Intuitions, Disagreement and Referential Pluralism
Mallon, Machery, Nichols and Stich (Philosophy and Phenomenological Research 79: 332–356, 2009) argue that the use of intuitions in the philosophy of reference is problematic as recent studies show intuitions about reference vary both within and between cultures. I use some ideas from the recent literature on disagreement and truth relativism to shed light on the debate concerning the appropriate reaction to these studies. Mallon et al. argue that variation is problematic because if one tries to use intuitions which vary to find the correct theory of reference one will end up endorsing an absurd position: referential pluralism. I argue that there is hope for intuition-based philosophy of reference. One can avoid endorsing referential pluralism (as Mallon et al. understand it). Furthermore, referential pluralism may not be so absurd after all.
In recent years, a number of arguments have challenged philosophers’ use of intuitions. One such argument is that of Mallon et al. (2009) (from now on MMNS).1 MMNS challenge the use of intuitions to theorise about reference in philosophy of language. MMNS present evidence that intuitions about the reference of certain terms vary cross-culturally.2 MMNS argue there are three ways the philosopher of language might try to accommodate this variation; none of them is attractive. First, the philosopher of language might choose to give up on the project of giving substantive theories of reference.3 Second, the philosopher of language might downplay their commitment to ‘the method of cases’ in theorising about reference. MMNS claim this is the dominant method in theorising about reference.4 To be committed to the method of cases is to think that, “[t]he correct theory of reference for a class of terms T is the theory which is best supported by the intuitions competent users of T have about the reference of members of T across actual and possible cases” (Mallon et al. 2009).5 An example of an intuition about the reference of a term is an intuition about whether ‘Gödel’ refers to the man historically identified as Gödel, or the man who satisfies all the descriptions we associate with the term ‘Gödel’. Third, the philosopher of language could endorse a position called ‘referential pluralism’: the idea that in some sense there is more than one correct reference relation for a class of terms.
In this paper, I discuss the third option. I consider whether referential pluralism is a viable position for philosophers of language who don’t want to give up the project of giving substantive theories of reference and who are committed to using the method of cases. MMNS claim referential pluralism is absurd. In particular, they think it has two absurd consequences. One is a claim I call The Co-membership Requirement. This is the claim that in order to agree or disagree with someone your intuitions about the reference of the relevant terms must be the same. The other is a claim I call Disagreement Agnosticism. This is the claim that we never know whether two parties agree, disagree or talk past each other.
I provide a limited defence of referential pluralism. First, I argue that MMNS fail to appreciate the variety of pluralist positions available — making use of a distinction between speaker relativism and assessor relativism about the truth of propositions. Second, I argue that the referential pluralist needn’t accept either The Co-membership Requirement or Disagreement Agnosticism—the positions which MMNS claim the referential pluralist should accept and which they claim are absurd. Finally, I challenge the idea that these claims are actually absurd; I consider both in light of clarifications made by MMNS in response to certain criticisms (see Ichikawa et al. 2012; and Machery et al. 2013). I argue that, so clarified, these two positions are not as absurd as MMNS suppose.
In presenting these objections to the challenge presented by MMNS, I provide a source of optimism that philosophers of language can accommodate evidence of cross-cultural variation in intuitions while retaining an ambition to provide a substantive theory of reference and a commitment to use the method of cases.6
2 Varieties of Referential Pluralism7
In order to accommodate variation in intuitions one might consider adopting some form of pluralist position. Let’s consider what this means, and why we might think it is the only option open to the philosopher of language who is committed to giving a substantive theory of reference using the method of cases.
Let’s say that what a non-pluralist theory of reference for a class of terms T does is provide a reference relation R. A reference relation, such as R, tells us what the terms in T refer to both in the actual world and all possible worlds. The basic idea of referential pluralism must be to deny that there is a single relation R with respect to the class of terms T. The pluralist about reference accepts that there are a number of distinct reference relations that tell us what the terms refer to.
We can see why one might think that using the method of cases will result in a pluralist position when the intuitions used vary between groups. To explain why, we need to use MMNS’s notion of an intuition group. An intuition group is a group of persons all of whose intuitions about a set of cases are identical. To model cross-cultural variation in intuitions, suppose that for some set of cases C involving terms from T there are two distinct intuition groups: I1 and I2. I1 and I2 comprise competent users of terms in T. All members of I1 agree about all the cases in C. All members of I2 agree about all the cases in C. I1 and I2 are distinct groups because there is some member of C about which members of I1 and I2 disagree with each other. To be committed to the method of cases is to hold that the correct theory of reference is that which is best supported by the intuitions of competent users. So, it does seem using the method of cases will result in some sort of pluralist position. By looking at the judgements of members of I1 one will get one reference relation, R1. By looking at those of I2, one will get another, R2.8
There is one reference relation per intuition group, R1 and R2. R1 is the relation best supported by the judgements of members of I1 (about terms in T and cases in C), and R2 is the relation best supported by those of I2.
RP1, plus the stipulation that the reference relation is relative to the context of the speaker (the relevant contextual feature being intuition group) such that, e.g., R1 gives the reference of terms in T when used by members of I1.
A result of RP2 is that the truth of sentences involving members of T is relative to the context of the speaker, as intuition group membership is part of an individual’s context.
MMNS think using the method of cases when intuitions diverge will result in a version of referential pluralism according to which RP2 is true.9 The point I want make clear here is that there are pluralist positions available which reject RP2. In particular, the referential pluralist doesn’t have to take a speaker relative view about the reference of terms in T. One alternative is for the referential pluralist to take an assessor relativist view. Consider the following case. Suppose that members of I1 have broadly descriptivist intuitions about mental state terms; they think that if there are no mental states satisfying all the descriptions we typically associate with, e.g., belief, then the word ‘belief’ fails to refer. Suppose that I2 have broadly causal-historical intuitions. In this case R1 will be descriptivist, and R2 will be causal-historical. Let’s also suppose that there are indeed no mental states satisfying all the descriptions members of I1 typically associate with belief, so ‘belief’ fails to refer according to R1. Finally, suppose that ‘belief’ does refer according to R2.
MMNS assume that it follows from this that the reference of mental state terms depends on who utters them and, accordingly, that the truth of sentences in which they appear depends on who utters the sentence. But, it doesn’t follow. It is theoretically open to the referential pluralist to say that the reference of mental state terms depends on who assesses the sentence. They can say that reference is relative to context of assessment and that, accordingly, truth is relative to context of assessment.10 To understand this, we should consider briefly what is meant by a ‘context of assessment’. A context of assessment is “a situation in which a (past, present, or future, actual or merely possible) use of a sentence might be assessed for truth or falsity” (MacFarlane 2005a). The idea that truth might be relativized to a context of assessment has received a considerable amount of attention in recent years. Some sort of assessment sensitivity has been argued to be essential to adequate semantic accounts of various phenomena, including future contingents, epistemic modals and statements of taste (MacFarlane 2005a).11
It is very important to appreciate that the referential pluralist can handle the relativisation of sentence truth (either to speaker or assessor context) in either of two rather different ways: indexicalism and non-indexicalism.12 Indexicalist views hold that the truth of sentences is contextually variant in virtue of the fact that the sentence expresses different propositions relative to different contexts. Non-indexicalist views hold that the same proposition is expressed relative to all contexts, but that the truth of the proposition varies.
In the following, my discussion primarily concerns the non-indexical assessment relativist view which the referential pluralist could adopt. To explain this type of view, let’s consider what a non-indexical assessment relativist view about taste predicates such as ‘is tasty’ would look like. Take the case in which I say ‘ginger beer is tasty’ and you say ‘ginger beer is not tasty’. Non-indexical assessor relativism says the following. You and I assert and deny the very same proposition p. The truth of p is relative to context of assessment, i.e., p as assessed by me is true and as assessed by you is false.13 The idea is that ‘is tasty’ makes the same contribution to the proposition expressed by a sentence regardless of context of assessment. This contribution is of a single relation whose extension varies with the context of assessment, i.e., a relation which tells us for each context of assessment the relevant reference relation (which tells us the extension of the predicate as assessed by a particular agent in this and other worlds).14
MMNS’s failure to consider that the referential pluralist could be a non-indexical assessment relativist is important. It is important because MMNS’s main worry about referential pluralism is that it leads to a view on which there can be no disagreement between members of different intuition groups. However, assessor relativist views about truth (especially, non-indexical views) have more power to accommodate disagreement than speaker relativist views (see MacFarlane 2014). In particular, they can accommodate a sense of disagreement which has been called ‘preclusion of joint accuracy’ (this and other senses of disagreement will be explained in the next section). The availability of assessor relativism about reference, therefore, promises to provide a more favourable picture of referential pluralism.15
3 The Co-membership Requirement
MMNS think this requirement is absurd. Let’s set aside the question of its absurdity for now and ask whether the referential pluralist has to accept it. I argue that they do not. In particular, I will argue that the referential pluralist can say that parties from different intuition groups can disagree.
- The Simplest Sense:
For two parties’ dispute to be a case of genuine disagreement, when they argue saying ‘beliefs exist’ and ‘beliefs don’t exist’ respectively, it is sufficient that they are respectively asserting and denying the same proposition, otherwise they talk past each other.
- The Doxastic Sense:
To disagree with someone’s belief that p is to have beliefs whose contents are jointly incompatible with p.18
According to the doxastic sense, disagreement is a doxastic phenomenon; in order to have beliefs that disagree with a belief of someone else it must be the case that you couldn’t consistently hold a belief with the same content as theirs (without dropping any current beliefs). If Donald believes the moon is cheese and Sarah believes the moon is yoghurt, they disagree in this sense, because the moon cannot be both cheese and yoghurt.
Is this the sense MMNS have in mind? Again, it can’t be. Again, the referential pluralist who goes for a non-indexical version of either speaker relativism (the only option according to RP2) or assessor relativism (as allowed by RP1) can say that members of different intuition groups can disagree in this doxastic sense. I’ll assume a simple theory of how our story about reference and language meshes with a story about mental content. The contents sentences express and the contents of our beliefs are the same in kind: propositional. A belief has the same content as a sentence expresses only if their content is the same proposition. Let’s consider A and B again. A and B belong to different intuition groups and appear to disagree saying ‘beliefs exist’ and ‘beliefs don’t exist’ respectively. According to the non-indexicalist (of either stripe), if A and B express their beliefs, the contents of those beliefs must be the same propositions they express when they speak. The contents are in this case clearly jointly incompatible, as on both types of view A and B respectively assert and deny the very same proposition. One has a belief with content < p > and the other with the content < ¬p>. The truth conditions are agent relative, so, while it is possible each has a belief which is true relative to his/her context, it is not possible for both beliefs to be held by the same individual in the same context without inconsistency. So both assessor and speaker non-indexical relativist views can accommodate the doxastic sense of disagreement very easily.
I suspect MMNS would claim that neither the Simplest Sense nor the Doxastic Sense of disagreement properly captures the sense of disagreement they had in mind. In particular, I suspect they would think that neither captures the sense in which they think the referential pluralist (or anyone) faces a problem if they have to claim that members of different intuition groups can’t disagree. They would probably say that doxastic disagreement isn’t good enough. MMNS may be willing to accept that A and B can’t take on each other’s attitudes, but nonetheless deny that A and B disagree in the sense they had in mind, because A and B could both be right.
If this is the thought, there are two senses of disagreement which might be being expressed. In order to understand them we need to understand MacFarlane’s intuitive notion of ‘accuracy’ (see MacFarlane 2014). The rough idea is as follows. An attitude or speech act has a propositional content. The content itself might be true relative to some contexts and false relative to others. For the attitude or speech act to be accurate is for its content to be true relative to the context that matters.
- The Preclusion of Joint Accuracy Sense:
A and B disagree if the accuracy of each of their beliefs or speech acts (as assessed from any context) precludes the accuracy of the other’s (as assessed from that same context).
To illustrate the sense of disagreement articulated here, consider the following case. A thinks ice cream is nice and B thinks it is nasty. One might think they disagree in the sense that joint accuracy is precluded, i.e., relative to any single context (which we can suppose determines a single standard of taste) ice cream can’t be both nice and nasty.
Is this the sense MMNS have in mind? Again, it can’t be, because the referential pluralist can accept that members of different intuition groups can disagree in this sense. This time the distinction between speaker and assessor relativism becomes important. The non-indexical speaker relativist cannot accommodate this sense of disagreement. They hold that ‘the context which matters’ is always that occupied by the speaker, i.e., the context of assessment is irrelevant to the accuracy of a belief or speech act. Therefore, for the non-indexical speaker relativist, if A believes p and p is true relative to A’s context, then A’s belief is accurate; likewise, if B believes ¬p and p is false relative to B’s context, then B’s belief is accurate. Since a proposition’s truth may vary between A and B’s contexts (in virtue of the reference relation being determined by speaker context), they may both have accurate beliefs, according to the non-indexical speaker relativist, so joint accuracy is not precluded.
However, the non-indexical assessor relativist can accommodate this sense of disagreement. They hold that relative to any single context of assessment, only that context is relevant to determining the accuracy of any speaker’s utterances or beliefs. They also hold that relative to any single context of assessment a proposition has only one truth value (in virtue of the reference relation, and thus the extension of the relevant terms, being fixed by a context). Therefore, A’s belief p and B’s belief ¬p cannot both be accurate relative to any context of assessment, according to the non-indexical assessor relativism. Consequently, the referential pluralist who opts for such a view can accommodate this sense of disagreement.
- The Preclusion of Joint Reflexive Accuracy Sense:
A and B disagree if the accuracy of A’s beliefs or speech acts (as assessed from A’s context) precludes the accuracy of B’s beliefs or speech acts (as assessed from B’s context)—and vice versa.
In this sense, it is more difficult to disagree. For example, John and I don’t disagree about the tastiness of ice cream, in this sense, because John’s attitudes may be accurate as assessed from his context while mine are accurate as assessed from my context.
Is this the sense MMNS have in mind? They give no explicit motivation for it. However, if they intended to understand The Co-membership Requirement in terms of this sense of disagreement, it would provide them some help; if preclusion of joint reflexive accuracy is necessary for disagreement, then no kind of referential pluralist can provide disagreement, regardless of the type of relativism they endorse. Moreover, if The Co-membership Requirement is to be understood in these terms, then it does follow from RP1, and a fortiori from RP2.
However, I think this cannot be the notion MMNS have in mind. The reason is that to insist that the non-indexical assessor relativist be able to accommodate this sense of disagreement—the preclusion of joint reflexive accuracy—is really to insist that the ‘context which matters’ always concerns the context of the speaker. It is to insist, contra assessment relativism, that accuracy is always determined by the context of the speaker. But this is simply to deny that non-indexical assessor relativism is a viable semantic account at all—regardless of the domain of discourse. If this is MMNS’s contention, they need to argue for it as it is somewhat implausible. One reason not to simply dismiss assessor relativism is that it has a great explanatory power, for example, in accounting for when one should retract an assertion in a number of domains (see MacFarlane 2014).19 A second reason is that it is terribly counter-intuitive to insist that the context which matters is always that of the speaker. The accuracy of speech acts and attitudes often seems to be properly evaluated relative to contexts other than the speaker’s. This is true even in matters of taste! Lasersohn (2005) describes a number of cases which help demonstrate this. For example, a case in which John tells Mary how their 2-year-old son Bill enjoyed a recent trip to the amusement park (p.25). Mary asks, ‘How did Bill like the rides? John replies, ‘Well, the merry-go-round was fun, but the water slide was a little too scary’. Lasersohn notes that, we think John’s utterance as true if the merry-go-round was fun for Bill independent of whether John found it fun. Likewise, to take another of Lasersohn’s examples (p.26), if John is trying to persuade Mary to do something saying, ‘Come on! It will be fun!’ intuitively it is whether the activity will be fun relative to Mary’s tastes which seems to matter most when asking whether what John says is accurate.
So, for now, it is safe to assume that this is not the sense of disagreement MMNS had in mind. If MMNS intend The Co-membership Requirement to be understood in terms of The Preclusion of Joint Reflexive Accuracy, then the onus is upon them to say a lot more.
I might seem to have been unfair. The worry to which The Co-membership Requirement gives voice is not always expressed in terms of disagreement. In the conversation between Ichikawa et al. (2012) and Machery et al. (2013), the worry which we have been thinking about in terms of disagreement is parsed in terms of communicative success. The worry might be that even if the relativist can give us disagreement, the pluralist picture and any accompanying relativism is one of communicative breakdown. For A and B to successfully communicate, MMNS might say, they must belong to the same intuition group. I am inclined to think this thought is mistaken; I am inclined to have the same optimism as Ichikawa et al. that there are ways for the relativist to account for successful communication, e.g., using Maitra’s notion of partially successful communication (Maitra 2007).
Ichikawa et al. (2012) respond to the worry that communication will fail between members of different intuition groups. The relevant part of Maitra’s paper concerns the idea that communication via semantic content might be comparatively easy in a relativist framework. She introduces the notion of partially successful communication. Partially successful communication can be possible even if the parties concerned do not have access to all the information necessary for full communication, e.g., one’s interlocutor’s intuition group. The basic idea can be illustrated using an example. Charlie tells Tara and Tamara, ‘I only play the trumpet’. Tara knows the trumpet to be a wind instrument, but nothing more. Tamara fails even to hear the final word of the sentence. Charlie’s attempt to communicate is in an important sense more successful with respect to Tara, than it is with respect to Tamara. A lot of information is conveyed; it would certainly be a push to declare that there has been a complete breakdown in communication between Charlie and Tara.
Something similar is possible in the domain of discourse MMNS care about. Suppose A and B say, respectively, ‘beliefs exist,’ and, ‘no, they don’t’. They are from different intuition groups. They each think that they both belong to a single intuition group. Partially successful communication is still possible. A and B, even if they have (or would have if asked) completely descriptivist intuitions and completely causal-historical intuitions respectively, still have enough common ground to avoid complete communicative breakdown. They agree on typical features of belief-attribution, commonly held ideas concerning the nature and extension of belief, and indeed, we might suppose, a great deal. They do not simply talk past each other.
It seems that you don’t need to be in the same intuition group as someone in order to avoid complete breakdown in communication when you speak with them. But perhaps this is not enough. Perhaps MMNS think that it is worrying if members of different intuition groups are incapable of anything but mere partially successful communication. But it is unclear why this would be quite so worrying, especially given that parties who communicate with only partial success can clearly disagree, e.g., if in our previous example Tara had known Charlie to play the piano and stated, ‘It is not true that Charlie only plays the trumpet,’ Charlie and Tara would surely disagree.
4 Disagreement Agnosticism
- Disagreement Agnosticism:
For any two parties A and B who appear to disagree, saying, for instance, ‘beliefs exist’ and ‘beliefs don’t exist’, we do not know when they agree, when they disagree and when they talk at cross-purposes.
- Additional Premise
“[W]e do not really know to which intuition groups any of us belongs” (Mallon et al. 2009)
Due to evidence of intra- and inter-personal variation “one cannot assume that people who share a language and a culture are members of the same intuition group.”
“Because lots of cases are relevant, intuition groups must be fine-grained …”
Therefore, “Numerous cases must be considered for each speaker”.
“Explicit views are not a good guide to intuition group membership…”
MMNS argue that these points strongly support the extra premise. I don’t want to go into the detail of the arguments given for these four points. I simply think that, far from being part of a good argument for the claim—that we never know whether two parties agree, disagree or talk past each other—they give us reason to think it is false.20
First, here’s an observation. By this point in their 2009 paper MMNS mean something a bit different by ‘intuition group’. When the notion of an intuition group was introduced one’s intuition group membership was relative to a class of cases—any class of cases. This can’t be what MMNS mean here. A class of cases can have a single member. It is wildly implausible to suggest that for all pairs of individuals there is no case relative to which we know their intuition group memberships. Even if language, culture and explicit theoretical views are bad guides to group membership (considering a class of cases containing one case), to be sceptical about our ability to know to which intuition group any of us belongs is to endorse a much more worrying scepticism. Remember, in the present debate, all that is meant by an intuition is a judgment about a case.
What MMNS clearly mean by ‘intuition group membership’ at this point is ‘intuition group membership with respect to [some particular sort of class of cases]’. I think the class in question is something like the following: that class of cases one would need to consider in order to work out whether one’s judgements best supported some particular highly specified theory of reference rather than all others (including distinct but very similar theories). This observation is not in itself a criticism.
It is easy to see why MMNS are talking about this particular type of intuition group. They are worried about the consequences of RP for arguments among philosophers about the reference of terms such as ‘belief’, and they think it is a consequence of pluralism that members of different groups, concerning any class, cannot disagree or agree. Which set of cases does one need to consider to allay such worries? Consider the following example. Imagine we have data concerning intuitions about a class of cases Csmall with five members x1 … x5. Suppose we find that individuals A and B make the same judgement about each of x1 … x5, and that they are thus members of the same intuition group with respect to Csmall. We are not in a position to determine that A and B can agree and disagree. If we consider intuition group membership with respect to some larger class of cases Clarger that not only includes x1 … x5, but also x6 and x7, we might find that A and B belong to different intuition groups, and so only talk past each other. It seems there will always be some pressure to ‘go large’ as it were—to consider more and more cases. To determine whether two individuals can agree/disagree or only talk past each other, you have to consider the largest class you can. As MMNS intimate, a lot of cases are relevant and so must be considered for every speaker.
MMNS are broadly right that the number of cases needed to distinguish members of an intuition group whose judgements best support one highly specified theory of reference, from another whose intuitions best support a distinct but very similar theory, is very large indeed.21 I think that, due to the fact that theories of reference for a class of terms can be (and have been) constructed that differ only in very small details, intuition groups (in the relevant sense) will be very fine-grained. I agree that, since neither language, culture nor explicit views are particularly good guides to intuition group membership, it will be very difficult to tell which exact group an individual belongs to.
Nonetheless, I argue, none of this precludes knowledge about whether two individuals are members of the same group and so the supposedly absurd consequence doesn’t follow from The Co-membership Requirement — even given the additional premise. My argument for this is that it seems fairly clear that, for any two individuals, the odds are that they belong to two different intuition groups (in this new hyper-fine-grained sense) concerning any class of terms T. Given the extreme (perhaps infinite) subtlety with which theories of reference can be crafted, it seems so likely that two individuals will belong to different intuition groups somewhere along the line that we can know that, for any two philosophers in a situation similar to A and B, they do not disagree (given the idea that one can only disagree with members of the same intuition group).22
- The Co-membership Requirement:
For two parties A and B to agree or disagree they must belong to the same intuition group; parties who belong to different intuition groups end up talking past each other.
- Disagreement Agnosticism:
For any two parties A and B who appear to disagree, saying, for instance, ‘beliefs exist’ and ‘beliefs don’t exist’, we do not know when they agree, when they disagree and when they talk at cross-purposes.
“Suppose that there are a number of different intuition groups and that it is unclear what intuition group philosophers A and B belong to. Then, if referential pluralism is correct, when philosopher A says ‘Beliefs exist’ and philosopher B says ‘Beliefs do not exist,’ it is unclear whether philosophers A and B disagree or whether they are speaking at cross-purposes. [I]f it is unclear which intuition group we belong to … referential pluralism leads to the absurd conclusion that we simply have no idea when proponents of these arguments agree, when they disagree and when they talk past each other.” (Mallon et al. 2009, p.348)
MMNS think it is absurd that when philosophers engage in these types of debate they are talking at cross-purposes. What characterises debates of this type is that they involve arguments, for and against strong ontological claims, which are made on the basis of a theory of reference.23 For instance, the eliminative materialist argues that beliefs do not exist, because nothing satisfies the descriptions which we associate with our term ‘belief’ and so the term fails to refer. Note that this argument is not remotely plausible unless a descriptivist theory of reference for folk-psychological terms such as ‘belief’ is assumed.
“While we do note in passing that our argument extends to ordinary discourse (Mallon et al. 2009, 348), the bulk of our discussion consists in arguing that relativizing reference to different intuition groups would cause a breakdown of philosophical argument as we know it. Since philosophical arguments often center precisely on cases where different fullfledged theories of reference allow one to draw different philosophical conclusions (e.g., “What does ‘good’ refer to?”, “Are there races?”, “Do beliefs exist?” etc.), there is little reason to think Ichikawa and colleagues’ remark offers much comfort.” (Machery et al. 2013, p. 14).
In the rest of this section, I consider what the referential pluralist ought to say about this type of philosophical debate. I shall argue that the referential pluralist who accepts The Co-membership Requirement and Disagreement Agnosticism should conclude that debates such as 1-4 tend to be spurious. Moreover, I shall argue that this is not an absurd conclusion.24
Let me make two important qualifications about what I am about to argue. First, I do think it would be absurd to argue that most philosophical debates are spurious. I do not deny this. What I deny is that it is absurd to claim that a certain class of philosophical debates are spurious: those which involve arguments for and against strong ontological claims on the basis of a theory of reference. Such debates hardly form the bulk of philosophical debate. Second, I don’t claim that such debates actually are spurious.25 Rather, I claim that it would not be absurd to claim they were spurious.
Why do I think that the referential pluralist who accepts The Co-membership Requirement and Disagreement Agnosticism should think that debates such as 1-4 are spurious? Recall my earlier argument that for any two individuals it is very likely that they belong to different intuitions groups, since intuition groups are so fine-grained (as argued by MMNS). If we endorse The Co-membership Requirement, this means that for any two individuals it is very likely that they are talking past each other rather than agreeing or disagreeing, even if we don’t strictly know which individuals are talking past each other and which are genuinely disagreeing. As noted, MMNS seem willing to concede that these worries do not extend to ordinary discourse, but still claim to be worried about debates such as 1-4. So the question we need to answer is: how absurd is the claim that most debates like those at 1-4 involve no real disagreement?
Let’s start off with a few unfair rhetorical questions. Of which sorts of debate in philosophy would we have been antecedently most likely to make the diagnosis that the parties in question were just talking past each other? Of which that the parties involved were failing to engage each other? Even before looking at the Mallon et al. (2009) paper, I was inclined to think that it was precisely these types of debate: cases where philosophers’ disagreement might be only about the correct reference of terms.
In each of these cases, and others like them, it seems like there is something very important at stake. An important point can be made, for instance, by drawing attention to the fact that the way that people typically think about race doesn’t fit with how the world is. This is not what is being denied when one claims the debate is spurious. It is the debate over who has the better ontology of beliefs, with respect to accurately representing the structure of reality, which one might want to deem spurious. A good way to make the point might be to say that, if the pluralist is right, the natural language semantics of ‘beliefs’ makes it a bad word for use in serious ontology (where used in its ordinary sense). In any case, it is important to emphasize that this doesn’t mean the referential pluralist who accepts The Co-membership Requirement and Disagreement Agnosticism is committed to any kind of serious ontological relativism. They may end up (i) committed to some form of relativism about the truth of the sentence ‘beliefs exist’, and (ii) saying – very loosely – that in some sense there are multiple candidates for how the reference of ‘beliefs’ gets fixed. However, specify a referent, and the referential pluralist can accept that there are non-relative facts about its ontological status.
I would like to have something better to say here, by means of argument, but I have not, for the moment. However, it seems appropriate merely to deny the claim that this conclusion is absurd. It does not seem so absurd to think that most disagreements such as these are spurious in the relevant sense.26 It would seem absurd to think that there was nothing at stake in such arguments. However, that conclusion won’t suffice for the argument against referential pluralism, for there is no reason to think the referential pluralist might have to accept that there was nothing at stake in such arguments.
I should make a quick note about the scope of my argument here. My claims concern the plausibility of referential pluralism and I hope to provide some source for optimism that we might be able to use the method of cases to theorise about reference even if intuitions vary across cultures. Mallon et al. (2009) note that one consequence of referential pluralism being absurd is that the arguments used in debates such as 1-4 are problematic.27 My arguments in this section clearly don’t provide any cause for optimism that such arguments might be defended.
MMNS argue that in the face of evidence of variation the philosopher of language who uses the method of cases to give substantive theory of reference is forced into referential pluralism: a position they claim is absurd. They argue that the use of the method of cases in theorising about reference is thus misguided; that intuitions are not so great a source of insight as we supposed. This paper’s main focus has been to provide a limited defence of referential pluralism. In particular, it comes in more varieties that MMNS appreciate and doesn’t have the absurd consequences which MMNS worry about. Consequently, it seems using the method of cases to theorise about reference ought not to be so quickly dismissed, even if intuitions about reference vary between cultural groups.
A number of responses to MMNS have argued that the data in question do not provide evidence of variation in (right kind of) intuitions (Cohnitz and Haukioja 2012; Deutsch 2009; Lam 2010; Martí 2009; Systma and Livengood 2008/2009). Machery et al. (2009) and (2010) respond to some of these objections. In this paper, I grant, for the sake of argument, that there is significant variation in intuitions about reference.
This descriptive claim may be in need of some defence. There are certainly those who argue that philosophers of language ought to be able to use other methods, e.g., as noted by an anonymous reviewer, Devitt (2012) argues fairly convincingly that corpus studies and elicited production can play an important role in theorising about reference.
It seems conceivable that one might encounter a tie. Two theories of reference might be supported equally by the intuitions of competent users. In such a situation, I suppose the user of the method of cases would either try to come up with some new cases, or resort to different means to decide the best theory.
Mallon et al. (2009) frame their argument as part of a challenge to a particular argument form which they call ‘argument from reference’. They argue that there is no way to accommodate variation in intuitions which permits such arguments. For a brief overview of the debate concerning arguments from reference, see Genone (2012). I do not consider this aspect of their paper here. MMNS’s main goal is “to challenge the methods philosophers of language have embraced in their efforts to develop a satisfying theory of reference” (Machery et al. 2013: 17). In this paper, I consider their arguments against referential pluralism and their challenge to the method of cases in philosophy of language. These arguments do not depend on further details of MMNS’s argument concerning ‘argument from reference’; MMNS simply argue that using the method of cases when intuitions diverge will lead to referential pluralism and that this is an absurd position.
This section owes much to discussion of relative truth and the possibility of faultless disagreement in relation to statements of personal taste (Egan 2010; Huvenes 2012; Kölbel 2004; Lasersohn 2005, 2009; MacFarlane 2005b, 2007a, b, 2014; Ponte 2011; Richard 2008; Schafer 2011; Stephenson 2007; Stojanovic 2007).
There are other options here which one might consider and so it isn’t clear that using the method of cases when intuitions vary forces one to accept referential pluralism. Thanks to an anonymous referee for pointing some of these out: (1) One might think that the theory which is best supported by the intuitions of competent users was that which described a single reference relation which was maximally accommodating on the intuitions of both I1 and I2. (2) Alternatively, one might think that a commitment to the method of cases doesn’t mean that one couldn’t find some reason to give preference to the intuitions of one group rather than another, e.g., perhaps one could put the intuitions of one group down to some sort of performance error. (3) Another alternative might be to think that different groups may be effectively responding to different cases or questions as they may import different information to the experimental material—a point similar to that made by Sosa (2009) in relation to a different challenge to intuitions. Although it is not obvious what difference in imported information could make one group descriptivist about a name and another causal-historical. (4) A final option might be simply to accept that only one group is tracking the true reference relation but remain neutral concerning which—a response which would certainly not provide much optimism for intuition-based philosophy of language.
Note the way Machery (2012) expresses the idea elsewhere “… philosophers of language interested in reference could maintain that intuitions are reliable guides to the semantic properties of names and go on to infer that names refer differently in different cultures” (p. 41, my emphasis).
It seems sensible to assume, like MMNS, that what one says about relativity of reference will be mirrored by what one says about relativity of truth.
MMNS do not explicitly address this distinction. But they indicate some awareness of the issue. Mallon et al. (2009, n.13) claim that they want to remain neutral as to how speaker relativism in natural language is best explained; I take this to mean that they are open to non-indexicalist forms of context-sensitivity given that they cite the work of MacFarlane in which relativism about propositional truth is discussed.
Contrast this with an indexicalist relativist account. Such an account might say that ‘is tasty’ expresses different contents as assessed from different contexts. For example, when I say ‘ginger beer is tasty’, there are no non-relative facts about the proposition I assert. As assessed by you, I assert p1 which is false; as assessed by me I assert p2, which is true.
Although, the non-indexicalist considered here is a pluralist about reference relations (as understood as mappings from terms/sentences to extensions/truth-values) there is something monistic about the account. There is a single relation expressed invariantly by ‘is tasty’, however, this is not the sort of reference relation which MMNS talk about, as its extension varies with the context of assessment.
To fail to consider assessor relativism seems odd given that MMNS cite the work of Lasersohn and MacFarlane who both articulate views about some sort of relativity (albeit not in terms of reference) to a context of assessment (See Lasersohn 2005, 2009; MacFarlane 2005b, 2007a, 2009, 2014). It may be that MMNS would find assessor relativism implausible for other reasons. I’ll look at some of these at the end of section 3.
E.g., “When asserting or denying that beliefs exist or that science progresses (or the conclusion of any other argument from reference), philosophers A and B would agree or disagree only if they belonged to the same intuition group” (Mallon et al. 2009: 347).
Some worry that this isn’t true: that relativism about the truth of propositions isn’t tenable and that any relativist view will collapse into relativism about the content of beliefs/sentences. I don’t share this worry and it isn’t a worry articulated by MMNS so I don’t address it here.
Suppose we accept a contextually shifty account of knowledge attributions such that truth of knowledge attributions depends on contextually variant epistemic standards. Suppose that I assert ‘I know p’ at t1 when the standard are low at my context. Suppose that at t2 I occupy a context in which the standards are very high. The relativist has an easy job of saying why it is correct to retract my earlier assertion: it is false at my present context. If it were always the context of use which mattered, then it is unclear why it would ever be correct to retract an earlier assertion.
If you are so inclined, you can read this argument as a reductio. This will probably lead you to reject the crude understanding of the method of cases which we have been working with so far (taken from MMNS), as it is this which leads to the idea that the philosopher using the method of cases needs to know about intuition group membership at such a hyper-fine-grained level. You would probably be right to do so.
Indeed this is simply an instance of the general problem of theory being underdetermined by the evidence, rather than a specific problem for referential theory. Thanks for an anonymous referee for pointing this out.
Perhaps MMNS rely on a very strong account of knowledge. In which case, my point here will not stand. However, (i) there are other reasons for thinking these authors wouldn’t want to rely upon any specific account of knowledge, and (ii) if they did require something like absolute certainty, then not having knowledge would not be particularly bothersome.
One might worry that the concern about spuriousness is not just restricted to those areas of philosophy in which strong ontological claims are made on the basis of a theory of reference. The worry might be that they will spread to any area of philosophical discourse, indeed any area of discourse, so long as intuitions about the application of the relevant terms vary. The worry might be that there is variation in intuitions about the reference of a great variety of the terms we use. Of course, if this worry is correct, then it is of little help for MMNS, as then their (2013) response to Ichikawa et al. (2012) is problematic. Nonetheless, if the worry is correct, then it is a problem for my arguments in this section. However, it is not a worry for my overall project in this paper as, see section 3, I argue that the referential pluralist doesn’t have to accept The Co-membership Requirement.
If I can show that the conjunction of the two views is not absurd, then it will be clear that neither position is absurd taken by itself.
I have already argued that MMNS’s argument fails at earlier stages, e.g., I don’t think the referential pluralist has to accept The Co-membership Requirement.
The conclusion that various similar philosophical debates are spurious is one which has been argued recently in Baz (2012). Thanks to an anonymous reviewer for pointing this out.
The reason being that in order to make such arguments one has to endorse a substantive theory of the reference of the relevant class of terms, e.g., folk-psychological terms, and the project of giving a substantive theory of reference runs into the problems we have discussed in this paper given cultural variation in intuitions.
Thanks to Carl Baker, Michael Bench-Capon, Greg Currie, Mark Jago, Phillip Percival, Stefano Predelli, Pendaran Roberts, audiences in Leeds, Manchester and Nottingham, and to a number of anonymous referees for very helpful comments. Acknowledgment is also due to the support of an AHRC-funded studentship.