Indeed the points that he makes in this connection are ones that we ourselves make in the book (see e.g. Cappelen and Hawthorne 2009, pp. 62–63-henceforth RMT).

At the end of his Section one, Richard oversimplifies the dialectical situation considerably. He says that the upshot of our discussion in Chapter two is that there is a very strong argument for relativistic content. The situation in all the core cases is much more complicated. Take predicates of personal taste—‘fun’ for example. There are indeed cases where standard contextualism predicts disparate content but where the disagreement diagnostic provides prima facie evidence for the claim of shared content promoted by a standard relativist view. But there are plenty of other cases where the predictions yielded by relativistic semantics do not sit well at all with the deliverances of that diagnostic (see e.g. RMT, p. 109). There is, thus, prima facie troubling data that each side has to explain away. Richard, like many relativists, is overly encouraged by a limited diet of examples.

Let us turn to the nature of disagreement, the topic of Richard’s Section two. As Richard anticipates we are sympathetic to a pretty simple view of disagreement: We disagree about a proposition if one of us believes it and one of us believes the negation.¹ Some relativists—notably MacFarlane—have thought that there are considerations that show that such a view must be wrong, even leaving the general issue of relativism to the side. One paradigmatic argument of this sort is to the effect that if someone in world w ₁ believes p and someone in world w ₂ believes not-p, they are not thereby disagreeing. Richard displays some sympathy to this line of argument. He shouldn’t. As we show in Chapter two, it is a terrible argument.

As we see it there is no good general argument against the simple view of disagreement. Suppose, however, one is a committed relativist, endorsing parameterized content for a range of subject matters while also endorsing non-relativity of belief reports (whereby those parameters become irrelevant when the relevant contents are embedded in belief reports). Is the simple view of disagreement a promising option in this setting? Richard’s thought—articulated also by MacFarlane and others—is that the analogue of the simple view ought not be endorsed by the lover of parameterization. To see why, consider the analogue of the simple view applied to a parameterized approach to tense and time. Someone who parameterizes for time (in the way that Kaplan does) holds the view that someone who believes what is expressed by ‘There is a cat on the mat’ on Tuesday believes the very same content that is rejected by someone who on Wednesday denies what is expressed by ‘There is a cat on the mat’. The fact that it would be silly to say they disagree is taken by Richard to show that the simple view of disagreement is not suitable in this context. We think that the dismissal of the simple approach, even in this setting, is a little quick, since it fails to be careful about tense in the disagreement report: The simple approach does not say that if x believes p and y used to believe not-p then x and y disagree. The best version of the simple view is one that ‘takes tense seriously’.

Given Richard’s rejection of the simple view, he is faced with a question we do not have to confront: ‘What does it take for disagreement beyond some person believing a content and someone else denying that very content?’ For what it’s worth, we think that it is a cost of any view that it is saddled with a distinction between disagreement on the one hand, and the relevant pattern of belief and denial on the other. Suppose one is a relativist who goes in for non-relativity of belief reports, a disquotational truth predicate and parameterized content but who, with Richard, thinks that some additional criteria need to be met for disagreement judgments to be correct beyond the relevant pattern of belief and denial. Then it is a short step to endorsing speeches of the following sort: ‘She believes that what I believe is false, but she doesn’t disagree with me’. But that speech sounds terrible. (Of course in the case of parameterized temporal contents one can save oneself from such speeches by being careful about tense, but, as we pointed out earlier, for that very reason such contents cannot be the grounds on which one breaks with a simple account of disagreement in the first place.)

Be that as it may, let us look at Richard’s solution. His thought is that while the content of a particular belief or assertion can be evaluated relative to all sorts of values along the relevant relativistic parameters, there will be particular values of those parameters at which the belief or assertion are ‘aimed’. In our lingo, crafted on behalf of the relativist: there are particular parameter values that are operative on a particular occasion of speech. In the Tuesday/Wednesday case, the content asserted is the content rejected, but a particular time on Tuesday was the operative time for the assertion, while a time on Wednesday was the operative time for the rejection. Now Richard gives a particular gloss on what it takes for a parameter value—or set of parameter values—to be operative: It is operative if it is appropriate to evaluate a belief or an assertoric performance relative to that parameter. Supposing there is just a single relevant parameterization, disagreement between x and y concerning some content is taken to require not only that one believes the content and the other denies it, but moreover, that there is some single parameter value along that parameterization that is operative for both the belief and its denial.

Take the case of ‘fun’. Whether we are contextualist or relativist about ‘The party will be fun’ we can all agree that there are cases where it is inappropriate to evaluate a speech by one’s own standards for fun. Suppose Jim has a terrible toothache and upon waking says ‘Today won’t be fun. I can’t see the dentist until tomorrow’. Meanwhile Jim’s friend, who is having a birthday party, say ‘Today will be fun. I’ll get a big cake’. Whether we are contextualists or relativists we will be uncomfortable using the friend’s speech as a basis for claiming that they have a real disagreement. This is a case where Richard wants to say that there is incompatibility without disagreement. But what Richard doesn’t properly notice is that in such cases, intuitions of incompatibility disappear along with intuitions of disagreement. After all it would be strange for Jim’s friend to say ‘Jim believes that what I believe is false’ or ‘Jim contradicted me’ or even to say ‘That’s false!’ when he overhears Jim’s speech in a setting where he is fully apprised of the relevant background facts. As we see it there is the following prima facie puzzle: While Jim’s friend will not have an intuition of contradiction, he will be willing to say ‘Jim believes that today won’t be fun but I do’. But that pattern of data is explained by contextualism in combination with the relevant ground clearing about how beliefs reports work. A ‘contradiction without disagreement’ approach, by contrast, does not ring true at all.

One also wonders whether Richard’s account of disagreement will be especially welcome to many contemporary relativists. Take MacFarlane’s ‘fish fingers’ case. A five year old child says ‘Fish fingers are tasty’. A grown up says ‘Fish fingers are not tasty’. This is supposed to be a paradigm of disagreement. According to Richard’s account, this disagreement judgment is correct only if it is appropriate to evaluate the child’s speech by the standards appropriate to grown ups. But how can that be appropriate? Insisting that adult and not childish tastes are the operative standards for the child’s speech smacks of the very sort of chauvinism that relativists are trying to distance themselves from. On the other hand, conceding that the standards are disparate and hence that there is no real disagreement would put Richard in the same boat as the contextualist on the matter of disagreement and thus erase one of the purported advantages of relativism, namely its ability to take initially appealing judgments of disagreement at face value.

This is pertinent to Richard’s discussion of disgust. Consider a case we discuss in the book. Certain tribes find kissing disgusting. We don’t. An ordinary episode of kissing occurs between Stephanie and Alex. Some tribe member says ‘That is disgusting’. We say ‘There is nothing disgusting about that’. Relativists typically want to say that we disagree. Our general view is that while there is an initial temptation to a disagreement verdict, reflection on the ways that disgust reactions are subject to various vicissitudes of culture and physiology—what we called Pyrrhonian therapy—tend to make disagreements judgments evaporate, and this in turn helps to pave the way for a contextualist treatment of the matter. Now what is Richard’s considered view supposed to be? If he thinks that it is appropriate to evaluate the tribesperson by the standards that are operative for us he is every bit as naïve and chauvinistic as the crudest village realist. If he concedes that it is not, then he will be forced to concede that, despite initial temptations to the contrary, there is no real disagreement between the kissing couple and the disgusted tribal onlookers. But that removes much of the dialectical machinery that relativists typically leverage against the contextualist. Indeed, speaking more abstractly, is there even such a thing as no fault disagreement—the alleged centrepiece in many pro-relativist monologues—on Richard’s view? Where the appropriate standards for disgust judgments are different, there is no disagreement at all. A fortiori there fails to be no fault disagreement. Meanwhile if Bill says x is disgusting and Ben says x is not, and by the standards appropriate to both speeches, one is wrong, then in what sense is the disagreement a no fault one? (Of course one might point out that one of them was in no position to know that their speech failed to meet the appropriate standards. But that is hardly the sense of ‘no fault’ that is at work in relativist apologetics.) In short Richard’s own version of relativism is not going to be one well served by any ‘argument from disagreement’.

Richard ends with a discussion of belief reports. He never tells us what, in his view, it takes to believe a relativistic content though we detect some sympathy with the non-relativity of belief, the idea that belief reports tend not to inherit the relativity of the contents they embed. His main focus is on driving a wedge between the felicitousness and correctness of such belief reports as ‘Frank believes that there is something delicious in the fridge’, said to a talking vulture on an occasion where the fridge has nothing but decomposing flesh in it. He denies that the belief ascription is correct and offers some pragmatic considerations to explain its felicity. It seems that his picture—at a rough first pass—is that for Frank to believe that there is something delicious in the fridge it is necessary and sufficient that Frank believe that there is something delicious by the standards appropriate to Frank in the fridge. The relevant piece of pragmatics runs like this:

• First point: Sometimes we felicitously say ‘X believes P’ when X does not believe the content expressed by ‘P’ but instead believes some Q which together with some conversational presuppositions entail P. If in such a scenario P is conversationally relevant ‘X believes P’ may well be felicitous even though false.

• Second point: Sometimes we conversationally presuppose propositions we don’t believe to be true. The felicity despite falsity of ‘Frank believes that there is something delicious in the fridge’ is then explained by combining these points, the relevant conversational presupposition being that whatever the vulture finds delicious is delicious.

Richard’s illustrations of the first point are not always compelling. One of his key examples is:

Cassius thinks your friends are at the senate

said in a setting where it is the speaker and hearer but not Cassius who think that the relevant people are friends. Richard is aware that some might think of this as a scope phenomenon—scope out ‘your friends’ and the ascription won’t semantically require that Cassius thinks of anyone as a friend of the addressee. Anticipating this kind of move Richard says:

“Those who think the example’s point can be avoided by reading the ascription de re should construct variants in which ‘friend’ plays a predicative role while not occurring within an expression that could be interpreted as a singular term.” (Richard, this vol.)

First note that this is not quite to the point since the wide scope manoeuvre does not require treating the expression in which ‘friend’ occurs as a singular term. All that is required is that a determiner phrase takes wide scope position with respect to an attitude verb and is then used as a vehicle for asserting or denying some type of singular thought (consider for example, the wide scope construal of ‘Cassius thinks no friend of yours is at the Senate’.)

Second, we can do quite a bit to by pass wide scope manoeuvres. Suppose we think that all and only the Romans with red hair are happy. Suppose Cassius is oblivious to this but thinks Caius is a Roman with red hair. Consider now an utterance by us of:

Cassius thinks Caius is happy.

Richard’s story seems to predict that it ought to be fairly easy to construct settings in which this ascription is felicitous under the assumptions outlined. Is it?

We shall end with three further considerations. First, the Richard strategy could be set in motion all over the place. Suppose Frank lives in San Francisco and plans to go to a bar local to where he lives. We say of Jim, who lives in Oxford, ‘Jim believes Frank is going to go to a local bar’. One option is to adopt a contextualist semantics where this is straightforwardly true on the natural completion. An alternative strategy would be to offer a strange semantics according to which this is true only if Jim believes Frank is going to go to a bar local to Jim, and that the felicity of the report is explained by a conversational presupposition (that we know to be false) that whatever is local to Frank is local to Jim. This second strategy, while a move in logical space, is altogether unappetizing. We can’t see what makes it so much more appetizing in the cases where Richard endorses it.

Second, if Richard’s view is something like (1):

• (1) “x thinks y is fun” is true iff x thinks y is fun for x,

then we will have to adopt pragmatic special pleading in a range of cases where the operative standard is a group that includes but is not restricted to x. But it strikes us as prima facie very odd to appeal to this complex pragmatic story in a case where Frank rejects ‘Whichever movie we see will be fun’ in a setting where the operative standard is himself and his partner (on the grounds that no movie among the options will be fun for both of them). Suppose instead that Richard doesn’t want to endorse (1). Then, pending further guidance, we have no idea at all about when we are supposed to regard belief reports as false but felicitous.

Third, in many ordinary cases where something is felicitous but false there are natural ways of indicating this. We can felicitously say that the swimming pool is a mile long, but then we can follow this up with ‘Of course it is not literally a mile long’. Where such indications are not readily available, ‘false but felicitous’ moves will strike us as prima facie costly. Richard’s special pleading seems to be of the latter variety.

3 Reply to Peter Lasersohn

Lasersohn discusses a type of case that we raise as problematic for a simple relativistic approach to predicates of personal taste: A waiter says ‘The party was not fun’. The host says ‘The party was fun’. We do not think they disagree. But, prima facie, it seems that a relativistic analysis will predict disagreement.

We were fully explicit that such cases could be handled by a mixed view that said that the relevant sentences have relativistic contents only some of the time. Moreover, this case was only a small piece of our overall argumentative arsenal. Nevertheless we did think it illuminating since a shift to a mixed view carries its own costs. Lasersohn’s main aim is to provide an answer to the case on behalf of someone who pushes a uniformly relativist account.

One relativist tactic here is to complicate the account of disagreement, distinguishing disagreement from inconsistency (see Richard contribution to this symposium). Lasersohn pushes a different strategy, which amplifies and refines on one that we briefly pursued in a footnote. The key thought is that ‘predication in natural language is often done on the basis of only certain parts or aspect of the object to which the predicate is applied.’ (Lasersohn, this vol.) The simplest version of such a manoeuvre (the one we discussed in the relevant footnote) is one that claims that ‘The party’ has different referents in the mouths of each speaker, the one occurrence referring to one part of the complex event, the other referring to a different part. Lasersohn is rightly wary of this. After all there is quite a bit of linguistic evidence against it. The host can say ‘It was fun’, anaphorically picking up on the use of ‘The party’ by the waiter. This is hard to make sense of on an analysis that posits disparate referents. Also, it is fine to claim that there is an event that the host thought was fun, but that the waiter did not think was fun. This does not square with the multiple referent approach.

Lasersohn intends instead that the ‘contextual effect be attributed to the predicate’ (Lasersohn, this vol.) Here is one toy model. Suppose that the deep structure of ‘The party was fun’ is something like ‘The party was fun in way w’. The ‘way’ slot is treated in line with standard contextualist semantics. Yet one can retain relativism by insisting that even when the ‘way’ slot is supplied the resulting content is one that can only be evaluated for truth relative to a standards parameter. On this picture, two interlocutors that say ‘The party was fun’ and ‘The party was not fun’ disagree only if they supply the same argument for the ‘way’ slot’. And whether or not they disagree, the contents expressed get a parameterized treatment, being incapable of being true or false simpliciter. With such an analysis in hand we can be relativist across the board while denying that there is disagreement in the waiter/host case.

A strategy of this sort is prima facie promising for the relativist, but we wish to flag two kinds of worries that the relativist is likely to run up against.

The first worry is that an analysis of this sort may still make some of the wrong predictions when it comes to anaphoric data. In this connection we wish to point to a contrast. Suppose x says ‘The watermelon is green not red’ in the midst of a discussion about its surface, and y says ‘The watermelon is not green but red’ in the midst of a discussion about its interior. In this connection the speech ‘x said/thinks that the watermelon is green but y said that it wasn’t’ sounds pretty bad. By contrast, to our ears the speech ‘The host said that the party was fun but the waiter said that it wasn’t’ sounds far better. Now someone who thought the structure of the initial speech was ‘The host said that the party was fun for him’ can perfectly well explain the felicity of the latter on the model of sloppy identity (cf. ‘x thinks that his parents were nice. But y doesn’t think so.’) But if one were to try to assimilate what is going on in the host/waiter case to the watermelon phenomenon, then it becomes far less clear why the first pair is bad but the second pair much better.

To illustrate our second worry let us turn to fun rollercoaster rides, one of Lasersohn’s favourite examples. Suppose x says ‘The rollercoaster ride was fun’ and y says that ‘The rollercoaster ride was not fun’. In any such case Lasersohn’s refined semantical framework predicts two possible diagnoses. One is that x and y are not disagreeing because they are talking about different aspects of the ride (the ‘way’ is different)—there are two contents being expressed (neither of which is true or false simpliciter). Another is that they are disagreeing about a single content (one that is neither true nor false simpliciter). But when are we to apply the first diagnosis and when are we to apply the second? Consider in this connection various versions of the case: (i) y is the paid rollercoaster driver, (ii) y has been riding the rollercoaster all day and is sick of it, (iii) rollercoaster rides make y feel nauseous, and (iv) rollercoaster rides make y too afraid to enjoy him or herself. Suppose x has a paradigmatically fun ride. In which of cases (i) to (iv) do we say ‘different aspects so no shared content’? After all, the contextualist parameter could, in principle, be invoked in defense of a judgment of disparate content in every case. (Most simply, one might say the way relevant to y is how y experienced it but that is not the way encoded in x’s speech.) Without some guidance on the matter there is a risk that the refined relativism will turn unruly.

Lasersohn’s discussion contains hints of partial guidance. He points out that in the host/waiter case the episode does not stop us saying that their tastes are the same. This gives us a heuristic:

If x says a ride is fun and y says it is not fun and x and y have the same tastes (as far as what they find fun), then they are speaking past one another – different resolutions of the contextualist parameter are in play.

This heuristic is rather limited, for three reasons.

• First, it gives us no guidance when x and y have different tastes (the waiter might hate parties whether or not he is a waiter at them). After all, even if x and y have different tastes it may on this occasion be differing resolutions of the contextualist parameter rather than real disagreement that is in play.

• Second, it may force judgments of ‘no real disagreement’ in cases where the relativist typically has intuitions of disagreement. Suppose y does not enjoy the ride because he has had a rough night and is in a bad mood. He says ‘The ride was not fun’ (not because he has different tastes to x, but because he’s exhausted after a long night.) This is the sort of case where relativists typically intuit disagreement. But the new heuristic indicates that such judgments are not to be trusted. This may in turn have a destabilizing effect on the programme.

• Third, judgments of ‘They have the same tastes’ are themselves somewhat context dependent. Suppose x finds a certain white wine tasty on account of having a habit of eating dates during the day, but y does not. One can imagine a conversation where an utterance of (1) is felicitous:

• (1) ‘x and y have different tastes owning to the fact that x eats dates’.

However, one can also imagine one a conversation where (2) is felicitous:

• (2) ‘x and y have the same tastes. The only thing making a difference is the fact that x ate dates before drinking the wine.’

Or suppose x has fun doing activity A because A produces phenomenology P in x. y does not enjoy A, A does not produce P in y, but y does enjoy activities that produce P in y. The speeches ‘x and y have different tastes—y doesn’t care for A’ and ‘x and y have the same tastes—they are both looking for activities which produce P’ are both felicitous in suitable contexts.

The relativist might think that there is a much more straightforward heuristic for detecting sameness of content in the relevant cases—just ask oneself whether there is a real disagreement. If one judges that there is, that is powerful evidence of common content, if one judges that there isn’t this is powerful evidence that the contextualist parameter is being resolved differently. This suggestion connects to larger issues in Chapter four of our book. There we argued that inclinations to disagreement verdicts are far more flimsy than relativists acknowledge. Matters are frequently distorted by a focus on generic claims like ‘Rollercoaster rides are fun’.² The focus, we argue, should be on fun verdicts about particular events. We hear x get off a ride and say to herself ‘That was fun’. We hear y get off a ride and say to herself, ‘That wasn’t fun’. Judgments of disagreement and contradictoriness are uniformly faltering and flimsy in a case like that. We are, thus, very sceptical indeed as to whether such judgments can lend adequate discipline to the kind of semantics that Lasersohn recommends.

4 Reply to John MacFarlane

Predictably, John MacFarlane claims that our discussion does not get to the heart of relativism and that his own framework does. We think he both underestimates the resources at work in our approach and overestimates the clarity and cogency of his. In what follows we shall briefly speak to both of these themes.

We say that relativists reject the idea of truth as a monadic property. MacFarlane disagrees—he doesn’t think the typical relativist rejects the idea of monadic truth. MacFarlane’s discussion of these issues confuses a monadic property with a monadic predicate. To see what we have in mind, let’s begin with the notion of a monadic property. Our grip on that notion is more or less in line with the tradition. Roughly speaking, a monadic property either does or doesn’t stand in the instantiation relation to a given object. For many purposes this can be represented as a particular function from worlds to extensions, and in a setting where intensional issues are irrelevant it may serve our purposes well enough to represent it as a set of objects. Abstract objects with a more complex profile are not what we have in mind by ‘monadic property’. In particular an abstract object with parameterized instantiation—something that is instantiated by Jones relative to one index on a parameter but not relative to another—is not what we had in mind by ‘monadic property’. In saying this, we are not acting as the terminology police. We are merely alluding to a tradition of thinking and talking about properties that we are immersed in and which is quite familiar. Aware that there might be some flexibility in how authors use ‘monadic property’, we tried to forestall misinterpretation with remarks such as the following:

One final point of clarification: to get our intentions right, think here of ‘instantiates’, as it figures in T1 [T1 = There are propositions and they instantiate the fundamental monadic properties of truth simpliciter and falsity simpliciter] as a simple binary relation between an object and a property. Suppose that one held that instantiation is a three-place relation between an object, a property, and a time, and one said that the property of being true was instantiated by a certain proposition at noon but not at 1 pm. That would not, on the intended construal, square with T1. (RMT, note 5, p. 3)

The standard relativist takes the content of an ordinary sentence to be an object that is true or false relative to parameter values. How about predicates? The standard relativist does not take the content of ‘is fun’ to be given by a particular function from worlds to extensions. Rather the content will deliver a function from worlds to extensions only relative to a parameter value. Relatedly, the content—even leaving intensional issues to one side—is not naturally thought of as a set, since then a proposition of the form ‘A is fun’ would be true or false simpliciter (assuming the standard connection between truth and true of.) Just as a sentence isn’t true simpliciter, so a predicate is not true of an object simpliciter. Predicates deliver extensions only relative to parameterizations. Now what goes for the predicate ‘is fun’ goes for ‘is true’. For relativists, the content of ‘is true’ is not given by a function from worlds to sets of propositions but instead yields such functions only relative to parameter values (standards of taste for example).³

The upshot is this. The relativist that we presented does indeed allow a syntactically monadic truth predicate that behaves in a disquotational way (roughly, ‘S’ is true relative to a parameter value iff ‘‘S’ is true’ is true relative to that value). But the standard relativist does not think of ‘true’ as expressing a monadic property.

That means that the relativist picture is not friendly to T1, the first component of Simplicity:

T1: There are propositions and they instantiate the fundamental monadic properties of truth simpliciter and falsity simpliciter

Of course, the relativist can pay lip service to T1 provided she is willing to do sufficient violence to the intended meaning of ‘monadic property’, but that is neither here nor there.

Given this, we are not willing to buy into MacFarlane’s way of setting things up. As he sees it, recognition of a monadic property of truth is common ground between relativist and non-relativist. He sees our main thrust as being to the effect that the two views push different orders of explanation between monadic truth and other notions. We did indeed voice a commitment to the fundamentality of monadic truth (more on that later), but we did not think that a commitment to a property of monadic truth was common ground between the relativist and ourselves. As we see things, the standard relativist does not think of himself as associating a monadic property with the truth predicate.

That said, there is an internal tension within relativism that may emerge. Just as the relativist gives herself a disquotational truth predicate, she will also likely give herself a ‘true of’ predicate that will license speeches about ‘true of’ that sound just like the realist ones (on the relevant use, ‘‘F’ is true of a’ is true relative to an index just in case ‘Fa’ is true relative to that index.) She can similarly contrive a use of ‘instantiates’ whose application to her parameterized contents of predicates mimics the realist’s application of ‘instantiates’ to non-parameterized properties (on the relevant use, ‘The content of ‘F’ is instantiated by a’ is true relative to an index just in case ‘Fa’ is true relative to that index). Insofar as the relativist goes along this path systematically enough, she risks depriving herself of any way of denying that ordinary sentences express complete propositions and of denying that ordinary predicates express traditional properties. She will thereby risk losing the ability to set herself apart form her putative opponents. It is only in connection with that instability that we entertained the relativist’s appealing to order of explanation issues as a candidate way of setting herself apart from the realist. But we left it as an ‘open question’ whether such an appeal could bear the explanatory burden placed on it and reminded the reader that ‘this way of proceeding risks conceding that there is nothing in pre-reflective ways of talking that points towards relativism rather than realism’ (RMT, p. 138). MacFarlane’s discussion reinforces our scepticism that appeal to order of explanation issues will enable one to stake out a distinctively relativist position once one has lost the ability to deny that ordinary sentences/predicates express complete propositions/properties. We take that not as evidence that we have misconstrued relativism but rather that we have identified an instability within that position.

A disappointing aspect of MacFarlane’s discussion is the extent to which he misrepresents our view of the relationship between contemporary analytic relativism and the proliferation of parameters of evaluation. He attributes to us the view that ‘assessment sensitivity is trivially implied by proliferation’ (MacFarlane, this vol.). We are quite explicit in our denial of this. We emphasised the point that standard relativists add disquotational truth and the non-relativity of belief and meaning reports to proliferation. And we were fully explicit that the latter moves were not trivially forced on one by the former. (For example, it is prima facie perfectly coherent to say that just as ‘Fiddle playing is fun’ varies in truth relative to standard parameters so ‘John believes that fiddle playing is fun’ varies in the same way.) Our relativist framework did make room for something like the phenomenon of assessment sensitivity and attempted to account for it by a combination of proliferation, disquotational truth, and non-relativity of belief and meaning reports. Indeed, we think it is very illuminating that the combination of those three elements give rise to something much akin to assessment sensitivity. The pattern of argument that MacFarlane criticizes and attributes to us is, by contrast, not recognizable to us. Further, his remark that we argued that there is no need to discuss assessment sensitivity is seriously misleading—we had a whole section on how something like that phenomenon arises within the framework that we describe (the section is hard to miss: it is called “Assessor Sensitivity” and starts of p. 17 of RMT.)

Now perhaps the assessment sensitivity that we discuss is still importantly different from the kind of sensitivity that MacFarlane thinks is crucial. Or perhaps, by certain relativists’ lights, we offer too shallow an explanation of that phenomenon. This is difficult terrain and we are open to constructive engagement along either of these dimensions.

MacFarlane’s own preferred strategy is to lean on a rather different ideology to characterize relativism. He says the relativist should take ‘the accuracy of some assertions or beliefs to vary with the context from which they are assessed’ (MacFarlane, this vol.) The philosophical lynchpin here is, thus, the concept of accuracy.⁴ Why did we not go along with his preference to characterize the key debates in those terms? We do not think it particularly helpful to frame debates using expressions that have the ring of ordinary concepts but which are in fact being used as terms of theoretical art. Indeed this commits two sins simultaneously—it fails to use the familiar to gain a conceptual anchor while creating an illusion that one is doing so.

Take the concept of accuracy. There is an ordinary predicate ‘accurate’ that we apply to particular belief episodes and assertions. But that predicate is straightforwardly tied to our one place notion of truth by the rule ‘An assertion is accurate iff what is asserted by that assertion is true’.⁵ The speech ‘His assertion is completely accurate but what he is asserting is false’ sounds like utter nonsense. One sign that MacFarlane is using a term of art is that we are supposed to be perfectly open to severing the connection between ‘accuracy’ in his sense and disquotational truth. Moreover, given the connection between ordinary accuracy and disquotational truth, the claim that ‘The accuracy of an assertion varies according to the context from which they are assessed’ will be just as prima facie puzzling as ‘The truth of what is asserted by a particular assertion varies according to the context from which it is assessed’, uttered using ‘true’ as a one place disquotational predicate.

When MacFarlane is being careful he does not use the monadic predicate ‘accurate’, but instead displaces it by the relational predicate ‘accurate as assessed by’. This no more expresses the ordinary notion of accuracy than the relational ‘is true as assessed by’ expresses the ordinary notion of disquotational truth. The apparent vacillation between the ordinary monadic notion of accuracy and a relational term of art is apt to create confusion all over the place. Consider for example the norm that MacFarlane articulates: ‘We should assert or believe something only if in doing so we would believe it accurately’. This is naturally interpreted as utilizing the ordinary one place notion, especially since, so interpreted, the norm is far from implausible. (For example, anyone who thought that knowledge was the norm of assertion would think of this norm as derivatively correct.) Yet MacFarlane then says that the view that ‘accuracy is assessment sensitive’ has consequences for this norm. It’s hard to see how this could be if the technical notion is relational and the notion that figures in the norm is the monadic one. So perhaps the quoted norm did not, despite appearances, use the ordinary notion. It is not easy to tell what is going on.⁶

MacFarlane believes that his own resources provide the means to answer a worry that we raise concerning how the relativist ought to distinguish herself from a naïve, chauvinistic realist—one briefly articulated above. The worry, roughly speaking, is that the relativist seems to make speeches that are remarkably similar to the naïve chauvinistic realist: ‘Gnocchi isn’t tasty. Giovanni thinks it is. So what he thinks is false.’ We recognized that the relativist can tack on remarks such as ‘Oh, but it is true for Giovanni’. But the worry is that there are very natural interpretations of additional remarks that the chauvinistic realist can buy into: ‘Oh, if your are talking about what meets Giovanni’s (false) standards for what is tasty, I’ll grant you that by those standards it is tasty’. We did not offer the objection as utterly decisive, but as one that is much more troubling than relativists acknowledge. MacFarlane, somewhat mysteriously, claims that his ideology of assessment relative accuracy somehow avoids the worry. But it seems that there are extremely natural ways for the chauvinistic realist to pay lip service to that relational ideology of accuracy at a context so as to buy into that as well. Nothing prevents the chauvinist realist from saying: ‘Oh, if you are talking about Giovanni’s standards, it is accurate as assessed relative to that context’. It is, at best, rather optimistic to suppose that slightly opaque terms of art will put the appropriate distance between the relativist and realist here. Might the difference show up in the patterns of assertion and retraction, or the patterns of normative evaluation of Giovanni’s utterances? It is hard to see that there will be any such difference. If the speaker’s tastes change in favour of gnocchi, he or she will switch to ‘I now think that, after all, gnocchi is tasty’ whether or not he is a chauvinistic realist who takes his current reactions as the guide to objective taste or a MacFarlanesque realist. MacFarlane’s discussion, thus, does not remove the worry—indeed, reliance on the term of art ‘accurate as assessed by’ merely serves to reinforce it.

MacFarlane apparently thinks that Chapters two and three in our book are largely irrelevant to contemporary relativism.

Consider first Chapter two, devoted to diagnostics of shared content, including say that, believe that, agreement and disagreement reports and anaphoric uses of ‘that’. One of the most cited defences of analytic relativism is Egan, Hawthorne and Weatherson’s ‘Epistemic Modals in Context’. Say-that and believe-that reports had a central place in that paper. More generally we noticed that both relativists and their precursors often rely on assumptions of shared content and we reckoned it useful ground clearing to compare various diagnostics for shared content. Along the way we made distinctions that we thought were both useful and seldom appreciated. Issues of shared content are pertinent to one of the central topics of the book—the time and world relativization of truth in formal semantics. These are often used as a basis for lending respectability to relativism. Dismantling views that displace monadic truth in favour of truth relative to worlds and times serves to put relativism in a radically worse light than it is typically thrown. (MacFarlane himself frequently alludes to time and world relativization as his primary strategy for softening readers up to his own views, so he can scarcely suppose such views are irrelevant.) Chapter two is very important in that connection. Our discussion reveals various arguments for time neutral contents (using collective belief reports and using anaphoric ‘that’ claims) to be pretty bad arguments. Moreover, the best supported diagnostics positively tell against those contents. The materials in Chapter two also figure quite prominently in Chapter four (a chapter that MacFarlane does not even mention). There we argue that while disagreement diagnostics are of evidential value, relativists have been altogether simplistic and one sided in their presentation of the relevant data when it comes to predicates of personal taste.

MacFarlane does briefly respond in a footnote to the line of argument that uses agreement and disagreement data against temporally neutral contents. Sometimes philosophers lapse into a use of ‘begging the question’ that sheds no lights whatsoever on a debate, teetering on the most pointless use whereby one automatically classifies an argument against a position as question begging when its premises support an unwelcome conclusion. MacFarlane’s complaint of question begging seems precisely one of this sort. The agreement and disagreement generalizations are well motivated. The time-neutral content lover has to engage in special pleading with regard to those generalizations when confronted with their application to his view. Special pleading of this kind is quite generally regarded as a cost for a view in syntax, semantics and science quite generally. If the relevant generalizations are well motivated and natural, an accusation of question begging will not make special pleading any more compelling or reasonable.

Consider next the ‘operator argument’ the topic of Chapter three. MacFarlane claims that he is ‘not aware of any analytic relativists who motivate proliferation on the basis of an operator argument’ (MacFarlane, this vol.) concluding that the ‘whole issue seems irrelevant to the debate over analytic relativism’. It is hard to reconcile his dismissive tone with the actual content of the relevant literature. Take Kölbel, one of the most cited contemporary relativists:

In the relativist semantics for taste predicates, we can introduce a class of operators that are analogous to Kaplan’s modal and temporal operators in that they shift the standard of taste parameter in the circumstance of evaluation, just as modal and tense operators shift the world and time parameter respectively. English seems to contain a construction that might intuitively be interpreted in this way, namely the “For S, p”- construction, as in “For Anna, whale meat is tasty.” (Kolbel 2009, p. 384)

How can we read this and not think that Kölbel is motivated, at least in part, by Kaplan’s operator argument? (Indeed Chapter three was the chapter with which Kölbel was most keen to engage when presented with a draft of our book.) Let us next consider MacFarlane himself. It is interesting here to focus on his 2007 paper “Relativism and Disagreement,” in which the operator argument plays a central role. MacFarlane’s presentation of relativism starts with a quote of Kaplan’s abbreviated statements of the operator argument viz:

A circumstance will usually include a possible state or history of the world, a time, and perhaps other features as well. The amount of information we require from a circumstance is linked to the degree of specificity of contents, and thus to the kinds of operators in the language. (Kaplan 1989, p. 502)

If we wish to isolate location and regard it as a feature of possible circumstances we can introduce locational operators: ‘Two miles north it is the case that’, etc…However, to make such operators interesting we must have contents which are locationally neutral. That is, it must be appropriate to ask if what is said would be true in Pakistan. (For example, ‘It is raining’ seems to be locationally as well as temporally and modally neutral.) (Kaplan, p. 504)

Commenting on this passage, MacFarlane says: ‘Taking this line of thought a little farther, the relativist might envision contents that are “sense-of-humor neutral” or “standard-of-taste neutral” or “epistemic-state neutral,” and circumstances of evaluation that include parameters for a sense of humor, a standard of taste, or an epistemic state.’ (MacFarlane 2007).⁷ MacFarlane does deny that the only thing that could ground such parameterization is the presence of operators of the sort Kaplan described (see e.g. MacFarlane 2009, p. 245). Now perhaps what he meant to say was ‘I not only wish to deny that operator-theoretic considerations of the sort that Kaplan describes are the only considerations that can motivate neutral contents. I wish to go further: Operator considerations cannot play any justificatory role at all’. But he didn’t say that and he can’t expect his audience to be mind readers.

Let us turn finally to the issue of fundamentality and explanatory priority. As we indicated earlier, our claim of fundamentality for monadic truth played a smaller role in our indictment of relativism than MacFarlane thinks. We did want to include something in our discussion by way of indicating a commitment to the idea that the property of truth is not ‘gerrymandered’—that it ‘carves at the joints’—and of indicating a commitment to the idea that it is more natural/less gerrymandered that relational properties such as true at which hold between propositions and worlds or times. That said, we concede that our remarks concerning the fundamentality of truth were pretty underdeveloped. At least part of the reason was that issues about fundamentality and priority are in general particularly tendentious and difficult issues, ones that would take us too far afield to discuss at length. Also, few of our arguments rely on considerations of fundamentality and so an extended discussion was for that reason less than vital.

We do wish to stress one point however. It is rather dangerous to suppose that the priority of one property family over another is established by defining predicates that express the second in terms of predicates that express the first. At least if one employs an intensional notion of definitional adequacy (whereby one succeeds in defining a predicate insofar as one captures its intension), one can define ‘green’ in terms of ‘grue’ and ‘bleen’: but that would show nothing about which holds in virtue of which or which is most fundamental. MacFarlane’s own way of connecting explanatory priority to definition is, thus, somewhat alien to our way of thinking.