Disputes and Defective Disputes

Chapter 1

INTRODUCTION

Cicero is credited with having said that there is nothing so absurd but some philosopher has said it. Bertrand Russell once remarked that the point of philosophy is to start with something so simple as to be hardly worth mentioning and to end with something so paradoxical that no one will believe it.[1] If we have a reputation for anything, it might seem, philosophers are known for flying in the face of common sense.

And yet there is a feeling amongst some philosophers that we ought to respect the pretheoretical commonsense intuitions of so-called ‘ordinary folk’. Some of these so-called ordinary folk, however, feel that the activity of philosophers is in an important way silly or absurd.

What does this charge amount to? Simply that a large number of philosophical disputes are defective. We are all familiar, from our nonphilosophical lives, with disputes that go nowhere. These disputes are interminable for one reason or another. Perhaps parties to the disputes are being irrational, refusing to see the other side’s point. Or maybe there is just nothing to argue over; each side is making some kind of mistake. Disputants might be talking past one another, arguing at cross purposes, or otherwise engaging in wrongheaded behavior.

Regardless of the details of specific disputes that are wrongheaded the following seems undeniable. If a dispute is wrongheaded, that is a reason for rational people to stop having it.

Many disputes from the history of philosophy have been alleged to be wrongheaded in this way. The charge of wrongheadedness, however, comes most often from philosophers themselves who think they have a principled reason to reject certain disputes. Defective philosophical disputes, we are told, are ones where there is no real disagreement. What exactly does it mean to say that there is no real disagreement? What does it mean for two or more persons to disagree? We will need to have a handle on this issue before we proceed to establish what makes a philosophical dispute wrongheaded.

What are the different ways in which disputes can be wrongheaded or defective? It is important that we understand that there are different sorts of defective dispute. For, different kinds of defective dispute will be reconciled in different ways. Consider: if disputants are talking past one another, they will do well to shift the dispute to a different vocabulary; whereas, if disputants are addressing an issue about which there is no fact of the matter, then resolving their differences of opinion may not even be possible.

It is the task of this project to establish criteria for telling when a dispute is defective and if so in what way it is defective. Part One beings with a discussion of disagreement. Chapter One seeks to answer questions like: What is it to disagree? Are disagreements necessary for disputes? In Chapter Two I distinguish a number of different ways that disputes can be wrongheaded or defective.

Part Two takes up a consideration of two test cases of disputes in philosophy. In Part Two I seek to determine whether these disputes are defective in any of the ways identified in Chapter 2. I choose as my two test cases the much-maligned dispute over mereological composition and the famous dispute between David Lewis and Alvin Plantinga over the nature of possible worlds. I argue that neither dispute is defective.

Along the way, there is much to learn about what makes for a silly or pointless dispute in philosophy. I contend that philosophical disputes should be repudiated when they do not serve to bear fruit. In short, we ought to abandon a philosophical dispute only when (i) it is defective in one of the ways I outline in Part One, and (ii) it is fruitless in the sense that consideration of it does not yield interesting and useful distinctions or theoretical machinery that could not be gotten in some other way.

As long as there have been philosophical disputes there have been philosophers who suspect that some of them are silly or misguided. To my mind, however, no one has stated in anything like a systematic way what it means for a dispute to be silly or misguided in the relevant way. This project is an attempt to state clearly the different ways a dispute can be defective or misguided and to give philosophers a method for determining which disputes ought to be set aside as pointless or without merit.

Chapter 2

INTRODUCTION TO PART ONE

Disputes tend to arise when two or more people disagree about something, are disposed to try to change each other’s mind, and think that it is, at least in principle, rational for them to do so. Of course, we are all familiar with disputes in which somebody is being pigheaded, willfully obtuse, insincere, or just plain disagreeable for the sake of getting a rise out of us. These kinds of disputes have an air of futility about them; they strike us as pointless, without merit, or otherwise silly.

There is another sort of way that disputes can be futile, pointless, or silly. This other sort of way essentially involves either a kind of miscommunication or mistake. Let’s consider miscommunication first. If I say, “I think it’s windy today” and you say, “No it is not, today is Thursday!”, then there is a sense in which any dispute over the matter would be silly. Our dispute wouldn’t be silly because one of us is being irrational, pigheaded, insincere, or disagreeable. In the above case, there is an unfortunate, and perhaps comical, misunderstanding. Unless cleared up, this misunderstanding renders subsequent dispute about the matter somehow defective.

Now let’s consider a case involving a mistake. Suppose that you and I are arguing about whether the person in the next room is wearing a red sweater or a blue one. We get involved in a heated dispute: you say that you saw someone enter the room in a blue sweater, and I say I saw someone enter the room in a red sweater. As it turns out, we’re both wrong; the room next door is empty. Again, there is a sense of futility about our dispute, but it does not have anything to do with miscommunication. I understood what you said perfectly well, and you understood me. We were just mistaken.

Part One of this project will deal with defective disputes. We will examine ways in which disputes can be defective either by involving a miscommunication or a mistake. I do not want to claim that the ways I shall enumerate and distinguish are the only ways that mistakes or miscommunications can render disputes defective, only that they are interesting, and indeed distinct, phenomena.

I said above that disputes tend to arise when people disagree. What is it to disagree with someone or something? This turns out to be a very challenging question. The first chapter of Part One is devoted to this question.

In Chapter Two I discuss some of the different ways that disputes can be defective. I argue that many philosophers have not sufficiently appreciated the distinctions that I draw between the different sorts of defective disputes.

Chapter Three is given over to a discussion of a particular kind of defective dispute: merely verbal disputes. Philosophers have had quite a lot to say on the subject of merely verbal disputes. Many philosophers have attempted to analyze the notion of a merely verbal dispute. In Chapter Three I argue that all the extant analyses I know of are incorrect.

Throughout Part One I make the following assumption. Defective disputes are natural language phenomena. That is, expressions like ‘merely verbal dispute’ should not be treated as philosophers’ terms of art. I assume that there is a phenomenon that philosophers have attempted to capture with the term ‘merely verbal dispute’. Non-philosophers have, and correctly diagnose themselves as having, merely verbal disputes even if they would not use the expression ‘merely verbal dispute’ to describe these disputes.

Philosophers, then, are not free to stipulate a meaning for ‘merely verbal dispute’ or ‘talking at cross purposes’ or any such term designed to capture a way that disputes can be defective. Should they choose to do so, they ought clearly to identify themselves as doing so, lest they engage in merely verbal or otherwise defective disputes about what counts as a merely verbal dispute or a defective dispute.

Chapter 3

disagreement

Disagreement is a pretty hot philosophical topic these days. Epistemologists are concerned with whether certain disagreements between so-called epistemic peers can be fully rational.[2] Contextualists and relativists about matters of taste are concerned with whether one view or the other can account properly for disagreements about matters of taste.[3] Lovers of ontology and critics of ontology are interested in whether or not certain ontological disagreements are misguided in some way.[4] While disagreement has become philosophically interesting of late, few philosophers, at least in the recent literature, have undertaken a serious exploration of just what disagreement is.[5]

What exactly is disagreement? Under what conditions is it appropriate to say that two or more people disagree? When do people count as having real or genuine disagreement rather than a merely apparent disagreement? How is disagreement related to agreement or consensus? I shall discuss these questions and related issues in this chapter. A characterization of disagreement that will suit our later discussion of disputes in general and of merely verbal disputes in particular will emerge from these considerations.

Let us begin with the seemingly trivial observation that disagreement is not merely the absence of agreement. The absence of agreement is not sufficient for disagreement, nor is the absence of disagreement sufficient for agreement. It would be incorrect to say of some ordinary first-grader, Sue, that she agrees with Brian Greene about string theory. Unless she is especially precocious, she does not know what string theory is and has no attitudes towards it one way or another. So, while Sue and Brian Greene do not agree about string theory, it does not follow that they disagree. Also, the speech, “Well, I don’t disagree with you, but I don’t agree with you either,” seems perfectly intelligible. This sort of speech seems appropriate in at least some cases where I haven’t made up my mind about the relevant topic.[6]

Does disagreement imply the absence of agreement? Is it possible for me to disagree with someone while being in complete agreement with them? This is a tricky question. At first blush, the following speech sounds odd: “I disagree with you, but it is not the case that I don’t agree with you.” While this speech sounds odd, it may still capture a genuine possibility. Compare the above speech to the Moore-paradoxical statement schema: “p, but I don’t believe that p.” Despite the oddness of uttering instances of the Moore-paradoxical statement schema, some instances of it are true.

Prima facie, the phenomenon of merely apparent disagreement would seem to provide cases of disagreements without the absence of agreement. But, whether or not such merely apparent disagreements count as disagreements is tricky. We shall discuss this issue below.

As above discussion suggests, saying that one does not agree often carries with it the implication that one disagrees, and saying that one disagrees often carries with it the implication that one does not agree. Take the following exchange:

Bill: I think that the moon landing was faked.

Jill: I’m sorry, but I just do not agree with you on that, Bill.

Most lookers-on to this exchange would probably assume that Jill disagrees with Bill. Compare this case with the following:

Bill: I think that the moon landing was faked.

Jill: I do not believe that, Bill.

Most lookers-on to this second exchange would take Jill to disbelieve that the moon landing was faked.

In some ways, belief and disagreement are similar. The absence of agreement is insufficient for disagreement; so too, the absence of belief is insufficient for disbelief. There is a further similarity: saying that one does not believe often carries with it the implication that one disbelieves. Consider: “I don’t believe that there are ghosts.” This speech, in the right circumstances, would create in my audience the impression that I disbelieve that there are ghosts. Similarly, “I don’t agree with Sam about whether there are ghosts” would, in the right circumstances, create in my audience the impression that I disagree with Sam.

These implications are a matter of pragmatics, not of semantics. This is to say that the proposition expressed by, “I do not believe that there are ghosts” does not imply that I disbelieve that there are ghosts. So too, the proposition expressed by “I do not agree with Sam that there are ghosts” does not imply that I disagree with Sam. While my audience would, in some cases, be entitled to infer that I disbelieve in ghosts or disagree with Sam upon my making the above speeches, this is no part of what the sentences I utter say.[7]

Similarities between belief and disagreement do not stop here. As far as all-or-nothing belief[8] goes, there are at least three distinct states I can be in with respect to some proposition p:

1. I believe that p.

2. I disbelieve that p (i.e., I believe that not-p)

3. I am indifferent to p (i.e., I neither believe that p nor disbelieve that p).[9]

Something similar applies to disagreement. Suppose Sam says, “There are ghosts”. One of the following is true of me:

1. I disagree with Sam.

2. I agree with Sam.

3. I neither agree nor disagree with Sam.

One theme that will emerge from this chapter is that disagreement and belief are interestingly related. One is tempted to say that whether Jill disagrees with Bill has only to do with what Jill and Bill believe. I shall argue below that this is not the whole story about disagreement.

Cappelen and Hawthorne observe that the verb ‘to agree’ has both a state and an activity sense.[10] John MacFarlane observes that the same is true of the verb ‘to disagree’: “When we characterize two people as disagreeing, we sometimes mean that they are having a disagreement—engaging in a kind of activity—and sometimes just that they disagree, which is a kind of state.”[11]

I think MacFarlane is correct. I believe that the moon landing was not faked. There are some moon landing conspiracy theorists who think the moon landing was faked. I am not aware of having met any such conspiracy theorists. In the state sense of ‘to disagree’, I disagree with all of these theorists, but there is no particular conspiracy theorist with whom I disagree in the activity sense.[12]

What is required in order for disagreement in the activity sense to take place? Is saying “I disagree with anyone who thinks the moon landing was faked” enough for me to count as disagreeing in the activity sense with each of the conspiracy theorists? I’m inclined to think that more than this is required, and I will spell out these requirements below.

That the state and the activity senses come apart is also evidenced by the phenomenon of merely apparent disagreement. What is merely apparent disagreement? Let me introduce the terms ‘apparent activity-disagreement’ and ‘apparent state-disagreement’. By ‘apparent activity-disagreement’ I’ll mean a pattern of verbal or other behavior that looks like (sounds like, etc.) activity disagreement. I introduce this locution in order that I shall have a way of talking about patterns of disagreement-like behavior without seeming to presuppose that those behavior patterns constitute real or genuine activity disagreement. By ‘apparent state-disagreement’ I’ll mean a case that seems like a case of state-disagreement. I introduce this locution so as not to presuppose that all seeming cases of state-disagreement are cases of real or genuine state-disagreement.

Following this usage, a merely apparent disagreement (state or activity) is an apparent disagreement the appearance of which is deceptive or misleading in some way. In cases of merely apparent disagreement, a rational person would be justified in taking there to be something going on that is not in fact going on. Some cases should make this clearer. Take the following case:

Will: I think human beings have free will.

Phil: I disagree; I do not think human beings have free will.

On the face of it, Will and Phil disagree in the activity sense. They needn’t disagree, however, in the state sense. Suppose their conversation continues as follows:

Will: Wait a moment, do you mean by ‘human beings have free will’ that human beings’ actions are not determined?

Phil: No; that’s not what I mean at all.

Will: Oh, because that’s what I mean.

Will and Phil, though they may have activity-disagreed, did not state-disagree.[13] But, they gave the appearance of state-disagreeing, so we will say they apparently state-disagreed. This appearance, however, is misleading. While a rational person who observed Will and Phil’s first speech would be justified in taking them to state-disagree, taking them to state-disagree would be a mistake.

What would a merely apparent activity-disagreement be like? Take the following case. Let Cell be a woman using an ear-bud assisted cellular telephone on a street corner, and let Belle be a woman standing next to her. Suppose that Cell’s hairdo conceals the ear-bud from Belle’s view. Now imagine the following speech:

Belle (to Cell): I think that it is going to rain today, what do you think?

Cell: No, I totally disagree.

Belle: Well, I think you’re wrong; just look at those clouds!

Onlookers who shared Belle’s inability to see Cell’s ear-bud might take Cell to be activity-disagreeing with Belle. But, unbeknownst to such onlookers, Cell didn’t hear Belle’s question at all, she was instead responding to a friend’s telephonic assertion of some non-meteorological claim.

Let’s further suppose that Cell thinks it will not rain. If she thinks this, then she state-disagrees with Belle. If this is the case, it seems appropriate to say that Cell’s apparent activity-disagreement with Belle was merely apparent, but that her state-disagreement is genuine.

As a further (perhaps more controversial) example, consider the behavior of some actors on a stage. The action goes like this

Thespian A: The world, I think, is flat!

Thespian B: No, I disagree! The world is not flat!

Barring any difficulties about whether the fictional characters these actors are portraying really do activity-disagree, one thing seems clear to me: the actors themselves do not activity disagree.[14] For all this, however, their performance looks just like activity disagreement. On my view, such performances are merely apparent activity-disagreements.

This observation seems true of acting in general; apparent activity-disagreements (and for that matter apparent agreements) on the stage, between actors in films, between people pretending to have arguments, and so on are merely apparent.[15] Contrast acting, however, with playing the devil’s advocate. When one plays the devil’s advocate, one behaves as though one adopts a position, even if one doesn’t hold that position.[16] I take it that devil’s advocates who engage in disagreement behavior are really activity-disagreeing with their interlocutors, though they needn’t be state-disagreeing. I will not hang too much on this view, so readers who object can just think of devil’s advocate cases as cases of merely apparent activity (and state)-disagreement.

I anticipate that my above description of the Cell-Belle case will cause discomfort on behalf of some readers. Why say that Cell apparently activity-disagrees with Belle at all? Why not just say there is no activity-disagreement, apparent or otherwise? In describing the above case the way I have, I betray a positive view about activity-disagreement. To disagree with someone in the activity sense is, on my view, to engage in a kind of behavior. It is to say things like “No, that’s not true” or “I disagree” or to write things like this down or to shake one’s head. It is to engage in such behavior, roughly, because one believes that the person at whom this behavior is directed holds some attitude or belief that one does not share. Of course, in order to count as really activity-disagreeing with some particular person, such disagreement behavior must be directed at them in some way. It is not enough to walk down the street muttering, “No, that can’t be! I disagree!” indiscriminately without a clear target in mind; the appropriate disagreement behavior must be directed at someone or something.

The above remarks should suffice to demonstrate that activity-disagreement and state-disagreement come completely apart. We shall examine below whether or not one of the activity sense or the state sense of ‘to disagree’ is conceptually more fundamental than the other. MacFarlane claims that “the state senses of disagreement are fundamental, and that any account of the activity sense will have to make reference to a state sense.”[17] This is not obvious to me. Given that activity-disagreement and state-disagreement come so neatly apart, I think we have a reason to suspect that one is not conceptually more fundamental than the other. I shall take up this question of fundamentality after having made some more preliminary remarks.

I want to make some comments on the logical grammar of the verb ‘to disagree’. Although the verb can appear grammatically without a direct object, it seems that some object of the verb is required in order to make sense out of claims to disagree. I can’t just plain disagree (though I can be just plain disagreeable). There has to be someone or something with whom or with which I disagree.[18] So much may seem painfully trivial. But what are the appropriate objects for disagreement?

So far, I have been talking mainly about interpersonal disagreement. That is, I have been treating disagreement (both state and activity-disagreement) as a relation that holds primarily between persons. I have also been using expressions like “I disagree with Bill”, “Susan disagrees with John” instead of expressions like “Bill and I disagree” or “Susan and John disagree”. This way of talking suggests that interpersonal disagreement is a non-symmetric relation that holds only between persons.

Is this correct? I think the answer depends on which sense of disagreement, state or activity, we happen to be talking about. Let’s start with activity-disagreement. Up until now, in connection with activity-disagreement I have been mainly discussing features of speech acts of disagreement and of performances of disagreement behavior. I now want to shift the focus momentarily to features of the speaker or performer. In order to disagree in the activity sense, something must be capable of disagreement-behavior. That is, it must be capable of saying things like “I disagree” or “That’s not the case” or of otherwise displaying a negative attitude towards some expressed content. Further, in order to activity-disagree, something must be capable of directing this behavior at a particular individual or expressed content. Also, in order to activity-disagree, something must be capable of believing, or otherwise holding, that some content has been either asserted or otherwise committed to.[19] I take it that persons, or sufficiently person-like creatures, are the only things capable of disagreeing in the activity sense.

Need both relata of the activity-disagreement relation be persons? I do not think so. I may vociferously activity-disagree with all sorts of things that are non-persons. For instance, all of the following seem intelligible when ‘disagree’ is taken in the activity sense:

I disagree with Bill’s assertion.

I disagree with this newspaper column.

I disagree with your attitude towards the art exhibit.

That activity disagreement can occur between a person and at least some non-persons suffices to demonstrate that activity disagreement is not a symmetric relation. For, while I activity-disagree with the claim that human beings have not landed on the moon, it is quite silly to say that this claim activity-disagrees with me. State-disagreement, however, intuitively seems to be a symmetric relation, though we shall question below whether in fact it is symmetric.

Activity-disagreement is non-symmetric even in the case where both relata are persons. I can activity-disagree with Bill when Bill does not activity-disagree with me and vice versa. Suppose Bill is standing on a street corner sincerely saying things like “Extra terrestrials rigged the presidential election!” I can say to the people around me, without Bill’s hearing me, “That’s hogwash.” In such a case, I activity-disagree with Bill, but it is arguable that Bill does not activity disagree with me, though he probably would were he to overhear what I said.

What are the suitable kinds of non-persons that are fit for the disagreement relation? It seems uncontroversial that statements and propositions are good candidates, as in “I disagree with the claim that no one has walked on the moon.” I am inclined to think that any sort of content capable of being expressed in a that clause is fit for the relation.

What about other sorts of things? The following, to my ear, is unacceptable:

I disagree with prawn vindaloo.

If ‘disagree’ here is given the state or the activity sense we’ve so far been talking about, this sentence seems pretty silly. There may be some sense of the verb ‘to disagree’ on which the above remark would be intelligible, though, on this sense it seems more appropriate to say, “Prawn vindaloo disagrees with me.” So, it would seem that nouns that do not refer to persons or to that-clauses are not suitable candidates for the disagreement relation. Is this right? Perhaps it is not.

The following do seem acceptable:

I disagree with foreign oil.

I disagree with conflict diamonds.

I disagree with Communism.

While these statements seem acceptable, they are acceptable, in either sense of disagreement, only if they are interpreted to mean, respectively (something like) the following.

I disagree with the practice of consuming foreign oil.

I disagree with the practice of purchasing conflict diamonds.

I disagree with something entailed by Communist doctrine.

It seems pretty clear that I can activity-disagree with a practice; I do so when I say things like, “No one ought to do that,” or “That’s a bad practice” or when I otherwise behave disapprovingly towards the practice or practitioners thereof. It also seems pretty clear that I can actitivity-disagree with some body of doctrine. I do so whenever I voice dissent from one of its implications or otherwise behave disapprovingly when I notice some implication of the doctrine being espoused, asserted, or discussed.

I can also state-disagree with a practice or body of doctrine. I do so whenever I disapprove of the practice or doctrine. One way to disapprove of a practice or a doctrine, say the practice of buying conflict diamonds, presumably, is to have a belief the content of which is (something like) what is expressed by

No one ought to purchase conflict diamonds.

Were one to have a belief with a content like this, one would count as state-disagreeing with the practice of purchasing conflict diamonds. One clearly state-disagrees with a doctrine, say Communism, if one has a belief the content of which is (something like) what is expressed by

Some claim entailed by Communist doctrine is false.

Were one to have a belief with such a content, one would count as state-disagreeing with an implication of Communist doctrine, and thereby count as state-disagreeing with Communism.

Perhaps one needn’t have a belief the content of which is that a practice or doctrine is worthy of disapproval in order to count as state-disagreeing with that practice or doctrine. We will come, shortly, to whether beliefs of a certain kind are required for state-disagreement.

So, disagreement, in either sense, is a relation that holds between a person and another person, a person and a proposition or statement, or a person and a practice or proposal. At this stage in the discussion, I anticipate the following philosophical move:

Wait just a moment. You say that activity-disagreement is a relation that holds between a person and another person, between a person and a proposition, or between a person and a practice or proposal. Surely one of these is conceptually more fundamental than the others? Can’t we think of activity-disagreement (and perhaps state-disagreement as well) as always a relation between a person and a proposition? Here’s how the analysis would go:

A person, S, activity-disagrees with a proposition p just in case they say that p is false, or otherwise behave as though it’s false (by shaking their head, etc. when someone asserts p). S state-disagrees with p just in case they believe that it is false. We can explain the other cases as follows. Whenever it sounds appropriate to say that S activity-disagrees with another person, S*, what’s really going on is that S says that some proposition believed or asserted by S* is false. Whenever it sounds appropriate to say that S state-disagrees with S*, what’s really going on is that S disbelieves some proposition that S* believes. Whenever it sounds appropriate to say that S activity-disagrees with some practice or proposal P, what’s really going on is that S asserts (or otherwise behaviorally endorses) the proposition that no one ought to adopt P. Whenever S state-disagrees with P, then S believes some proposition like the proposition that no one ought to adopt P. There, so the only thing that a person really disagrees with, in both the activity and state senses of ‘to disagree’, is a proposition.

This speech is fine so far as it goes, but a parallel move could be made by another theorist as follows:

Hold on now…you’re saying that a person disagreeing with a proposition is fundamental, but you’re making a mistake! Disagreeing, both in the activity and the state sense, is a relation that holds fundamentally between persons. Here’s how things really go. Whenever it sounds appropriate to say that S disagrees with p, what’s really going on is that S would disagree with any person, S*, who were to believe that p. When S disagrees with some proposal or practice P, what’s really going on is that S would disagree with any person, S*, who were to advocate P. At bottom, then, disagreement in either the activity or the state sense is fundamentally a relation between persons.

Let’s call a philosopher who makes the first speech McX and a philosopher who makes the second speech McY. Who is right? When I disagree with a person, do I disagree with them in virtue of disagreeing with some proposition they believe? When I disagree with a proposition do I disagree with it in virtue of being such that I would disagree with anyone who were to affirm it? I think both views are mistaken.

McX’s view is apparently a nonstarter. For take this case:

Bill: I like Smith.

Jill: I hate Smith.

Prima facie, Bill and Jill disagree. It is not clear, however, that there need be a proposition that the one affirms and the other denies.[20]

Bill and Jill might also disagree about a proposal as follows:

Bill: Let’s go to the circus.

Jill: No, I don’t want to do that.

In this case, there is no need for Bill and Jill to disagree about any proposition.[21]

Further, McX’s view can be taken to an extreme. If what McX says is correct, why not say that disagreement is fundamentally something that takes place, not between a person and a person, or even between a person and a proposition, but instead only between propositions? Here’s how the proposed analysis would go.

When S disagrees with some proposition p, this is not because S stands in some relation to p; what is really going on is that S believes some proposition or set of propositions G such that the union of G and {p} entails a contradiction or entails something that is otherwise inconsistent.[22] The proposition or propositions that S believes that are inconsistent with p stand in a relation to p, and it is in virtue of their standing in this relation that “S disagrees with p” is accurate. Something like this is suggested by McX’s proposal, but I think this view is mistaken for the following reason. Disagreement is something that persons or sufficiently person-like entities do. Of course we often say things like:

Jones’s report disagrees with Smith’s report.

Jones’s claim disagrees with Smith’s claim.

Smith’s claim disagrees with the facts.

These locutions, while felicitous, seem to involve some kind of personification.

I do not have a very strong argument that this is the case, but the following seems to be all the argument that is required. In order for x to disagree with y in the state or activity sense, x must be capable of having or avowing an attitude towards y. By ‘attitude’ here I mean some kind of mental state with a positive or negative valence. Call such an attitude a charged mental state. A proposition, report, or claim cannot be in such a charged mental state or avow such a mental state. It follows that these items cannot disagree in the state or activity sense.

Is there some other sense of ‘to disagree’ in which propositions can disagree? Propositions can be incompatible, logically inconsistent, or otherwise inconsistent with other propositions. I see no harm in calling this a kind of disagreement, but it is not the kind of disagreeing that persons do. Can interpersonal disagreement be analyzed in terms of this kind of disagreement, call it propositional-disagreement? We shall take up this question below.

McY’s view seems implausible as well. For whether or not I disagree with a proposition seems to have something to do with my attitude towards the proposition, not towards people who believe it. Also, McY’s view may seem to require possible persons. For, if McY is correct, then when I disagree with p, I stand in a relation to a person who agrees with p. But suppose there is no such person. Who am I related to? Some possible person who agrees with p? Proponents of McY’s view have some explaining to do here, where proponents of McX’s view can just point to the proposition as the relatum.

Activity-disagreement seems to be a non-symmetric, two place relation between a person and another person or between a person and a suitable non-person, like a proposal or proposition. What about state-disagreement? Like activity-disagreement, state disagreement can occur between persons and between a person and a suitable non-person. As with activity-disagreement, in order to state-disagree, one must be capable of occupying a charged mental state directed at an object: liking gruyere, disliking rock music, believing that there was a moon landing, hoping for rain, expecting a bear attack, etc. Below we will explore in some detail what sorts of charged mental states can give rise to state-disagreement. For now it suffices to say that it is a necessary condition for being in state-disagreement that a creature be capable of having charged mental states.

Interpersonal state-disagreement is different from interpersonal activity-disagreement in an important way. The former is a symmetric relation while the latter is not.

When person S disagrees with person S* in the state sense, it is typically appropriate to say “S and S* disagree.”[23] When S disagrees with S* in the activity sense, however, it is not always appropriate to say “S and S*” disagree. Perhaps this does not convince.

Here is, perhaps, a more convincing argument. When S activity-disagrees with S*

S is having a disagreement with S*.[24]

Now, ‘is having a disagreement with’ functions asymmetrically; the following is intelligible:

S is having a disagreement with S*, but S* is not having a disagreement with S.

Take a case: John McCain is speaking at a rally, and Bill vociferously objects to what McCain is saying. McCain proceeds with his speech, ignoring Bill’s remarks. It seems appropriate to say “Bill is having a disagreement with McCain, but McCain is not having a disagreement with Bill.”

Of course, in most ordinary circumstances when S is having a disagreement with S*, it will be appropriate to say “S and S* are having a disagreement,” though, as the McCain case shows, it will not always be appropriate to say this.

Contrast this with state-disagreement. When I state-disagree with Bill, the following is true:

GM has a disagreement with Bill.

The phrase ‘has a disagreement’, though, functions symmetrically. If S has a disagreement with S*, then S* (whether she knows it or not) has a disagreement with S. Of course, sometimes saying “Bill and I have a disagreement” carries with it the implication that Bill knows about the disagreement. In such cases it would be inappropriate to say that Bill has a disagreement with me if he doesn’t know about the disagreement. Despite the impropriety of saying that Bill has a disagreement with me in such cases, nevertheless, Bill does have a disagreement with me if I have one with him.

Let’s consider the McCain case again. While it is incorrect that McCain is having a disagreement with Bill[25], it seems correct that McCain has a disagreement with Bill. Of course, McCain may not be aware that he has a disagreement with Bill, indeed, in our example McCain needn’t know who Bill is.

So, when S activity-disagrees with S*, we will say:

S is having a disagreement with S*,

while, when S state-disagrees with S* we will say:

S has a disagreement with S*.

Now, in order to count as having a disagreement with someone, it is not necessary that I be currently engaging in disagreement behavior. Take, for example, the activity-disagreement that I am having with my neighbors about how loud they ought to play their music. I have been having this activity disagreement for quite some time, and though my neighbors are not currently playing music, and I am not pounding on the walls and shouting at them to turn it down, I still count as having an activity disagreement with them. When a case is like this we will say

S is having an ongoing disagreement with S*

When S is currently displaying disagreement behavior, we will say,

S is having an occurrent disagreement with S*.[26]

There is no occurrent-ongoing distinction to be made when it comes to state-disagreement. This is what we should expect. For S and S*’s having a state-disagreement is a matter of their occupying certain charged mental states, not of their engaging or being disposed to engage in disagreement behavior. Most charged mental states, like believing, hoping, wishing for, denying, rejecting, etc. are dispositional in character.

So, state disagreement can occur between a person and another person and between a person and a proposition or other suitable content. When state disagreement occurs between a person and another person, the relation is symmetric. If I state-disagree with you, then you state-disagree with me. Our analysis of state-disagreement will have to respect both of these facts about the logical grammar of state-disagreement.

Trouble looms for our account so far of the logical grammar of the verb ‘to disagree’. So far we have allowed that one of the relata in the disagreement relation, in both the state and activity senses, can be a suitable non-person. This may cause some trouble for the following reason.

When I disagree, in either the state or activity sense, with a proposition, it is inappropriate to say that there is something I disagree with the proposition about. Consider:

*I disagree with the claim that the Earth is flat about its truth.

This sounds peculiar. On the other hand, when I disagree in either the state or activity sense with a person there is always something I am disagreeing about. For example:

I disagree with Jones about whether the Earth is flat.

When I claim to disagree with a proposition, it is never appropriate to ask “What about?”[27] When I claim to disagree with Bill, however, it is typically appropriate to ask “What about?” unless context makes it very clear what my disagreement with Bill is about.

This asymmetry suggests that there are different relations involved in the two sorts of case. When I disagree with a person (in either the state or activity sense) let’s call the disagreement an interpersonal disagreement. When I disagree with a suitable non-person, let’s call the disagreement a content disagreement. Four distinct relations emerge:

Interpersonal activity-disagreement

Interpersonal state-disagreement

Content activity-disagreement

Content state-disagreement

The logical grammar of content-disagreement differs from that of interpersonal disagreement. As the above remarks suggest, interpersonal disagreement is best thought of as a three-place relation between a person, a person, and a suitable object of disagreement (e.g., a proposition, proposal, doctrine, etc). Content-disagreement, on the other hand, is best thought of as a two-place, asymmetric relation between a person and a suitable content.

There are four disagreement relations one can stand in. In what follows, I plan to focus mainly on interpersonal state-disagreement, though I shall have some things to say about the other relations. In this section, I want to take up a worry about the logical structure of interpersonal state disagreement that we have developed so far. Discussion of this worry will help to sharpen the notion of interpersonal state-disagreement preparatory to an analysis of it.

MacFarlane limns the logical structure of interpersonal state-disagreement as follows:

x disagrees with y’s fing in context c.

where f can be replaced by a verb phrase describing an attitude—for example, believe that Mary is smart, or hate the taste of grape jelly.[28]

As we have parsed interpersonal state-disagreement, however, we have left out context: it is a relation that holds between a person, another person, and some content. Is a context required? It may seem so. Take the case of John and Mary:

John: I’m tall.

Mary: John is not tall.

John is talking to some friends in his philosophy department; Mary is talking to some friends at a basketball game. In the context of John’s conversation, the appropriate comparison class for the gradable adjective ‘tall’ consists of John’s philosophy colleagues, most of who are of average height. In the context of Mary’s conversation, however, the appropriate comparison class consists of professional basketball players. Do John and Mary disagree? Intuitively, no. What John says is true at his context of utterance, and what Mary says is true at her context of utterance.

Let’s call John’s context of utterance c_j and Mary’s c_m. The following seems to be the case. John does not disagree with Mary’s having uttered her claim in c_m and vice versa. Contrast this with another case in which John and Mary are talking to each other at a party:

John: I’m tall.

Mary: No you’re not; there are plenty of people here way taller than you.

Now it looks like John and Mary disagree. John has asserted that some content is true at his context of utterance and Mary has denied this.

So, it seems that a context of utterance or evaluation has to get into the mix somehow. Our explication of the logical grammar of interpersonal state disagreement, then, will be somewhat similar to MacFarlane’s. We will say that interpersonal state-disagreement is a four-place relation that holds between a person, a person, a suitable content, and a context. MacFarlane’s precise gloss on the logical structure will not work for our purposes, however. For, as MacFarlane describes the logical structure of state-disagreement it is a relation that holds between a person and a speech act, performance, or belief state (y’s fing in c) and a context. But this loses the sense in which interpersonal disagreement is a relation that holds between persons. When I disagree in the interpersonal state sense with Bill, I disagree with Bill, not with some attitude or behavior of Bill’s.

MacFarlane’s gloss of the logical grammar of disagreement, though, seems to work just fine for content state-disagreement. When I disagree with a proposition, I typically disagree with anybody’s affirming that proposition in some context or other. As with interpersonal state-disagreement, context must be specified in cases of content state-disagreement. Take this case. Suppose I disagree with the proposition expressed by

John McCain is President of the USA.

Of course, I disagree with this proposition as evaluated at this world. McCain might have won the election, however. Take a world w_j where McCain did win. I do not disagree with the proposition that McCain is President when it is evaluated with respect to w_j. When I disagree with some proposition p in some context c, then, we will say that I disagree with anybody’s affirming p at c.

In cases of activity disagreement (both interpersonal and content), the context is pretty obvious; it is just the context in which the disagreement occurs (unless the conversation somehow affects a shift in context, as in cases of talking about counterfactual possibilities). When we specify the relata of the activity disagreement relation, then, we will often leave off explicit mention of a context of disagreement.

Now that we are clear on the logical grammar of disagreement, we shall attempt to state conditions for disagreement. When does disagreement occur? What is required for disagreement to occur? We have already looked at some basic requirements: S can disagree with S* (in the interpersonal state-sense) only if S and S* are suitably person like creatures capable of occupying charged mental states. What else is needed for disagreement? Here I shall focus on interpersonal state-disagreement.

Perhaps the most obvious analysis of interpersonal state-disagreement (hereafter just ‘disagreement’) is the following:

Affirm-Deny: S and S* disagree iff there is a proposition p such that S affirms p and S* denies p.

By ‘affirms p’ here one means ‘believes that p’ and by ‘denies p’ one means ‘believes that not-p’. When S affirms p and S* denies p, we will say that S and S* have a difference of opinion with respect to p.

Something like Affirm-Deny is operative in a number of discussions about disagreement. It is what Richard Feldman has in mind when he discusses whether peer disagreements are ever rational: “In saying that there were disagreements among the students I am saying only that there were propositions that some of them affirmed and some of them denied.”[29]

Affirm-Deny is inadequate for a number of reasons, some of which we considered above, but which will bear reexamining now. Some cases of disagreement, while they are properly characterized as disagreement about a proposition p, do not involve one party believing p and the other party disbelieving p (i.e., believing not-p). I have three types of cases in mind (though there may be more).

Case 1: Belief and Rejection

Take the case of Peg and Neg:

Peg: Bill’s bald.

Neg: That doesn’t seem right.

Peg: Oh, so you believe it is not the case that Bill is bald?

Neg: No, that’s not what I said. I reject the proposition that Bill is bald, but I do not do so in virtue of believing its negation.

I assume that the above discussion is intelligible where Bill is a borderline case of baldness. In such a discussion, no one needs to be confused about what they believe or disbelieve.[30] I think it makes sense to characterize Peg and Neg as disagreeing about the proposition that Bill is bald. But, it looks like it is not the case that Peg affirms the proposition and Neg denies it.

Perhaps it will here be objected that while Neg does not deny the proposition that Bill is bald, there is some proposition that Neg denies which Peg affirms. Perhaps this proposition is (something like) what is expressed by

The proposition that Bill is bald has a determinate truth value.

Does Neg need to deny something like this? I do not think so. Neg might be an epistemicist about the predicate ‘is bald’. So, Neg might think that the proposition that Bill is bald must have a determinate truth value, but be unsure that he knows what that truth value is.

But, suppose that Neg is an epistemicist, doesn’t he believe something like:

Peg doesn’t know that Bill is bald.

Presumably, Peg denies this. I don’t think this is required. I have cast the above discussion as though both participants are of a philosophical bent. But we can imagine a similar case involving non-philosophers as follows:

Peggy: Bill is bald.

Neggy: I disagree, I’m uncomfortable saying that Bill is bald.

Peggy: Oh, so you think it’s false that Bill is bald?

Neggy: No, I don’t; I’m just uncomfortable saying that he is bald.

Given that Neggy does not have a positive belief about why it seems wrong to say that Bill is bald, it looks like there needn’t be some proposition p such that Peggy believes p and Neggy believes not-p, but nonetheless, Peggy and Neggy disagree.[31]

Case 2: Belief and Agnosticism

Another sort of case of disagreement without difference of opinion is provided by cases in which one party believes that p and another party is agnostic with respect to p. Take the case of Theist and Not-sure:

Theist: God exists.

Not-sure: I’m not convinced either that God exists or that God doesn’t exist. I’ve made
up my mind to suspend judgment on the issue.

Intuitively, Theist and Not-sure disagree. Note that there is a difference between Not-sure’s attitude towards the proposition that God exists and the attitude of three-year old Sally who has never even considered the proposition. Sally is indifferent to the proposition while Not-sure isn’t. Not-sure’s attitude is the result of serious and thoughtful consideration about the proposition, not merely the result of inattention to it.

Is there a proposition such that Theist affirms it and Not-sure denies it? Perhaps there is. Maybe a proposition like the one expressed by:

It is possible to know whether God exists.

Perhaps agnostics about God’s existence typically deny a proposition like this while theists believe it. But does this have to be the case? I am inclined to think not.

Case 3: Difference of Attitude

Perhaps the above two cases are controversial. This case seems less so. Take the case of Like and Dislike:

Like: I like Bill.

Dislike: I do not like Bill.

I think it is appropriate to say that Like and Dislike disagree. They disagree about Bill. Need there be some proposition that Like affirms and Dislike denies? I do not think so.[32] Take another case:

Cinema: I want to go to the cinema tonight.

Park: No, I do not want to do that; I want to go to the park instead.

Cinema and Park disagree. But, it seems that there needn’t be any proposition that they disagree about. They have conflicting desires, and maybe desires ought to be analyzed in terms of an attitude towards a proposition like the proposition that we go to the cinema tonight, but it doesn’t need to be the case that Cinema believes such a proposition and Park disbelieves it. In fact, suppose that Park urges Cinema to provide a reason for preferring the cinema to the park, “Why do you want to go to the cinema instead?” Cinema answers, “I just feel like it!” In such a case, it looks like Park and Cinema can agree on every proposition, but still disagree about where to go.[33]

Affirm-Deny, therefore, ought to be rejected. What analysis should take its place? The following seems to get us past Case 1 and Case 2:

Accept-Reject: S and S* disagree iff there is a proposition that S accepts and that S* rejects.[34]

Here by ‘accepts p’ I have in mind ‘believes that p’ and by ‘rejects p’ I have in mind ‘either believes not-p, suspends judgment on p, or thinks p does not have a truth value’. [35] By ‘suspension of judgment’ I mean that one has thought about p and, for whatever reason, has come to the opinion that both belief that p or belief that not-p are somehow inappropriate.

Accept-Reject, however, has a hard time handling Case 3. The problem here is that in cases like Case 3, the attitudes that support disagreement are not directed at propositions at all.[36] In the case of Like and Dislike, they are directed towards a person; in the case of Cinema and Park they are directed at proposals (let’s go to the cinema vs. let’s go to the park). Accept-Reject, therefore, is incapable of handling all of those cases we intuitively want to call disagreement.

Part of the problem with the above analyses is that there are many sorts of phenomenon we are willing to call disagreement. Following MacFarlane, I will talk about types of or kinds of disagreement. This talk about kinds of disagreement is not meant to imply that each different kind of disagreement corresponds to a distinct disagreement relation. Rather, we will say that each kind of disagreement is a distinct way that the interpersonal state-disagreement relation can hold.

In this section, I want to discuss the types of disagreement MacFarlane identifies.[37] MacFarlane identifies three main types of conditions that seem sufficient for disagreement to occur: non-cotenability, preclusion of joint satisfaction, and preclusion of joint accuracy.[38] We shall discuss each in turn.

Non-cotenability

People will be said to disagree when they have incommensurable attitudes. As MacFarlane puts it, I can be said to disagree with someone when they adopt some attitude and I “could not coherently adopt that same attitude—an attitude with the same content and force—without changing my mind, that is, without dropping some of my current attitudes.”[39] Let’s formulate this as follows:

Non-cotenability: S and S* disagree if S adopts some attitude a and S adopts some attitude a* such that S could not coherently adopt a* without giving up one or more of S’s attitudes and S* could coherently not adopt a without giving up one or more of S*’s attitudes.

By ‘attitude’ here I mean charged mental state. When I am happy, cantankerous, or suicidal I have a certain attitude. But, I don’t (need to) disagree with folks who are unhappy, good natured, or who have a lust for life, even though these attitudes are non-cotenable with mine. By ‘coherently’ here I mean coherent in a broad sense that covers logical consistency as well as pragmatic or practical consistency. When I desire to eat a ham and I desire not to eat that ham, I am being incoherent in some way, but it is not clear that I am guilty of believing a contradiction or something that entails a contradiction.

We must sharpen this sufficient condition on disagreement a bit to avoid an obvious counterexample. Take the case of Larry and Carrie:

Larry: I think that two plus three equals six.

Carrie: Me too! That’s exactly what I think!

Now, it seems clear that Carrie and Larry do not disagree. But, Carrie cannot consistently adopt Bill’s belief, and vice versa, without dropping one of her own, namely the belief that two plus three equals six. We should understand that a and a* in the above formulation do not pick out one and the same attitude.

This fix, however, doesn’t exactly work. Consider the following case. Suppose that Chuck holds the view that two and two are six and William holds the view that there is a round square. Suppose that Chuck and William share all the same attitudes otherwise. So, the only difference between the two is that William believes a necessary falsehood about addition but does not hold any necessarily false or inconsistent beliefs about geometric objects, while Chuck holds a necessarily false belief about geometry but does not hold any necessarily false or inconsistent beliefs about arithmetic. Intuitively, Chuck and William do not disagree about anything. But, they hold attitudes that are non-cotenable. Chuck could not consistently take on William’s attitude and vice versa. Indeed, this will be the case any time one or more disputants holds any attitude that is itself inconsistent or any set of attitudes that are jointly inconsistent.

Of course, one way to fix things here would be to exclude cases in which one or more of the attitudes in question are necessarily incorrect. As MacFarlane observes, however, “then we would exclude too much: in noncontingent domains like mathematics, disagreement always involves an acceptance or rejection that is necessarily inaccurate.”[40]

Intuitively, Non-cotenability gestures at a sense in which people can be said to disagree, so we oughtn’t to abandon it entirely. How, though, should we fix the above sufficient condition on disagreement? We want to allow for people who have inconsistent attitudes about unrelated matters to agree about something or other. So, suppose that I have the inconsistent belief that Obama is and is not President of the USA. On the above account, I disagree with anybody about anything! For, no matter what the attitude in question is, I can’t consistently take it on board without changing my mind, because I already believe something inconsistent. We don’t want to rule out cases where one or more parties to a disagreement believe something inconsistent, for most of us probably do hold beliefs that are inconsistent, and we still manage to disagree.

I suggest recasting the sufficient condition in terms of relevant attitudes. The new condition goes as follows:

Non-cotenability*: S and S* disagree if S adopts some attitude a and S adopts some distinct attitude a* such that S could not coherently adopt a* without giving up one or more of S’s relevant attitudes and S* could not coherently adopt a without giving up one or more of S*’s relevant attitudes.

What makes an attitude relevant to a disagreement? This is a very hard question. I can give some clear cases of relevant attitudes and some clear cases of irrelevant attitudes, but I cannot specify jointly necessary and sufficient conditions for relevance. The following should suffice to demonstrate what I have in mind. Suppose that Jim thinks Obama was not legally elected President. Tim thinks Obama was legally elected President. If Tim holds the belief that grape jelly is typically made of puppies, this is presumably an attitude that is irrelevant to his disagreeing with Jim about the Obama election. If, however, Tim believes that Obama was born on American soil, this attitude is presumably relevant to his disagreement with Jim. That we can understand and classify the clear cases indicates that some sense of relevance is operative in our assessments. In what follows, I will understand ‘relevance’ to pick out whatever (perhaps vaguely understood and semantically indeterminate) notion of relevance is operative in these assessments.

No doubt a clever philosopher can contrive to counterexample Non-cotenability*, but it should suffice for our purposes. Non-cotenability* (hereafter, just non-cotenability) can account for the disagreement between Like and Dislike about Bill. Like cannot, on pain of pragmatic or practical inconsistency, both like and not like Bill, nor can Dislike.[41] Non-cotenability also captures the belief and agnosticism case. For, while it is not the case that the believer affirms the proposition that God exists and the agnostic denies this proposition, their attitudes are non-cotenable. Theist cannot consistently take up a suspension of judgment about God’s existence without dropping his belief that God exists, nor can Not-sure consistently adopt the belief that God exists without abandoning his suspension of judgment on the matter.

Non-cotenability, therefore, can account for many cases we intuitively want to call disagreement. But, non-cotenability is not necessary for disagreement. Take the case of Amy and Ari who both want to eat the same cupcake. Both share the same attitude: wanting to eat the cupcake.[42] Amy and Ari intuitively disagree, but their attitudes are cotenable, indeed, given our attempt to rule out cases where a and a* are identical, Non-cotenability does not even apply in this case.

Preclusion of Joint Satisfaction

Amy and Ari disagree in the following sense: it is not the case that their attitudes can be jointly satisfied. This gives us another sense in which persons can be said to disagree:

Preclusion of Joint Satisfaction : S and S* disagree if S holds some attitude a and S* holds some attitude a* and it is not the case that a and a* can be jointly satisfied.

Preclusion of joint satisfaction and Non-cotenability come apart. The case of Amy and Ari is a case of preclusion of joint satisfaction but not a case of non-cotenability (at least, given a certain view of the content of desires). The following case is a case of non-cotenability but not a case of preclusion of joint satisfaction. Willard wants to eat just the frosting of a particular cupcake; Dillard wants just the cake. Their attitudes are non-cotenable, for it is inconsistent to want just the icing and just the cake, but they are jointly satisfiable, for both Willard and Dillard can each get what he wants.[43]

A problem analogous to the problem of necessarily incorrect belief arises here. Necessarily unsatisfiable attitudes make trouble for Preclusion of joint satisfaction. Suppose that Mary wants to eat a cupcake and not eat that cupcake. Her attitude cannot be satisfied. So, if Preclusion of joint satisfaction is a sufficient condition for disagreement, Mary disagrees with anybody, say Larry, about everything. This is not a welcome result. People with incoherent attitudes can still disagree, however, so it is not clear what to do about cases of unsatisfiable attitudes. Preclusion of joint satisfaction, however, yields intuitively correct results for many cases, so it should not be abandoned. I do not, however, have a good fix for the problem.

Preclusion of Joint Accuracy

Intuitively, people disagree when they cannot both be correct. Following MacFarlane, we will say that one’s acceptance or rejection of a proposition is ‘accurate’ just in case “the proposition accepted is true (false) at the circumstance of evaluation that is relevant to the assessment of the acceptance (rejection) in its context (or at all such circumstances if there is more than one).”[44] For one’s non-propositional attitude to be accurate is for it to be veridical at the relevant circumstance of evaluation.[45]

So, I disagree with you if I hold an attitude or attitudes that preclude the accuracy of one or more of your attitudes. The sense of ‘preclude’ here is somewhat tricky. MacFarlane has the following to say of it:

I am not going to try to spell out more precisely what I mean by “preclude”; instead, I’ll rely on an intuitive grasp. Certainly “preclude” can’t be spelled out modally: saying that it is impossible for A and B both to be accurate falls short of saying that the accuracy of A precludes the accuracy of B, because it may be that A and B cannot both be accurate because B cannot be accurate, quite independently of any relation to A. It is difficult to say what preclusion amounts to in other terms, but I think we have a tolerable grasp of the notion (otherwise we would not be so confident about the counterexamples we can easily construct to various modal explications).[46]

Many cases of disagreement are cases of preclusion of joint accuracy. When I believe that a human being walked on the moon and Jones believes that the moon landing was faked, the accuracy of my belief precludes the accuracy of Jones’s and vice versa. Preclusion of joint accuracy captures the commonsense notion that in cases of disagreement both parties cannot be right.

MacFarlane observes that there are two sorts of preclusion of joint accuracy. In order to bring out the difference between these two kinds of preclusion, I will avail myself of the notion of a centered proposition or a de se proposition. Suppose that Mary S. is suffering from amnesia and does not believe that she is Mary S. Mary might still believe a number of things about herself: that she is located in Arizona, that she is in front of a bank, that she is eating a sandwich, and so on. The content of her beliefs cannot be, it would seem, that Mary S. is thus and so, for she does not believe that she is Mary S.

Let us say that Mary believes the centered propositions

That I am in Arizona,

That I am in front of a bank,

and

That I am eating a sandwich.

Larry, however, believes the centered proposition I am not eating a sandwich. There is a sense in which Larry and Mary can both be right. Their beliefs are accurate. At the world, person (and perhaps time) tuple at which Mary has her belief, it is accurate and vice versa for Larry.

Following MacFarlane, when we relativise the notion of accuracy to a context, there are “two things we can mean by ‘preclusion of joint accuracy’.”[47] MacFarlane states these as follows:

“1. The accuracy of my attitudes (as assessed from any context) precludes the accuracy of your attitude (as assessed from that same context).

2. The accuracy of my attitudes (as assessed from my context) precludes the accuracy of your attitude (as assessed from your context).”[48]

We shall follow MacFarlane in calling (1) preclusion of joint accuracy and (2) preclusion of joint reflexive accuracy. So, while Mary’s attitude precludes the accuracy of Larry’s attitude when assessed from one and the same context, their attitudes do not jointly reflexively preclude accuracy.

10.

I want now to return to the question of conceptual fundamentality. MacFarlane takes the state sense of disagreement to be conceptually prior to the activity sense. I will here offer a qualified agreement with this claim.

Disagreement discourse, like just about any kind of discourse, can be bent to a variety of uses. This is to say that one can use the words ‘I disagree’ in order to perform a number of speech acts. Perhaps the most basic speech act associated with ‘I disagree’ and related expressions is the making of an assertion. In many cases, when I say that I disagree with someone or something, I am asserting that I am in a particular state, namely, I am claiming that I state-disagree with that person or thing. If this were the only use of the words ‘I disagree’, then I would unqualifiedly agree with MacFarlane that the state sense of disagreement is conceptually prior to the activity sense.

Sometimes, however, disagreement discourse is used for a different purpose altogether. The expression ‘I disagree’ has what Austin called a performative use. Saying these words sometimes does not amount to asserting that one state-disagrees, but instead amounts to an attempt to pull off a certain speech act whose illocutionary point is to engage one’s interlocutor in a dispute.

I shall have quite a bit to say about disputes in subsequent chapters. For now a simple gloss will suffice. A engages B in a dispute (and B reciprocates) about subject matter M just in case A gives reasons for having some attitude about M and B gives reasons for having some other attitude about M. In short, to engage in a dispute is to have an argument, in the philosopher’s sense of the term. It is to exchange reasons for one’s views with another party.

When I say “I disagree”, there are at least two performances I could be making. One of these performances is just to disagree in the activity sense. Saying ‘I disagree’ in the right conditions just is disagreeing in the activity sense. Let us call this use of ‘I disagree’ the assertive use. Sometimes, though, when I say “I disagree” I mean to be challenging my interlocutor, intending to provoke a defense of her position, to instigate a dispute. When I use the locution “I disagree” in this way, we will call my use of it provocative. The assertive use and the provocative use correspond to different speech acts, respectively: activity-disagreeing and inviting a dispute. These speech acts have different felicity conditions and can misfire in different ways.

Let us take up the assertive use first. Like any speech act of assertion, the act misfires or fails if I do not mean what I say. Here the notion of meaning is perhaps obscure, so an example will help to clarify it. Suppose that Bill says, “I think it is going to rain tomorrow,” and I, testing out a new microphone, say “I disagree.” In such circumstances I would not count as having asserted that I state disagree with Bill about the prospects for rain. So much is familiar from discussion of the speech act of assertion. I also fail to disagree if the target of disagreement is inappropriate. The thing with which I disagree must be a person or a propositional content or a proposal or some other such suitable thing.

The assertive use can also be infelicitously performed. When this is the case, although the illocutionary act is a success, it is deformed in some way. The simplest case of such infelicity is a case of insincerity. Suppose Bill says, “I think it will rain tomorrow.” Wanting to be disagreeable, I say “No, that’s not the case. I disagree.” I succeed in asserting that I state-disagree with Bill, but if I do not hold the requisite meteorological belief, then my disagreement is insincere.

Infelicity occurs in a different sort of case. Suppose that Bill says “It will rain tomorrow,” but he happens to be talking about the meteorological conditions in Portugal. Were I to say, “Nonsense, I disagree,” taking Bill to be making a claim about Seattle, my speech act is infelicitous. Despite my sincerity, there is something amiss with my saying that I disagree with Bill.

Let us contrast the assertive use with the provocative use. The provocative use is intended to invite dispute. Like the assertive use, the provocative use misfires if I do not mean what I say. Actors on a stage fail of performing the illocutionary act of challenging each other, drunkards walking down the street muttering “I disagree” also fail (provided they do so indiscriminately, without a clear target in mind). In the provocative use, I also fail if I, alone in my apartment, I say to the television screen during a political speech “I disagree!”

The provocative use is infelicitous when I do not intend to carry through on having a dispute. Suppose I say to a friend who just uttered p, “No! I disagree,” without any intention of giving reasons for my disagreement. In such a case, I infelicitously provoke a dispute. My friend may begin to defend her claim that p, taking me to be inviting her to do so. But, if I do not take up my part in the dispute by answering her arguments or providing my own, I have not performed the illocutionary act in good faith.

Notice, it is no part of the success or felicity conditions for the provocative use that I state-disagree with my interlocutor. We can have a (quite successful) dispute without our state-disagreeing about anything. Any philosophy professor who has advocated naïve act utilitarianism for the sake of a good classroom discussion knows this. Devil’s advocate cases can lead to productive disputes. What is required for felicity, however, is that I take the dispute to be, in principle, resolvable. Suppose my friend says, “I think that the number of atoms in the Universe is odd,” and I respond, “Rubbish! I disagree.” If my uttering this sentence is intended as a provocative use of ‘I disagree’, I have not performed the illocutionary act in good faith (provided I know how absurd it would be to try to resolve this dispute). I shall have quite a bit more to say about infelicity of disputes below.

In order for my provocative act of disagreement to be felicitous, I must at least take my interlocutor to commit to some attitude. The sense of commitment here must be somewhat broad, in order that the devil’s advocate cases will count as cases of felicitous provocation. What do I mean by ‘commitment’ in this connection? We will say that S commits to an attitude just in case S sincerely holds that attitude or provisionally adopts that attitude. There are many ways to adopt an attitude provisionally. I do not have an analysis of provisional adoption in terms of necessary and sufficient conditions, but perhaps some examples will serve to explain what I mean by this term. One provisionally adopts an attitude when one takes up a position for the sake of argument, makes a conjecture or suggestion, argues on behalf of something or someone, or any such thing. One does not provisionally take up an attitude merely by supposing it, however. When I make a supposition, I take some proposition to be true, in most cases, to see what follows from it. My audience cannot hold me responsible for committing to the proposition, in our sense, when I do this.

There is another illocutionary act I want to talk about. Using the words ‘I disagree’ is also a way of gainsaying. If I do not intend to provide a reason for disagreeing with someone, nevertheless I can succeed in felicitously gainsaying. In such a case, I merely voice dissent. To do so is to activity-disagree without the intention of developing the activity-disagreement into a dispute.

Specifying precisely all of the different speech acts one can perform with ‘I disagree’ together with their success and felicity conditions would be quite a task. It is not our task here. My crude handling of these speech acts, however, should suffice for our purposes.

So, when ‘I disagree’ is intended in the assertive use, state disagreement is indeed conceptually prior. This is because one of the felicity conditions for the assertive use is that the speaker state-disagree with her interlocutor. In the provocative use, however, state-disagreement does not enter the picture. Now, typically, speakers will state-disagree when they have disputes, but the case of the devil’s advocate and the philosophy professor should suffice to show that this needn’t be the case.

11.

When I state-disagree with someone, this so in virtue of our being in charged mental states that satisfy one of the four conditions to which MacFarlane draws our attention: non-cotenability, preclusion of joint satisfaction, preclusion of joint accuracy, or preclusion of joint reflexive accuracy. Does any charged mental state count? I am inclined to say yes. Disagreements can arise when parties take up any of the following attitudes: liking someone or something, believing something, conjecturing something, hoping that something will occur, wishing for something, desiring something, and so on.

There are, perhaps a few cases that are not amenable to disagreement. Suppose I fear spiders and you do not fear spiders. Do we disagree? This is tricky. Some philosophers will tell us that when I fear spiders I represent them as being dangerous.[49] If you do not fear spiders, you do not represent them as dangerous. If something like this is correct, then when I fear spiders I disagree with you if you do not.

Of course, not every way of representing the world can give rise to disagreements. Take a case. Smith is colorblind. His visual system does not represent a ripe tomato as red. Jones is not colorblind. Her visual system does represent a ripe tomato as red. Do Smith and Jones state-disagree when they view a ripe tomato? They needn’t. For while Smith’s visual system represents in a particular way, Smith needn’t represent that way.

If our emotional states involve some kind of representing of the world that has correctness conditions, that is, if our emotional states involve commitments to some content’s being veridical, then emotional states can give rise to disagreements (at least of the non-cotenability sort).

Our focus will be mainly on cases of state-disagreement with respect to some proposition or other and activity disagreements purportedly about the truth of some proposition or other. Belief, disbelief, and suspension of judgment will therefore be the relevant attitudes. It will do us well to remember, however, that disagreement may arise in connection with other charged mental states.

12.

So, as we will analyze state-disagreement, S disagrees with S* just in case S has some attitude a and S* has some (distinct) attitude a* and:

(i) S and S*s attitudes are non-cotenable,

(ii) S and S*’s attitudes preclude joint satisfaction,

(iii) S and S*’s attitudes preclude joint accuracy,

(iv) S and S*’s attitudes preclude joint reflexive accuracy.

No doubt a clever philosopher could probably contrive to find an example of a case in which we intuitively want to say that S and S* disagree, but in which none of (i)-(iv) are met. At this point, at least, I cannot. I therefore advance the above analysis as a tentative hypothesis the refutation of which I allow as an epistemic possibility. This analysis, however, will serve as our working theory about disagreement for the remainder of our project.

13.

Now that we have an analysis of what it is to disagree in the state sense, we can characterize the speech act or activity sense of disagreement more carefully. When I say “I disagree” sometimes I’m not just stating a fact, but I am doing something, namely disagreeing in the activity sense. I shall state two kinds of conditions for this speech act: constitutive conditions and non-defectiveness conditions.[50] The constitutive conditions must be met in order to the speech act of disagreement to have occurred; the non-defectiveness conditions must be met in order for the speech act to be non-defective, or as Austin would say, felicitous.

Let us start with the constitutive conditions. I have alluded to some of these above. S activity-disagrees (i.e. performs the speech act of disagreement) with S* if, and only if:

(i) S is capable of having charged mental states and engaging in disagreement behavior,

(ii) S directs some bit of disagreement behavior at S*,

(iii) S takes S* to be committed to some attitude that S intends to single out by means of
the disagreement behavior mentioned in (ii).

(iv) S* is capable of having charged mental states.

If any of (i)-(iv) are not met, S does not activity disagree with S*. Condition (i) is intended to rule out non-persons or non-personlike creatures from having activity disagreements. If I program a machine to emit sounds that resemble “I disagree”, intuitively it does not activity disagree with anyone. Condition (ii) is intended to rule out cases where S indiscriminately says “I disagree” or says “I disagree” just to test out a microphone. In order for activity disagreement with S* to occur, disagreement behavior must be directed at S*. Recall the Cell-Belle case from above. Although Cell says “I disagree”, her remark is not directed at Belle. So, Cell does not activity disagree with Belle. Condition (iii) is meant to rule out cases where I shout “I disagree” at Smith, but Smith hasn’t said anything for me to disagree with. I direct my comment at Smith, but if I don’t take him to hold an attitude with which I can disagree, then my remark does not succeed; I fail to have a disagreement with Smith. Condition (iv) is intended to rule out cases where I have an activity disagreement with a tomato or other inanimate object. When all of (i)-(iv) are met, we will say that S is having a disagreement with S*.

Activity disagreement can go wrong, however. We will say that an activity disagreement is defective when the following condition is not met. S’s activity disagreement with S* is non-defective just in case:

(a) S state-disagrees with S* about the purported topic of the activity-disagreement.

The provocative use has similar constitutive conditions, each of (i)-(iv) is required for the provocative use, but in addition the following is required:

(v) S* notices S’s disagreement behavior and takes it as disagreement behavior directed
at S*.

I do not succeed in challenging you to a dispute if you are oblivious to my remark. Condition (v) is supposed to rule out cases where I disagree with John McCain by shouting at my television. In such cases surely I do not challenge McCain to a dispute. A challenge is defective if any of the following are not met:

(a¢) S commits to an attitude that is non-cotenable with, precludes joint satisfaction, precludes joint accuracy, or precludes joint reflexive accuracy, of some attitude that S* is committed to.

(b) S intends to give reasons for the attitude she commits to.

(a¢) is intended to capture cases where I invite you to a dispute, but I’m only playing the devil’s advocate or engaging in the dispute for pedagogical reasons. Again, the commitment involved here needn’t be a belief; it could be some other attitude like conjecturing, assuming, or taking up for the purposes of argument. Condition (a¢) is also intended to count as defective disputes that are merely verbal. In the next chapter we will consider a number of ways disputes can fail to meet condition (a¢). Condition (b) commits me to some further behavior. My challenge to you is defective if I have no intention of following through on having an argument with you in the discursive sense. Condition (c) counts as defective disputes over issues where there is no fact of the matter or where the dispute is irresolvable for some other reason. We will look at a number of ways that disputes can fail to meet (c) in the next chapter.

14.

The above distinctions between different sorts of disagreement may seem like so much philosopher’s word play. Do non-philosophers really have in mind these recherché notions of disagreement? Perhaps not. I do think, however, that something like the above notions serve to guide attributions of disagreement. Perhaps most competent speakers of English would not be able to articulate such conditions, but this does not mean that those conditions are not operative in their use of disagreement language. An extensive sociolinguistic study or series of studies would be needed in order to confirm or disconfirm the hypothesis that the above are senses of disagreement that guide ordinary linguistic practice. This sort of study is beyond the scope of the present project. I put it forward, then, as a tentative hypothesis, amenable to empirical confirmation or refutation, that the above are all kinds of disagreement captured by the ordinary language expression ‘to disagree’.

Where does this leave us? With the above distinctions in hand, we are now set to consider a number of different ways that activity-disagreements can go wrong. Intuitively, an activity-disagreement is somehow defective if it is an activity disagreement that rational people should not have for some reason or another. This is of course, quite vague. There are all sorts of reasons why particular activity-disagreements about some topic are or would be irrational. Not all of these reasons are reasons to reject out of hand any disagreement about that topic. We will use the above distinctions in order to sharpen the notion of a defective activity-disagreement.

The next chapter will be given over to discussion of the notion of a dispute. This notion is operative in many philosophers’ repudiation of certain issues in metaphysics. Many philosophers disparage as merely verbal disputes some of the apparent disagreements among ontologists, theorists about free will, proponents of various ethical theories, and proponents of sundry metaphysical views including, but not limited to endurantism, presentisim, Platonism, anti-Platonism, four-dimensionalism, and three-dimensionalism. In order to determine what the charge of having merely verbal disputes amounts to, some attention will have to be paid to the notion of a dispute. The distinctions we have made in this chapter will help us get clear on what disputes are and under what conditions they are defective, misguided, or otherwise irrational.

Chapter 4

DISPUTES AND DEFECTIVE DISPUTES

One of the activities for which philosophers are perhaps best known is having disputes with one another. Some philosophers, and many non-philosophers, think a lot of these disputes are silly, misguided, or otherwise defective. The task of the present chapter is to get clear on what a dispute is and to articulate and distinguish many ways that disputes can go wrong. Under what conditions does S have a dispute with S*? What is the relationship between disputes and activity-disagreement? What is the relationship between disputes and state-disagreement? When is a diagnosis of defectiveness in S and S*’s dispute correct? These and other questions will come up for consideration in this chapter.

One very simple and attractive view is that disputes are simply activity disagreements. Perhaps we will be willing to call an activity disagreement a dispute if it is particularly heated, or if it is about some controversy of legal ramifications, or so on, but a dispute is nothing fundamentally different in kind from an activity disagreement. So goes the simple view.

Is this view correct? This is tricky. For interpersonal disputes are typically cases where some argument occurs. The expression ‘argument’ can be taken in one of at least two ways:

the pugnacious sense: S and S* are having an argument just in case they are vehemently activity-disagreeing, perhaps shouting, S saying (things like) “No! That isn’t so!” and S* saying (things like) “Yes! Yes it is so!” or otherwise engaging in extended (and usually vehement) disagreement behavior.

the discursive sense: S and S* are having an argument just in case S advocates some position a, S* advocates position a* and (at least one of ) S and S* give reasons for advocating their position.

Philosophers will most likely characterize disputes that turn on arguments in the pugnacious sense as somehow defective; one ought to give reasons for one’s view rather than just gainsaying one’s interlocutor. Competent language users who are not trained in philosophy, however, will happily characterize as disputes cases that involve arguments in either sense.

For our purposes, we will count as disputes only those cases of activity-disagreement in which both parties argue in the discursive sense. In the last chapter I suggested that the locution “I disagree” can be used to perform a particular kind of illocutionary act. When “I disagree” is uttered in the right circumstances, it counts as the performance of the illocutionary act of challenging.[51] One challenges one’s interlocutor typically with respect to some view or attitude that they have articulated or avowed. The perlocutionary uptake of this speech act is the having of a dispute, in the following sense:

DISPUTE: S and S* have a dispute just in case S and S* have an activity-disagreement characterized by the mutual giving of arguments in the discursive sense.

I want to note some consequences of this view of disputes. Since activity-disagreement can occur without an underlying state-disagreement, it follows that a dispute can occur without an underlying state-disagreement. This seems intuitively correct to me. Take the case of the devil’s advocate who does not hold a particular view, say libertarianism, but supports that view anyway. Such a person can have a dispute with her interlocutors even if they agree about the truth value of all the claims entailed by libertarian doctrine. If I accept your conclusion, but do not buy your argument, we can have a dispute. I can even accept the truth of all of your premises, but reject your argument on the grounds that it is invalid or dialectically unsound. If this is the case, I needn’t disagree with your view in order to have a dispute with you.

Of course, some philosophers will not be happy with this characterization. Such philosophers, I suspect, take state-disagreement to be required for having a dispute, and further such philosophers are probably in the grips of something like the Affirm-Deny conception of state-disagreement. We have shown, however, that this conception of state-disagreement is untenable. But what is there to the claim that a state-disagreement is necessary for a dispute?

Perhaps even in the devil’s advocate cases I mentioned there is a state-disagreement hiding somewhere. After all, doesn’t the devil’s advocate at least think that her interlocutor is making a bad argument? It seems like our disputants have a difference of opinion about the goodness of a certain argument. If this is correct, then there turns out to be state-disagreement underlying their dispute after all.

Perhaps a different sort of case will suffice. Take a legal battle between the McFs and the McGs over who should inherit some piece of land. Suppose that Wheezle is the legal council for the McFs and Wheedle is the legal council for the McGs. Wheezle and Wheedle are close personal friends. They both agree that the McGs claim is legitimate. Wheezle, nevertheless, is bound by professional obligation to make a case for the McFs. He doesn’t believe that the law is on their side, but he makes the most convincing arguments he can. It seems appropriate to say that, when they are staring each other down in the courtroom, Wheezle and Wheedle are having a dispute or they are engaged in the act of disputing each other. They are engaged in a legal dispute over the entitlement to a particular piece of land.

This sounds like the right way to describe the case, but in the Wheezle-Wheedle case, there is no state-disagreement between the two advocates. Is there some attitude that Wheezle has that Wheedle disagrees with or vice versa? I do not see that there has to be.

Perhaps philosophers fond of the view that state-disagreement is required for disputes would be happier calling the Wheezle-Wheedle case a case of a merely apparent dispute. They are welcome to do so if they wish. I do not, however, want to encourage a merely apparent state-disagreement about what counts as a dispute, so we will understand that by ‘dispute’ I mean to refer to a case in which there is activity-disagreement and argument in the discursive sense, but not necessarily state-disagreement.

There is a type-token distinction to draw for disputes. Suppose I come upon a moon-landing conspiracy theorist having the following activity disagreement with her interlocutor:

No-Landing: There was no moon landing.

Landing: What are you talking about? Haven’t you seen the pictures?

No-Landing: Those pictures are faked, anybody can see that.

I take this to be a dispute in our above sense. Landing argues for his view and No-Landing argues for hers. This is a token dispute. It is a token dispute of a particular type, though. Consider the following conversation.

Landy: Human beings landed on the moon.

Nolandy: I don’t think so, it’s too far away. How did they get there?

Landy: Easy, they hitched a ride on the back of a leprechaun.

Landy and Nolandy are having a dispute, even if quite a silly one. The Landy-Nolandy dispute is of the same type as the No-Landing-Landing dispute: it is a dispute over the truth of the claim that there was a moon landing. But, it is a different token of that dispute: different reasons are being given, different disputants are involved, the dispute takes place at a different time, and so on.

This token-type distinction will become important later on when we discuss whether certain disputes are defective. For, as the above indicates, two token disputes of the same dispute type can have dramatically different properties: one can be reasonable and another can be unreasonable.

It is tempting to say that a type dispute is defective just in case any disputants to engage in tokens of it needn’t have a state-disagreement. Perhaps something like this is correct. However, there are a number of different ways that two parties can fail to have a state-disagreement. These different ways can give rise to different kinds of defective dispute.

How can disputes be defective? Let us count the ways. First, though, we need to say something about what it means for a dispute to be defective. As a good first-pass account, let’s say that a token dispute is defective, roughly, when it is a dispute that rational persons should not be having. A dispute is type-defective when no token of it is non-defective.

Let’s look at how token and type defectiveness can come apart. Suppose that Mary and Larry are having the following dispute:

Mary: Barack Obama will win another term in office.

Larry: I don’t think so, he’s failed to accomplish his platform goals so far.

Mary: Yes, but the Republicans will probably run an extreme right-wing candidate.

This dispute does not seem like it is type-defective. That is, there are some tokens of it that are perfectly reasonable disputes. But let’s suppose that Mary and Larry are having this dispute in the stairwell of a burning office building when they should be evacuating. Instead of evacuating, they stop on a landing, smoke all about them, and have the above dispute. There is something irrational about their having the dispute in this context.

There are all sorts of contextual features that make certain disputes irrational or inappropriate. When features of the context are such that a token dispute is irrational in that context, we will call the dispute context-defective. Now suppose that Mary and Larry, in the stairwell of the building, have this dispute:

Mary: We should use the east stairwell instead of this one because it lets out in the courtyard where there is fresh air and this one lets out in the garage where there isn’t.

Larry: No, that’s not right, I heard that there are rescue workers in the garage, we should use this stairwell.

So long as their dispute is not so protracted that they eventually die of smoke inhalation before it is resolved, it looks like this token dispute is not irrational (or at least not irrational for the same reason as the dispute about Obama).

Let’s take the case of the Landy-Nolandy dispute from above. Even if their dispute occurs in a context in which it is appropriate (i.e., they are not in grave personal danger, they have enough time for the dispute, there aren’t direly important matters that need to be settled, etc.), still it looks defective in some way. The arguments that Landy and Nolandy are giving for their respective positions are just plain absurd. No rational person should find these arguments even prima facie convincing. The problem with the Landy-Nolandy dispute is not that it is context-defective, but rather that it is content-defective. It is the content of the dispute that makes it defective.

Let us further distinguish two ways a dispute can be content-defective. The Landy-Nolandy dispute is what we will call token content-defective. It is because of the content of the particular arguments that Landy and Nolandy are making that their dispute is defective. Contrast this with a dispute that is type content-defective. Such a dispute is one that all tokens of which are defective in virtue of the content of the attitude or attitudes being disputed. Perhaps an example will make this clear. Suppose that Bill is a borderline case of baldness and Jane and Elaine have the following dispute:

Jane: Bill is bald, he’s got a patchy comb-over and all such people are bald.

Elaine: No, Bill isn’t bald, because he’s got more hair than Joe, and Joe isn’t bald.

Provided that Bill is a borderline case of baldness, it looks like there is something wrong with this type of dispute over the truth of the claim expressed by “Bill is bald.”[52] Further, the content of particular token-dispute that Jane and Elaine are having (i.e., the particular arguments they are making) do not adequately explain the defectiveness of their dispute; there is instead something wrong with anybody’s having a dispute about this matter, regardless of what arguments are in play.

Most people who think that some philosophical dispute is defective probably do not have in mind context defectiveness or token content-defectiveness. Most people who think that some philosophical dispute is defective probably think that anybody who were to have a token of that dispute in any context would be doing something silly, irrational, or mistaken. Accordingly, in what follows, we will focus mainly on type content-defectiveness. In this and subsequent sections, I will adumbrate a number of different kinds of type content-defectiveness.

Let us call the first kind of type-content defectiveness a no fact of the matter dispute. In a no fact of the matter dispute, it is inappropriate to say that one party to the dispute is wrong and another is right. This is because, in such disputes, there is just nothing for the disputants to be right or wrong about. Let’s take a (perhaps controversial) case.

Imagine two literary critics trying to crack the puzzle about Sherlock Holmes’ sexuality. Critic A thinks that Holmes was a confirmed bachelor, while critic B thinks that Holmes was heterosexual. A advances what he takes to be textual evidence for his view; B advances what she takes to be evidence for her view. Is there a fact of the matter as to who is right? Perhaps Conan Doyle never gave the matter any thought at all. Maybe this is hard to swallow. Take a (slightly) different case. Literary critic C thinks that Holmes had a hairy mole above his left shoulder blade, while critic D denies this. Both think they have good reasons for their view and the issue is hotly disputed between them in a series of letters published in a literary magazine. I’m inclined to think that there just isn’t a fact of the matter here, and that C and D are just being silly. Indeed, I’m inclined to think that anybody who were to have a token of the Holmes-mole dispute would be doing something patently absurd.

Some philosophers claim that certain disputes in metaphysics are like this.[53] In Part Two we will take up the question as to whether certain disputes in physical object ontology and certain disputes over possible worlds are like this.

Some disputes are silly, not because there is no fact of the matter as to who is right and who is wrong, but because no one is in a position to know who is right and who is wrong. Call such disputes epistemically defective disputes.[54] Consider a dispute between Mordecai and Methuselah about the number of atoms in the universe. Mordecai thinks the number of atoms is odd, while Methuselah thinks it is even. Suppose both give some sort of reason for their view. Their dispute is silly. It is not silly because there is no fact of the matter; presumably either Methuselah or Mordecai is correct. The problem with their dispute is that nobody is in a position to know who is right.

Here there is an important distinction to draw between two different kinds of epistemically defective dispute. The Methuselah-Mordecai dispute is type epistemically defective because, presumably, nobody is in a position to know what the right answer is. Contrast this with a dispute that is merely token epistemically defective. Take the following dispute between Jack and Jill. Jack and Jill are viewing a scene of a classroom through a black and white closed-circuit TV monitor. They get into a dispute over the color of a particular chair in the room. Let’s say that neither Jack nor Jill is in a position to know the color of the chair. So, their dispute is epistemically defective. However, there are some people in the room who, presumably, are in a position to know what color the chair is.

Some philosophers claim that certain disputes in metaphysics are type epistemically defective.[55] Below we will examine the charge of type epistemic defectiveness with respect to some disputes in ontology and about possible worlds.

Sometimes both parties to a dispute turn out to be wrong. When this is the case, there is something defective about their dispute. We’ll call this a both wrong dispute. Both wrong disputes, like epistemically defective disputes, come in both a type and token variety. Take the following case. Let us suppose that Bill is a borderline case of baldness and that, as a borderline case of baldness, it is not true that Bill is bald and it is not true that Bill is non-bald. Now, suppose Larry and Barry have a dispute in which Larry advocates the view that Bill is bald and Barry advocates the view that Bill is non-bald. Both Larry and Barry are incorrect. Further, any parties who were to have a token dispute of this type would be wrong in the same way.[56]

Contrast this with a case of a token both wrong dispute. Suppose Gary and Sheri are arguing about whether the man in the next room is bald. Sheri says, “The man in the next room is bald” and Gary says, “The man in the next room is non-bald”. Each gives supporting arguments for their respective view. But, unbeknownst to either of them, there is no man in the next room; the room is empty of men. Taking ‘wrong’ very broadly here, it looks like both Gary and Sheri are wrong.[57]

Perhaps another case will be more convincing. Suppose that Jim argues that an acquaintance, Phil, is an engineer but not an artist while Jon argues that Phil is an artist but not an engineer. Suppose Phil is both an artist and an engineer. Jim and Jon’s dispute is a both wrong dispute.

Both-wrongness has an analogue for disputes that are not about the truth or falsehood of a proposition. Suppose that both Bill and Jill want the same cupcake. Bill wants to eat it, and so does Jill. Bill makes his case for why he is entitled to the cupcake, Jill makes hers. But, unbeknownst to them, Bob has already eaten the cupcake. So, neither Bill nor Jill can have it. Neither Bill nor Jill’s attitude can be satisfied. When a dispute turns on some (nonpropositional) attitudes neither of which can be satisfied, we will call it a joint nonsatisfaction dispute.

8.
Sometimes two disputants are both right. Take the time-worn case of Half-full and Half-empty. Half-full thinks that a particular glass of water is half full and Half-empty thinks that the glass is half empty. They each have some argument for their view. It seems like both of them are right.[58]

Perhaps this case doesn’t convince. Consider the following case. Let Iffy and Jiffy be introductory logic students having the following dispute:

Iffy: It’s true that if Barack Obama is not President, then he lives in the White House.

Jiffy: But wait, Barack Obama is President.

Iffy and Jiffy both intend to be using the expression ‘if,…then’ in the way the material conditional is used. Both, however, turn out to be right.

We will call cases in which both disputants are right both right disputes. Some philosophers suggest that merely verbal disputes are special instances of both right disputes.[59]

We have all probably heard someone say, when having a dispute that seems to go nowhere, “I think we’re talking past each other here.” What does talking past one another amount to? For the moment, let’s consider cases of miscommunication not involving disputes that can be characterized as instances of two persons talking past each other. Talking past each other involves, roughly, talking about two different things without realizing it. Take a tired and well-worn case. Bill notices some children playing by the side of a river and says to his friend Jill, “There’s a game of jump-rope happening by the bank.” Jill takes Bill to be talking about the Wells Fargo around the corner and she says, “That’s a pretty strange place to be having a game of jump rope.” Bill was talking about a river bank, while Jill was talking about a financial bank. Since their comments were intended to be about different places, there is an important sense in which they failed to communicate with each other; they were talking past each other.

Cases of talking past each other do not always turn on interlocutors’ using a word in two different senses. Sometimes cases of mishearing result, usually fairly comically, in instances of talking past one another. Take the case of Robert and Julie:

Robert: It’s Thursday.

Julie: Yes, me too, I could use a glass of water.

Julie took Robert to be saying “I’m thirsty,” so her reply failed to connect with what he was really saying.

Talking past each other has an analogue in cases of disputes. When a dispute arises or persists due to a misunderstanding like the ones above, we will say that the disputants are arguing past each other. Notice, that cases of arguing past each other can overlap with both wrong and both right disputes. Suppose that I have a dispute with you about whether there are children playing near the bank. You mean financial bank and I mean river bank. But, we could both be right. You say “There are no children playing near the bank” and I say, “Of course there are.” We both argue for our position, but we could both turn out to be right. We could also both turn out to be wrong when we’re arguing past each other. Suppose you say, “Tomorrow will be Wednesday,” and I say, “Nonsense, the weather report is not calling for windy conditions at all!” Our dispute is a case of talking past each other, but provided today is Monday and I’m mistaken about the weather report, we’re both wrong.

When we are arguing past each other, we typically do not have a state-disagreement, but this is not necessarily the case. Let’s look at a modified version of the bank dispute. Barry says, “There are children playing by the bank” and Larry says, “There are not children playing by the bank.” Their argument goes on for a few moments before they come to learn that they were using ‘bank’ in different senses. But, suppose that their further conversation uncovers the following facts about their beliefs:

Barry: believes that there are children playing near Wells Fargo

disbelieves that there are children playing near the river

Larry: disbelieves that there are children playing near Wells Fargo

disbelieves that there are children playing near the river.

Suppose that Barry means by ‘bank’ a financial bank, but Larry means by ‘bank’ a river bank. As it turns out though, Barry and Larry do state-disagree about whether there are children playing near the financial bank, but they were still talking past each other because Larry took Barry to be talking about a river bank and not a financial bank.

It is tempting to think that cases of arguing past each other are merely verbal disputes. I think this is a mistake. For cases of mishearing can give rise to disputants’ arguing past each other without their disputes being merely verbal. A merely verbal dispute is best characterized in terms of speakers being disposed to apply some term or terms in divergent ways. In the windy-Wednesday case, however, this is not what is going on. You and I are both disposed to use the terms ‘windy’ and ‘Wednesday’ in exactly the same way; I just misheard you (maybe because it was windy when you spoke).

It is also tempting to think that merely verbal disputes are cases of arguing past one another. This is also a mistake. Cases of arguing past each other essentially involve disputants arguing for or about different things. But some merely verbal disputes, as we will see later, involve disputants arguing about precisely the same facts, but just electing to describe them differently.

10.

Many of us have probably heard some people say, when having a protracted and unproductive dispute, “I think we’re talking at cross purposes here.” What does talking at cross purposes amount to? Talking at cross purposes is subtly distinct from talking past one another. Indeed, I suspect many people are inclined to run these different phenomena together, taking the expressions ‘talking at cross purposes’ and ‘talking past one another’ to be roughly synonymous. This is a mistake.

Talking at cross purposes is a kind of miscommunication that essentially involves two conversants having different plans, goals, or starting with different presuppositions or assumptions. Let’s take a case. Suppose that your plan is for us to go to the cinema first and then the shopping mall, while my plan is exactly the reverse of this. If we get into an argument about the best route to take, we will most likely be arguing at cross purposes. But isn’t this a case of talking past each other as well? For, after all, we are using the expression ‘best route’ differently. I think this is a stretch. I needn’t misunderstand what you say in this case. What’s going on is that I misunderstand some presupposition that you’re making, namely the presupposition that we’ll be headed to the cinema first. I haven’t misheard you, nor have I failed to understand in what sense you’ve used the terms that you’ve used. What’s happening here is that we have divergent attitudes with respect to our plan for the evening.

A lot of political disputes are instances of arguing at cross purposes. Members of different political parties often have different ways of valuing, assume different things to be true, and accept as evidence different claims. When this is the case it is not necessarily impossible to avoid arguing at cross purposes and to have reasonable disputes, but these different presuppositions and the like must be explicitly laid on the table. If parties to a dispute are not clear about what they presuppose or what they take to be evidence or what the structure of their system of values is, the likelihood of an argument at cross purposes is very good.

Perhaps a lot of philosophical arguments are like this. When philosophers do not share each other’s presuppositions, starting points, methodologies, or assumptions, and they are not aware of this, then they are probably disposed to argue at cross purposes. Take the case of analytic philosophers and continental philosophers. Perhaps many disputes that occur between members of these camps are instances of arguing at cross purposes. While it is hard to be precise about the differences between philosophers of these respective schools, they clearly seem to have different methodologies, make different starting assumptions, and place different values on different styles of argumentation.

It is important to note that arguing at cross purposes and having a merely verbal dispute are two different phenomena. You and I can have a merely verbal dispute when our goals, plans, and presuppositions are all the same. We can be explicit about what we’re presupposing, what we intend to accomplish, and how we intend to accomplish it, but, owing to a divergence in our dispositions to apply some term or other, we can have a merely verbal dispute. Perhaps this is impossible, one might think, for when I use a term t in one way and you use it in another, don’t we make different presuppositions? Namely you presuppose that t means one thing and I presuppose that it means something else? This is tricky. Whether this description in terms of divergent presuppositions is correct depends on controversial theses in the philosophy of language and the philosophy of mind. We will broach some of these issues but our discussion of them will be far from what is required in order to settle the controversy.

11.

Another kind of defective dispute is perhaps the counterpart to epistemically defective disputes. I have in mind here disputes over a claim that is trivial. Call these easy disputes. When A and B are arguing over some claim that is either trivially true or trivially false, they are having an easy dispute. The right answer should be clear, or at least, should be clear to anyone who is a competent user of the language or languages in which the dispute is conducted.

What exactly is meant by trivial here? This is a bit tricky. When a claim is trivial in the sense I have in mind, it is easily knowable a priori or by a quick empirical investigation. So, suppose that, having nothing better to do, you and I get into an argument about whether my cufflinks are in a drawer only a few feet away. You say they are there, and I say they are not there. This is a silly dispute. We could easily get up and go look. Notice, however, that our dispute is only a token easy dispute. For we could have a dispute of the same type in circumstances that precluded a quick look in the drawer (say we’re having the dispute on the way to the opera and the drawer is too far away for us to turn back and have a look without being late).

Disputes can also be type easy disputes, however. Take the case of Billy and Willie who are disputing whether a bachelor can be married or not. Billy says there are some married bachelors and Willy denies this. There’s an easy answer, though. All that is required is the trivial bit of knowledge that a bachelor (in the operative sense) is an unmarried man. Note, that this easy dispute needn’t be a case of arguing past one another. For both Billy and Willy are (at least trying to be) using the word ‘bachelor’ in the same sense. Is this easy dispute a merely verbal one? This is a bit trickier. Ostensibly, the dispute is a dispute about bachelors, not about the word ‘bachelor’. But, doesn’t the dispute turn on the disputants having divergent patterns of use? Perhaps, but this needn’t be the case. Suppose Billy and Willy both use the term bachelor in exactly the same way, to refer to unmarried men, but Billy is just momentarily confused. Perhaps this doesn’t convince.

Cases of easy a priori disputes abound where simple calculations are concerned. Take the case of Mark and Clark who are arguing about whether it has to be the case that some members of Congress were born on the same day (not on the same day and year, though). Mark says it is impossible for there to be over four hundred members of Congress all born on a different day of the year, while Clark thinks it is possible for all members of Congress to have a different birthday. There is an easy answer here that turns on a simple mathematical calculation. Mark and Clark’s dispute is defective, because one of them is making a simple mistake of arithmetic. Many disputes that take place in restaurants over the check are of this sort.

Some philosophers think that ontological disputes, when understood as disputes over the question of whether entities of a certain type exist, are easy disputes. Jonathan Schaffer, for example, argues that the dispute about whether numbers exist is an easy dispute in our sense.

Start with the debate over numbers. Here, without further ado, is a proof of the existence of numbers:

1. There are prime numbers

2. There are numbers.

1 is a mathematical truism. It commands Moorean certainty, as being more credible than any philosopher’s argument to the contrary.[60]

Schaffer’s point is that the existence of numbers follows trivially from a truism. So, there is an easy answer to the debate over whether numbers exist. Of course, this does not mean that there is no substantive, nontrivial dispute to have over what numbers are like, but if Schaffer is right, there is no non-easy dispute to have over the bare existence of numbers. In Part Two we will discuss whether certain disputes in ontology are easy disputes.

12.

Let us now turn our attention to merely verbal disputes. William James tells us the following story:

Some years ago, being with a camping party in the mountains, I returned from a solitary ramble to find every one engaged in a ferocious metaphysical dispute. The corpus of the dispute was a squirrel—a live squirrel supposed to be clinging to one side of a tree-trunk; while over against the tree’s opposite side a human being was imagined to stand. This human witness tries to get sight of the squirrel by moving rapidly round the tree, but no matter how fast he goes, the squirrel moves as fast in the opposite direction, and always keeps the tree between himself and the man, so that never a glimpse of him is caught. The resultant metaphysical problem now is this: Does the man go round the squirrel or not? He goes round the tree, sure enough, and the squirrel is on the tree; but does he go round the squirrel? In the unlimited leisure of the wilderness, discussion had been worn threadbare. Everyone had taken sides, and was obstinate; and the numbers on both sides were even. Each side, when I appeared therefore appealed to me to make it a majority.[61]

James intuited, as indeed most of us would, that the dispute in the above story is defective. James concluded that the dispute was merely verbal in that it turned on a difference of opinion as to how ‘going round’ is to be made precise.

If James’ diagnosis is correct, there is an important sense in which the operative dispute is not metaphysical at all; it does not concern, primarily, the way things are, but instead turns on the meanings of words. Of course whether a word means one thing rather than another has something to do with the way the world is. But in the above case, it looks like the ordinary language meaning of the expression ‘goes round’ is not sufficiently precise to settle the dispute. In cases like these, what is required is some kind of semantic decision. In order to settle the above dispute, the disputants must make it explicit what they are to mean by ‘goes round’. If the dispute survives this stipulation, then perhaps there was a non-verbal issue at stake after all. James suspects that the dispute will not survive such precisification: “Make the distinction, and there is no occasion for farther dispute. You are both wrong and both right according as you conceive the verb ‘to go round’ in one practical fashion or the other.”[62]

David Hume puts the following words into the mouth of Philo:

So little do I esteem this suspense of judgment in the present case to be possible, that I am apt to suspect there enters somewhat of a dispute of words into this controversy, more than is usually imagined. That the works of nature bear a great analogy to the productions of art is evident; and according to all the rules of good reasoning, we ought to infer, if we argue at all concerning them, that their causes have a proportional analogy. But as there are also considerable differences, we have reason to suppose a proportional difference in the causes; and in particular ought to attribute a much higher degree of power and energy to the supreme cause than any we have ever observed in mankind. Here then the existence of a Deity is plainly ascertained by reason; and if we make it a question, whether, on account of these analogies, we can properly call him a mind or intelligence, notwithstanding the vast difference, which may reasonably be supposed between him and human minds; what is this but a mere verbal controversy? No man can deny the analogies between the effects…[63]

Philo thinks that all parties to the argument from design will accept a certain description of the facts: namely that the intricacy of natural systems shows there to be a designer. He laments, however, over the irksome propensity that disputants have to describe this designer in different ways, some using anthropomorphic terms like ‘mind’ or ‘intelligence’ and others refraining from using such terms. The dispute, Philo thinks, is merely verbal, for everyone (at least everyone he’s concerned about) agrees about the argument. What is there left to dispute? A bare form of words, claims Philo.

Cases of alleged merely verbal disputes abound. Can all of these disputes be characterized in the same way? In this section we will attempt to get clear on what it is to have a verbal dispute with someone. We will examine a number of different proposed analyses of this concept. I will argue that the extant analyses are inadequate in various ways. At the end of the section, I will propose an account that, I will maintain, is not defective in any of these ways. We shall then take up, in Part Two, the question of whether certain disputes in ontology and about possible worlds are merely verbal in any meaningful sense of that term.

12.1.

First, some stage setting. Some philosophers will be tempted to say that merely verbal disputes are not real disputes, perhaps because they do not involve disagreements. For such philosophers, ‘merely verbal dispute’ functions semantically like ‘forged bank note’ or ‘fake passport’.[64] Given what I mean by ‘dispute’, I am not willing to say this. There is a sense in which the ‘merely’ in ‘merely verbal dispute’ does function to diminish the importance of merely verbal disputes, but this word functions more like ‘only’ in ‘He’s only a child’ rather than like ‘fake’ in ‘He has a fake passport’. The former locution is not supposed to carry the implication that the referent of ‘he’ is not a real child, but rather that he is a mere child. Being a child, in whatever context is operative here, is somehow to fall short of some conversationally relevant mark, as in “He can’t get into an R-rated movie, he’s only a child” or “You can’t expect him to take care of his sister, he’s merely a child.” The ‘merely’ in ‘merely verbal dispute’ works like this. Merely verbal disputes are disputes that fall short of some kind of standard of interestingness, substance, or whatever. They fall short in virtue of some facts about the linguistic practice of the disputants.

Can a pair of disputants be having a dispute about language without thereby having a merely verbal dispute. I think so. You and I can get into a heated argument that is explicitly about the meanings of words. As David Chalmers correctly observes

Sometimes words matter. Verbal disputes are often important disputes, when something important rests on matters of linguistic usage. In linguistics and the philosophy of language, for example, words are the primary domain of concern, so that a disagreement over the meaning of ‘round’ cannot simply be set aside as merely verbal. The same applies in cases where we are studying the users of words. In literary criticism or history, for example, it might be crucial to know how a given word is used by a given individual or community.[65]

Whether to count a particular action as falling under the extension of the word ‘murder’ or ‘marriage’ can be a very serious matter indeed, despite the issue being about the application conditions for a bit of verbiage. When a dispute turns on the meaning of words in such a way that the dispute is not rendered defective, we will call it a verbal dispute as opposed to a merely verbal dispute. The former locution is understood so as not to carry any implication that disputants having a verbal dispute are engaged in behavior that is irrational or silly.

In what follows, I am going to restrict my attention to two-person disputes. If what I have to say is correct, then presumably it can be extended to cover multi-person disputes or disagreements one may have with oneself (e.g., a disagreement I have with what I used to think). In what follows, we will be talking mainly about token merely verbal disputes, though my remarks should be extendable to cover type disputes as well.

12.2.

Perhaps the simplest and initially the most attractive view about merely verbal disputes is what I’ll call the Equivocation Model. This is the sort of view of merely verbal disputes suggested by remarks like the following of Ted Sider’s:

To say that an apparent dispute over sentence f is merely verbal is to say that the disputants do not mean the same thing by the sentence f, and that what one says by uttering f is consistent with what the other says by uttering ⌐f. Imagine an American and a Briton debating whether a certain quantity of water has a volume of one ‘gallon’. In fact, the American has in mind an American gallon, while the Briton has in mind an imperial gallon. This debate is pointless. The disputants agree on the volume of the water; they simply need to settle on whether to speak in the British or the American way about the agreed-upon facts.[66]

The idea here is pretty simple. If I say s and you say not-s, then our dispute is merely verbal just in case (i) s means one thing in my mouth and another thing in your mouth and (ii) what you mean by not-s is consistent with what I mean by s. Formally, this is as follows:

Equivocation Model: S and S* have a merely verbal dispute iff S asserts t and S* asserts not-t, t in S’s mouth expresses a different proposition than t in S*’s mouth, and the proposition S expresses by uttering t is consistent with the propositions S* expresses in uttering not-t.

This account works fairly well, up to a point. The Equivocation Model is very good at handling cases where disputants come from different sociolinguistic backgrounds. When two disputants are literally speaking different languages or different dialects of the same language, merely verbal disputes in the Equivocation Model sense occur pretty frequently. Below we will see that the Equivocation Model captures what David Chalmers calls narrowly verbal disputes.[67]

Notice, however, that disputes that are merely verbal on the Equivocation Model are better described as instances of two disputants arguing past each other, in our above sense. Also, the Equivocation Model may be insufficient to capture merely verbal disputes in which both parties to the dispute are speaking the same language or dialect of the same language. Take the case of a dispute between Mark and Clark about whether Edna has arthritis. Mark uses ‘arthritis’ the way most of us do, but Clark mistakenly thinks that arthritis is a condition of the muscles. If Tyler Burge and other content externalists are correct, then Mark and Clark both express the same proposition when they say “Edna has arthritis” (presuming that they mean to use ‘Edna’ in the same way).[68] The Mark-Clark dispute strikes me as merely verbal, but if externalism is correct it does not involve equivocation.

A further problem for the Equivocation Model is that it delivers the verdict that a dispute is not merely verbal if one party to the dispute affirms a necessary falsehood. For the Equivocation Model requires that the disputants hold views that are consistent. But any proposition is inconsistent with a necessary falsehood, so if one disputant affirms a necessary falsehood, the criteria outlined by the Equivocation Model are not met. But surely I can get into a merely verbal dispute with McA even if one of us is necessarily wrong? Putting the Burge point to one side for a moment, suppose that McA thinks it is impossible for there to be round squares. By ‘round square’ I take McA to mean something that is both round and square in the geometric sense. But, unbeknownst to me, McA intends ‘round square’ to refer to objects, like some drinking coasters, that are four-sided with rounded corners. If I say, “No, it is possible for there to be round squares,” what I say is necessarily false.[69] So, our dispute is not merely verbal on the Equivocation model because the proposition that I express is inconsistent with the proposition that McA expresses.

Perhaps it might be suggested that we somehow exclude cases of disputes in which one side affirms something necessarily false. But, as we’ve seen above, this would exclude too much, for in many disputes in mathematics and the other a priori sciences, one side is often affirming something necessarily false. But surely we cannot conclude that merely verbal disputes are impossible in these subject areas.

Perhaps we can drop the consistency clause from the Equivocation Model. Even if we do this, however, we are still left with the Burge concern. If natural languages work in the way that Burge and other externalists say they do, then if the Equivocation Model correctly captures merely verbal disputes, it will typically be impossible for me to have a merely verbal dispute with an interlocutor who speaks my language, unless the sentences that we affirm contain some kind of context sensitive vocabulary like ‘I’, ‘here’, ‘now’, or perhaps some gradable adjectives like ‘tall’ or (on a contextualist semantics of knowledge ascriptions) ‘knows’. This is an unwelcome result.[70]

12.3.

Another intuitively very obvious view counts a dispute as merely verbal just in case the dispute would not survive the disputants’ making explicit what they mean by the terms operative in the dispute. Take James’ famous case. James thought that if each party made it clear what they meant by ‘goes round’, then the dispute would dissolve. Call this the Naive Dispute Resolution view.

James’ case, however, is both a good illustration of and an obvious counterexample to the Naïve Dispute Resolution view. For, as James describes the case some disputants obdurately persisted in having the dispute even after it was made clear what they meant by ‘goes round’. What we need in a dispute resolution view is a clause that handles such obdurate, pig-headed refusal to abandon the dispute. Let us try the following:

Dispute Resolution View: S and S* are having a merely verbal dispute just in case S and S* ought to (on pain of irrationality) stop having the dispute once it is made clear how they intend to be using the operative terms in the dispute.

A couple of things need to be made clear here. First, a pattern of activity disagreement may involve a number of different disputes. For example, in the course of arguing about whether God exists, we can have a dispute about each of the following: (i) whether God is omnipotent, (ii) whether human beings have free will, (iii) whether it is metaphysically possible for God to create a world without evil in which human beings have free will. The way I am using the word ‘dispute’, we have a distinct dispute about each of (i)-(iii) if they come up in the course of our discussion and we argue discursively about them. So, it might be that our dispute about (ii) turns out to be merely verbal, but our dispute about (i) does not. If this is the case, then, according to the Dispute Resolution View, we ought to stop having a dispute about (ii) once it is made clear what each of us means by ‘free will’. It may be hard to say, in practice, whether two parties are having a dispute over p and a distinct dispute over q or whether they are having just one dispute, but I take it that we can, in principle, make a distinction here.

As the above example about God makes clear, one dispute’s mere verbalness can infect another dispute. So, if our dispute about (ii) is merely verbal, it is very likely that our dispute about (iii) will be verbal as well.

Second, what is intended by the expression ‘operative terms’ in the above formula? What does it mean for some term t to be operative in some dispute? This is tricky. One might think that t is operative in a dispute just in case t is used in that dispute. This won’t do, however. For we can have a dispute about free will without our ever using the expression ‘free will’. We can have a dispute about whether Jones is a murderer without ever using he expression ‘murder’ or ‘murderer’. I propose the following gloss of ‘operative’. It is somewhat crude, but it should suffice for our purposes:

t is operative in some dispute d just in case the disputants’ beliefs about the meaning of t influence their position in the dispute.

So, ‘murder’ is operative in a dispute between Smith and Jones just in case Smith’s and Jones’ beliefs about what ‘murder’ means influence their attitudes about the disputed subject matter. Let’s take an example. Suppose that Smith thinks ‘murder’ means ‘unlawful killing’ while Jones thinks murder means ‘immoral killing’. This difference in belief about the meaning of ‘murder’ will doubtless influence their respective positions in a dispute about whether somebody is a murderer. I am here using ‘influence’ in a sort of intuitive sense that I will not try to make too precise. Suffice it to say that my position in some dispute is influenced by my metalinguistic beliefs just in case I would most likely have a different position if I had different metalinguistic beliefs.[71]

As initially attractive as it may be, the Dispute Resolution View cannot capture what it is for a dispute to be merely verbal. For, there are cases in which getting clear on the meanings of terms will (or ought to) dissolve a dispute, but in which the dispute is not merely verbal. So, the Dispute Resolution View counts some non-verbal disputes as merely verbal. Let’s look at some examples. Suppose that you and I are arguing over whether or not a particular person is bald. You say she is bald, I say she is non-bald, and each of us gives some kind of reasons for holding our respective positions. But, when we begin to get clear on what we mean by ‘bald’, it occurs to us that it is a vague term that admits of borderline cases. We further (and let’s say correctly) identify the person in question as a borderline case of baldness. It looks like we should give up on our dispute; she is not bald nor is she clearly not bald.

Let’s take another case. Suppose you think that a comment that Bill made is sarcastic and I claim that it was not sarcastic, instead it was sardonic. Our dispute needn’t be merely verbal. For what I say implies that Bill’s comment was mocking and what you say does not carry this implication. But, our dispute might dissolve if we get clear on what ‘sardonic’ and ‘sarcastic’ mean in English. Many arguments over whether a law has been broken or not also might not survive a clarification of some key term in the debate. Surely not all such disputes ought to be counted merely verbal.

The problem with the Dispute Resolution View is that it fails to specify why parties to merely verbal dispute ought to stop having their dispute once it is made clear what they mean by what they say. One way of specifying the relevant reason would be viciously circular: disputants ought to stop having their dispute after making clear the operative terms and they ought to stop having the dispute because it is merely verbal. This won’t do. Also, it won’t do to say that disputants ought to stop having their dispute because there is no fact of the matter as to who is right, for there are no fact of the matter disputes that are not merely verbal. It is unclear just how to fill in the lacunae in this account without either (i) rendering the account viciously circular, or (ii) conflating merely verbal disputes with some other kind of defective dispute.

12.4.

A more nuanced view comes from Carrie Jenkins.[72] Jenkins calls her view MVD; I shall reserve the name.

MVD: Parties A and B are having a merely verbal dispute iff they are engaged in a sincere prima facie dispute D, but do not disagree over the subject matter(s) of D, and merely present the appearance of doing so owing to their divergent uses of some relevant portion of language.[73]

A few remarks are in order here. By ‘do not disagree’ here, I take it that Jenkins means ‘do not state-disagree’. On Jenkins’ usage, a ‘sincere dispute’ is a dispute in which the disputants do in fact hold the positions they appear to advocate. The restriction to sincere disputes is supposed to exclude cases in which actors are having an apparent activity disagreement characterized by the apparent giving of arguments in the discursive sense. Suppose that there are two actors portraying a pair of philosophers having a merely verbal dispute, surely “[w]e don’t want to classify the actors as engaged in a merely verbal dispute”, but if we dropped the sincerity clause from the above biconditional, then “MVD would so classify them.”[74]

By calling a dispute a ‘prima facie’ dispute, Jenkins means to characterize it as one that looks like a genuine dispute. Since Jenkins wants to “leave open…the question of whether or not a merely verbal dispute is really a dispute” her biconditional cannot require that a merely verbal dispute be a dispute.[75] So, Jenkins uses the expression ‘prima facie dispute’ in a way similar to the way I’ve been using ‘apparent activity disagreement’. It bears mentioning here that Jenkins does not require for a prima facie dispute the apparent giving of reasons. So, her expression ‘prima facie dispute’ is equivalent to my ‘apparent activity disagreement’.

The ‘subject matter’ of a dispute is somewhat tricky to specify. Intuitively, the subject matter of a dispute is what that dispute is about. But this is not very clear. We can get clearer on what subject matter is by first considering what it is not. First, a subject matter of a dispute is not to “be identified with the subject matter(s) of the statements sincerely made during the course of that prima facie dispute by the parties to it.”[76] Why not? This is largely because some parties to a dispute may make statements designed to support their side in the dispute that are not properly to be taken as disputed. Suppose that we agree upon p, q, and r. If we are having a dispute about whether s, I might marshal some of our agreed-upon propositions as evidence for my view. If I do this, we oughtn’t to take our dispute thereby to be about p, q, or r, for these are agreed upon and not disputed.[77]

A further reason for thinking that the subject matter of a dispute is not identical to the statements made in the course of having the dispute is that “some parts of the subject matter may not be explicitly mentioned by either party.”[78] I have an ongoing dispute with my significant other about the temperature at which the thermostat ought to be set. On a particular occasion, we might have the following exchange:

Her: You changed it again, didn’t you?

Me: But I was hot!!

Her: Our bill is going to be outrageous!

Me: A small price to pay for comfort!

Here we are having a dispute about the claim expressed by “The thermostat should not be set lower than 75 degrees Fahrenheit during the summer,” but this statement is never explicitly made during the course of our present skirmish.

We have observed what subject matter is not, but what is there to be said about what the subject matter of a dispute is? According to Jenkins, “the parties’ intentions, interests, and other mental states seem very important for determining the subject matter of a prima facie dispute.”[79] If my significant other and I are having a dispute about whether the thermostat should ever be set below 75 in the summer, this is because we “are occurrently interested in that question, and/or because [we] intend to dispute that question, and/or because of other mental states of [ours], such as [our] desires to dispute it, [our] desires to prove that [we] are right on the matter, or perhaps even [our] merely attending to it.”[80] Jenkins admits that this is a “deliberately vague” way to specify what the subject matter of a dispute is. Despite its vagueness, it should suffice for our purposes.

I think that MVD delivers the wrong results due to its inclusion of what I call the ‘no disagreement’ clause. MVD requires that in a case of a merely verbal dispute, the disputants do not disagree over the relevant subject matter. This requirement excludes some merely verbal disputes, however.

Take the case of George and Sally. Both are philosophers. George mistakenly thinks that the word ‘philologist’ means what most of us mean by ‘phlebotomist’ and vice versa. He’s switched the words around in his idiolect and continually misuses them accordingly, often to comic results. When Sally informs George that there will be a philologist giving a talk at the Eastern Division APA conference this year, George is incredulous. Here’s their exchange:

Sally: There will be a noteworthy philologist giving a talk at the APA this year.

George: I think that’s false, what would a philologist have to do with the APA?

This dispute is intuitively verbal. Moreover, this dispute is intuitively verbal even if Sally and George have the following beliefs:

Sally: Believes that there will be a phlebotomist giving a talk at the APA (though she doesn’t mention having this belief)

Believes that there will be a philologist giving a talk at the APA.

George: Disbelieves that there will be a phlebotomist giving a talk at the APA (though he’s describe this belief by saying “There will be no philologist giving a talk at the APA”)

Disbelieves that there will be a philologist giving a talk at the APA (though he’d describe this belief by saying “There will be no phlebotomist giving a talk at the APA.”)

It looks like Sally and George state-disagree about the relevant subject matter. For the following seems reasonable to me: either their dispute’s subject matter is whether there will be a philologist giving a talk or it is whether there will be a phlebotomist giving a talk. But George and Sally disagree about both of these.

Let’s take another case. Take the famous bank-bank case. I mean ‘financial bank’ and you mean ‘river bank’. We have this exchange:

Me: There is a donut shop by the bank.

You: No there isn’t.

If we go on to give arguments for our positions without it becoming obvious that we mean ‘bank’ in two different senses, we’ll be having a merely verbal dispute. But, for all of that we can state-disagree about whether there is indeed a donut shop by the financial bank. I think that the shop has been torn down, and you do not think this. Intuitively, though, our dispute is still merely verbal. For this reason, I think we should reject MVD.

12.5.

Perhaps one of the most well-defended and philosophically sophisticated accounts of merely verbal disputes is the one given by Eli Hirsch.[81] Hirsch maintains that “[p]erdurantists, endurantists, mereological essentialists, four dimensionalists, and sundry nihilists have engaged each other in lengthy and often highly theoretical disputes, though many of these disputes are…merely verbal” (Hirsch 2009, 232). He calls those philosophers “revisionists” who suggest that many of our commonsense beliefs about composition and the persistence conditions of physical objects are mistaken. He says,

the revisionists are in effect merely choosing to use a language different from ordinary language, and, insofar as they are not aware of this (and take themselves to be using ordinary language), they are making a certain kind of verbal mistake (2009, 232).

Put rather roughly, Hirsch’s view is the view that “an issue in ontology (or elsewhere) is ‘merely verbal’ in the sense of reducing to a linguistic choice only if. . . [e]ach side can plausibly interpret the other side as speaking a language in which the latter’s asserted sentences are true” (Hirsch 2009, 231).

This way of putting the view is unsatisfactory for a couple of reasons. First, the above statement reads as a necessary condition on verbal disputes. Given that Hirsch goes on to argue that the disputes in ontology are verbal in virtue of satisfying this condition, it is better to formulate the above as providing a sufficient condition as well. Second, it is not clear that the target disputes in ontology satisfy this condition. As Hirsch is aware, “[Tyler] Burge and others have claimed that, if a person belongs to a certain linguistic community, then the language of that community is the person’s language” despite any idiosyncrasies in that person’s usage (Hirsch 2009, 239; Burge 1979). But, since many rival ontologists whose views Hirsch discusses are members of the same language using community, if Burge’s view is correct, it would appear to follow that it is not plausible for them to interpret each other as speaking different languages.

In order to escape this difficulty, Hirsch uses the expedient of possible communities. On Hirsch’s preferred sense of ‘merely verbal’ a dispute between X and Y is merely verbal in the following way. Suppose X and Y are both members of the same language using community and that X assertively utters S₁, S₂,…, S_n, while Y assertively utters ~S₁, ~S₂,…, ~S_n. There is a possible community, the X-community, where everyone accepts sentences orthographically and phonetically identical to the disputed sentences X accepts: S*₁, S*₂,…, S*_n. There is a similar community, the Y-community, where everyone accepts sentences orthographically and phonetically identical to the disputed sentences Y accepts: ~S´₁, ~S´₂,…, ~S´_n. In all other respects, these communities are just like the language using community to which X and Y belong. Call the language spoken by the X-community ‘X-lish’ and the language spoken by the Y-community ‘Y-lish’. In the Possible Communities sense of ‘merely verbal’, what makes X’s dispute with Y merely verbal is that, on the most plausible interpretation of X-lish, S*₁, S*₂,…, S*_n are true, and on the most plausible interpretation of Y-lish, ~S´₁, ~S´₂,…, ~S´_n are true. Accordingly, we shall call Hirsch’s view the Possible Communities Model.

Let us formulate Hirsch’s condition as follows:

Possible Communities Model: S’s dispute with S* is merely verbal iff S ought to admit (given the correct theory of linguistic interpretation) that a community typical members of which affirm sentences orthographically and phonetically identical to the disputed sentences S* affirms speak a language in which those sentences are true and vice versa for S* and S’s disputed sentences.

To take an example, suppose there are two English-speaking friends, McA and McB, who agree that there are two drinking vessels on the table which sits between them. McA claims, however, there are two cups on the table, while McB claims there to be one cup and one glass. McA claims “Just as a cat is a kind of animal, a cup is a kind of glass”, while McB denies this.[82]

Hirsch intuits that McA’s dispute with McB is merely verbal (Hirsch 2005). I agree that this is the correct assessment. On Hirsch’s view, their dispute is merely verbal because the most plausible interpretation of the imagined language we might call A-lish is one on which sentences orthographically and phonetically identical to McA’s are true and the same is true mutatis mutandis of B-lish and McB’s disputed sentences.

According to Hirsch, the most plausible interpretation of a community’s language is one that is maximally charitable to the way members of that community speak. This means that there is a defeasible presumption in favor of interpretations on which a large number of sentences that members of a community utter assertively come out true. The most charitable interpretation of a language is one that does not attribute widespread error, whether perceptual or conceptual, to its speakers. Hirsch takes linguistic interpretation to be guided by a cluster of charity principles including:

Charity to perception: one oughtn’t interpret a language in such a way that attributes widespread perceptual mistakes to its speakers.

Charity to understanding: one oughtn’t interpret a language in such a way that attributes widespread a priori or conceptual errors to its speakers.

Charity to retraction: one oughtn’t interpret a language in such a way that attributes to speakers widespread errors of retraction in light of further evidence.[83]

Hirsch’s idea here is that we shouldn’t interpret a community’s linguistic behavior in such a way that has them making false or ridiculous assertions. We should assume that speakers (i) will say true things about what they see right in front of them, (ii) are not grossly mistaken about what their words mean, and (iii) will retract their earlier assertions in light of new evidence in a reasonable way (i.e., they will tend to retract falsehoods in favor of truths and not vice versa).

While Hirsch notes that “considerations of semantic compositionality, complexity, and property-naturalness may play a role” in interpretation (Hirsch 2005, 72), he is chiefly concerned that interpretations respect the above charity to use principles: he says that while “[c]harity to use may not be the only relevant interpretive principle” it is nevertheless, “by far the dominant one” (Hirsch 2009, 243).

An interpretation of A-lish on which

(1) There are two cups here,

came out false would violate charity to perception. For, in order to interpret the A-community as saying something false here, one would have to assume that members of that community are grossly mistaken about a highly visible feature of their environment. Similarly, to interpret

(2) Here is a glass that is a cup

in such a way that it came out false would attribute to speakers of A-lish an a priori or conceptual mistake. For, according to A-lish speakers,

(3) A glass is a kind of cup

is true a priori. So, (2)’s denial is not merely false in A-lish, it is absurd in a way similar to that in which “Here is a dog that is not a mammal” is absurd in English. Similar considerations apply to B-lish and the disputed sentences members of the B-community utter assertively.

Since the A-community’s assertions of ‘there are two cups’ (and related sentences) are true on the most plausible interpretation of A-lish, and the B-community’s assertions of ‘there’s a cup and there’s a glass’ (and related sentences) are true on the most plausible interpretation of B-lish, then, by Hirsch’s lights, McA’s dispute with McB is merely verbal.

In what follows, I want to raise a general problem for Hirsch’s view, independent of the ontological cases against which he takes it to cut. I argue that Hirsch’s criterion for the mere verbalness of disputes is neither necessary nor sufficient for a dispute’s being merely verbal.

Recall that we observed that it is possible for two parties to have a dispute in which both sides are wrong. We elected to call these kinds of disputes both wrong disputes. Both-wrong disputes that are also merely verbal disputes make trouble for Hirsch’s view. Take an adapted case from Hirsch.[84] Phillip and Janelle are having the following argument about the drinking vessel that sits on a table between them.

Phillip: There’s one cup here and it’s a glass.

Janelle: Nonsense, there is a cup there that is not a glass.

Their dispute turns on their idiosyncratic usages of ‘cup’ and ‘glass’. This dispute seems merely verbal. However, suppose that there is no drinking vessel there at all, but a clever holographic projection of the image of a drinking vessel. In such a case, both Phillip and Janelle, though they’re having a merely verbal dispute, turn out to be wrong. Let’s further suppose that Phillip and Janelle know or ought to know (they have very good evidence) that there is no drinking vessel before them, but they ignore their evidence and have the silly cup-glass dispute anyway. Then it looks like each ought not, given their evidence, admit that the other’s disputed sentence is true in the imagined public language associated with their position.

Some cases of both wrong disputes that are also merely verbal disputes show that the condition specified by the Possible Communities Model is not necessary for a merely verbal dispute. Here is the argument:

1. If the Possible Communities Model is correct, then both parties to a merely verbal dispute ought to admit that the other speaks the truth in the imagined language associated with her position.

2. If both parties to a merely verbal dispute ought to admit that the other speaks the truth in the imagined language associated with her position, then Phillip and Janelle do not have a merely verbal dispute.

3. Phillip and Janelle do have a merely verbal dispute.

4. The Possible Communities Model is incorrect.

If Hirsch’s view is correct, then when I have a merely verbal dispute with you, I ought to admit that you speak the truth in the imagined public language associated with your position. But, if what you say involves a mistake about some matter of fact (like there being a drinking vessel before you) and I ought to know that you’re making a mistake about some matter of fact, then I ought not to admit that you speak the truth. In Janelle and Phillip’s dispute, each side ought to admit that the other side speaks falsely. So, on Hirsch’s view, their dispute is not merely verbal. But, intuitively this dispute is merely verbal.

Recall as well that we considered cases where both parties to a dispute are correct. We elected to call these both right disputes. Some both right disputes make trouble for Hirsch’s view. Suppose that you and I are both competing for a fellowship. Suppose the fellowship committee decided to split the fellowship between us. We each get a letter informing us that we’ve won the fellowship. I say, “I won the fellowship” and you hotly dispute my claim saying, “I won the fellowship”. As it turns out, we’re both correct. Supposing that we each have good evidence for thinking that the fellowship was split between us, we each ought to admit that the other speaks truly. It does not seem that we are having a merely verbal dispute, however.

What should I conclude about a possible community that speaks a language in which your sentences are true? Well, our actual community is such a possible community. So, it looks like I should conclude that what you say is true. And the same is true of you. You should conclude that what I say is true in the language associated with my position. But if this is correct, then our dispute meets the sufficient condition specified by the Possible Communities Model. But our dispute is intuitively not a verbal dispute.

The argument is as follows:

1. If the Possible Communities Model is correct, then if S and S* ought to admit that the other speaks the truth in their respective language, then S and S* are having a verbal dispute.

2. If it is the case that if S and S* ought to admit that the other speaks the truth in their respective language, then S and S* are having a merely verbal dispute, then (in the fellowship case) you and I are having a merely verbal dispute.

3. It is not the case that you and I are having a merely verbal dispute.

4. The Possible Communities Model is incorrect.

If the above arguments are sound, then satisfying Hirsch’s criterion is neither necessary nor sufficient for two disputants to be having a merely verbal dispute.

Hirsch anticipates the objection from both-wrong disputes in the following passage:

One may be initially tempted, then, to define a verbal dispute as one in which each side speaks the truth in its own language. That cannot be right, however, since a dispute can be verbal even when the disputants are mistaken about the non-linguistic facts. There may be a verbal dispute about whether Harry is running around the squirrel or merely running around the tree containing the squirrel, when in fact Harry is home in bed and the runner is someone who looks like Harry.[85]

Hirsch formulates his condition so as to avoid the Harry-in-bed counterexample. He says

I would therefore define a verbal dispute as follows: It is a dispute in which, given the correct view of linguistic interpretation, each party will agree that the other party speaks the truth in its own language. This can be put more briefly by saying that in a verbal dispute each party ought to agree that the other party speaks the truth in its own language.[86]

This reformulation avoids the Harry counterexample because both parties to the dispute, while they are mistaken, each ought to admit that the other speaks the truth. After all, in the Harry case, Harry is home in bed and the disputants (presumably) are unaware of this. So, given what they know, each ought to admit that the other speaks the truth.

The reformulated condition, however, does not escape my Phillip-Janelle case. For in that case both disputants ought to know that they are making a mistake about the nonlinguistic facts, but they bullheadedly persist in having their dispute. Notice, even if Hirsch’s condition is formulated as “each party will agree” rather than “each party ought to agree” the counterexample still goes through. Their dispute is verbal even if Phillip and Janelle won’t agree that each speaks the truth in their respective language. So, Hirsch’s condition is not necessary for a merely verbal dispute.

Also, given the stipulation that, in the fellowship case, you and I ought to know that the fellowship has been split, we ought to think that each side speaks the truth. So, even on the revised version of Hirsch’s condition, the argument for insufficiency goes through.

Couldn’t a proponent of the Possible Communities Model accept everything I’ve said so far and still maintain that certain disputes in ontology are defective? I think this is correct. However, the proponent of the Possible Communities Model would have to argue that this model describes a way in which disputes can be defective. Furthermore, this argument could not rely on the claim that such disputes would be merely verbal, in the intuitive sense of that term. Perhaps the Possible Communities Model does describe a way in which disputes can be defective, but this is not immediately clear.

I want to consider a reply on behalf of the Possible Communities Model.[87] I argued above that it is not the case that Phillip and Janelle, while they are having a merely verbal dispute, each ought to admit that the other speaks the truth. A critic might claim that I’ve cheated. There are two senses of ‘ought’ one might intend in stating the Possible Communities Model. If I believe that p, there seems a sense in which I ought to believe that p or q even if my belief that p is irrational. Let’s take a case. Suppose I believe (against all my evidence) that the moon is made of green cheese. Then I ought (in some sense) to believe that either the moon is made of green cheese or aliens rigged the presidential election. I would believe something inconsistent were I to believe both that the moon is made of cheese and it is not the case that either the moon is made of cheese or aliens rigged the presidential election. Given what I believe (irrationally), it would be rational for me to update my beliefs by adding the disjunctive belief. Call this the ‘wide scope’ sense of rationality. There is a sense in which, however, I ought not to believe the moon is made of green cheese at all! It goes against my evidence. On this sense of ‘ought to believe’, I ought not to believe that either the moon is made of green cheese or aliens rigged the presidential election (assuming, of course, that I do not have evidence for this second disjunct). Given the evidence I have, it would be irrational for me to update my beliefs by adding the disjunctive belief. I ought instead to abandon my unjustified belief that the moon is made of green cheese. Call this the ‘narrow scope’ sense of rationality. The wide scope sense prohibits belief in contradictions; the narrow scope sense prohibits belief against one’s evidence.

Here goes the reply on behalf of the Possible Communities Model. Given that Phillip believes that there is a cup in front of him that is a glass he ought to believe, in the wide scope sense, that in the imagined language associated with Janelle’s position what she says is true, and vice versa for Janelle and the imagined language associated with Phillip’s position. It doesn’t make a difference whether Janelle and Phillip are both being irrational insofar as they are ignoring evidence that indicates that there is no drinking vessel between them at all. For given what they believe, they would be irrational in the wide scope sense to believe that the other side speaks falsely in the imagined public language associated with their position.

There are two things to say about this reply. First of all, Janelle and Phillip needn’t believe that there is a drinking vessel between them at all in order to have their above dispute. Both Janelle and Phillip could be playing the devil’s advocate, Janelle arguing for the position that there is a cup there that is not a glass and Phillip arguing for the position that there is a cup there that is a glass. Suppose neither believes the position they are arguing for; they support their respective positions just to give them a fair hearing. On our view of disputes, they would be having a dispute, and it seems intuitively to be a merely verbal dispute.

Perhaps this does not convince, however. Some philosophers will be uncomfortable taking devil’s advocate cases to be real disputes, and would prefer to characterize them as pretend or merely apparent disputes.[88] I find this description implausible. I find it implausible especially in cases where only one party to a dispute is playing the devil’s advocate.

There is a second thing to say, however, about the reply on behalf of the Possible Communities Model. The wide scope sense of ‘ought’ does not require that Phillip and Janelle admit that the other speaks the truth in a case where they both believe contradictory claims ‘going in’ to the dispute. Suppose that Phillip believes that there is no drinking vessel on the table, because he thinks there is a clever holographic image there instead, but he also believes the contradictory claim that there is one glass there and it is a cup. Janelle too believes that there is no drinking vessel but instead there is a clever holographic image of one. She believes, inconsistently, that there is a cup there that is not a glass. What does wide scope rationality here require? Janelle and Phillip ought not to update their beliefs by adding another belief, namely that the other speaks the truth, for this would just add a belief that is inconsistent with something else they believe, namely that there is no drinking vessel on the table before them. Wide scope rationality here clearly requires that Janelle and Phillip get rid of at least one belief, but it does not require that they acquire the further belief that each speaks the truth in the imagined public language associated with their position.

Perhaps cases in which disputants believe inconsistencies are too strange to count as counterexamples to Hirsch’s condition. I think they are not too strange. When disputed claims are complex enough, parties to the dispute will often believe things that are inconsistent without this inconsistency being apparent. In my above case, the inconsistency ought to be transparent given that each disputant understands what a drinking vessel is. But many disputes over complicated claims (perhaps over whether free will is compatible with divine foreknowledge) may involve one or more parties to the dispute believing a contradiction without being aware of it. Surely we do not want to say that in such cases it is impossible for disputants to be having a merely verbal dispute. To get a feel for how this could happen, just imagine a complicated dispute about divine foreknowledge and free will in which one or more parties believe something inconsistent and in which both parties use the term ‘free will’ differently. Surely such disputes occur in undergraduate philosophy classes quite frequently.

Let’s look at an actual case from an undergraduate philosophy class. The names have been changed to protect the innocent. Suppose that Sarah and Paul are having a dispute over whether God can create a stone so heavy he himself cannot lift it. Their dispute goes like this:

Sarah: Of course an essentially omnipotent God could create a stone so heavy he himself could not lift it!

Paul: No way! An essentially omnipotent God could not create such a stone.

On the face of it, this is a substantive dispute about the nature of omnipotence. But, as it happens, Sarah thinks ‘omnipotent’ means something like ‘can bring about any logically possible state of affairs’, whereas Paul thinks it means something like ‘can bring about any state of affairs whatsoever’.

Both Sarah and Paul think that the other is saying something inconsistent. But, as it happens, given what they think ‘omnipotent’ means, if we imagined communities that spoke just like they do and mean by ‘omnipotent’ what they think this word means, we would be imagining communities that say something inconsistent. The Sarah community would be saying that an essentially omnipotent God can do something that renders himself not omnipotent, while the Paul community would be saying that a creature that can bring about any state of affairs whatsoever could not bring about some particular state of affairs.

So, here we have a problem for the Possible Communities Model. Each party to the dispute believes that the other side says something inconsistent. Further, each party does say something inconsistent in the imagined public language associated with their position. So, it looks like it is not the case that each party ought (in the wide sense) to admit that the other speaks the truth in the imagined public language associated with their position. What’s going on here is that each party believes something contradictory.[89] What they believe is not transparently contradictory, like what is believed in the revised Phillip-Janelle case above, however. Wide scope rationality demands that each party update their beliefs in such a way as to avoid contradiction. But, Sarah cannot avoid contradiction by acquiring yet another belief that is contradictory, namely the belief that Paul speaks the truth in the language associated with his position. Nor can Paul avoid contradiction by acquiring yet another belief that is contradictory, namely the belief that Sarah speaks the truth in the language associated with her position. So, wide scope rationality demands that Paul and Sarah give up a belief rather than acquire another belief. So, each ought not to admit that the other speaks the truth. Hirsch’s condition blows a fuse. But, their dispute, since it turns on a difference of opinion about what ‘omnipotent’ means, seems intuitively to count as merely verbal.

I anticipate a reply on behalf of the Possible Communities Model. Would Sarah think that Paul’s claim was inconsistent even if she knew what he meant by ‘omnipotent’? Would Paul think that Sarah’s claim was inconsistent even if he knew what she meant by ‘omnipotent’? If the answer here is yes, then the dispute is not verbal. If the answer is no, then in the wide scope sense each side ought to admit that the other speaks the truth.

I want to stipulate that in my above case the correct answer is ‘yes’. Each side is disposed to think, even upon learning what the other side means, that what they say is inconsistent. What reason could one have for thinking that, if the answer is ‘yes’, the dispute is not verbal? The most obvious reason is that one presupposes the truth of something like the following:

RESOLVABILITY: A dispute is verbal iff the disputants ought to stop having it were each to become aware of what the other means by some operative term in the dispute.

I suspect that something like RESOLVABILITY is driving the intuition that Sarah’s dispute with Paul is not verbal if she is disposed to think Paul’s claim inconsistent even upon learning what he means by ‘omnipotent’, and vice versa for Paul and Sarah’s claim.

Are there any good reasons to accept RESOLVABILITY? The only reason I can think of is that the truth of RESOLVABILITY would account for the truth of the intuitively very plausible claim that merely verbal disputes are disputes that rest entirely on each disputant’s being confused about what the other means by some key term.

Unfortunately, there are more reasons to reject RESOLVABILITY than there are to accept it. We examined some of these reasons above. RESOLVABILITY gives neither a necessary nor a sufficient condition on merely verbal disputes. There are many instances in which getting clear on a key term in a dispute ought to dissolve the dispute, even when the dispute is, intuitively, not merely verbal. Take disputes about the application of a vague term to a borderline case. When disputants get clear on what they mean by a vague term, they sometimes come to realize that it admits of borderline cases and, if the case in question is a borderline case, then they ought to give up their dispute.[90]

Consider a legal case. Jones and Smith are arguing over whether a gun was used in a particular crime. Jones say it was, Smith says it wasn’t. Jones argues that since some criminals were using a gun to prop open a door during a crime, then a gun was used during the crime. Smith says this doesn’t count. If each gets clear on how the term ‘used’ is being used, then their dispute might evaporate. But the dispute does not seem merely verbal.

RESOLVABILITY is not sufficient for mere verbalness of disputes, but it also seems not to be necessary either. Take the (by now very well-worn) case of two people arguing past one another about whether there is a bat in the garage. Tim says there is; Jill says there isn’t. Tim means by ‘bat’ a flying rodent; Jill means a piece of sporting equipment. This dispute is merely verbal. But, given Tim and Jill’s beliefs, they might be rationally disposed to keep arguing even once it’s made clear what each means by ‘bat’ in this context. Suppose Tim thinks that there is a flying rodent in the garage and a baseball bat there as well. Jill thinks there is a flying rodent but there is no baseball bat. It looks like each will keep arguing (and ought, in the wide sense, to keep arguing) even after ‘bat’ gets disambiguated.

Perhaps what we should say here is that Bill and Jill’s dispute is resolved. The dispute they go on to have after they sort out what they mean by ‘bat’ is a dispute about the truth of a different (though phonologically identical) sentence, and thereby a different (and not merely verbal) dispute. I am happy to say this. If this is what is going on in the Bill-Jill case, then it seems like RESOLVABILITY is satisfied in the Sarah-Paul case. Sarah and Paul ought to abandon their dispute over the truth of “An essentially omnipotent God could create a stone so heavy that he himself cannot lift it” once it becomes clear how each is using ‘omnipotent’. But, a dispute over a distinct sentence would arise. So, Sarah and Paul’s dispute is merely verbal. But, each side ought not to admit that, in the original dispute, the other spoke the truth, for each side is affirming something inconsistent (in the imagined public language associated with their position).

If disputes are individuated in terms of the phonetic and orthographic contours of disputed claims, then RESOLVABILITY is false. If, on the other hand, disputes are individuated non-phonetically and non-orthographically, then RESOLVABILITY cuts against the claim that Sarah and Paul’s dispute is not verbal. If RESOLVABILITY is false or disputes are not individuated phonetically and orthographically, I can see no reason to think that Sarah and Paul’s dispute is not verbal.

I think there is another reason to reject the Possible Communities Model. The Possible Communities Model conflates cases of arguing past one another with merely verbal disputes. Take the following case:

Bob: I’m thirsty.

Rob: No it’s not, it’s Wednesday.

Are Bob and Rob having a merely verbal dispute? It is clear that they are arguing past each other, but it may seem infelicitous to characterize their dispute as merely verbal. After all, each agrees about what ‘thirsty’ and ‘Wednesday’ mean in English. The prima facie dispute arises merely because Rob misheard Bob. Their dispute turns on a mistake, of course, but it is unlike the mistakes in other clear cases of merely verbal disputes.

I do not want to hang too much on this objection, however. Suffice it to say that the Possible Communities Model does not allow for a distinction in principle between cases of arguing past one another and merely verbal disputes. If this is the sort of distinction that we ought to preserve, as I have argued above, then so much the worse for the Possible Communities Model.

I therefore conclude that we ought to reject the Possible Communities Model, as it supplies neither a necessary nor a sufficient condition for merely verbal disputes and conflates types of defective disputes we ought to distinguish.

12.6.

Another view of merely verbal disputes is adumbrated by David Chalmers.[91] Let’s call this view the Disagreement about Meaning view. Chalmers states this view as follows:

Disagreement about Meaning: “A dispute over S is (broadly) verbal when for some term T in S, the parties disagree about the meaning of T, and the dispute over S arises wholly in virtue of this disagreement regarding T.”[92]

By ‘broadly’ here, Chalmers means to be talking about a dispute that is merely verbal despite the fact that the disputants do not disagree over the truth value of some proposition or other. Sensitive to Burge’s view about content, Chalmers wants to allow that there is a sense in which I can have a verbal dispute with you about whether Edna has arthritis even if ‘arthritis’ in my mouth has the same semantic value as it does in yours given our membership in the same linguistic community. Suppose that Burgean facts about community-wide usage patterns suffice to fix the meaning of ‘arthritis’ for the both of us. Still, if our dispute goes on like this:

Me: Edna does not have arthritis; she doesn’t have a condition of the muscles!

You: Edna does so have arthritis! She has a joint condition!

It looks like we’re having a merely verbal dispute. Chalmers says,

[w]e can simply note that the dispute between A and B is a pointless dispute in the same way that paradigmatic verbal disputes are pointless, and that resolving the verbal issue about what counts as [‘arthritis’] will resolve their disagreement. We might say that this is a broadly verbal dispute, while the definition in terms of propositions defines a narrowly verbal dispute.[93]

What is a narrowly verbal dispute? A narrowly verbal dispute is, roughly, a dispute that is counted as merely verbal by the Equivocation Model. As we have been using the expressions, narrowly verbal disputes are merely verbal disputes that are also cases of disputants arguing past each other. So, the dispute between Smith and Jones about whether there are some children playing by the bank is a narrowly verbal dispute. On the other hand, a broadly verbal dispute can arise between us even when you affirm some proposition and I deny that very proposition.

Note that Chalmers’ condition above is a condition on verbal disputes and not merely verbal disputes. When is a verbal dispute a merely verbal dispute? Chalmers says the following: “I would say (to a first approximation) that a mere verbal dispute is a verbal dispute in which usage of the key term T (the locus of metalinguistic difference or disagreement) is not itself a central object of concern.”[94] The idea here is that a merely verbal dispute is one in which the disputants are having a verbal dispute, in the above sense, but purport to be arguing about something other than the meanings of words.

What about ‘in virtue of’ in the above condition? This is somewhat tricky. Chalmers urges us to understand this as “an explanatory ‘in virtue of’: the idea is that the metalinguistic disagreement explains the apparent first-order disagreement” and goes on to say that “the relevant sort of explanation should require something stronger than an arbitrary causal explanation.”[95] So, for example, if you and I have a difference of opinion about the meaning of some key term that then causes us to feel hostility towards one another which is then manifested in our having a bitter dispute, our dispute won’t necessarily count as merely verbal.

Chalmers anticipates an objection at this point. He says, “[a]n opponent might reject the in-virtue-of characterization of putative broadly verbal disputes” on the grounds that the opponent “denies that language use requires substantive beliefs about meaning.”[96] Such a theorist might hold that first-order disagreements about applying a particular term themselves explain metalinguistic disagreements, “or they might hold that neither dispute is explanatorily prior to the other.”[97]

It is important to note Chalmers’ reply to this kind of objection, for it betrays a tacit commitment to a ‘no disagreement’ account of merely verbal disputes like the clause explicitly contained in Jenkins’ MVD condition. Chalmers says, “even if we accept the objection, a certain counterfactual variant of the in-virtue-claim appears to be satisfied: if the parties were to agree on the meaning of the key term, their disagreement over S would be removed.”[98] In a footnote, Chalmers spells out a condition on broadly verbal disputes in terms of the counterfactual variant as follows: “The counterfactual variant might say that a dispute over S is broadly verbal when for some term T in S, if the parties were to agree over the meaning of T, then they would (if reasonable) agree over the truth of S.”[99] This counterfactual condition, however, has a no disagreement clause tacitly built into it. Take case of Sally and George from above. Recall that George is confused about the meanings of ‘phlebotomist’ and ‘philologist’. Sally and George believe as follows:

Sally: Believes that there will be a phlebotomist giving a talk at the APA (though she doesn’t mention having this belief)

Believes that there will be a philologist giving a talk at the APA.

George: Disbelieves that there will be a phlebotomist giving a talk at the APA (though he’s describe this belief by saying “There will be no philologist giving a talk at the APA”)

Disbelieves that there will be a philologist giving a talk at the APA (though he’d describe this belief by saying “There will be no phlebotomist giving a talk at the APA.”)

Even if Sally and George were to come to agree about the meanings of the terms ‘philologist’ and ‘phlebotomist’, their dispute would not disappear. This is because they have a difference of opinion about the subject matter of their dispute. We have already argued that there is good reason to think that merely verbal disputes can involve genuine state-disagreement. Chalmers’ apparent commitment to a no-disagreement clause, then, is an intuitive cost of his picture of merely verbal disputes.

Given the above, Chalmers’ condition should be understood as follows:

Disagreement about Meaning: S and S* are having a merely verbal dispute over sentence f iff (i) S and S* are having a dispute over f, (ii) for some term t in f, S and S* state-disagree about t’s meaning, (iii) S and S*’s activity-disagreement over f arises wholly in virtue of their state-disagreement about t’s meaning, and (iv) the usage of t is not a central concern in S and S*’s dispute.

Let us see how the Disagreement about Meaning view handles a case. Suppose that Eli and Ted have a dispute about whether Anyssa’s remark was sarcastic or sardonic. Eli says it was sarcastic and not sardonic, and Ted says it was was sardonic. Suppose that Eli and Ted both agree that Anyssa’s remark was not mocking, but Ted thinks that ‘sarcastic’ and ‘sardonic’ are synonyms. Ted doesn’t know that a remark’s being sardonic implies that it is mocking. It looks like Eli and Ted are having a merely verbal dispute, and the Disagreement about Meaning view adequately explains why. Eli and Ted’s dispute arises wholly in virtue of their disagreement about the meaning of ‘sardonic’. Eli thinks a remark’s being sardonic implies that it is mocking, and Ted disagrees.

Notice, however, that the Disagreement about Meaning view does not characterize the following dispute at merely verbal. Take the Eli-Ted dispute from above. Suppose that Eli thinks that Anyssa’s remark was not mocking, but Ted thinks it was mocking. If Eli and Ted have a dispute over whether Anyssa’s remark was sardonic, then their dispute, presumably, will not arise wholly in virtue of their disagreement about the meaning of ‘sardonic’. Their dispute arises, in part, in virtue of a difference of opinion as to whether Anyssa’s remark was mocking. So, the Difference of Meaning view does not imply that their dispute is merely verbal. Intuitively, this seems like the right result.

I want to suggest a problem for the Difference of Meaning view. The Difference of Meaning view seems unable to account for merely verbal disputes that are not disputes over the truth of some sentence. Let’s take a case. Suppose that Lisa and Frank are at a party and they get into an activity-disagreement. Lisa says, “I loathe Bill” and Frank says, “Bill? I do not loathe Bill.” They go back and forth, each trying to change the other’s attitude about Bill. But, as it happens, Frank is confused. He thinks ‘loathe’ is another word for love. Let’s stipulate that Lisa and Frank both detest Bill. Their dispute is merely verbal, but there doesn’t seem to be any sentence f in which the expression “loathe” appears about which they apparently disagree. One might object, saying that “I loathe Bill” as Lisa utters it is such a sentence. We can reformulate their dispute, however, as follows:

Lisa: Lisa loathes Bill.

Frank: Bill? No way! Frank does not loathe that guy.

The reformulated dispute is, admittedly, pretty stilted, but there is no sentence f in which ‘loathe’ appears that is the object of Lisa and Frank’s dispute.

Perhaps this doesn’t convince. It may seem like Lisa and Frank are not really having a dispute at all. It might be objected that their case is not a dispute because they are not having an activity-disagreement. I think this is wrong. Lisa has an attitude about Bill, she takes Frank to have a non-cotenable attitude about Bill, and she is disposed to try to change his mind. Frank has an attitude about Bill, he takes Lisa to have a non-cotenable attitude about Bill, and he is disposed to try to change her mind. This looks like an activity disagreement. If Lisa and Frank start making arguments for holding their attitudes, then they are having a dispute. But there needn’t be any one sentence containing the expression ‘loathe’ that they activity-disagree about. What is going on in the Lisa-Frank dispute is that Lisa and Frank are having an activity disagreement about Bill, not about some sentence. The Difference of Meaning view fails to capture this because it is formulated in such a way as to presuppose that all disputes are disputes over the truth of some sentence. This is a mistaken view about disputes, however.

It may be objected, however, that so long the Difference of Meaning view is understood as a view only about disputes over the truth of some sentence, it is adequate. Indeed, this seems to be the way Chalmers intends the view, for he builds in ‘A dispute over S’ into the condition. Is Chalmers’ condition correct when understood as restricted in this way?

This depends on how the ‘in virtue of’ locution is made precise. If something like the counterfactual version of ‘in virtue of’ is intended in the Disagreement about Meaning view, then I think the view is incorrect. The view is incorrect on this interpretation of ‘in virtue of’ because this interpretation smuggles in a no-disagreement clause into the account of merely verbal disputes, and we have good reasons to reject such a clause.

12.7.

Where does all of this leave us? I am inclined to think that some version of the Disagreement about Meaning view can be made to work. This view, however, will have to be modified to count as merely verbal disputes some disputes that are not explicitly about the truth of some sentence. Also, the view will have to be understood as not incorporating a no disagreement clause. I suggest the following:

The Divergent Uses View: S and S* are having a merely verbal dispute iff (i) S and S* are having a dispute, (ii) there is some operative term t such that S and S* are using (or are disposed to use) t in divergent ways, and (iii) the divergence of use explains S and S*’s dispute.

Some clarificatory remarks are in order. Clause (i) requires that S and S* be having a dispute. On our preferred view of disputes, it is not required that there be some sentence f such that S affirms f and S* denies it in order for S and S* to be having a dispute. S and S* have a dispute just in case they are having an activity-disagreement characterized by the mutual giving of arguments in the discursive sense.

Clause (ii) requires a divergence in use of some relevant term. This is a bit tricky. The relevant term needn’t be used by both disputants in having the dispute. Consider the George-Sally dispute. Suppose their exchange goes as follows:

Sally: There will be a philologist at the APA.

George: No there won’t.

Sally and George do not diverge in their uses of ‘philologist’ here in the sense that Sally tokens the word and intends it one way and George tokens it and intends it in another way. George doesn’t token the word at all. So, ‘using t in divergent ways’ ought to be understood as follows:

S and S* are using t in divergent ways in dispute D iff (a) S and S* both token t in the course of D and S intends it to mean something different from S*, or (b) S and S* are disposed to token t and intend it to mean something different. (

Here b) is intended to capture those cases in which a dispute turns on divergent uses of some term that is not explicitly used in framing the dispute. Take a dispute about whether God can coexist with evil that turns on a verbal difference with respect to the meaning of ‘freedom’. In such a dispute, the word ‘freedom’ and its cognates might never appear. But, if the dispute turns on a verbal difference, intuitively it should be counted as a merely verbal dispute.

The notion of explanation in clause (iii) is not to be understood in the counterfactual way that Chalmers suggests in reply to his imagined critic. How then, is this notion to be understood? Perhaps something of the counterfactual account can be salvaged. It will not do to say that a dispute arises in virtue of a divergence in use of t just in case were the disputants to agree on the use of t, they would abandon their dispute. But, something in the vicinity of this condition seems accurate. When a dispute arises in virtue of a divergence in use of some term t, were the disputants to agree on how t is to be used, they each ought to admit that there was something wrong with their original dispute. They may continue to have an activity disagreement, but they should acknowledge that the original dispute was defective. The idea here is that the relevant notion of ‘in virtue of’ is something that explains the disputants’ counterfactual predilection towards having an ‘Oh my gosh, that was silly’ feeling upon each of them learning what the other meant by some operative term in the dispute.

This counterfactual condition is not too precise, but it is able to handle cases where a merely verbal dispute about subject matter M occurs despite a genuine state-disagreement about M.

No doubt a clever philosopher will be able to think of counterexamples to the Divergent Uses View. I do not present the view as immune to counterexamples. Rather, I suggest that it is able to handle all of the cases I have suggested the other views cannot. So, we should prefer the Divergent Uses View.

Chapter 5

INTRODUCTION TO PART TWO

We now turn to a discussion of two test cases of disputes in philosophy. First we shall examine the much maligned dispute over the nature of mereological composition. Just when is it the case that some things make up or are parts of some further thing? Many philosophers have claimed that there is something fundamentally wrong with disputes over the proper way to answer this question. We will look at one particular dispute: the dispute between the Universalist and the Nihilist about material composition.

After having looked at this dispute in some detail, we will next turn our attention to the famous dispute between David Lewis and Alvin Plantinga about the issue of transworld identity. Lewis famously said that each concrete individual is a part of one world only, while Plantinga famously defended the view that one and the same concrete individual can exist in a number of different worlds.

I do not know of any literature in which philosophers have argued that the Lewis-Plantinga dispute is defective in some way. Accordingly, our discussion of it shall be somewhat shorter than our discussion of the Universalist-Nihilist dispute.

It is important to keep in mind that though we will be examining the details of two highly specific disputes in philosophy, one of our main goals is to make some observations about philosophical disputes in general. When is a philosophical dispute defective? When ought we to abandon a philosophical dispute? Are all defective philosophical disputes ones that we should abandon? These are the sorts of questions that we want to answer.

Chapter 6

DISPUTES OVER COMPOSITION: AN INTRODUCTION

We are now prepared to begin applying some of the developments of Part One to our first text case. I will start with an example of a dispute that many philosophers and non-philosophers think is defective: the dispute between so-called Nihilists and Universalists about mereological composition. Many philosophers have argued that there is something wrong with this dispute.[100] Further, philosophers who have argued that there is something wrong with this dispute typically have taken themselves to have shown that there is something type defective about it. In what follows, we shall examine the allegation that the Nihilism-Universalism dispute is type defective. There are a number of arguments that purport to establish this claim. We shall sort these arguments in terms of the distinctions from Chapter Two between different sorts of defective disputes.

Some stage setting is required before we begin our treatment of the Nihilism-Universalism dispute. It is very natural to see this dispute as arising from two seemingly different ways to answer what Peter Van Inwagen calls The Special Composition Question (SCQ).[101] Van Inwagen states the SCQ as follows:

SCQ: Suppose one had certain (nonoverlapping) objects, the xs, at one’s disposal; what would one have to do—what could one do—to get the xs to compose something?[102]

The intuitive idea is supposed to be this. On my desk I have a copy of Perry, Bratman, and Fisher’s introductory philosophy textbook. My copy of this book is made up of some things: the pages, the front cover, the back cover, and the binding. The pages, the covers, and the binding are all parts of my copy of the book. These several things make up, or compose, my book. Think of the various parts of the book before assembly by the bookbinder. Here are some chapters, over there some binding materials, and over there are the front and back covers. Pretheoretically, these disparate things do not (yet) compose anything. The bookbinder’s act of binding them together in the appropriate way, it seems, brings something composed of these things into existence. The SCQ asks, what, in general, are the conditions under which some things compose some other thing?

One (perhaps surprising) answer to the SCQ is that there is nothing one could do to get some non-overlapping xs to compose something. We will say that something is mereologically simple just in case it has no proper parts. We will say that something is mereologically complex just in case it is not mereologically simple. According to the view suggested by the above answer to the SCQ, there are no mereologically complex objects; everything that exists is mereologically simple. Call this view Nihilism.

Nihilism is a pretty radical view. Think about most of your favorite objects of everyday acquaintance. These are all things that, if they exist at all, have proper parts. The table I’m sitting in front of now, for example, has some legs, a glass table top, some filing drawers underneath, some wheels on its legs, and so on. All of these things are parts of the table. But, the table is divisible into further parts. The wood fibers that make up the legs have a complex microphysical structure; these fibers are made up of countless tiny parts, molecules, which themselves are made up of even tinier parts, atoms, which themselves are further made up of even tinier parts, subatomic particles. According to the Nihilist, the only objects that exist are these fundamental, subatomic particles that themselves have no further parts. This is pretty hard to believe! After all, it seems that what I’m sitting in front of is an object; I interact with it causally; it affects my sense organs, and so on. It looks like there is a single object in front of me made up of all those parts; anybody can see it, right?

Not so says the Nihilist. For there is not even a me to do the seeing.[103] After all, if I existed, I’d be the sort of thing that has proper parts: hands, feet, a brain, etc. But, nothing that exists has proper parts, according to the Nihilist, so I do not exist. This is pretty had to swallow.

The Nihilist instead holds that there are mereological simples that are arranged in particular ways. One of the ways in which simples can be arranged is person-wise.[104] While it is true, according to the Nihilist, that these simples do not compose anything, nevertheless their collective action can explain certain features of the world. It is in virtue of the collective behavior of some simples arranged brain-wise that are appropriately related to some more simples arranged human-skull wise and also related to some simples arranged spinal-column wise, leg-wise, etc., that it is appropriate to say that the simples arranged human-being wise that we collectively describe by using the proper name ‘Gerald Marsh’ engage in some complex activities that we describe using expressions like ‘sitting before a table’ or ‘believing that there is a table’ and so on.

Talking in terms of the simples arranged thus and so vocabulary would be a phenomenal pain. So, the Nihilist allows, there is a sense in which we can say that fast and loose talk about persons, tables, mountains, rocks, and other medium sized goods is correct, despite that, strictly speaking, none of this thought or talk is true. So, when I say “I am sitting in front of a table”, according to the Nihilist, what is asserted is false because it implies the existence of a table and of me. Nevertheless, there is a sense in which it is still correct for me to say this, given that my sentence is translatable into an idiom that does not carry these false existential implications.[105]

Contrast Nihilism with another sort of answer to the SCQ. Some philosophers say that the right way to answer the SCQ is to say: “There’s nothing at all you have to do to get some xs to compose something. Any time some xs exist they compose something: the mereological fusion of the xs.” So, let’s take the example of the bookbinder. Before she begins work, there is already something that the chapters, binding materials, and cover compose. While it is true that we would not call this thing a book, the mere fact that our parochial language does not provide a general count noun for things of this sort is irrelevant to their existence. The mereological fusion of the pages, covers, and binding exists even before these objects are fastened together. It is only after the fastening, however, that we call this thing a book.

Universalism is also a pretty radical view. It implies that any objects, be they mereologically simple or complex, uniquely compose some object. So, consider your entire body, the Eiffel Tower, every can of Coca Cola, and Barack Obama. According to Universalism, there is some unique object that has these, and only these objects as parts.[106] What’s this object like? How much does it weigh? Why don’t natural languages have terms to refer to it and things like it? These are some pretty challenging questions for the Universalist. It is just hard to believe in all of the objects that the Universalist thinks there are.

The Universalist can explain away some of the counterintuitiveness of her view by saying that natural languages contain quantifiers that admit of domain restriction. Since human beings can get around fairly well without naming or talking about most of the objects that Universalists think exist, it comes as no surprise that our languages do not have names for them and that we tend to leave them out when we quantify. For the most part, we just ignore these objects. But, the fact that we ignore some object does not entail that it does not exist.

So far our theoretical choices are between a feast of famine and an embarrassment of riches. Many would be inclined to say, “I’ll have neither.” I do not want to try to adjudicate between some third alternative and Nihilism or Universalism here. Our goal is to determine whether the dispute between Nihilists and Universalists is defective in some way. There are many theoretically interesting, well motivated, and well defended alternatives to these two views, but limitations of space preclude a thoroughgoing treatment of all of the possible views on mereological composition.[107]

Another kind of obvious thing to say about these two views is that they are both non-starters. They arise as attempts to answer the SCQ, but this question is silly. There are (at least) two sorts of ways to think that the SCQ is silly. On the first way, the SCQ is like the Special Tootsie-Pop Question (STQ):

STQ: How many licks does it take to get to the tootsie roll center of a Tootsie-Pop?

If STQ is intended as “On average, licking just once at a time, how many times would you have to lick a Tootsie-Pop in order partially to expose the tootsie roll center?” then it looks like it is pretty hopeless to try to answer it.[108] Given what we know and could be expected to know, there is no way to adjudicate between rival answers to STQ.[109] Perhaps we could conduct a series of experiments to answer STQ, but the likelihood of anyone’s doing this is pretty minimal.[110]

Another way that the SCQ could be silly is if it is relevantly like the Special String Question (SSQ):

SSQ: How long is a piece of string?[111]

The SSQ seems to have no answer unless it is interpreted to mean something like “What is the shortest length a sample of string could be?” or “How long is this particular piece of string?” or “what is the average length of a piece of string?” If SSQ just means “In general, how long is a piece of string?” it seems to have no answer.

I want to say two things to the theorist who is inclined to give either of these sorts of response to the SCQ. First, Universalism and Nihilism do not find their only motivation in their both being answers to the SCQ. One way to get Nihilism and Universalism going is to ask, “Are there tables?” The intuitive answer to this seems to be ‘yes’. But now consider a paradigm specimen of a table, perhaps your favorite table. Call it Tably. Now imagine Tably loosing a few small parts, say I gouge some of the material that forms one of the legs away with a screwdriver. Does Tably still exist? Presumably yes. But suppose I keep on removing parts in a continuous sort of fashion, chipping away ever more pieces of the table until all we are left with is a pile of table shavings. Further suppose that none of the table’s microscopic parts are destroyed. At the end of this procedure Tably presumably is no more. But what stage in the process did Tably fail to survive? Specifying a single point or even a range of points would seem arbitrary. Neither the Nihilist nor the Universalist has to do this, however. The Nihilist says that there was no table to being with, just the microscopic bits of matter. The Universalist also says there is no net change in the number of existents, for each of the microscopic parts survives the process, and so, even at the end they still compose something. So, Nihilism and Universalism can be motivated by their both providing non-arbitrary answers to the question “When did the table cease to exist?”

Second, even if it is correct to say that the SCQ is silly, examining the Universalist-Nihilist dispute is still worthwhile. Why? Because some philosophers get the feeling that Universalists and Nihilists are not even disagreeing (in the state sense) about anything at all, while others think there is a genuine disagreement. Even if it is true that one of the sources of the Univeralist-Nihilist dispute is an attempt to answer a silly question, the question of whether there really is a disagreement is not answered.

In what follows, we will examine several arguments designed to show that disputes between Universalists and Nihilists are defective disputes. We will begin with those arguments designed to show that there is simply no fact of the matter as to which view is correct.

Chapter 7

UNANSWERABLE QUESTIONS

Amie Thomasson has recently argued that the dispute between Nihilists and Universalists is not an example of a “real debate about genuine issues of substance”.[112] If Thomasson is correct then disputes between theorists of these stripes are defective disputes. Thomasson argues that the Special Composition Question, when properly understood, turns out to be an unanswerable question in the sense that “no straightforward answer to it, stated in the same terms as the original question, is truth evaluable” and “where this failing is in principle, not a reflection of mere epistemic shortcomings but of deficiencies in meaning.”[113] According to Thomasson, ontologists who attempt to answer the Special Composition Question are like people who try to answer the question “How long is a piece of string?” or “How long did your idea sleep?” There is something wrong with these questions: no answer to them, stated in terms of the original question, is truth evaluable.[114]

Thomasson’s argument does not merely target the SCQ. Rather, she purports to show that there is no way to motivate either Nihilism or Universalism without asking questions that are unanswerable in the above sense. If Thomasson is right, then disputes between Nihilists and Universalists are what we have been calling no fact of the matter disputes.[115] If this is correct, then these disputes are defective, and they ought to be abandoned. If Thomasson’s argument is sound, these disputes are not merely token defective, they are type defective. In what follows, we will examine Thomasson’s argument for the conclusion that the disputes between Nihilists and Universalists are type defective no fact of the matter disputes.

In framing Thomasson’s master argument, it is best to look at a specific example of a dispute between a Nihilist and a Universalist. Then we can generalize her results to cover all cases of such disputes. So, suppose that our two theorists are faced with the following scenario. There is a vacuum chamber in which there are two, and only two, mereologically simple particles. Let’s call our theorists Peter V. and David L. Peter V. says “There are just two things there: the two simples.” David L. says, “No way, there are three things there: the two simples and their mereological fusion.”

Thomasson intuits that there is something wrong with this dispute and disputes like it. Her master argument is something like this. If Peter V. and David L. are having a real dispute over an issue of substance, then there must be some way to make sense out of questions like “How many things are there in the vacuum chamber?” or “Is there something in the chamber composed of the two simples?” or “Under what conditions do the two simples make up some further thing?”. Let’s call these questions and questions relevantly like them Thinghood Questions. Thomasson thinks Thinghood Questions, when interpreted in the way that the ontologists want to interpret them, are unanswerable.

Here is the structure of her master argument:

If Universalists and Nihilists are having real disputes about genuine issues of substance, then Thinghood Questions are answerable questions.
Thinghood Questions are not answerable questions.
Universalists and Nihilists are not having real disputes about genuine issues of substance.

In the sections that follow, we shall fill in some of the details. Particularly, we shall look at Thomasson’s reasons for accepting premise (2) in the above.

Before examining Thomasson’s argument for premise (2), we must first get clear on some views about reference and meaning that undergird her position. Thomasson holds a view on which conceptual analysis has a central role to play in answering metaphysical questions. She remarks that

[s]ome might think that the very idea that conceptual analysis should play a central role in metaphysics is implausible, as causal theories of reference show that our terms may refer without the need for speakers to have any concept in mind about what kind of thing is to be referred to, making conceptual analysis irrelevant to the real truths about the natures of the objects and kinds referred to.^{^[116]}

The causal theory of reference Thomasson characterizes in this passage is one motivated by considering examples of terms whose reference is fixed by a language using community’s exposure to instances of natural kinds such as samples of H₂O, cases of arthritis, or members of biological kinds such as tigers, mammals, and so on. On such a causal theory of reference, that the English term ‘water’ serves to refer to conglomerations of H₂O molecules is a fact determined by the nature of the samples to which early users of this term intended to refer. In an important sense, the nature of the world (or the bits of it to which early users of ‘water’ meant to refer) did all the work in determining the reference of the term ‘water’. It is pretty hard to see, if something like this picture best describes the way general terms (like ‘tiger’, ‘water’, ‘table’, ‘person’, and so on) come to refer, how it could be the case that conceptual analysis could tell us anything about the nature of the referents of such terms. Since metaphysics is concerned with the nature of reality, it seems like conceptual analysis is of no help to the metaphysician.

Take an ontological case. Advocates of Nihilism say that there is nothing mereologically complex for the word ‘table’ to refer to. Universalists say that ‘table’ does refer to some things that are mereologically complex. If a causal theory of reference for terms like ‘table’ is correct, then it looks like conceptual analysis has little role to play in determining who is right. In order to determine whether the referents of ‘table’, if there are any, are mereologically complex or not, all we need to do is look at the conditions that prevailed in the instances that serve to ground the reference of ‘table’. If the stuff present in these grounding situations is mereologically complex, then the term refers to things that are mereologically complex. If not, then the term does not refer to anything mereologically complex.

Thomasson thinks that “while causal theories of reference may have persuasively established that causal and contextual relations play a key role in establishing reference, we have good reason to think that this can’t be the whole story.”^{^[117]} Why not? Thomasson suggests that the ‘qua problem’ forces causal theorists to accept what she calls a ‘hybrid’ account of reference. What is the qua problem? Thomasson illustrates the problem by means of a thought experiment:

So, suppose a new term, ‘fillow’ is introduced. Even if grammar makes it clear that ‘fillow’ is to be a noun rather than an adjective or an adverb, it may remain completely indeterminate whether ‘fillow’ is to refer to a piece of copper, sum of copper atoms, a sculpture, an exhibition, a location, etc., unless speakers somehow disambiguate the ontological type of entity to be referred to. In the standard cases considered by the causal theories of reference, in which we ask what determines which person or which species a term refers to, this problem is hidden, since in formulating the question this way we already presupposed that the term is, e.g., a name for a person.^{^[118]}

If speakers in a grounding situation do not disambiguate, say, by having in mind an ontological category of things to be referred to by a general noun or noun phrase, then either the ontological category of the things referred to is indeterminate or reference simply does not occur. If the ontological category of the referents is indeterminate, then, according to Thomasson, a number of metaphysical questions about the referents will be unanswerable questions.

Suppose that in our grounding situation, our speaker points to a region of space before her and says, “Let that and things like it be called ‘fillows’”, but does not have in mind any ontological category to which the referents belong. If reference succeeds, there will be some questions about fillows that are unanswerable: “Can a fillow survive being heated to 1400 degrees Fahrenheit?”, “Can a fillow survive the loss of a few small parts?”, “Is a fillow mereologically complex?” and so on.

How mistaken can a speaker be about the ontological category of thing to which she refers before we say that reference is unsuccessful? According to Michael Devitt, “we have to draw the line somewhere, saying that some sort of error invalidates reference. Reference failure is possible and, in the case of illusion at least, it is actual.”^{^[119]} So, in order for reference to occur in a grounding situation, it seems that the grounder needs to have in mind some “very general category” to which the intended referent belongs.^{^[120]} Of course, as Devitt points out, “[t]here is an element of arbitrariness in our determination of these categories.”^{^[121]} So, whether we count a grounding situation as establishing reference will depend on a number of features of the context in which the attempted grounding takes place. Depending on the intentions and goals of the speaker, we might be willing to say that reference occurs, even though the referent does not belong to the very general category the speaker took it to belong. But, the speaker cannot be too wildly mistaken about the ontological category of the entity to which she intends to refer.

How are we to make up our minds as to whether reference occurs? Does reference occur when a speaker is mistaken? Perhaps there is no good answer to this question. What seems to be the case, however, is that a speaker have some general category of thing in mind in order to secure reference.

Let’s suppose that something like Thomasson’s picture of reference is correct. What are the implications of this view for ontology? What does this hybrid causal theory of reference imply about the dispute between Peter V. and David L.? Before we can tease out the implications of Thomasson’s remarks on reference, we need to fill in a few more of the details.

According to Thomasson, in cases where the “causal and physical structure of the world alone cannot disambiguate among potential referents,” there are “two sorts of rules of use for our nominative terms [that] aid in ontological disambiguation.”^{^[122]} Thomasson calls these sorts of rules “frame-level application conditions” and “coapplication conditions”. Frame level application conditions are “very basic conditions under which the attempted grounding would or would not take place.”^{^[123]}

Consider an example. Suppose ‘shmelkie’ is introduced as a general noun term for medium-sized lumps of stuff of a certain kind (say medium sized lumps of copper), ‘shmelkie’ will refer on a given occasion only if such a lump is present. If it is introduced as a name for statues of a particular kind (say, statues made of copper), then it will refer on a given occasion only if there is a lump of copper arranged or selected by someone with artistic intentions present.

The other sort of rules that govern our use of nominative terms are coapplication conditions. These are rules of use that determine when it is appropriate to say that two applications of a general term, F say, are used to refer to one and the same F on both occasions. Suppose ‘shmelkie’ is introduced as a general name for statues made of copper. Given the application conditions for ‘shmelkie’, it correctly applies only if a mass of copper has been shaped or selected by someone with artistic intentions. Suppose this application condition is met at a time t₁ so that “here’s a shmelkie” is correct. If the statue referred to at t₁ is melted down at t₂, given the rules of use for terms that refer to statues, it is incorrect to say of the melted mass of copper at t₂, “this is the same shmelkie” as the one at t₁.

Thomasson uses the term ‘sortal’ to refer to terms that are associated with both of the above kinds of rules.^{^[124]} Questions about the persistence and existence of entities that fall under a sortal term, e.g. ‘table’, are answerable only insofar as the application and coapplication conditions succeed in individuating the ontological kind that the term refers to. So, if the rules of use for ‘table’ leave it ambiguous whether ‘table’ refers to mereological fusions of microscopic parts or to relatively stable, cohesive lumps of stuff, then certain questions about tables, e.g. could a table survive the loss of some parts, will be unanswerable in the sense that no answer to these questions will be truth evaluable.

So, if Thomasson is correct, then sortal terms come along with some minimal conceptual content. Let F be a sortal term. Where F’s minimal conceptual content leaves it ambiguous whether some xs could compose an F, the question “do these xs compose an F or not?” is simply an unanswerable question.

We are now in a position to see what Thomasson thinks is wrong with Thinghood Questions. Answering Thinghood Questions requires one to specify some conditions under which some things compose some further thing or to say when some situation contains a thing composed of various parts, and so on. These questions are intelligible, and answerable, only if ‘thing’ is intelligible. So, how is ‘thing’ being used in Thinghood Questions? Thomasson’s argument that Thinghood Questions are unanswerable is an argument by exclusion. She specifies three different ways that ontologists could be using ‘thing’ in asking Thinghood Questions: a sortal use, a covering use, and an (allegedly) neutral use. Thomasson then argues that ontologists could not be employing the sortal or covering uses of ‘thing’, so (if their question is intelligible), they must be employing an allegedly neutral use of ‘thing’. But, Thomasson argues, on the allegedly neutral use, Thinghood Questions, including the SCQ, are unanswerable. Thomasson’s argument, then, proceeds in three stages. First: she argues that ontologists cannot be employing the sortal use of ‘thing’, then she argues that they cannot be employing the covering use of ‘thing’. Having disposed of these two uses, she argues that Thinghood Questions are unanswerable on the neutral use of ‘thing’. We shall examine each stage of the argument in turn.

How are we to interpret ‘thing’ as it appears in Thinghood Questions? One answer is that ‘thing’ is being used as a very general, high level sortal term. So, like ‘animal’, ‘material object’, ‘chair’, and so on, ‘thing’ is being used as a general noun. Thomasson argues that ontologists do not have a sortal use of ‘thing’ in mind when they ask Thinghood Questions. For, if ‘thing’ takes a sortal use, then it presumably comes along with frame level application and coapplication conditions. But, if ‘thing’ is being used in accordance with rules of use that specify application conditions, then if two theorists disagree about whether it applies in a given case (say, when there are some atoms arranged table-wise), then either they are using ‘thing’ with different application conditions or one of them is just mistaken about the nonlinguistic facts that obtain. If vying theorists use ‘thing’ with different application conditions, then

all disputes about composition would turn out to either involve participants talking past each other (by using ‘thing’ differently) or at best be conceivable as shallow disputes about the best way of turning the term into a sortal (not as deep disputes about what there is).^{^[125]}

Let’s return to our Peter V.-David L. dispute from above. Peter V. says “There are two and only two things in the vacuum chamber” while David L. says “There are exactly three things in the vacuum chamber.” What could explain the difference of opinion here? It cannot be that Peter V. or David L. is making a mistake about whether the application conditions are satisfied; for both know precisely what the non-linguistic facts are: there are two (and only two) mereologically simple objects in the vacuum chamber. So, if the application conditions for ‘thing’ require that an object is mereologically simple, then they are not met by something that has both simple particles as parts. If the application conditions for ‘thing’ apply whenever there are any simple particles at all, then the conditions are clearly met. Suppose that both theorists are using the term ‘thing’ in such a way that ‘thing’ applies to each simple taken by itself and to the two simples when taken together. It is unreasonable to think that such talented philosophers as Peter V. and David L. could be mistaken about whether this application condition is met in the above case. On the other hand, if both theorists are using the term ‘thing’ with application conditions that are met only by each simple taken apart from the other, it seems hard to think that one of them is making a mistake about whether the application conditions are met, for they know that the vacuum chamber contains only two things that are simples.

The difference must be that Peter V. and David L. are following different rules of use for applying the term ‘thing’. Peter V. only applies ‘thing’ in cases where its purported referent is mereologically simple. David L. applies ‘thing’ in some cases where its purported referent is mereologically complex. But now, it looks like either we have a case of two disputants talking past each other or arguing over the best way to make their use of ‘thing’ precise. If the former is the case, what we have is a defective dispute; if the latter is the case, we do not have an ontological dispute, for it is a dispute about how best to use a word, not about what exists.

The first stage of Thomasson’s argument can be stated more formally as follows:

If ‘thing’ is used as a sortal by Peter V. and David L., then either they associate the same application conditions with ‘thing’ or they do not.
If they associate the same application conditions with ‘thing’, then at least one of them is making a mistake about whether the application conditions are satisfied.
It is unreasonable to think that either Peter V. or David L. is making a mistake about whether the application conditions are satisfied.
If Peter V. and David L. do not associate the same application conditions with their use of ‘thing’, then they are talking past one another or they are not having an ontological dispute.
If ‘thing’ is used as a sortal by Peter V. and David L, then their dispute about how many things are in the vacuum chamber is a case of talking past each other or it is not an ontological dispute.

If Thomasson’s argument is sound, her result should generalize to any ontological dispute between any Nihilist and any Universalist. All we need do is replace the name ‘Peter V.’ with the name of some Nihilist, replace the name ‘David L.’ with the name of a Universalist, and replace ‘their dispute about how many things are in the vacuum chamber’ with a phrase that picks out the relevant dispute. If Thomasson’s argument is sound, then each such substitution instance of it will also be sound.

So, if Thomasson’s argument is sound, we ought to conclude that Nihilists and Universalists cannot be grounding their disputes by appealing to Thinghood Questions interpreted as involving a sortal use of ‘thing’. Perhaps there is some other way to interpret ‘thing’ in these questions.

Perhaps the Nihilist and the Universalist intend ‘thing’ in a covering sense. What is a covering sense? According to Thomasson, words like ‘object’ or ‘thing’ can be used “as a place-holder for any genuine sortal term” in such a way that they are “guaranteed to apply given the application of any genuine (first-order) sortal term…”[126] So, on a covering use of ‘thing’ or ‘object’

[I]f there is an eyelash, a six-pack of beer, or a domestic dispute, we are licensed to infer that there is something in my eye, something in the fridge, or something going on at the Johnson residence.[127]

So, on this use of ‘thing’, the Peter V.’s claim that there is nothing composed of the two simples can be interpreted as “There is no table, chair, mereological fusion, etc…, composed of the two simples.”

Thomasson thinks that rival ontologists cannot appeal to a covering use to make sense of Thinghood Questions, however. Her argument for this claim proceeds in two stages. First, she sets out to prove that ontological disputes over specific existence questions (e.g., do tables exist?) must be grounded in disputes over generic existence questions (do particles arranged table-wise compose some thing?). What reason is there to think this?

According to Thomasson, specific existence questions are to be resolved by means of a two step process. Let’s take the specific existence question “Are there tables?” The first stage in answering this question is to determine the frame level application conditions for the sortal term ‘table’ as used by speakers of English. Completing this step involves no more than a bit of conceptual analysis. The second step is to determine whether “the history of uses of that term lead back to a grounding situation in which the term’s application conditions are fulfilled.”[128] If the answer here is ‘yes’, then the answer to our existence question “Are there tables” is ‘yes’. If the history of uses does not lead back to a grounding (or regrounding) situation in which the application conditions are satisfied, then the answer is “No, there are no tables.”[129]

Here the serious ontologist encounters a problem, according to Thomasson. Let’s take our case of Peter V. and David L. David L. says that there is a mereological fusion in the vacuum chamber. Peter V. denies this. Both are making a specific existence claim. But, either they are using ‘mereological fusion’ with the same application conditions or they are not. If they are not using this term with the same application conditions, Thomasson claims, they are talking past one another. If they are using this term with the same application conditions, then they are not having an ontological dispute, for the proper way to determine the truth value of the claim “There is a mereological fusion here” is just to check whether the application conditions are satisfied, not to give an ontological argument.

Thomasson concludes that, if there is an ontological dispute going on at all between Peter V. and David L., then it cannot be a dispute about the truth value of the specific existence claim “There is a mereological fusion here”. It must be a dispute about the generic existence claim “Is there some thing here composed of the simples”. Here is the argument in schematic form:

If Peter V. and David L.’s dispute in the vacuum chamber case is grounded in their disparate opinions about how to answer the specific existence question “Is there a mereological fusion here”, then their dispute is either a case of talking past one another or it is not an ontological dispute.
Neither disputant would admit to talking past the other or to having a non-ontological dispute.
Peter V. and David L.’s dispute in the vacuum chamber case is not grounded in their disparate opinions about how to answer a specific existence question.

If the above argument is sound, then there is a problem with the covering use of ‘thing’ in the David L.-Peter V. dispute. Covering uses of ‘thing’ are place holders for specific sortal terms. But, we have just seen that David L. and Peter V. cannot be having a genuine ontological dispute about whether some particular sortal term (in this case ‘mereological fusion’) applies. So, their dispute cannot be grounded in a covering use of thing. Again, these results are supposed to generalize.

If the above arguments are sound, we are left with just one option for making sense out of the question “Is there some thing composed of the two simples?” that is at bottom of the dispute between Peter V. and David L. The use of ‘thing’ here must be neutral. That is, there can be no frame level application conditions associated with ‘thing’ (otherwise it would be a sortal term), nor can it be a stand-in for some genuine sortal term or range of sortal terms (otherwise its use would be a covering use). But, when ‘thing’ is stripped of frame level content, Thomasson claims, it cannot contribute in the appropriate way so as to render “Is there any thing here composed of the simples?” an intelligible question: “existence questions stated using such a ‘neutral’ use of ‘thing’ or ‘object’ are defective and unanswerable questions.”[130]

Why should we think this? On Thomasson’s view, “existence claims of the form ‘there is a P’ or ‘Ps exist’ are true just in case the frame-level application conditions for the term ‘P’ are fulfilled in the grounding situation(s).”[131] But, if this is the case and ‘thing’ has no application conditions, then any answer to an existence question stated using ‘thing’ will not be truth evaluable. But this is just what it means for an existence question to be unanswerable.

Here is the final stage of the argument in schematic form:

If David L. and Peter V. are using ‘thing’ neutrally, then ‘thing’ has no application conditions.
If ‘thing’ has no application conditions, then any answer to ‘Is there some thing here composed of the simples’ is not truth evaluable.
If no answer is truth evaluable, then the question is unanswerable.
If the question “Is there some thing here composed of the simples” is unanswerable, then David L. and Peter V. are not having a real dispute about issues of substance.
If David L. and Peter V. are using ‘thing’ neutrally, then they are not having a real dispute about issues of substance.

But, if Thomasson’s above arguments are all sound, then David L. and Peter V., if they are having a genuine ontological dispute at all, must be using ‘thing’ neutrally. So, their dispute, if it is a genuine ontological one, cannot be a real dispute about issues of substance. This result, since it did not depend on any specific features of David L. and Peter V.’s dispute, will generalize to any dispute between a Nihilist and a Universalist over whether some composite material object exists.

One line of resistance that will no doubt appear obvious at this point is that Thomasson’s above arguments mistakenly interpret the English expression ‘some thing’ as a general noun phrase when in fact it is a quantifier. Her arguments, if sound, only show that ‘thing’ in Thinghood Questions cannot be a noun phrase or a surrogate for a class of noun phrases.

The dispute between Peter V. and David L. can be construed as follows. Suppose that Peter V. and David L. were discussing a world at which the only existents were the contents of the vacuum chamber. How many objects would each say existed at such a world? Peter V. would say “only two” and David L would say “only three”. But, these claims needn’t be formulated in terms of the objectionable word ‘thing’, for both theorists have at their disposal the resources of quantified logic.[132] Existence claims, when stated entirely in logical vocabulary, do not appear to depend on there being a sortal, covering, or neutral use of the English expression ‘thing’. So, ontologists can relax, secure in the knowledge that they can make sense out of the Thinghood Questions that serve to ground their ontological disputes.

Thomasson is sensitive to this objection.[133] She argues that claims made in terms of the quantifier of first order logic rely on uses of the English expression ‘thing’ (or ‘object’). Here is how the argument goes. The formal quantifiers of first order quantified logic are supposed to regiment ordinary expressions in English like ‘all things’ and ‘at least one thing’. As Thomasson puts it (paraphrasing Van Inwagen): “…the source of the meaning of ‘"x(…x)’ is the English phrase ‘it is true of everything that it is such that…’ and the source of the meaning of ‘$x(…x…)’ is ‘It is true of at least one thing that it is such that…’ So, if ‘"’ and ‘$’ depend for their meaning on ‘thing’ in the above way, it seems clear that the ontologist may only appeal to the language of quantified logic to ground ontological disputes if she may appeal to ‘thing’ in order to ground ontological disputes.[134]

There is another way. Ted Sider has forcefully argued that expressions like ‘something’ and ‘everything’ do not get their meanings in virtue of frame-level application and coapplication conditions associated with these terms by competent speakers. Instead, the logical structure of reality supplies the meaning for these expressions. On Sider’s proposal, there is a unique, perfectly natural candidate meaning for quantificational talk. Words like ‘thing’, ‘object’ and so on, Sider argues, are intended by competent speakers to glom onto the world’s quantification structure.^{^[135]} Compare the situation with our use of the word ‘thing’ to the situation with respect to early uses of ‘water’. Early users of ‘water’ intended their use of this term to pick out a kind of stuff. It just so happened that there was a highly natural kind of stuff out there to be picked out: conglomerations of H₂O molecules. Sider contends that users of words like ‘thing’ intend to capture some quantificational aspect of reality in virtue of which claims like:

A) Something is red,

B) Something is human,

are about facts that are similar.^{^[136]} Similarity between facts is perhaps a slippery notion; some examples should help to clarify what is meant by it. Take the facts that obtain whenever

C) Jones is sitting,

D) Smith is sitting,

are both true. There is a sense in which these facts are similar; they are both instances of somebody’s sitting.^{^[137]} These facts are similar in a way in which the facts that obtain when the following are true are not similar:

E) Smith is sitting

F) Jones is wearing a dinner jacket.

Similarity between facts makes for property naturalness in the following way. Natural properties are those properties the joint instantiation of which makes for genuine similarity between facts. Sharing the “property” being an instance of somebody’s sitting or wearing a dinner jacket does not make for genuine similarity, so property sitting or wearing a dinner jacket is not a natural property.

Suppose, however, that the kind of reference magnetism Sider points to is insufficient to fix the meaning of ‘thing’ and ‘object’ in English. Sider maintains that, should English and other natural languages prove unsuitable in this way for framing ontological disputes, serious ontologists can always abandon these languages in favor of Ontologese. Serious ontologists can just stipulate that they mean to be using ‘there are’ and related expressions in such a way that they have their unique, joint-carving sense. [138]

Thomasson thinks that, even when philosophers are speaking Ontologese, the truth conditions for their “formal claims rely on the meanings of natural language terms”.[139] Thomasson argues that this is the case because “on the standard ways of doing semantics, quantified claims are only truth-evaluable when some or other domain is specified.”[140] This presents a problem for speakers of Ontologese, because the domain is not specified using a formal apparatus; instead it is specified “using a natural language like English, asking us, for example, to consider the domain of all the natural numbers, or all the people in Michael’s office.”[141]

Here is the argument:

1. Claims formulated in formal quantified languages are truth evaluable only if a domain of discourse is specified.

2. A domain of discourse can only be specified using natural language expressions like ‘all the things there are’ or ‘all the things in Arizona’ or ‘all the desks’.

3. If (1) and (2), then claims formulated in formal quantified languages (including Ontologese) are truth evaluable only if a domain of discourse can be specified using natural language terms.

4. Claims formulated in formal quantified languages (including Ontologese) are truth evaluable only if a domain of discourse can be specified using natural language terms.

So, it would appear that, in order for sentences involving formal symbols for completely unrestricted quantification to be truth evaluable, then there must be a way to make sense out of expressions like ‘all the things that there are’. But, Thomasson has argued that ‘things’ cannot take a sortal, covering, or neutral use. So, it would appear that the serious ontologist is sunk.

At this stage I think it is crucial to note that Thomasson’s remarks on the proper way to answer existence questions have wide reaching implications for the practice of philosophy. According to Thomasson, answering existence questions is a two step process. First: do some conceptual analysis in order to determine the frame level application conditions for some term (e.g., ‘person’, ‘table’, ‘mereological fusion’, etc.). Once the frame level application conditions are in hand, all that remains is to check in order to see whether they are satisfied. This second step is conducted at the level of empirical research of the sort that can be undertaken by scientists and investigative journalists as well as philosophers and with as much hope of success.

Let us now imagine applying this model more broadly in philosophy. Is freedom compatible with determinism? Well, let’s find out. What are the application conditions for ‘free act’? Can these ever be satisfied by an act that also satisfies the application conditions for ‘causally determined act’? If the answer is ‘yes’, then freedom is compatible with determinism. If ‘no’, then freedom is not compatible with determinism. Is a priori knowledge possible? Let’s find out. First we need to discover the frame level application conditions for ‘a priori knowledge’, then we determine the frame level application conditions for ‘knowledge’, then we check to see if they can be jointly satisfied. If the answer is ‘yes’, a priori knowledge is possible, if the answer is ‘no’, then it is not. Are human persons identical to human bodies? Let’s apply Thomasson’s methodology. First discover the frame level application and coapplication conditions for ‘human person’, then find out the application and coapplication conditions for ‘human body’, and finally check to see if they match up. What makes an action a morally right action? All we need do to answer this question, if Thomasson is right, is divine the frame level applications for ‘action’ and ‘morally right action’, then pass things off to the nearest empirical researcher. This seems wrong.

Where the frame level application and coapplication conditions for terms like ‘free act’, ‘causally determined act’, ‘knowledge’, ‘a priori knowledge’, ‘human person’, and ‘human body’ are vague or ambiguous or just not specified completely by the rules of use for these terms, then answers to certain questions framed using these terms will be what Thomasson calls ‘unanswerable questions. There are good reasons to think that our rules of use do not specify frame level application and coapplication conditions that will enable us to answer all of the philosophical questions we can frame using the above terms (e.g., “Can human persons survive the death of their bodies?”, “Can there be free acts if God has divine foreknowledge?”, etc.). If this is the case, then there are good reasons to think that many philosophical questions outside of ontology will be rendered unanswerable given Thomasson’s preferred methodology.

Thomasson’s arguments, then, threaten to prove too much. Consider questions like, “How should I live my life” and “What things should I care about?” Are these questions unanswerable in Thomasson’s sense? The latter seems to be, for it does not seem plausible that ‘things’ takes either the sortal or covering use.[142] Things seems to be intended in the neutral way. But then, if Thomasson is right, we should stop asking them. This seems wrong to me.

Perhaps this is not a reason to reject Thomasson’s view, though she does seem to be arguing philosophers out of a job. What Thomasson calls ‘conceptual analysis’ does not seem like it should be uniquely the province of philosophers. For finding out frame level application and coapplication conditions of our nominative terms seems like a task that empirical researchers like linguists or behavioral scientists should be able to accomplish. For, these frame-level conditions are to be read off rules of use, which in turn, it seems, ought to be read off our linguistic behavior and our dispositions to linguistic behavior.

I suggest that Nihilists, Universalists, and other ontologists should not feel so threatened by Thomasson’s arguments, for if these arguments are sound, there is good reason to think that ontologists are not alone in trying to answer unanswerable questions. I will have more to say about this below, for the present it shall suffice to observe that, if Thomasson’s arguments are sound, there is room in the boat for practitioners of other subdisciplines in philosophy.

10.

So much has been exposition. Now we come to evaluation. Should serious ontologists be threatened at all by Thomasson’s master argument?

Thomasson’s argument proceeds in four stages. First, she argues that ‘thing’ in Thinghood Questions is not a sortal, then she argues that ‘thing’ does not take a covering use, then she argues that it cannot take a neutral use, and finally, she argues that Thinghood Questions cannot be understood quantificationally. Let us consider the first stage:

If ‘thing’ is used as a sortal by Peter V. and David L., then either they associate the same application conditions with ‘thing’ or they do not.
If they associate the same application conditions with ‘thing’, then at least one of them is making a mistake about whether the application conditions are satisfied.
It is unreasonable to think that either Peter V. or David L. is making a mistake about whether the application conditions are satisfied.
If Peter V. and David L. do not associate the same application conditions with their use of ‘thing’, then they are talking past one another or they are not having an ontological dispute.

5. If ‘thing’ is used as a sortal by Peter V. and David L, then their dispute about how many things are in the vacuum chamber involves talking past each other or it is not an ontological dispute.

Premise (1) looks unassailable. Premise (2) however, is dubious. Let’s suppose that something like Ted Sider’s account of the semantics of ‘thing’ is correct. This word, let us suppose, is a highly general sortal; it is the most general sortal. Its frame level application conditions, however, are not supplied by the rules of use for this term, for there is a highly intrinsically eligible property in the vicinity: being some existent. So, let’s suppose that Peter V. and David L.’s use of ‘thing’ (and related words) is fixed by the logical structure of the world. And let’s just suppose, for the sake of argument, that the frame level application conditions for ‘thing’ are just these: ‘x is a thing if and only if x is mereologically simple’. Now, ex hypothesi, Peter V. and David L. associate the same frame level application conditions with ‘thing’, due to the overpowering reference magnetism of the perfectly natural and intrinsically eligible candidate meaning for this expression. So, the antecedent of (2) is satisfied. What about its consequent? Does either Peter V. or David L. make a mistake about whether the application conditions are satisfied? No. Recall the case: in a vacuum chamber there are two mereologically simple particles a and b. Both David L. and Peter V. are correct in believing that a and b satisfy the condition: x is mereologically simple. No mistake here. Both David L. and Peter V. also correctly believe that the (alleged) mereological fusion of a and b does not satisfy the application condition: x is mereologically simple. So, neither is making any mistake about the satisfaction of the application conditions for ‘thing’.

What explains their disagreement, then? In the above example, at least one disputant is mistaken about something. As I’ve told the story, David L. is mistaken about what the application conditions are. He is not mistaken about whether they are satisfied or not, but he is mistaken in thinking that the application conditions permit ‘thing’ to apply to objects that are mereologically complex. David L. uses the term ‘thing’ mistakenly, to apply to mereologically complex objects. Peter V. and David L. are not talking past one another, for since they both mean the same by ‘thing’, one of them asserts a proposition that the other denies. However, they use the word ‘thing’ divergently, and this explains their dispute. So, on our above account of ‘merely verbal dispute’, David L. and Peter V. are having a merely verbal dispute.

How does this help ontologists? Isn’t it bad to be having a merely verbal dispute? Perhaps it is. Rational people shouldn’t do this. However, if David L. and Peter V. are having a merely verbal dispute, then there is hope for resolution of this dispute. If Peter V. and David L. are having a no fact of the matter dispute, however, there is no hope for resolution; the best they could achieve is realization that there is no fact of the matter. If the dispute is merely verbal, however, there is a substantive dispute that Peter V. and David L. could have instead: they could have a dispute over what the application conditions for ‘thing’ are. There is (we are assuming) a fact of the matter as to who is right in this dispute. According to the above story Peter V. is right.

Perhaps taking this line would be cold comfort for the serious ontologist. But, if the disputes are verbal in the way I sketch above, it follows that they are not attempts to answer unanswerable questions. So, Thomasson’s conclusion can be resisted at the cost of admitting that the Nihilist-Universalist disputes are merely verbal. This is a cost that most serious ontologists, however, will be unwilling to pay.

11.

The second stage of Thomasson’s argument itself proceeds in two stages. First Thomasson argues that ontological disputes must be grounded in attempts to answer generic existence questions:

If Peter V. and David L.’s dispute in the vacuum chamber case is grounded in their disparate opinions about how to answer the specific existence question “Is there a mereological fusion here”, then their dispute is either a case of talking past one another or it is not an ontological dispute.
Neither disputant would admit to talking past the other or to having a non-ontological dispute.
Peter V. and David L.’s dispute in the vacuum chamber case is not grounded in their disparate opinions about how to answer a specific existence question.

One strategy for resisting this argument is to reject premise (1). Admit that the dispute over the truth of ‘There is a mereological fusion here’ is merely verbal with one party using the term ‘mereological fusion’ incorrectly. Both parties mean the same by this term, but one is mistaken about whether it correctly applies in virtue of having a mistaken belief about its application conditions.

Again, this strategy will probably not appear too attractive to serious ontologists. Perhaps there is an ontology friendly way to resist the second stage of Thomasson’s argument:

If Peter V. and David L. are employing a covering use of ‘thing’ in asking ‘Is there some thing composed of the simples?”, then their dispute over the right answer to this question must be grounded in a specific existence question (e.g., Is there a mereological fusion here?).
Their dispute cannot be grounded in a specific existence question (for the reasons adduced above).
Peter V. and David L. are not employing a covering use of ‘thing’ in asking ‘Is there some thing composed of the simples”.

The serious ontologist must reject either (1) or (2). The only reason for rejecting (2) that we have considered requires ontologists to admit their disputes are verbal. This option will not seem too savory to practicing ontologists.

What about the prospects for rejecting (1)? I think these may seem rosier to the serious ontologist. Suppose that (1)’s antecedent is true. Both parties to our hypothetical dispute are using ‘thing’ in the covering sense. This means that their question, “Is there any thing here composed of simples a and b?” is best understood as (something like): “Is there any [book, table, person, mereological fusion…] composed of simples a and b?” David L. says ‘yes’ and Peter V. says ‘no’. Need their dispute be grounded in a difference of opinion with respect to a specific existence question? I do not think so. Recent work by Ted Sider suggests that there is a theoretical way to ground the Universalism-Nihilism dispute that doesn’t rely on taking that dispute to be an attempt to answer questions like “Do mereologically complex things exist?” or “Is there a mereological fusion of the two simples a and b in the vacuum chamber?”[143] The dispute can be grounded in a dispute about the proper application of a principle of parsimony. The Nihilist, argues Sider, has a view that is simpler: she is not committed to a part-whole relation, while the Universalist is. Call this difference a difference in ideology, as opposed to ontology. As Sider remarks, “Ideologically simpler theories aren’t just more convenient for us. They’re more likely to be true, since the worlds they posit are simpler, contain less structure.”[144]

How does this help us construe the question “Is there any thing that has a and b as parts here?” as involving a covering use of ‘thing’? Suppose that David L. and Peter V. ask the question “Is there any thing here that has a and b as parts?” and they intend ‘thing’ in the covering sense. So, they mean to be asking (something like) “Is there a sortal that, when substituted for ‘S’, generates a true substation instance of: ‘An S has a and b as parts?” The Nihilist says ‘no’, the Universalist says ‘yes’, but not because they disagree about whether any particular sortal applies, but rather because the Nihilist thinks no sortal applies on the grounds that such application carries an unacceptable commitment to the part-whole relation, while the Universalist thinks that some sortals do apply because she takes the commitment to a part-whole relation to be unproblematic and not ideologically profligate.

This way of grounding the dispute characterizes it not as an attempt to answer the (allegedly) unanswerable question “How many things are there?”, but instead as a dispute over the correct way to apply a principle of ideological parsimony:

IP: Ceteris paribus, the best theory is the one with the simplest ideology.

Recasting the dispute in this way may help to give new life to disputes over mereological composition, shedding light on new arguments, and avoiding the charge that such disputes are, at best, silly attempts to answer ill-formed questions.

12.

What of the alleged neutral use of ‘thing’? Can Thomasson’s argument against such a use in ontological disputes be resisted? Here is the argument:

If David L. and Peter V. are using ‘thing’ neutrally, then ‘thing’ has no application conditions.
If ‘thing’ has no application conditions, then any answer to ‘Is there some thing here composed of the simples’ is not truth evaluable.
If no answer is truth evaluable, then the question is unanswerable.
If the question “Is there some thing here composed of the simples” is unanswerable, then David L. and Peter V. are not having a real dispute about issues of substance.
If David L. and Peter V. are using ‘thing’ neutrally, then they are not having a real dispute about issues of substance.

The lynchpin premise here is (4). Premise (1) is a consequence of the definition of ‘neutral use’; premise (2) is true (provided that we are granting Thomasson’s semantic methodology); and premise (3) is a consequence of the definition of ‘unanswerable question’ as Thomasson is using the term.

Why should we accept (4)? (4) is plausible only if there is some bridge between a question’s being unanswerable (in Thomasson’s sense) and a dispute over how to answer that question’s being defective (in our sense). How could there fail to be such a bridge? After all, isn’t “trying to answer an unanswerable question” just silly? This is not fair; for ‘unanswerable question’ here is a technical term. A question is unanswerable, in the operative sense, if and only if no answer to it formulated in terms of the original question is truth evaluable. Let’s take an example. Suppose that you and I are attempting to answer this question, “What do you propose we do tonight?” To answer this question in terms of the question is to make a proposal, not to make an assertion which is truth evaluable. Suppose you answer by saying, “I propose we go to the cinema,” and I answer by saying “I propose we go to the park.” We then get into an argument (in the discursive sense) in which I try to get you to endorse my proposal, and you try to get me to endorse yours. Given the way we’ve characterized disputes, this pattern of behavior counts as a dispute. Moreover, it is a real dispute over a matter of substance. But, it arises from an attempt to answer an “unanswerable” question.

Let’s apply this to the ontology cases. Suppose that vying ontologists are making distinct proposals, “Let’s use a language on which we describe the vacuum chamber as containing three things,” and “No, let’s not use such a language,”[145] but that they perform the speech act of disputing these proposals by uttering sentences like, “There is no mereological fusion there” and “There is something composed of a and b.” If this is the case, then their “answers” to the question “Is there some thing composed of a and b?” are not truth evaluable (for they are not assertions, but instead proposals). If we understand the question, “Is there some thing composed of a and b?” not as a request for an answer that takes the form of an assertion, but instead as an invitation to give a proposal, then its being unanswerable (in Thomasson’s sense) is no bad thing.

No doubt many practicing ontologists will not want to see the matter this way; but our task has not been to provide an answer to Thomasson’s argument with which ontologists would be happy. As we established in Part One, disputes can occur over matters that do not involve disputants asserting or defending some truth evaluable claim (consider so-called disagreements in attitude). If this is the case, then a dispute over an unanswerable question may turn out to be a real dispute over an issue of substance.

13.

Let’s consider a different reply to Thomasson’s argument. The first thing to notice in this connection is that there is an important distinction to draw between different kinds of unanswerable questions. There are a number of ways that a question can fail to be answerable. The first way a question can meet this condition is in virtue of being nonsensical. Take the question:

Q1: How long did your idea sleep?

This question is nonsensical, and so no answer to it formulated in terms of the question (i.e., “My idea slept n hours.”) will be truth evaluable. Further, this deficiency is not epistemic, it has to do with what the question means. It is not because no one could ever be in a position to know how many hours their idea slept that this question cannot be answered. Thomasson’s example about the piece of string is somewhat like this. A piece of string (not a particular piece of string) just isn’t the sort of thing that can have a length. The question “How long is a piece of string”, then, is nonsensical in virtue of its committing a category mistake. It is plain that we do not even know what the world would have to be like in order for some answer to these questions to be truth evaluable.

Contrast questions like Q1 and the piece of string question with questions that are not nonsensical, but nevertheless unanswerable. Consider:

Q2: Is Jones bald?

Suppose that Jones is not just a borderline case of baldness, but that he is a borderline borderline case. That is, it is vague whether he is a borderline case of baldness.[146] Q2 is an unanswerable question, and not because no one is in a position to know the answer, but rather because our word ‘bald’ just isn’t precise enough to render any answer truth evaluable. But, unlike the above nonsensical questions, we know what an answer would be like. We know what the world would have to be like in order for the answer to Q2 to be ‘yes’, and we know what the world would have to be like in order for the answer to be ‘no’.

Contrast Q1 also with cases of presupposition failure like:

Q3: Is the present king of France bald or not bald?

Q3 is unanswerable, and not merely because no one is in a position to find out how hirsute the king of France is. Q3 is unanswerable because of a semantic defect: one of its terms fails to refer. No answer formulated in terms of the question is truth evaluable. But, we can envision what the world would be like if this question did have a truth evaluable answer formulated in its own terms.

Thinghood questions do not seem to be nonsensical. If they are unanswerable questions, then, they are relevantly disanalogous to Q1 and the piece of string question. This is important, for nonsensical questions are questions that we have no idea what a truth evaluable answer would look like. It is patently absurd to have a disagreement over the right way to answer a question if you have no idea what a truth evaluable answer would even be like. What would the world have to be like in order for ‘six hours’ to be a truth evaluable answer to Q1? No one is in a position to say. On the other hand, we are in a position to say what the world would have to be like if there were a truth evaluable answer to Q2. In one such world in which Q2 is answerable, Jones has no hair at all on his head; in another world he has long, flowing locks. We have a pretty clear idea of what the world would have to be like in order for answers to Q2 to be truth evaluable; it just so happens that the world isn’t like this.

This observation is crucial, for it is not patently absurd to have a disagreement over the correct way to answer a question if you have some idea what the world would have to be like in order for there to be a truth evaluable answer.

I want to suggest that the Nihilist and the Universalist really do have application conditions associated with ‘thing’, or at least, if they do not, that it is pretty easy to imagine them associating such conditions with ‘thing’. Suppose that Peter V. and David L. make the following speech:

“What we want to do is have a dispute about whether there is any material object in the vacuum chamber that has a and b as parts. What is a material object? Well, we cannot be perfectly precise, but the following remarks should serve to pick out what we mean. Something is a material object if it occupies space, endures through time, can move about it space, has a surface, a mass, and is made of some kind of stuff or stuffs.[147] We want to know, is there anything like that in the vacuum chamber of which both a and b are parts?”

If there is no fact of the matter as to what the right answer is here, then this is in virtue of the vagueness of some of the conditions for ‘material object’. But, this makes the dispute over the truth of “Is there some material object that has a and b as parts?” more like disputes over Q2 than like disputes about Q1. There is no fact of the matter as to the correct answer and this is because some terms operative in framing the dispute are vague. It is not patently absurd to have a dispute over the question, for theorists can say something about what the world would have to be like in order for some answer to the question to be truth evaluable.

So, if Peter V. and David L. are responding to a question that is unlike Q1, but like Q2, then it looks like their dispute is not patently absurd. However, since they are attempting to settle an issue about which there is no fact of the matter, they are having a defective dispute in our sense; theirs is a no fact of the matter dispute, which we have argued that rational people should not have. Also, since the unanswerability of the question to which each disputant is responding obtains in virtue of a deficiency in meaning (vagueness), this question is unanswerable in Thomasson’s sense as well.

It is illuminating to ask why rational people should not engage in no fact of the matter disputes. What is crucial to the claim that rational people should avoid no fact of the matter disputes is the fact that sincerely having them, insofar as they are disputes, carries with it the implication that these disputes are, in principle, resolvable. Recall that in Chapter One we looked at one of the speech acts that a person can perform with disagreement discourse: the speech act of challenging their interlocutor to a dispute. We stated that S’s issuing a challenge to S* is infelicitous if the following conditions are not met.

(a¢) S commits to an attitude that is non-cotenable with, precludes joint satisfaction, precludes joint accuracy, or precludes joint reflexive accuracy, of some attitude that S* is committed to.

(b) S intends to give reasons for the attitude she commits to.

Let us take the Peter V. and David L. case. Presumably this dispute involves some kind of disagreement locution like, “No, that is not the case” or “I disagree with your view that…” or just plainly “No.” If these disagreement locutions are being used by Peter V. and David L. to perform the speech act of challenging their interlocutor to a dispute, then the speech act is felicitous only if the resulting dispute is in principle resolvable. So, now we can explain why it is that disputants who sincerely engage in no fact of the matter disputes are having disputes they shouldn’t have. No fact of the matter disputes are in principle irresolvable. But, if David L. and Peter V. are having a sincere no fact of the matter dispute, then presumably they think their act of challenging the other comes off felicitously. But, if they think this, and their dispute is a no fact of the matter dispute, they are mistaken. To continue to have such a dispute, both parties would have to be operating under the impression that their dispute is in principle resolvable when in fact it is not.

Notice, however, that there is nothing wrong with having a no fact of the matter dispute insincerely. For, when I knowingly make an insincere challenge to my interlocutor, if I do not represent myself as sincere, I do not represent the proposed dispute as meeting all of the felicity conditions for the speech act of challenging. So, I do not represent the proposed dispute as resolvable.

What sort of logical space does this create for the serious ontologist who wants to continue having ontological disputes? Even if Thomasson is right and there is no fact of the matter as to whether the Nihilist or the Universalist is correct, even if they are attempting to answer an unanswerable question, still it might be worth their continuing to have a dispute. How could this be?

Even if no answer to the relevant ontological question is truth evaluable, engaging in a serious attempt to answer it may teach us something important about our language, our concepts, and perhaps even our world. Here I think it is useful to draw an important distinction between two kinds of argumentation: (what we have been calling) disputes and debates. A dispute is only felicitous if the dispute can, at least in principle, be resolved. Debates are not like this. Parties to a debate needn’t think that the issue they are debating is resolvable. Debate is a discursive practice that does not necessarily aim at consensus, nor does it necessarily aim at truth. Debate is about looking at arguments, assessing reasons, and getting clear on concepts. In debates, participants can take themselves to be articulating positions they do not accept or do not even think are truth evaluable.

Consider the differences between having a dispute about God’s existence and having a debate about God’s existence. When A and B have a dispute over God’s existence, they take up conflicting attitudes about the matter, they are trying to change each other’s mind, they think it is rational to try to change the other’s mind, and (if their dispute is felicitous) they think the issue could, in principle, be resolved. But, suppose C and D are having a debate about God’s existence. C says God exists and D says God doesn’t exist. Both believe there is, in principle, no way to determine who is right. But they do not care about this; all they care about are the arguments.

Perhaps this line of resistance to Thomasson will strike serious ontologists as a kind of giving in. Perhaps it is, in a way. But consider the indictment of philosophy one often hears from non-philosophers: “You folks are just arguing over questions that have no answers, or at least questions the answers to which we could never know!” This sounds like a serious objection that serious practitioners of philosophy need to answer in order to avoid being intellectual laughingstocks.

I reject the assumption that this objection needs an answer before philosophers can look themselves in the mirror. What is wrong with a philosopher flat footedly saying “You’re right. That is exactly what philosophers do. But it doesn’t matter.” Compare what philosophers do with what historians and literary critics do. The latter have arguments over questions like, “Was Raskolnikov acting out of some Fruedean desire for his mother when he killed the pawnbroker?” Laypeople might think, “Is there even a right answer here?”, while literary critics propose all sorts of evidence for answering the question one way rather than another. Historians ask, “What were Napoleon’s reasons for failing to send in the Prussian cavalry at the Battle of Waterloo?” Non-historians might demur, saying “How could anybody every know!”

Now, perhaps philosophers will be hesitant to cast in their lot with historians and literary critics, but there seems to be here a way to defend the view that it is not absurd to try to answer unanswerable questions. Doing so teaches us something about our concepts, hones our acumen, helps us distinguish internally inconsistent views from coherent ones, and can generate knowledge about our world. So long as we take ourselves to be having debates rather than disputes, can’t philosophers continue rationally to put forward and defend answers to unanswerable questions? I do not see any reason why we cannot.

At this stage it is important to make a distinction between two sorts of no fact of the matter dispute. The sorts of cases to which Thomasson draws our attention are no fact of the matter disputes that involve deficiencies in meaning. These are disputes for which there is no fact of the matter as to who is right or wrong owing to a problem with the language in which the dispute is cast. Vague terms, terms with incompletely specified boundaries, cases of presupposition failure, nonsensical claims, and the like all can give rise to no fact of the matter disputes. Call these Semantically Defective No Fact of the Matter Disputes.

There is another sort of no fact of the matter dispute. Disputes of this sort are cases in which there is no fact of the matter as to which disputant is right or wrong (or if both are right or wrong) owing to a lack of structure in the world. These are cases in which the disputes are framed in non-problematic language, but in which the world fails to supply sufficient structure in order to settle the matter one way or another. Some examples will help to illustrate what I have in mind. Consider the case of two literary critics disputing whether or not, in the Holmes stories, Sherlock Holmes was gay. One critic maintains that Holmes was a confirmed bachelor, the other denies this. Each has what they take to be evidence for their view. Now, suppose Conan Doyle never gave the matter any thought. He never wrote any sentences or had any intentions that could determine the facts about Holmes’s sexuality.[148] But, since the only features of the world relevant to determining which literary critic is right are those that have to do with Conan Doyle and what he thought and wrote, then there is no fact of the matter as to which critic has it right.

Further, the lack of any fact of the matter has nothing to do with the language used to frame the dispute. Each theorist is using terms that are (let’s suppose) perfectly well defined, that come along with Thomasson’s frame-level content, and that can be unambiguously applied.

Let us consider one more example. Suppose that Quine got it right about theories of translation being underdetermined by the facts.[149] If this is the case, then there are (possible) instances in which two field linguists propose empirically equivalent but divergent manuals of translation for some language, and there is no fact of the matter as to which translation manual is correct. Suppose two linguists, Fred and Ted, are in such a situation with respect to their manuals of translation for a language L. Imagine Fred and Ted having a dispute over whose manual is correct. This dispute would be a no fact of the matter dispute, but the lack of a fact of the matter is, if Quine was right, a function of the world’s lacking the appropriate structure to determine a correct answer.

Call disputes like these Metaphysically Defective no fact of the matter disputes. The distinction between Semantically Defective and Metaphysically Defective no fact of the matter disputes is that the former can be explained by a failing of our language, while the latter can be explained by a failing of the world. Sometimes the world just isn’t up to the task of making a question decidable. If what I have said above is correct, then there are good reasons to think the dispute between Peter V. and David L., if it is a no fact of the matter dispute, is better characterized as a Semantically Defective dispute, but not a dispute over a patently absurd question. Rather, the semantic defect arises out of the presence of vagueness in the vocabulary in which the dispute is stated, especially in the term ‘material object’.

Where does this leave us? I take the above to have established a disjunctive conclusion. Either the structure of the world is sufficient to establish frame-level application conditions for ‘thing’ that are independent of our linguistic practices or it is not. If the structure of the world is sufficient, then Thomasson’s argument fails. If the structure of the world is not sufficient, then we should take ontologists at their word: they are having disputes over whether a certain situation contains n or n + 1 material objects. If there is something wrong with their dispute, it is the vagueness of ‘material object’, or some of the terms used to define ‘material object’. But, if this is the case, then ontologists are not having disputes over nonsensical questions (like “How long is a piece of string”?), instead they are having Semantically Defective disputes over questions framed in a vague language (like, “Is Smith really bald or not?”). If this is the case, then Thomasson has failed to establish that ontologists are not having disputes over genuine issues of substance. I conclude that, in either case, Thomasson’s argument fails.
Chapter 8

NO FACT OF THE MATTER

In this section, we shall examine some other arguments designed to show that there is no fact of the matter as to whether Nihilism or Universalism is true. What does this claim amount to? Some theorists say that there is no fact of the matter in the following sense: Nihilists and Universalists “agree on all the facts”, they just disagree about how to describe them. It should be clear from Part One that this is not enough for there to be no fact of the matter. Half-Fullists and Half-Emptists disagree about how to describe a case in which a drinking vessel is partly filled with water. Half-Fullists say, “The glass is half full”, while Half-Emptists say, “No, the glass is half empty.” However, there is, presumably, a fact of the matter as to whether these views are true or not. It just so happens that both are true or both are false.[150] A dispute characterized by agreement on “the facts” coupled with disagreement about how to describe them counts as a merely verbal dispute. Having a merely verbal dispute, however, is neither necessary nor sufficient for having a no fact of the matter dispute.

What is required for there to be no fact of the matter as to whether some particular view is true or false? There has to be underdetermination. That is, the world must leave the question “Is this view true or false?” unsettled. This talk of underdetermination and unsettledness is somewhat imprecise, however. There are a number of ways to try to make it precise. One could talk about truth-makers, saying that there is no fact of the matter as to whether view V is true if, and only if, neither V nor not-V has a truthmaker. Whether or not truths require truthmakers is, however, a hotly contested issue which I do not want to try to settle here.[151] The leading idea is fairly intuitive, however. Truths require some state of the world that makes them truths rather than falsehoods. A truth-maker for a truth T is some state of the world that suffices for T.[152] This way of talking presupposes a realism about truth, and it is not clear to me that adherents of Nihilism or Universalism need to accept such a view.

Are there other ways to make the notion of underdetermination or unsettledness precise? Perhaps one might, following Quine, take the data that serve to determine the truth or falsehood of a view to be the sum of all (actual and possible) sensory stimulations. The claim that there is no fact of the matter as to whether Nihilism or Universalism is true, on this understanding, would then amount to the claim that the sum total of sensory stimulations does not impel assent to either view or its negation and the sum total of sensory stimulations are the only data relevant to determining the truth of the matter. This sort of view is somewhat unsatisfying, for it is not clear that a priori matters should be held accountable to it. Is it reasonable to think that whether there is no fact of the matter as to whether Goldbach’s conjecture is true or false depends on whether some set of sensory stimulations would impel assent to it? Perhaps not.[153]

In what follows, I will assume the truth-maker account of no fact of the matter disputes. So, a dispute is a no fact of the matter dispute if, and only if, two disputants, A and B, hold divergent views and there is no truth-maker for the claims “A’s view is correct” or “A’s view is not correct”, or for the claims “B’s view is correct” or “B’s view is not correct.”

Recently Alan Sidelle has argued that there is no fact of the matter as to whether Universalism or Nihilism is true.[154] It is important to note that Sidelle uses ‘Nihilism’ and ‘Universalism’ in different senses than we have been using those terms here. On Sidelle’s usage, Universalism is the view that “whenever you have some matter, you have an object which that matter composes” while Nihilism is the view that “there are no objects—there’s just the matter.”[155] This subtle difference between our respective uses of these terms should not cause too much trouble, but it bears mention at the outset.

Sidelle maintains that truth-makers for Nihilism and Universalism, if there are any, are to be found by asking:

Q1. Is Nihilism/Universalism consistent with our ordinary judgments/intuitions about what there is and how things persist?

Q2. Is Nihilism/Universalism consistent with plausible theoretical judgments/principles?

Q3. Is Nihilism/Universalism internally consistent?

Q1-Q3 give us a scorecard for the Nihilism-Universalism dispute. How well these views fare with respect to this scorecard is all, Sidelle thinks, that is relevant to determining their truth, “beyond this, there are no truth-makers to distinguish among the positions.”[156] Sidelle’s position is not to be confused with a position that might be called epistemic underdetermination. According to epistemic underdetermination there are no truth-makers that could help us decide whether Nihilism or Universalism are true or false. Sidelle’s position is what might be called metaphysical underdetermination. There are no truth-makers that could make it the case that Nihilism or Universalism are true or false.

Here is Sidelle’s master argument for metaphysical underdetermination;

Universalism and Nihilism each answer Q1-Q3 equally well.
Beyond answering Q1-Q3 there are no further facts that could suffice to make either Universalism or Nihilism true or false.
If (1) and (2), then there is no fact of the matter as to whether Universalism or Nihilism are true or false.
There is no fact of the matter as to whether Universalism or Nihilism are true or false.

In what follows, we shall examine Sidelle’s reasons for accepting each of these premises.

Sidelle does not present an argument for premise (3). Given what he means by ‘no fact of the matter’, I do not think he needs to provide such an argument. The key premises, then, in the above argument are (1) and (2). Sidelle provides reasons for accepting both. Let us begin with (1).

Proponents of Universalism and Nihilism present what Sidelle calls “packages”. That is, each view has some leading or motivating idea, provides an account of how it is consistent with the correctness of our ordinary beliefs, and provides an account of how it is consistent with a number of theoretical principles like the following:

No Coincidence Thesis: There cannot be two material objects wholly located in the same place at the same time.

No Arbitrary Distinctions Thesis: judgments about which portions of the world do and which do not contain material objects should not be arbitrary.[157]

Add to these two theses a number of theoretical principles extolling the virtues of simplicity, theoretical elegance, and so on.

Both the Universalist-package and the Nihilist-package fare equally well with respect to all of the desiderata of a theory about which material objects exist, according to Sidelle. Sidelle maintains that Q1-Q3 should be answered affirmatively. Why should we believe this? After all, aren’t Universalism and Nihilism both wildly inconsistent with our ordinary beliefs about what exists?

Sidelle argues that this is not the case. Universalism, while it may affirm the existence of a great number of objects we ordinarily do not think of as existing, does not fly in the face of common sense. For, “it isn’t clear that common-sense so much denies the bulk of scattered objects, as it ignores them.” Universalists can even account for the straightforward truth of claims like, “There’s nothing composed of my nose and the Eiffel Tower.” All the Universalist need do is avail herself of the formal device of quantifier domain restriction. Most ordinary quantified claims about what there is and what there is not involve tacit restriction of the quantifiers to range only over the domain of objects that are of interest or importance to us, leaving out scattered or gerrymandered objects.

What of the Nihilist? Aren’t her claims wildly counterintuitive? After all, she claims there are no tables, chairs, persons, mountains, etc. Sidelle thinks that the Nihilist can plausibly account for the correctness (if not the literal truth) of all of our commonsense beliefs about what exists by appeal to a suitable paraphrase of our ordinary claims about material objects. Sidelle maintains that the Nihilist has “a number of options available to him” including

the ‘feature-placing’ language proposed by Hawthorne and Cortens (1995), and the idea that each apparent objectual expression in a paraphrase can be seen as just a place holder. If this is right, the Nihilist can continue to speak with the vulgar, and acknowledge all facts about the distribution of matter through space and time, while denying ‘in his heart’ that there really are objects.[158]

Let us suppose that Sidelle is correct about this. Further, let’s grant that the Universalist and the Nihilist can each accommodate every commonsense or intuitive judgment about what exists. Still, this leaves us with internal consistency. Sidelle maintains that both package views are internally consistent. I think we should grant this. There is a great amount of inductive evidence for this claim: the history of the dispute over composition. Many extremely bright and talented philosophers have been trying very hard for quite some time to show that one or other of these views suffers from some kind of internal inconsistency. All these philosophers have failed. I take it that this is evidence that the theories are not internally inconsistent.

This leaves us with just one place where differences between Universalism and Nihilism can be found. Sidelle argues that both views provide packages that accord equally well with all of the theoretical principles on the table. This claim has some plausibility. Both Nihilism and Universalism reject coinciding objects.[159] Both views fare equally well on the No Arbitrary Distinctions Thesis. Both views have equally well-defended answers to a number of problem cases.[160] So, it looks like both theories do equally well at respecting theoretical intuitions and principles, right?

I think this is not the case. Nihilism fares better with respect to two parsimony principles:

Ontological Parsimony: all else being equal, the theory that posits the fewest objects is more likely to be true.

Ideological Parsimony: all else being equal, the theory that posits the fewest properties and relations is more likely to be true.

Nihilism clearly posits fewer objects. Take the object that consists of your nose and my left hand. Nihilists do not believe in this object, while Universalists do. Ted Sider recently argues that Nihilism fares better with respect to Ideological Parsimony as well.[161] For the Nihilist needn’t posit a part-whole relation. On the Nihilist’s view, there are no objects that have proper parts. So, it is open to the Nihilist to reject this relation wholesale.

Given that there are at least two theoretical principles with respect to which Nihilism and Universalism perform differentially, I think we are justified in denying Sidelle’s premise (1).

What of premise (2)? Are there any reasons to accept it? Sidelle gives two arguments. The first argument I call the Noseeum Argument. Sidelle gives it in the following passage:

Now, I have to admit to being unhappy with many of the packages, but there is a minimum constraint that they all meet. Their leading ideas are well-motivated, they have something to say to each desideratum, we can understand them…and crucially, any distribution of matter in space-time can be coherently described by them…there can be no fact of the matter as to which truly describes the material ontology and persistence of things. They can only be understood as different ways of articulating, extending and making coherent the combination of our ordinary judgments and theoretical ideas. But short of showing that really, all but one are incoherent, I don’t see what in the world can make one true; or equivalently, while the theories plainly differ, I don’t see how that with respect to which they differ can be understood as a factual matter.[162]

Sidelle’s argument is the following:

I do not see how Nihilism and Universalism differ with respect to some factual matter.
If (1), then Nihilism and Universalism do not differ with respect to some factual matter.
If Nihilism and Universalism do not differ with respect to some factual matter, then there are, beyond answering Q1-Q3, no truth-makers relevant to determining whether Nihilism or Universalism are true or false.
There are, beyond answering Q1-Q3, no truth-makers relevant to determining whether Nihilism or Universalism are true or false.

I do not find this argument very convincing. As I have argued above, there are theoretical constraints with respect to which Nihilism fares better than Universalism. A theory’s faring better with respect to theoretical constraints, like parsimony, is relevant to determining its truth or falsehood. Since there is a relevant difference between Nihilism and Universalism along a dimension relevant to determining truth or falsehood, the Nihilist and the Universalist can happily grant that the above argument is sound.

There is a more serious criticism. We have reason to reject (2). Isn’t it just obvious that Nihilism and Universalism differ on some factual matter? After all, they differ with respect to whether some simples, say the ones that appear to make up my computer monitor, compose something. Isn’t this a factual difference? And don’t Nihilists and Universalists disagree about allegedly metaphysically necessary facts like the one expressed by, “Anytime there are some simples, there is something they compose”?

Sidelle’s second argument targets this kind of reply. I call this argument the Modal Argument. Sidelle argues that Nihilism and Universalism, if true, would have to be necessarily true. This much would not be resisted by Nihilists and Universalists. But then Sidelle goes on to argue:

…insofar as we have any grasp at all on factual differences between the views, we cannot think of one as necessarily true and the rest at necessarily false. The factual differences between the views would consist in the ‘extra’ truths that, say, such-and-such a relation between parts was (or wasn’t) necessary, or sufficient, for objecthood…Insofar as we can make sense of one ‘package’ of superfacts obtaining, we can make sense of any of them obtaining, just as if laws of nature are something supra-Humean, we can equally make sense of any of a host of such laws being the ‘real ones’ governing motion and change. And this, I think, makes it impossible to see any of them as actually necessary in the widest sense.[163]

Sidelle’s Modal Argument is as follows:

If we can make sense out of factual differences between Nihilism and Universalism, then they consist in one theory’s positing some package of ‘superfacts’ that the other theory rejects.
We can make sense out of the package of superfacts that Nihilism posits obtaining, if and only if we can make sense out of the package of superfacts that Universalism posits obtaining.
If we can make sense out of both packages of superfacts obtaining, then neither package is necessary or impossible.
If we can make sense out of factual differences between Nihilism and Universalism, then Nihilism, if true, would be contingently true and Universalism, if true, would be contingently true.
Nihilism, if true, is not merely contingently true, and Universalism, if true, is not merely contingently true.
We cannot make sense out of factual differences between Nihilism and Universalism.
If we cannot make sense out of factual differences between Nihilism and Universalism, the best explanation for this is that there are no such differences.
If there are no such differences, then beyond answering Q1-Q3, there are no truth-makers relevant to determining whether Nihilism or Universalism are true or false.
There are no truth-makers relevant to determining whether Nihilism or Universalism are true or false.

Here a ‘superfact’ is a fact above and beyond facts about the qualitative distribution of matter. It is a fact like the (alleged) fact that some simples, x₁,…x_n do not compose anything.

What of Sidelle’s argument? The lynchpin premise here is (3). Does our ability to conceive of both sets of facts obtaining imply that both are possible? Sidelle anticipates the reply that “the necessity here is metaphysical, not logical, so we cannot take the equal conceivability of each view to establish their equal possibility in any but an epistemic sense.”[164] Suppose that the Nihilist has got things right. Then the package of superfacts posited by the Universalist is metaphysically impossible. So, our ability to ‘conceive’ of that set of facts obtaining is like the introductory student’s ability to ‘conceive’ of water’s not being H₂O.

Sidelle responds to this objection, saying

The ‘necessity’ of water’s being H₂O doesn’t rule out the other cases we might have thought of as non-water H₂O or non-H₂O water—it only keeps them from counting as non-water, or water, respectively. There is still a robust sense, then, in which the possibilities—H₂O that behaves very unwaterly, and non-H₂O that is like water in all other relevant respects—are still acknowledged...But in the current case, what possibilities can be acknowledged for the other views? None—and I think this is indicative of the fact that the only handle we have on the supposed factucal difference between the views is given by the very descriptions themselves. That is why no ‘redescription’ of the possibilities they postulate is left.[165]

If I understand Sidelle’s argument here, it is something like the following. The above rejection of (3) only works if the allegedly metaphysically necessary facts about material composition are relevantly analogous to the metaphysically necessary facts about water. Water is H₂O necessarily, and (the ontologists tell us) material objects are thus-and-so necessarily. But, in the water case, we can countenance the possibility of samples of H₂O behaving in some unwaterly ways, perhaps at worlds where the laws of chemistry and physics are somewhat different than they are at our world. We can also imagine the possibility of twin water, a substance that is just like H₂O with respect to its qualitative properties, but differs in its chemical composition. Consider an analogy with the ontology case. Suppose the Nihilist is right. Necessarily, no simples ever compose anything. Can we imagine non-composites that behave compositely? Can we imagine twin composites that are identical to non-composites with respect to their qualitative features, but differ with respect to the deep metaphysical facts about composition? Sidelle thinks that these descriptions do not carve out distinct possibilities in the way that the above descriptions about water and H₂O do describe distinct possibilities. This is a reason, Sidelle maintains, for thinking the case of H₂O and water is relevantly disanalogous to the case of material composition. The only factual differences between “Some simples always compose something” and “Simples never compose anything” are given by these very descriptions of these (alleged) facts.

Suppose we grant Sidelle this point. It does not follow from this that there is no factual difference between these two superfacts. For the lack of a rediscription of the differences between these superfacts is attributable to the phenomenon of vocabulary exhaustion rather than to the lack of any genuine difference.[166] Consider descriptions of (alleged) facts like “It is wrong to lie” and “I am a conscious being”. These facts (if facts they are) are particularly resistant to rediscription in terms that do not involve cognates for ‘wrong’, ‘ought’, ‘permissible’, and so on, for the moral case, or in terms that do not involve ‘conscious’ for the phenomenal consciousness case. Our vocabulary for describing these cases is exhausted once we are disallowed use of the operative terms ‘wrong’, ‘ought’, ‘conscious’, etc. Are we to conclude from this that alleged superfacts like “Whether something is morally wrong or not supervenes on its qualitative properties” or “Whether something is a conscious being supervenes on its microphysical properties” cannot be meaningfully held to be necessary or impossible? This seems implausible.

What is going on in the above cases is that we have reached conceptual bedrock. There just are no conceptually more fundamental terms other than ‘wrong’ (and its cognates) or ‘conscious’ (and its cognates) for redescribing facts about morality or consciousness. But this should not be taken to imply that there cannot be metaphysically necessary facts about morality or consciousness. Rather, all it shows is that these concepts are, in an important way, primitive.

I hereby suggest a dilemma in reply to Sidelle’s Modal Argument. Either the argument fails because the best explanation of our failure to redescribe the composition facts is the conceptual primitiveness of composition language, or the argument succeeds, but also shows that there is no fact of the matter in many other philosophical disputes that involve the use of conceptually primitive terms like ‘right’, ‘wrong’, ‘permissible’, ‘conscious’, etc.

As with Thomasson’s view, we see that the no fact of the matter view about ontology, if correct, seems to cut against disputes in several other areas of philosophy (e.g., ethics, metaethics, philosophy of mind). Perhaps this is no bad thing. Perhaps disputes in all of these areas are defective because there are no facts of the matter as to which views are correct. This counsel of despair would have it that much of philosophy involves attempts to settle issues about which there are no facts of the matter. Perhaps there are some who will find this conclusion satisfying. I do not, nor do I think do proponents of the no fact of the matter view with respect to ontology. The goal of many no fact of the matter theorists is to undermine ontology, specifically the kind of ontology done by disputants over positions like Nihilism and Universalism, while leaving most of philosophy intact. I maintain that this goal is problematic. Ontological disputes share some important features with many other disputes in philosophy. One such feature is that these disputes are cast in a language which is conceptually pretty fundamental. Notions of part, whole, and mereological overlap are pretty hard to explicate in nonmereological terms. So too, moral concepts like right and wrong are very hard to explicate in nonmoral terms, and concepts like phenomenal consciousness are very hard to explicate in non-phenomenal terms. Disputes in philosophy of mind and ethics, then, will be subject to arguments analogous to the ones that have been leveled against ontology. Should the latter succeed, then prospects look grim for many areas in philosophy.

Some readers have probably been wondering when our discussion would get around to Carnap. I have so far avoided Carnap for two main reasons. First, he does not specifically target the Nihilism-Universalism dispute. Second, he does not, at least in “Empiricism, Semantics, and Ontology”, seem to give an explicit argument that there are no facts of the matter in ontology. Rather, Carnap gives a model for understanding how it could be the case that there are no facts of the matter. This is a model which is fairly well motivated, but no argument seems to be given that compels us to accept it.

Some philosophers have attempted to provide arguments for accepting a Carnapian model of ontological disputes. In this section, we will look at some of these arguments. First, however, a brief description of the model Carnap proposed is in order.

Central to the Carnapian picture is the notion of a linguistic framework. What is a framework supposed to be? A thought experiment should bring out the relevant notion. Let us imagine a community of speakers members of which use a natural language L. Further, let’s suppose that L speakers want to talk about a new kind of thing they have discovered. Suppose L speakers have recently discovered that ordinary material objects have a complex structure, and that they want to introduce into L a way to talk about that structure.

L speakers, should they want to talk about objects’ molecular structure in terms of entities that they had not previously acknowledged, will have to “introduce a system of new ways of speaking, subject to new rules.”[167] The introduction of this system of new ways of speaking is what Carnap calls “the construction of a linguistic framework for the new entities in question.”[168] This way of describing a linguistic framework is somewhat problematic. For, Carnap uses the notion of a framework in order to show how certain ontological questions, “Are there really entities of this new kind?” are devoid of cognitive content. However, in describing the construction of a framework as the introduction of linguistic rules for talking about a new kind of entity, Carnap seems to presuppose that the new kind of entity is there to be talked about in the first place. This will not do.

Carnap later gives us a better description of the process of introducing a linguistic framework. There are two essential steps in introducing a linguistic framework:

1. Introduce a new general term, a predicate “permitting us to say of any particular entity that it belongs to this kind (e.g., ‘Red is a property’, ‘Five is a number’).”[169]

2. Introduce variables of a new kind, the values of which are to be those things (if any) satisfying the new predicates and which permit general statements about the things (if any) that satisfy the new predicates (e.g., “There is an x such that x is greater than five”, “For all x either x is an odd number or x is an even number.).

Along with the introduction of new predicates in (1) comes the introduction of new linguistic rules governing the application of the predicates. These are rules that specify what is to count as “confirming or disconfirming evidence for possible answers” to questions about whether something falls in the extension of the new predicates or not.[170]

Let us take an example. Suppose that our L speakers, hitherto not too keen on glancing at the heavens, suddenly discover that there are bodies discernible in the sky. They decide to introduce a new way of talking in order to discuss these things. They introduce the general term ‘planet’ and the predicate term ‘is a planet’ in order to talk about these new entities. Further, they introduce variables, p₁, …p_n, to range over things that satisfy the predicate ‘is a planet’, so that they may formulate general claims like, “There is a p_i such that p_i is a planet and p_i is red’ and “For all p_i, if p_i is a planet, then p_i is only visible at night.”

The new predicate ‘is a planet’ must be introduced alongside rules for confirming or disconfirming whether something falls under it.[171] Now, these rules needn’t be specified in any great detail in order for L speakers to count as having introduced a planet-framework. Perhaps the rules are loosely specified. The rules are also subject to negotiation. For example, at first ‘is a planet’ may be taken to apply if and only if the thing in question is visible only at night, while later users of the term may decide to modify the rules of use such that the predicate correctly applies to things also visible in the daytime.

So far, we have been using a kind of just-so story to describe the introduction of a new framework. However, speakers are often thrust into a language using community in medias res. This is to say that our language already contains rules of use for predicate terms like ‘is a number’, ‘is a material object’, ‘is a chair’, and so on. This is not to say that these rules do not constitute linguistic frameworks, however.

Central to Carnap’s picture is the notion of existence questions that are internal to linguistic frameworks and existence questions that are external to linguistic frameworks. An example will help to illustrate this distinction. Let’s take the planet framework introduced by our newly heavenward-looking speakers of L. One asks an existence question internal to the framework if one asks, using the new rules, whether there is something that satisfies the new predicate or predicates. So, a child coming to learn the new fragment of L that refers to planets will ask, “Is there a planet in the sky now?” and will be answered affirmatively if, and only if, the rules of use for ‘is a planet’ determine that there is something in the sky meeting the application conditions for this predicate. So, let’s suppose that the rules of use for ‘is a planet’ in L sanction the application of this term if, and only if, one observes a shiny speck in the night sky that does not twinkle. Upon observing the Sun, a child may ask “Is there a planet in the sky?” and be answered “No, that is not a planet.” Such questions are formulated using the grammatical rules for the new predicates and answered by appeal to the rules of use for those predicates.

What, then, is an external question? External existence questions are those “concerning the existence or reality of the system of entities as a whole”.[172] Suppose a speaker of L, having learned to apply the predicate ‘is a planet’ when she observes shiny, non-twinkling things in the night sky may come to wonder, “Are planets real or are they just illusions of some kind?” Such a question is external to the planet framework. According to Carnap, external questions can be given one of two interpretations. First, they can be seen as pragmatic questions like, “Should we continue to use the fragment of our language that contains the expression ‘is a planet’, or should we reject this way of talking in favor of some other idiom?” The second kind of interpretation of external existence questions is a metaphysical interpretation on which they ask (something like) “Independently of the rules of use that give meaning to the word ‘planet’, are there planets?” Carnap maintains that the metaphysical interpretation of external existence questions results in questions that are unanswerable.

Why should we go along with Carnap about this? Can’t we make perfectly good sense out of the metaphysical question (asked by an L speaker) as to whether planets are real or not? Isn’t she asking something that could be answered by some kind of further investigation? Carnap’s answer is that one who asks an external metaphysical question (or makes an external metaphysical claim) is asking (saying) something without cognitive content. Carnap does not explicitly give an argument for the claim that external metaphysical questions are devoid of cognitive content, though an argument for this claim can be extracted out of what he does say explicitly.

Here is the argument, as far as I understand it. External metaphysical questions must be framed using the terminology of a particular framework, e.g., “Are planets real?” But if these questions are asked outside of the framework, since the rules of the framework give meaning to its terms, these questions must contain terms that are meaningless. This meaninglessness infects the entire question, so that it does not ask anything at all.

Here it will be objected that external metaphysical questions can be asked, not by using the terms of the framework whose entities are in question, but instead by mentioning these terms. So, the skeptic’s questions, “Are planets real?” can be reframed as “Does ‘planet’ as it is used in L refer to something real?” What is the problem with this? The trouble here is that there are no framework neutral ways to frame questions about reference. The question:

QE: Does ‘planet’ as it is used in L refer to something real?

must itself be asked in some language or other. Let this language be English. English has a system of rules for the correct application of expression like ‘something real’. There is a predicate in English, ‘is a real thing’ that comes along with rules for its correct application. This predicate applies to living persons, to chairs, mountains, and so on, but does not apply to illusions, hallucinations, fictional characters, and so on. Now, QE is a question that is internal to the framework of ‘real things’. What is the correct answer to QE? In order to determine this we need to consult the rules of application for ‘is a real thing’ in English and determine whether the referents of ‘planet’ in L fall under that term or not. If they do, the answer to QE is ‘yes’, if not the answer to QE is ‘no’. However, this construal of QE makes it an unproblematic, but internal, existence question.

Suppose that one wishes to step outside the framework of ‘real things’ as specified by the practices of English speakers and ask instead:

QE¢: Does ‘real thing’ in English refer to anything at all? Is there a world of so-
called ‘real things’?

QE¢ is meaningful only if it is asked in some language that has rules of correct use for expressions like ‘anything at all’. But then, we’re back to asking an internal question again, a question that will have an answer that can be determined by comparing the rules of use of ‘real thing’ and of ‘anything at all’ in their respective frameworks. The external question cannot be meaningfully formulated, according to Carnap, because to do so we would have to form a question in something like a non-language. But this is impossible.

Just how threatening is this to the Nihilism-Universalism dispute? The case looks pretty bleak. For let’s look at what Nihilists are actually doing when they present their view (in what Sidelle would call its full-blown “package” sense). Let me introduce a fictional character called Schman Inwagen.

In sketching and defending his view, Schman Inwagen makes use of a device he calls paraphrase. How does paraphrase work? Schman Inwagen introduces some new variables, plural variables like ‘the xs’, that are to be used in connection with currently existing singular variables like ‘y’. Schman Inwagen specifies rules of use for constructing general claims involving these plural variables:

One may form sentences containing plural referring expression by combining these expression with so-called variably polyadic or indefinitely polyadic predicates: ‘are in a minority’, ‘are quarreling’, ‘are carrying a beam’, ‘outnumber’, ‘run the risk of alienating’, and so on. We have, for example, such sentences as ‘Tom, Dick, and Harry are carrying a beam’ and ‘The Democrats run the risk of alienating those voters who live on a fixed income’.[173]

After having introduced these variables, Schman Inwagen introduces a new stock of predicates: ‘are arranged chair-wise’, ‘are arranged rock-wise’, ‘are arranged mountain-wise’, and so on. There are also a number of other predicates, ‘is a simple’, which applies to something just in case that thing has no mereological parts; there is also, ‘is a an object’ which applies to something if, and only if, it is a simple. Call the language-fragment that constitutes Schman Inwagen’s proposed manner of speaking Inwagish.

What is Schman Inwagen doing? The Carnapian answer is that he is introducing a linguistic framework. He gives us a new stock of predicates together with rules for determining when they apply, and he gives us some variables that range over the things that satisfy these new predicates. This is unobjectionable, as far as Carnap is concerned, for he abides by an injunction disallowing intolerance when it comes to linguistic forms: “let us grant to those who work in any special field of investigation the freedom to use any form of expression which seems useful to them”.[174] What Carnap does object to is Schman Inwagen’s insistence that the only objects that are real are the ones that fall under the predicate ‘is simple’.

Now, let’s imagine another fictional character, David Schlewis, who gets into an argument with Schman Inwagen. Schlewis says, “You’re wrong Schman Inwagen, there are material objects that have proper parts. Furthermore, for any simples whatsoever, there is a material object that is their mereological fusion.” What Schlewis says here is intelligible only if ‘mereological fusion’ is meaningful. Schlewis introduces the term ‘is the mereological fusion of a and b’ by saying that it is satisfied if, and only if, a and b both exist. Schlewis then introduces some variables ranging over things that satisfy this new expression and formulates the general sentence, “For any two objects, a and b, there is an x such that x is the mereological fusion of a and b.” Again, Carnap would have no problem here, for we should be “tolerant in permitting linguistic forms”.[175] Schlewis is free to talk any way he thinks would benefit his special area of inquiry. Call Schlewis’s proposed framework Lewish.

But, there is a problem when Schlewis and Schman Inwagen get together. They will be inclined (provided that they are not good Carnapians) to think that there is a serious dispute to be had about who is right. They will bicker over answers to the following:

QE¢¢: Are there really any mereological fusions?

Schlewis says, ‘Yes there are’ and ‘Schman Inwagen says, ‘No there are not’. What is going on here?

One of two things is happening, according to Carnap. Either Schlewis and Schman Inwagen are having a pragmatic dispute about which language to adopt, or they are having a dispute over an external metaphysical question. If the former is going on, then Carnap has no complaint. If the latter is going on, then there is no fact of the matter that could possibly settle the dispute. So goes the Carnapian story. Is there any reason to believe it? We will examine this question below.

Before determining how acceptable Carnap’s picture is, I want to consider an objection to it that, I think, totally misses the mark. Explaining how this objection misses the mark will help us to understand exactly what Carnap’s position is and how it is to be appropriately criticized.

This line of resistance to Carnap’s picture comes from Ted Sider’s introduction to his Four Dimensionalism, and can be found in passages like the following:

Carnap’s challenge to ontology is that there are many possible linguistic frameworks containing different rules of use for the symbol ‘$’. These frameworks agree on a core inferential role of ‘$’ that is appropriate to existential quantification but differ in other respects…If ‘$’ has different but equally legitimate meaning in all the frameworks, then the ontological enterprise is trivialized for no one framework is correct to the exclusion of the others.

Given the best-candidate theory we can avoid trivialization. The inferential role played by the symbol ‘$’ in our use of that symbol is only part of what secures its meaning. Another part is the intrinsic eligibility of candidate meanings. Suppose the world comes equipped with ‘logical joints’ as well as extra logical ones. In particular, in addition to there being distinguished classes of objects that count as genuinely similar, the world comes ‘ready-made’ with a single domain D of objects: the class of all the objects there are.[176]

On Sider’s proposed reply, there is a framework neutral way to give meaning to existence questions. Simply specify that one is using the quantifier to range over only those things in the preferred domain D, and external existence questions become meaningful.

This proposal does not work against Carnap. It does not work for it betrays a fundamental misunderstanding of what Carnap took himself to be doing when he proposed the internal-external distinction in terms of linguistic frameworks. What is it that Carnap took himself to be doing, exactly?

Carnap took himself to be making a proposal. This proposal he did not take to be truth evaluable, for it takes a form like the following: “Let’s talk like this about ontology”. Carnap is himself best thought of as introducing a linguistic framework. Here’s how it goes. Let’s use the predicate term ‘is a linguistic framework’ in accordance with the following rules of use. Anytime a theorist introduces a way of talking by appeal to the two steps (1) and (2) above, we will say that the resulting language or language fragment is a linguistic framework. We can now introduce some new variables ranging over frameworks that allow us to make the general claim, “For any sentence S, S is meaningful if and only if there is some f such that f is a linguistic framework and f specifies rules of use for S.’

Carnap’s proposal, then, is not a claim. It is instead an invitation to talk in a certain way. What could count as a criticism of his proposal? The only thing that could tell against this proposal is an argument that shows there is some other proposal that meets our goals better.[177]

I want to make an objection to Carnap on these sorts of prudential or pragmatic grounds. What kind of benefits accrue to the philosopher who adopts Carnap’s proposal? Allegedly, ontological disputes evaporate or become more tractable. For, they are recast instead as disputes about which language to adopt, and such disputes are resolved by appealing to notions like elegance, simplicity, and so on.

Does this move to Carnapianism really make ontological disputes more tractable? It is hard to see how this is the case. Imagine Schman Inwagen and Schlewis coming to adopt Carnapianism. Doesn’t it seem like their dispute will be just as interminable? Schman Inwagen will say, “Inwagish is a simpler, more elegant language that Lewish, and it does a better job at meeting our goals!”, while Schlewis will disagree vehemently. It seems like the dispute will be just as intractable.

There is another sort of problem analogous to the one that I raised for Thomasson’s view. If we accept Carnap’s proposal with respect to ontology, it seems to command our acceptance more broadly in philosophy. As it turns out, there will be no fact of the matter as to whether Carnapianism itself is true or not, for it is merely a proposal. Even worse, however, there will be no fact of the matter as to whether any philosophical view is true or false, for all such views can be interpreted as suggestions or proposals to use language in a certain way.

This result would not have been alarming to Carnap. However, it may be very alarming to those contemporary philosophers who wield Carnap against ontologists. If these philosophers want to buy into Carnapianism about ontology, then they should go all in, accepting that philosophical views generally (including Carnapianism itself) are merely proposals to use language in a particular way.

Once appreciated correctly, I think, Carnap’s view requires a dramatic reappraisal of what it is that philosophers do. We do not propose theories; instead we make proposals. This sort of view will not be very palatable for many philosophers. Above I asked, with respect to Carnap’s view, “Is there any reason to believe it?” This, I hope, can now be revealed as a very misleading way to approach Carnap’s view. For, Carnap is not making a claim that one can accept or reject, believe or disbelieve. Instead he is offering us a proposal which we can find expedient or not.

I maintain that there is no more reason to find Carnap’s proposal expedient than there is to find Sider’s proposal expedient. For it seems to me that both views will lead to ontological disputes that go on and on, with no hope for resolution. Perhaps one view is better for certain purposes (e.g., teaching ontology to undergraduates), but I do not think there is any reason to prefer Carnap’s view to Sider’s in general. Perhaps Carnap would be happy with this assessment. His present-day followers, I think, would not.

At this point I anticipate the objection that I have given contemporary neo-Carnapians short shrift indeed. For after all, isn’t it a theoretically live option to adopt what Carnap said about linguistic frameworks while shedding his commitment to the principle of tolerance? Can’t a neo-Carnapian say consistently that her view is not just one proposal among many, but instead that it is the God’s honest truth? What would be wrong with this?

Perhaps there is nothing wrong with such a move; however far removed it may be from what Carnap intended. After all, Carnap has been criticized on other grounds as well, e.g. his commitment to a disreputable analytic-synthetic distinction. In what follows, then, I will look at two theorists who depart from Carnap in the letter of his view, but maintain something like the spirit of it: Steven Yablo and David Chalmers.[178]

Yablo defends the view that the contemporary practice of ontology rests on a mistake.[179] The mistake is adhering to a Quinean methodology for doing ontology. According to Yablo, the Quinean method is embodied by (Q):

(Q): count a thing as existing iff it is a commitment of your best theory, i.e., the
theory’s truth requires it.[180]

Yablo argues that following (Q) will be ultimately unsuccessful in answering ontological questions. His master argument is as follows:

If the ontologist’s existence questions (e.g., are there really chairs or just simples arranged chair-wise?) had answers, then Quine’s method of applying Q would answer them.
Quine’s method of applying Q cannot answer the ontologist’s existence questions.
The ontologist’s existence questions do not have answers.
If (3), then there is no fact of the matter as to whether Nihilism or Universalism are true or false.
There is no fact of the matter as to whether Nihilism or Universalism are true or false.[181]

The argument as I have represented it differs somewhat from Yablo’s own formulation of the argument, for he does not specifically target Nihilism and Universalism; instead he takes his remarks to apply to the post-Quinean practice of ontology generally. I take it Yablo would have no objection, though, to this particularized version of the argument.

Should we accept this argument? It is certainly valid. (3) follows straightforwardly from (1) and (2). (1) seems to be presupposed by many philosophers currently working in ontology. Many practicing ontologists seem to follow a broadly Quinean methodology; arguing that some theory is best on the grounds that it is consistent, captures our intuitions better than its competitors, is simpler, more elegant, more in line with scientific findings than its competitors, and so on. So, let’s grant premise (1). Premise (4) looks unassailable, for if ontological questions in general have no answers, then it seems pretty hard to see how Nihilism or Universalism could be true or false. One caveat bears mention here, however, with respect to our acceptance of premise (4). If either Nihilism or Universalism turns out to be internally inconsistent, then premise (4) is false. So, premise (4) should be read as having a “provided both views are internally consistent” clause tacked on. For my part, it is an epistemic possibility that one of these two views involves a subtle, difficult to detect inconsistency. I do not want to rule this possibility out.

Everything, then, turns on premise (2). What reasons are there to believe it? Yablo argues persuasively that applying (Q) in practice requires a fairly clear and principled distinction between literal existence claims and figurative, metaphorical, or make-believe existence claims. No Quinean worth his salt would think that our acceptance of “He left me in the lurch” on some occasion seriously commits us to the existence of entities we might call lurches. This is because the apparently existentially committal sentence can be paraphrased away. We do not literally mean, when we say that we’ve been left in the lurch that there is an x such that x is a lurch and we have been left in x.

Yablo’s remarks on metaphor and make believe are extensive and intricate. I do not wish to deal intimately with the details. The following, then, will depict Yablo’s reasons for accepting (2) in very wide brushstrokes. Here is, roughly, the argument. Applying (Q) will only yield answers to our ontological questions if our best theory does not contain quantified sentences (or sentences that imply quantified sentences) that range over variables, but that aren’t meant to be taken literally. But, there is no reason to believe that our best theory will be bereft of non-literal, ontologically loaded sentences in such a way. So, (Q) does not yield answers to ontological questions.

Here is the argument in schematic form:

Q answers all ontological questions only if all metaphor, make believe, and non-literalness ultimately drops out of the language in which we state our best theory.
It is not the case that all metaphor, make believe, and non-literalness will ultimately drop out of the language in which we state our best theory.
Q cannot answer all ontological questions.

Why should we accept (1)? The answer here seems pretty obvious. Suppose our commitment to: “There is an x such that x is a composite material object” involves some kind of non-literalness. Suppose that we do not really take it to be straightforwardly true, but instead it operates in a complicated game of make believe. We act and talk as if there were composites, but we do not take this talk seriously. If this is the case, then (Q) cannot solve the ontological question, “Are there really mereologically composite tables?”

The best hope for the practitioner of Quinean ontology, then is to deny (2). Yablo “see[s] only two ways of avoiding this result.”[182] Let us consider each in turn. The first way is to “say that the make-believe elements are never going to make it into our theories in the first place” because “[a]s theorists we are in the business of describing the world; and to the extent that a statement is something to be pretended true, that statement is not descriptive.”[183]

Here is the argument:

Q1. Claims that are (merely) pretended or made-believe to be true do not
describe the world.

Q2. Every claim made by our best theory describes the world.

Q3. Metaphor, make believe, and non-literalness will drop out of our best theory.

Yablo rejects Q1. In arguing against Q1, Yablo draws our attention to prop oriented games of make believe:

…a game whose content reflects the game-independent properties of worldly props can be seen in two different lights. What ordinarily happens is that we take an interest in the props because and to the extent that they influence the content [of the game]; one tramps around London in search of 221 B Baker street for the light it may shed on what is true according to the Holmes Stories.

But in principle it could be the other way around: we could be interested in a game’s content because and to the extent that it yielded information about the props. This would not stop us from playing the game, necessarily, but it would tend to confer a different significance on our moves.[184]

The leading idea here is that some games of make believe involve commitments (tacit or avowed) to treating certain statements as if they were true. Consider some children playing ‘mud pies’. The children treat the claim “this lump of mud is a pie” as true for the purposes of the game. When they do so, the lump of mud is a prop in their game.[185] Now, for the most part, we are interested in features of props because of how they influence the content of (i.e., what we take to be true in) the games we play with those props. But Yablo notices that things could go the other way around. We might become interested in what is true in the game because we want to learn something about the props themselves.

Take, as an example, the make-believe assertion “Jim’s pie is too big for the oven!” We might become interested in this pretend-claim insofar as it can tell us some fact about the world (a certain lump of mud is larger than the aperture of a certain hollow stump, say).[186] Yablo remarks that it is true that “to pretend is not itself to describe” but that, “the reason for the pretence may be to portray the world as holding up its end of the bargain, by being in a condition to make a pretence like that appropriate.”[187]

So, it is just not true that claims we only make believe to be true do not describe the world. Consider: Italy is a boot and Crotone is in the arch of the boot.[188] This bit of make believe tells us something about the world, namely where Crotone is located.

This is all well and good. However, the Quinean can object that the descriptive content that certain make believe assertions carry can always, in principle, be got at by sentences that are intended literally. To take this line is to attempt the second strategy for resisting Yablo’s argument. This line of response involves the concession that “[o]ur theories may start out partly make-believe (read now as metaphorical),” but to insist that “as inquiry progresses the make-believe parts gradually drop out.”[189] This is to suggest that the ideal scientific language that is the limit of inquiry will not contain any statements intended metaphorically.

Yablo argues that this line cannot be successfully maintained. Yablo argues that is plausible that certain metaphors will not drop out. What reason could there be to accept this? According to Yablo, “[t]he most obvious reason is a lack of a literal alternative.”[190] There may be insufficient linguistic resources to capture the content that is accessed metaphorically. Second, there may be metaphors, perhaps “conventionally ‘picturesque’ metaphors that pack a cognitive punch no literal paraphrase can match” such as “scientific metaphors like feedback loop, underground economy, and unit of selection”.[191]

Metaphor, Yablo thinks, is with us to stay. If that is the case, then the non-literal truth about what our best theory says exists may be inaccessible.

Should we go along with Yablo here? I have no quarrel with the particulars of Yablo’s argument per se. It does seem, however, that if his argument is sound, its results do not condemn ontology alone. If there is a principled problem with making a distinction between sentences that we take to be literally true and sentences that involve some modicum of make-believe, metaphor, or other non-literalness, it seems hard to see how the practice of philosophy can continue. For, philosophers (at least most of them) purport to be getting at the strict and literal truth about their areas of interest. Consider ethics. Ethicists want to know what kind of lives we should be living, what it means to be flourishing as a human being, what is morally right and what is permissible. But couldn’t it be the case that metaphor infects each of these areas of inquiry? Do we mean ‘flourishing’ literally or do we mean it metaphorically? Do we mean ‘permissible’ literally or is it some kind of legalistic metaphor? If there is no clear answer to these questions, then many of the most important questions in ethics will not be answerable.

Take the philosophy of mind. Perhaps talk of intentional states is an irreducible metaphor. Perhaps there are no clear answers to questions like “Do we mean talk of beliefs and desires literally or only metaphorically?”, “Do we intend talk about phenomenally conscious states to be taken seriously?” and so on. If this is the case, then many of the most interesting questions in philosophy of mind will not be answerable.

Perhaps this is as it should be. Philosophers make careers out of vying for answers to unanswerable questions. This is not a very happy result, but perhaps it is the God’s honest truth.

I want to maintain that even if Yablo is right and there is no hope for ontological disputes, including the Nihilist-Universalist dispute, to be resolved, and this is because there are no facts of the matter, it does not follow either that (a) Nihilists and Universalists are not having genuine disputes (as would seem to be the case if Carnap is correct) or (b) rationality dictates that ontologists should give up the ghost. Perhaps if Yablo is right ontologists should stop having disputes, for these disputes would be defective. Nevertheless, ontologists should go on having debates.

Why? In the last chapter I made some remarks about this, but I will expand upon these somewhat here. Ontological debates are worth having even if there are no answers. This is because the discursive practice of debate over ontological issues can teach us things that we might not be able to learn elsewhere. We can come to learn facts about our mereological concepts and we can have our pretheortic assumptions laid bare. Debate, insofar as it is a discursive practive, involves the participation of a community (of at least two persons). When we subject our philosophical views to the scrutiny of the community in a public forum for debate, presuppositions that we might have found initially very plausible or cognitive biases of which we may have been unaware come into the light of inquiry where they can be exposed, tested, and rooted out if they prove to be irrational. Taking a cue from Yablo, perhaps philosophers (when they are practicing well) are not even taking themselves literally! Perhaps we are engaging in a centuries old game of make believe in which we pretend that there are answers to questions like “Is there a God?”, “What does it take for a creature to be conscious?’, “What should I care about?”, and “What is there?” Why do we continue to engage in this act of pretense? There are a number of reasons. Asking the questions is stimulating, interesting, and fun. If we behaved as though the questions had no answers, we wouldn’t get anything out of doing philosophy. Consider the stick-in-the mud students who refuse to take philosophical questions seriously at all. What are they poised to learn from studying philosophy? Not very much at all.

There is another reason why we might engage in philosophical make-believe. Because we want to learn something about the props. What are the props in philosophical games of make believe? They are concepts, mostly, but also claims, models, thought experiments, intuitions, and so on. By playing the philosophical game, we come to learn things about these props. To wax Wittgenstinean, one might say that the whole purpose of the game of philosophy is to learn, in the end, that we needn’t be trapped by it; to show the fly the way out of the fly bottle. Wittgenstein never meant to assert that there was anything analogous to a fly bottle; instead we pretend that there is one. Philosophical problems, on this way of thinking, are not real problems. They are ‘made up’ problems. Why do we find them so damned interesting, puzzling, and worth our time? For the same reason we play games in general. They sharpen our wits, entertain us, and (sometimes) even teach us things.

Now, the view of philosophy as make believe is not likely to attract many adherents. After all, philosophers tend to think they’re after the truth and there’s no pretense about it. I am inclined to take such philosophers at their word: they do not think they are pretending.[192] But, put one of these ‘serious’ philosophers in front of a classroom and see what they do. Won’t they “argue for” skepticism? Won’t they try to “prove” that there aren’t any chairs? That we have no free will? Now, perhaps these classroom pretenses are merely pedagogical tools that the serious philosopher will shed when she sits down to the practice of serious philosophical research aimed at getting the truth. Maybe, however, there is no principled distinction to make between the make believe of the classroom and the serious philosophical work of writing journal articles.

I have no argument for the view that philosophy is best seen as a game of make believe in which the pieces are concepts, intuitions, arguments, and the like. However, such a view does seem to make sense out of what we do and out of how what we do is perceived by many laypersons. Take the non-philosopher who doesn’t see the point of philosophy. How is such a person any different from the philosopher who doesn’t see the point of basketball, curling, or rowing crew (I am such a philosopher, and I suspect there are many more like me)? Such a layperson just doesn’t get it. Is that because there is something wrong with them? I do not think so. They are not moved to play by the philosophical rules. We are all famililar, as philosophy instructors, with students who refuse to play by the rules (e.g., students who try to wriggle out of thought experiments). They don’t find the game we’re playing meaningful or useful.

In an important sense, however, we philosophers have a bit of the layperson in us as well. There comes a time in the course of every philosopher’s day when she will put away philosophical questions. When at the supermarket, I—serious ontologists that I am—will just stop worrying about whether there are really cantaloupes and I will buy some. This doesn’t quite make sense if I seriously doubt whether there are any.

I do not expect that many philosophers will be willing to accept the view of philosophy suggested by the above remarks. That is fine. I merely suggest that it is consistent with this view that ontologists are having real debates about issues that are really important. They are just pretending that there are answers to be had. If something like this model is correct, then Yablo’s remarks that target ontology can be welcomed by any serious ontologist. In order to carve out a logical space for doing serious ontology, the model I describe needn’t be accurate, only consistent. I maintain that it is.

David Chalmers takes the cantaloupe-buying phenomenon I mention above as evidence for the claim that there is a way of making existence statements (e.g., there are canteloupes) that does not threaten the ontologist’s practice of making existence statements (e.g., there are no composite objects, including canteloupes). Chalmers calls this the distinction between ordinary and ontological existence assertions.[193] He uses this distinction to breathe new life into Carnap’s much abused distinction between internal and external existence questions. In this section we will examine the neo-Carnapian view that Chalmers states in terms of this distinction. If the view Chalmers defends is correct, then there are no facts of the matter that could in principle settle the Nihilism-Universalism dispute.

The view Chalmers defends is complex and intricate. My criticisms of his proposal will be largely big-picture sorts of worries that do not treat the details. Chalmers begins by reflecting on what looks like an obvious piece of data: assertions of existence made in ordinary, non-philosophical contexts are not sensitive to ontological arguments. Take the above example involving cantaloupes. Suppose that I ask my grocer where I can find some cantaloupes. She responds saying, “There are some cantaloupes over there, beside the watermelons.” What she said seemingly commits her to the existence of cantaloupes. It would be inappropriate, however, for a Nihilist who overheard our conversation to say, “No there are not; cantaloupes are composites and there are no composites, so cantaloupes do not exist.” The impropriety of such a claim is not just a function of its being rude or out of place, rather it seems like the Nihilist misunderstood what the grocer meant to do with her assertion.

Chalmers characterizes this distinction as one between two types of existence assertion: ordinary and ontological. According to Chalmers, an ordinary existence assertion “is an existence assertion of the sort typically made in ordinary first-order discussion of the relevant subject matter” and it is “insensitive to ontological matters”.[194] Contrast this with an ontological assertion, which is “an existence assertion of the sort typically made in broadly philosophical discussion where ontological considerations are paramount.”[195] According to Chalmers, ontological assertions are sensitive to ontological concerns such as philosophical arguments, accord with philosophical principles, and so on, while ordinary existence assertions are not. Consider how inappropriate it would be for me to object to my grocer’s assertion on the grounds that cantaloupes, if they existed, would have vague persistence conditions. Ordinary and ontological assertions have what Chalmers’ calls different correctness conditions. An ordinary assertion of ‘there’s a table’ is obviously correct in situations that contain a tablish arrangement of stuff. Ontological existence assertions are not obviously correct in such situations.

It should be clear that this distinction between ordinary and ontological existence assertions can do much of the work that Carnap’s distinction between internal and external existence assertions can do. Chalmers’ distinction scores over Carnap’s in two important respects. First, “the terminology of ‘internal’ and ‘external’ is too closely tied to Carnap’s theoretical apparatus involving frameworks to serve as a neutral starting point”, and second, “‘internal question’ and ‘external question’ may suggest two different sorts of sentence” whereas it may be more appropriate to draw the distinction in terms of “different uses of sentences (or perhaps, between different evaluations of sentences).”[196]

I agree with Chalmers that his distinction scores over Carnap’s. I think, however, that the distinction is not needed in order to explain the data. It is true that we do not take philosophical arguments as counting against our ordinary claims of existence. When we leave the philosophy seminar room, even the staunchest Nihilists among us will still talk about chairs and tables, persons and so on. Further, we will not take even our own philosophical arguments to count against the correctness of what we say outside the seminar room. What is going on here?

Perfectly orindary features of conversations can explain why ordinary assertions may look insensitive to ontological matters.[197] Consider the following case. While setting the table, we have this conversation:

You: There are three glasses on the table.

Me: No, there are no glasses. After all, glasses, if they existed, would be vague
objects, and there are no vague objects.

My denial of your assertion is inappropriate, and likely to seem incorrect, given our joint goal of getting the table set. My introducing philosophical concerns into the conversation in the way I do does not help us achieve our shared goal of setting the table and threatens to put us off by getting us bogged down in an ontological dispute. Notice that this sort of feature generalizes. Take a conversation about whether Jones accepted a bribe freely or not:

You: Jones accepted the bribe of his own free will.

Me: No he didn’t, because causal determinism is true.

If the purposes of our conversation are to resolve certain practical issues (should we trust Jones to chair the finance committee or not), then the philosophical consideration I raise is likely to get us off track, and will therefore be ignored or rejected as relevant. This does not show that it is irrelevant to the correctness or truth of what you said, however.

Another explanation of the phenomenon Chalmers’ points out is to be found by looking to our practice of making and evaluating assertions in general. There are some good reasons to believe that the practice of assertion is governed by our tacit acceptance of certain norms. Some philosophers have convincingly argued that one such norm is (something like) the following:

KNorm: Assert that p only if you know (or at least justifiably believe) p.[198]

Typically we hold speakers accountable for knowing, or at least justifiably believing, what they assert. This practice undergirds another assertoric practice: that of challenging an assertion.

Suppose that A asserts, “It is not going to rain tomorrow.” We may challenge her assertion, asking what evidence she has for the truth of this claim. Whether we count A as having met the challenge, however, will depend upon our standards for assessing the strength of her evidence. Which standards are in play is a contextually determined feature of the conversation. So, suppose that not too much hangs on whether it will be fair tomorrow. We may count A’s having read a weather forecast in yesterday’s paper as sufficient for her to meet a challenge to her assertion. However, if quite a lot hangs on whether it is fair tomorrow (say we have a football match or some such thing), we may require more evidence.

How is this relevant to ontological assertions? The norm, “Assert only what you know (or at least have good evidence for),” together with a sort of sliding scale for what counts as knowledge or good enough evidence can explain our pattern of linguistic behavior when it comes to existence assertions. Much ink has been spilled on the details of how this works.[199] I do not want to commit to a particular view about the semantic machinery. For our purposes it suffices to say that this picture explains our dispositions to assent to certain so-called ordinary existence assertions when we would otherwise balk at those same ontological assertions. Let’s take an example.

Suppose that we are picking out our seats in a restaurant, and ontology has so far not been a serious topic of conversation. I say, “Here’s a chair,” and no one objects. Why not? Most of my companions at dinner are philosophers, so why shouldn’t they produce some kind of argument. Given the conversational parameters, this assertion, if challenged, could be defended by an appeal to unaided vision: I see a distribution of matter arranged chairishly. Under the conversational circumstances that prevail at the context of my utterance, I count as possessing adequate justification to discharge my conversational duties with respect to KNorm. I needn’t be able to rule out certain ontological arguments against my assertion, for these are not relevant, given the standards for justified belief that obtain at the context of utterance. Compare the case of my asserting that there is a chair with the case of my asserting that I know where my car is parked. In the context of our present easy-going conversation, I needn’t be able to rule out the possibility that I am a brain in a vat and thereby cannot possibly know where my car is parked. The standards for my knowledge assertion are relatively low. But, these standards could be inflated were certain skeptical scenarios to become conversationally salient.[200]

What is going on in the cases that Chalmers mentions, I maintain, is not that conversational context produces a shift in the semantic value of existence claims, nor is it the case that there are two separate uses of existence claims, but rather there is contextually determined variability in the standards required for justified belief or knowledge.[201] In ordinary contexts, evidence about the qualitative distribution of matter is sufficient to meet challenges to existence assertions. In philosophical contexts, however, certain arguments and considerations become salient, thereby driving up the standards for justified belief so that one cannot count as justifiably believing, and thereby being entitled to assert, an existence claim without having a counterargument or response to certain ontological arguments and considerations. Take an example. Suppose that, in the restaurant case, the conversation shifts to talk of philosophy, in particular to talk of ontology. Issues about vagueness come up for consideration. It now seems that my interlocutors will hold me to a higher standard if I say, “Here is a chair.” They will want to know what answers I have to various philosophical problems involving vagueness, and will not count me as warranted in asserting that there is a chair absent some kind of evidence that I have a ready reply to these problems. This is all to the good, but it does not show that I am using the sentence “Here is a chair” in a different way than I did earlier when the standards were lower; instead it shows that the salience of certain phenomena (e.g., vagueness, indeterminacy, arbitrariness, etc.) suffice to raise the standards for what counts as justified belief that there is a chair here.

The account in terms of the context sensitivity of standards for assessing assertions in accordance with KNorm explains more than the distinction between ordinary and ontological existence assertions. This model explains a wider pattern of assertions and retractions of assertions than Chalmers’ distinction alone can, so it is to be preferred.

If I am right about this, then there is no good reason (or, at least the data to which Chalmers points provide no good reason) for accepting a distinction between two different uses of existence sentences.[202] If this is correct, then there is no ground for Chalmers’ analogue to the internal-external distinction. Already the neo-Carnapian view Chalmers sets out to defend seems to be built upon shaky foundations. In what follows, however, I will provisionally grant Chalmers his distinction between ordinary and ontological existence assertions.

Chalmers argues that a robust ontological realism of the sort required to defend the view that there is a determinate, non-trivial fact of the matter as to whether Nihilism or Universalism is true requires that there be a concept of absolute quantification.[203] What is a concept of absolute quantification? Chalmers has the following to say about what such a concept would be if one existed:

[T]here is prima facie reason to think that participants in ontological debaters use terms such as ‘exist’ to express a single, common, primitive concept. We might call this the concept of absolute existential quantification. Since this concept, if it exists, is primitive, there may be little one can do to give an informative analysis of it in independent terms. But one can at least say some things to elucidate the concept. Uses of the concept attempt to quantify over absolutely everything—that is, over everything that exists, in the most fundamental sense of ‘exists’. Of course these elucidations are circular, but they may at least be helpful.[204]

Chalmers does not “think it is obvious that there is a concept of absolute existential quantification,” but it seems clear that the serious ontologist who maintains that there is a determinate, non-trivial answer to ontological questions at least attempts to employ such a concept.[205]

Absolute existential quantification requires that the world come “pre-carved” with a unique domain of objects. It is this assumption of a unique domain which Chalmers finds implausible. Let’s consider our dispute between Peter V. and David L. about how many objects are in the vacuum chamber. Is there a determinate answer to the question whether Peter V. or David L. is correct? It might seem that there has to be. Call the view that there is no determinate answer to this ontological question ontological anti-realism with respect to that question. According to Chalmers:

The central objection to ontological anti-realism…goes as follows. Ontological anti-realism holds that ontological existence assertions can lack an objective and determinate truth-value, even when the nonquantificational vocabulary is unproblematic. But the (absolute, unrestricted) existential quantifier is a logical constant, with a logically defined semantics that is objective and determinate. So this quantifier cannot fail to yield objectivity and determinacy for the assertion as a whole, when the other vocabulary is nonproblematic. So ontological anti-realism is false.[206]

We have encountered this sort of argument before in connection with Thomasson. Chalmers’ reply is somewhat similar to the one Thomasson gives: there is no non-question-begging way for the ontologist to specify a unique domain of discourse. But, the semantics for the existential quantifier “tells us only how to evaluate a quantified statement in a model.”[207] A key component of any model sufficient for yielding a truth value for quantified formulae is a domain. But, according to Chalmers “[f]or assertions of quantified statements to have a truth-value, we have to evaluate them at a world,” and “[w]orlds are not models.” Worlds “do not obviously come with built-in domains.”[208]

Given Chalmers’ remarks, one can formulate the following argument:

1. If ontological assertions (including the assertion that Nihilism is true and the assertion that Universalism is true) have determinate, objective truth values, then the world must come ready-made with a domain.

2. The world does not come ready-made with a domain.

3. Ontological assertions (including the assertion that Nihilism is true and the assertion that Universalism is true) do not have determinate, objective truth values.[209]

The leading idea is pretty simple. Ontological views involve commitment to existence claims. Existence claims are evaluable for truth only if some domain is specified. Since the ontologist means for the domain to be unrestricted, to contain everything that exists, this means that the ontologist must appeal to a single, unique domain against which to evaluate her existence claims. But, the domain cannot be specified by appeal to our interests or activities or whatever (e.g., we cannot say, “Consider the domain consisting of all the things on Earth”); the domain has to be uniquely determined by the structure of the world. Chalmers doubts that the world has enough structure to do the job. Ontologists (notably Ted Sider) disagree: the world is up to the task.[210]

Obviously the key premise in the above argument is (2). Are there good reasons to accept it? One might worry here. Suppose that Chalmers is correct. Existence questions do not always have objective, determinate, truth evaluable answers. How then should we go about answering the question as to whether (2) is true? (2) seems to involve commitment to (something like):

NODOMAIN:   It is not the case that there is an x such that x is the unique
                          domain of discourse for evaluating quantified claims at this
                          world.

Sider and other serious ontologists deny this, affirming instead:

DOMAIN: There is an x such that x is the unique domain of discourse for
evaluating quantified claims at this world.

But both NODOMAIN and DOMAIN are ontological existence assertions. If ontological anti-realism is true, we cannot be sure that these questions have determinate, objective answers.

Perhaps this seems like a cheat. But let us ask a more practical question. What could possibly count as evidence for or against one of these two theses? Is it plausible to think that there is an objective, determinate fact of the matter as to which of NODOMAIN or DOMAIN is true? If first-order ontological questions like “Are there mereological fusions or not?” have no determinate answers, how can we make sense out of the claim that there is a determinate fact of the matter as to whether NODOMAIN or DOMAIN is true? I cannot quite get a grip on what the world would have to be like if there were a determinate answer here. Perhaps this is a failing on my part, but it makes me suspicious that there is even a genuine issue here at all.

10.

I do not claim to have refuted the position Chalmers defends. There is much that is theoretically elegant and persuasive about Chalmers’ semantics for existence assertions (both ordinary and ontological). I do not have the requisite technical skill to assess competently the details of his proposal. I leave this task to semanticists and philosophers of language.

While I do not think I have refuted Chalmers, I do think I have raised a serious concern for his and other no fact of the matter views about ontology. Many philosophers claim to detect a whiff of superficiality in the ontology room. I think that this is a datum that serious philosophers would be intellectually remiss to ignore. However, it seems to me that there is a suspicious odor of superficiality in the meta-ontology room. How could it be that there are facts of the matter as to whether there are facts of the matter about first-order ontology? What would the world have to be like in order for there to be something that fixes the truth of theses like DOMAIN and NODOMAIN? I find these meta-ontological questions just as suspect as some philosophers claim to find the first-order ontological questions. Some philosophers have a hard time seeing what facts could possibly determine who is right in certain ontological debates; I have a hard time seeing what sorts of facts could determine who is right in meta-ontological debates.

Perhaps the best picture is the thoroughgoing Carnapian view that there are no facts of the matter at all. What we have are proposals all the way up the iterative hierarchy. There are no facts that fix the truth of ontological or even metaontological positions, for these positions themselves are not (and are not even intended to be) truth apt. Instead they are proposals. This view may seem very unattractive to many. Accepting it might seem like adopting a counsel of despair. I feel the pull of this way of thinking. Accepting the thoroughgoing Carnapianism does seem, at first, like giving up when the ontological (and meta-ontological) going gets tough.

But the Carnapian view, when taken to its extremely radical and ultimate conclusion, can also appear liberating. I maintain that if we are going to be Carnapians, then we should be Carnapians all the way. We should admit that even our commitment to Carnapianism is itself a proposal that is not truth apt. Where is the shame in doing this? There are still truths that we can discover; we can still do philosophy; we just have to rethink how we’ve been approaching our discipline.

There are some who will regard the rational pull towards taking this extreme Carnapian line that seems to gain hold once we accept a more watered-down Carnapianism (if there is indeed any rational pull here at all) as providing a reductio of the watered-down Carnapianism. Perhaps this is correct. I then present a disjunctive conclusion for neo-Carnapians: either be Carnapians all the way or go home. It is just not plausible, I think, to tow the Carnapian line against ontology but then draw the line at meta-ontology. If you can’t stand the heat, get out of the kitchen.

Here I reiterate my skepticism as to whether Carnapianism, understood as the project of rendering all philosophical positions including itself as proposals and not theories, can be defended on prudential grounds against competitors. It does not seem clear to me that adopting the Carnapian proposal is eminently more practical than adopting some other proposal. As far as I know, no one has given a thoroughgoing defense of Carnap’s view in these terms. I invite defenders of neo-Carnapianism to produce such a defense and let us evaluate it.

I do not claim to have exhausted all of the arguments for a no fact of the matter view about the Nihilism-Universalism dispute. I harbor the suspicion, however, that any no fact of the matter view will either (a) indict many more subdisciplines in philosophy than ontology or (b) be such that there is not clearly a fact of the matter as to whether it is true.

Chapter 9

EPISTEMICALLY DEFECTIVE

I now want to turn our attention to some arguments designed to show that the Nihilism-Universalism dispute is epistemically defective. Recall that S’s dispute with S* is epistemically defective just in case there is no available evidence that could make it rational to accept or reject either S or S*’s position or, there is evidence that could make it rational to accept or reject S or S*’s position and this evidence is trivially easy to come by. The first sort of epistemically defective dispute we will call an impossibly hard dispute; the second sort of epistemically defective dispute we will call an absurdly easy dispute. We shall first consider whether there is any reason to think that the Nihilist-Universalist dispute is impossibly hard.

Karen Bennett has recently argued that when it comes to debates over mereological composites, “there is little justification believing either side” has the correct answer.[211] The “sides” she is talking about here are, what we have been calling the Nihilist view and what Bennett calls the view of “believers” in composites.[212] It is important to note here that Bennett lumps together as believers Universalists and so-called commonsense ontologists (who believe in tables but not scattered objects). This point is worth mention at the outset, for we shall see below that a careful consideration of it undermines Bennett’s argument when that argument is leveled against participants in the Nihilism-Universalism dispute.

Another thing that bears mention at the outset is Bennett’s preferred methodology for answering metaphilosophical questions. According to Bennett, “rather than making broad generalizations about the Status of Metaphysics, we need to look at the details of particular disputes.”[213] We shall see below that despite the emphasis Bennett places on this methodological point, she does not sufficiently cleave to her own methodology. My main criticism of her argument will turn on a particular feature of the Nihilism-Universalism dispute that Bennett ignores in virtue of her broadly casting the relevant ontological dispute as one between Nihilists and ‘believers’.

What is Bennett’s argument? Bennett argues for the following view:

            Epistemicism: Disputes about the truth-value of ‘there are Fs’ are not verbal
                                    disputes. But there is little justification for believing either that it
                                     is true or that it is false.

Two points: first, Epistemicism, as formulated above, is a general thesis. We shall only

be concerned with a particularized version of it:

            EpistemicismC:   disputes about the truth value of ‘there are composite material
                                         objects’ are not verbal disputes, however, there is little
                                         justification for believing either that this claim is true or that it
                                         is false.

Second: we shall not concern ourselves with Bennett’s arguments that the composition dispute is not merely verbal. We shall occupy ourselves with the question as to whether the dispute is verbal later.

Following Bennett, let’s call the Nihilist side of the dispute the ‘low ontology’ side and the believer’s side the ‘high ontology side’. Bennett’s argument for EpistemicismC is as follows:

There is a low ontology side (occupied by the Nihilist) and a high ontology side (occupied by the believers) in the material composition dispute.
The high-ontologist insists that her extra ontology is nothing over and above what the low-ontologist already accepts.
The low-ontologist tries to recapture most of the claims that the high-ontologist accepts.
If (3) and (4), then the low-ontologist’s view is not simpler than the high ontologist’s and the alleged problems for the high-ontologist rearise for the low-ontologist.
If the low-ontologist’s view is not simpler than the high ontologist’s and the alleged problems for the high-ontologist rearise for the low-ontologist, then there is no available evidence that could make it more rational to accept the Nihilist’s view instead of the believer’s or vice versa.
If there is no available evidence that could make it more rational to accept the Nihilist’s view than the believers or vice versa, then EpistemicismC is true.
EpistemicismC is true.[214]

The argument looks valid. In what follows, then, we shall assess its premises.

(1) is hard to deny, and no participant in the dispute would deny it. There are some philosophers who might deny (2) and (3), but, as Bennett correctly observes, most participants in the dispute over composition are difference minimizers. That is, proponents of high ontology go to great lengths to argue that their ontology is not bloated, and proponents of low ontology go to great lengths to ensure that their views are consistent with the truth (or at least the correctness or assertability) of everything we pre-theoretically want to say about composites.

The lynchpin premise here is (4). Is there any reason to accept it? Let’s first look at the way Nihilists try to establish that their view is not inconsistent with what commonsense has to say about composites. Nihilists typically employ a paraphrase device in terms of plural (and pluplural) quantification and a host of predicates like ‘arranged chair-wise’. Bennett observes that this paraphrastic move involves the Nihilists’ “postulating a certain amount of machinery” in virtue of which their “view reflects the complexity of the high-ontologist’s view”. In virtue of this symmetry in complexity, Bennett concludes, the Nihilist “cannot automatically claim victory on the simplicity score.”[215]

There is something to Bennett’s argument here. Consider the Nihilist who wants to acknowledge that the ‘man in the street’ says something true, or at least correct or acceptable, when he says, “There’s a table here”. According to the Nihilist, the man in the street’s assertion is correct in light of the truth of the closely related paraphrase, “There are some simples arranged table-wise here.” So, it looks like, for every object the believer believes in, the Nihilist will have to postulate “complex structured plural predicates”, while “[t]he believer need not countenance either these highly structured plural predicates, nor any properties that answer to them.”[216] So, it would seem that if the Nihilist successfully captures everything that commonsense wants to say about composites, then “the high-ontologist multiplies objects while the low-ontologist multiplies properties”; this is to say that the Nihilist “buys her way out of ontology with the coin of ideaology,” so that “even if the low-ontologist wins the battle of ontological commitment, [she] does not win the war of simplicity.”[217]

This is all well and good so long as we take the relevant dispute to be one between the Nihilist and the believer who does not accept scattered objects. For, it is reasonable to think that there is a one-one correlation between the objects posited by the believer and the highly structured plural properties postulated by the Nihilist. But, the game is changed somewhat when we take the relevant dispute to be the one between the Nihilist and the Universalist. How so? In addition to all of the ordinary composites the commonsensical believer holds to exist, the Universalist also posits sums of arbitrarily chosen material objects. So, take the object that the Universalist says consists of my nose, the Eiffel tower, and your head. The Nihilist, it seems to me, not only rejects such objects, but holds that there are no highly structured complex plural properties that correspond to them. This is because the Nihilist is under no obligation to account for the truth (or assertability or whatever) of claims about such things. So, while there may be a one-one mapping between the composite objects beloved of common sense and the Nihilist’s highly structured properties, the same cannot be said for the objects the Universalist believes in and the Nihilist’s structured properties.

This is crucical, for it shows that, so long as the target dispute is the one between Nihilists and Universalists, we needn’t go along with Bennett in thinking that the Nihilists lose the war over simplicity. The Nihilist’s overall view is simpler than that of the Universalist. So, when the relevant ‘believer’ is the Universalist, we may safely reject Bennett’s premise (4).

What we see in Bennett’s argument, then, is a failure to appreciate the particulars of the Nihilism-Universalism dispute. So long as the dispute over mereological composition is described as a dispute between high-ontologists and low-ontologists, or between Nihilists and believers, important details are glossed over. The Universalist view is less simple than the Nihilist’s view once we consider all of the scattered and gerrymandered objects accepted by the Universalist.

I think we should therefore reject Bennett’s argument, at least insofar as it applies to the Nihilism-Universalism Dispute. However, even supposing Bennett’s argument to be sound, is there a serious problem for practitioners of ontology? I do not think there is. Suppose that the composition disputes are epistemically underdetermined. It is a live epistemic possibility that, despite the lack of any available evidence that could make a decision between competing ontological packages rational, nevertheless there is some possible evidence that could do the trick. Practicing ontologists, then, should not give up the search for such evidence. And, even supposing no such evidence is found, it is still rational for ontologists to seek an end to the dispute by “reflection on broader theoretical and methodological questions.”[218]

Even supposing such reflection will not be up to the task of determining which view is eminently more rational than the others, there is still not a serious problem for ontologists. Compare the state of the ontologists’ dispute with that of certain disputes between historians and investigative journalists. Historians and investigative journalists often ask interesting questions about the mental states of certain historical figures. Take, for example: “What were Napoleon’s reasons (if any) for failing to send in the Prussian cavalry at the Battle of Waterloo?” or “What was on or before Nicole Brown Simpson’s mind in the hours before her untimely demise?” These are questions that seem to have answers; they seem to have answers that some people care about; but there seems to be little or no evidence that could make one answer eminently more rational to believe than another. Should historians give up on questions like these? Should investigative journalists stop trying to answer such questions? The answer seems plainly to me to be ‘No.’

There is a broader connection with practical concerns, however. I am moved to cite the wisdom of Gandalf the Grey in this connection. Concerning the Ring of Power, Gandalf says, “We should seek a final end of this menace, even if we do not hope to make one.”[219] Gandalf and the other members of the Council of Elrond found disposing of the One Ring a task that was nearly impossible. They held very little hope of their Quest succeeding. But, it was still rational for them to try to destroy the Ring. Something similar applies in the case of massively epistemically underdetermined questions in philosophy, history, and other areas of human interest. We should seek answers to these questions even if we do not hope to find them. Why? In some cases very much of practical importance hangs on the answers to questions that are arguably epistemically undetermined. Consider questions like, “What is the best course of action to bring peace to the Middle East?” or “What policies will keep Americans safest from the threat of terrorist attacks?” These questions probably do not have answers that are uniquely epistemically determined by the available (or even the possible) evidence. However, we should still seek answers to them. In other cases, however, we are just plain curious, regardless of whether much of importance hangs on finding an answer.

Of course, ontological questions do not have the kind of serious practical import that questions about national security or peace in the Middle East have. These questions, however, have practical significance of a somewhat different kind. It matters to us (or at least some of us) what the right answers are. It is of some significance whether a complete and accurate account of the world must include a relation of part to whole. Granted, having an answer to questions about mereological composition is not something that even most philosophers would regard as paramount, but we still want an answer. Continuing to look for one, even after we have acknowledged the vanishingly small hope of finding one, is still rational.[220]

Further, doing ontology is not totally pointless, for we have a fairly well established metric for deciding whether one ontology is better than another. Choice of a theory in ontology is not completely without constraints. Given the fact (at least, I take it to be a fact) that it is an epistemic possibility that one or more theories in material object ontology be shown to be internally inconsistent, doing ontology should still be a live research option.

Neo-Carnapian naysayers[221] will not find remarks such as these very convincing. There is another reason, however, to think that Bennett’s argument, if sound, does not pose any special threat to ontologists. A problem for everyone, so goes this line of resistance, is a problem for no one. If Bennett’s argument is sound, there is at least some reason to worry that no disputes in philosophy are epistemically decidable. For, what could the relevant differences be between disputes over composition and disputes, for example, between Rationalists and Empiricists, Deontologists and Consequentialists, Quasi-realists and realists (in meta-ethics), scientific realists and anti-realists, contextualists and relativists (about knowledge discourse and predicates of personal taste), and so on. Is it reasonable to think that there is evidence that could settle these disputes but not any evidence that could settle ontological disputes over composition? This is not clear.

Perhaps, then, all or most disputes in philosophy are epistemically defective. This is a conclusion that most philosophers would find unsavory indeed. But, in each of the above disputes, there is good reason to think that the views in question are difference minimizing in the sense that advocates of all the views will have ready explanations to account for any counterintuitive results of their theories, and theorists on all sides of the above disputes will have something to say about how their preferred view enables them to capture our pretheoretic intuitions. Perhaps all of the above disputes are epistemically defective, then. So, the ontologist can plead that she is in good company.

There is good reason to think that compelling evidence is pretty hard to come by in philosophy. As evidence of this I cite the lack of consensus on just about any philosophically interesting claim. We should not despair, however. As Russell observed,

There are many questions—and among them those that are of the profoundest interest to our spiritual life—which, so far as we can see, must remain insoluble to the human intellect unless its powers become of quite a different order from what they are now. Has the universe any unity of plan or purpose, or is it a fortuitous concourse of atoms? Is consciousness a permanent part of the universe, giving hope of indefinite growth in wisdom, or is it a transitory accident on a small planet on which life must ultimately become impossible? Are good and evil of importance to the universe or only to man?[222]

I that Russell is right insofar as answers to questions like the ones he mentions are probably not epistemically decidable. This does not mean, however, that we should stop thinking about them or even stop trying seriously to find answers. The value of doing so, as Russell claims, is not in securing answers, but in “remov[ing] the somewhat arrogant dogmatism of those who have never travelled in the region of liberating doubt”, in “keep[ing] alive a sense of wonder by showing familiar things in an unfamiliar aspect,” and “suggest[ing] many possibilities which enlarge our thoughts and free them from the tyranny of custom.”[223] I do not think it would be wrong to place ontological questions in the list of questions Russell mentions.

I close this section by posing another kind of problem for EpistemicismC. Suppose that ontological disputes are not solvable. There is no evidence that could make belief in Nihilism more rational than belief in Universalism or vice versa. One might worry, if worries like Bennett’s are well-grounded, about the epistemological status of metaontology or metaphilosophy itself. How can questions like “Is philosophical question Q answerable?” or “Is there evidence that could make it rational to accept philosophical view V?” be questions that are themselves epistemically decidable? What kind of evidence counts for one metaontological view of the matter? Are there facts of the matter about ontology? What could count as evidence for or against this metaontological claim? It seems that the epistemic foundations of metaontology are in just as much doubt as those of ontology.

I conclude that Bennett’s argument is problematic. There are good reasons to reject its premise (4), and there are good reasons for thinking that, even if her argument is sound, things do not look that bleak for the ontology of composition. I offer the tentative conclusion, then, that if Bennett’s argument makes trouble for disputes in material object ontology, then most of philosophy is in trouble.

We now turn to the charge that disputes over mereological composition are absurdly easy disputes. Jonathan Schaffer argues that “contemporary existence debates are trivial, in that the entities in question obviously do exist.”[224] Schaffer’s claim about the trivialty of existence claims applies to existence claims in general, even those made outside of physical object ontology.

Schaffer’s argument is fairly simple. There has been ongoing controversy over whether numbers, properties, things with proper parts, and fictional characters exist. Schaffer argues that these matters are trivially easy to settle. Take the following arguments:

1. There are prime numbers.

2. Therefore there are numbers.

3. There are properties that you and I share.

4. Therefore there are properties.

5. My body has proper parts (e.g. my hands).

6. Therefore there are things with proper parts.

7. Arthur Conan Doyle created Sherlock Holmes

8. Therefore Sherlock Holmes exists.

In each case, we move from a claim that is supposed to command some kind of intuitive obviousness to a controversial claim about what exists.

Take the argument for the existence of numbers. According to Schaffer, “(1) is a mathematical truism. It commands Moorean certainty, as being more credible than any philosopher’s argument to the contrary”; try to deny (1) and you have “ipso facto produced a reductio” of your premises.[225] But (2) follows from (1) by means of a “standard adjective-drop inference”.[226] “Prime number’, unlike ‘forged bank note’ or ‘fake passport’ permits adjective drop inferences. For example, “Bill is thinking of a prime number” entails “Bill is thinking of a number.”

According to Schaffer, (3) “is an everyday truisim” and (4) clearly follows; “(5) is a biological banality, and 6 follows”; “(7) is a literary fact, and (8) follows.”[227] If these inferences go through, then “the contemporary existence debates are trivial.”[228] Of particular interest to us is the inference from (5) to (6), for if this goes through then there is an obvious, trivial answer in the Nihilism-Universalism dispute.

What is a serious ontologist to do? Schaffer identifies three strategies for blocking each of the above inferences: the paraphrase strategy, the equivocation strategy, and the ontological neutrality strategy. We shall examine each in turn along with Schaffer’s objections to them.

Each of the above inferences proceeds from a truism to a existence claim. An obvious reply on behalf of the serious ontologist is that of paraphrasing away the existential commitment of the truism. Take the mathematical case. An ontologist might say that according to the fiction about numbers, there are prime numbers, and that is all one is committed to in accepting (1). But, Schaffer objects, “presumably this is a way of saying that (1) is false, and only some suitable paraphrase is true. But (1) is obviously true, as stated. Whatever philosophical concerns might motivate this paraphrasing fictionalist have met their reductio.”[229]

Similar remarks should apply to the other cases. So, take a Nihilist who paraphrases away claims about proper parts. Such a theorist denies (5). But (5) is a biological banality and thereby commands more credence than any philosophical attempt at paraphrase.

What of the equivocation strategy? One might claim that ‘there are’ or ‘exists’ shift in meaning from the truism to the existential claim in each of the above arguments. There are a number of different ways to make out this kind of shift. If one accepts a difference in correctness conditions between ordinary and existence assertions (a la Chalmers), then one can resist the move this way, saying that ‘there are’ picks out a contextually determined content in the conclusion that differs from that picked out in the premise.

Schaffer is not impressed by this move. He says, “[t]here is no linguistic evidence of any ambiguity in our idioms of existential quantification. Indeed, if there were such meaning shifts then no adjective-drop inference would be valid”, but since such inferences are valid, there are no such meaning shifts.[230]

A last ditch effort would involve claiming that “all quantification is ontologically neutral” so that one can accept each of the above truisms without accepting the existence claim. According to Schaffer this fails because “the neutralist seems committed to the following unfathomable conjunction: “Numbers do not exist, and there are numbers.”[231]

So, we have a series of fairly compelling-looking arguments for controversial existence claims. Three obvious lines of objection seem to come to nothing. What is the ontologist to do? One strategy Schaffer does not consider is that of accepting the inferences. That is, suppose the serious ontologist accepts that the above inferences are trivial when both premises and conclusion are ordinary existence assertions. These arguments, however, do not go through if the conclusions are taken to be ontological existence assertions.

It is important to note that this line of resistance is not merely a version of the equivocation strategy. For it is compatible with ontological and existence assertions having different correctness conditions that they mean the same. So, there need be no shift in meaning postulated between premise and conclusion in the above arguments. Rather, when the conclusion is an ontological assertion, it is evaluated for correctness relative to ontological standards, while when an assertion is an ordinary assertion, it is evaluated relative to ordinary standards.

I have introduced some reasons above for rejecting the view that ontological and existence assertions vary in standards for correctness in favor of the view that it is knowledge claims that have contextually variable standards. .

But I do not think serious ontologists need even this much machinery to reply to Schaffer. Let’s consider an example. Imagine a fictional philosopher named Jonathan Schmaffer. Schmaffer says, “Epistemological skepticism is trivially false, here’s my proof: I know I have hands, so there are some things I know. End of story.” Is this really an answer to skepticism? Doesn’t it just beg the question? The problem with the above arguments is not that their premises do not follow from the conclusions, but that they are question-begging. The ontologist who wants to answer existence questions wants arguments for existence claims that do not beg the question. Inspired by Schaffer’s strategy for showing that existence questions are trivial, one might try to show that many more philosophical issues are trivial. Take these inference patterns:

9. It’s a moral fact that torturing babies for fun is wrong.

10. Therefore there are moral facts.

11. I believe I have a hand.

12. Therefore there are folk psychological states (eliminative materialism is false).

13. There is a possibility that it will rain.

14. There are (mere) possibilia.

None of these arguments should convince us of the truth of their conclusions, however, for they beg the question. But in each case, they move from an everyday truism to a philosophical conclusion.

Why is it that serious ontologists tend to balk at the move from Schaffer’s truisms to the existential claims he draws from them? Schaffer’s diagnosis of this failing attributes to ontologists the mistake of asking the wrong question. The question of ontology shouldn’t be just what exists, but rather, it should be given what exists, what is fundamental?[232] A permissive ontology, Schaffer thinks, is not objectionable if the entities it postulates are grounded in basic, more fundamental entities: “there is no longer any harm in positing an abundant roster of existents, provided it is grounded on a sparse basis.”[233] The permissive ontologist can help herself to the so-called Doctrine of the Ontological Free Lunch familiar from Armstrong:

[w]hatever supervenes, or, as we can also say, is entailed or necessitated,…is not something ontologically additional to the subvenient, or necessessitating, entity or entities. What supervenes is no addition to being.[234]

I have never understood the ‘no addition to being’ claim in the doctrine of the ontological free lunch. It seems like a theorist who accepts entities of kind A offers an addition of being over the theorist who does not accept these entities. Perhaps this is a failing on my part, however.

Schaffer’s argument for a permissive ontology only looks plausible when we ignore first-order ontological considerations that serve to motivate eliminitivism about material objects. The Nihilist, after all, can plausibly block the argument from vagueness against material objects, has a ready answer to the problems of coinciding objects that does not carry a commitment to collocated objects or temporal parts, and can accept causal overdetermination arguments.[235] Nihilists can also, without commitment to any strange objects or strange kinds of objects, adhere to theoretical principles like the No Arbitrary Distinctions principle and various parsimony principles.[236] In light of these considerations, it is hard to see how the commitment to composite material objects that the Nihilist avoids is somehow an “ontological free lunch”; there seems very little about it that is free.

Of course Schaffer might suggest that these costs are not real costs. This sort of reply suggests that what it is we find counterintuitive is not the existence of vague objects per se, but rather the existence of vague fundamental objects, not the existence of co-located objects, but the existence of co-located fundamental objects. These claims are empirically verifiable claims about our dispositions to assent to various ontological principles, and as such they require some empirical support.[237] When approaching such problems for composite objects, we certainly might have thought that we were finding vague objects, in general, counterintuitive. In order to substantiate the hypothesis that it was instead vague fundamental objects we found counterintuitive, some kind of psychological evidence would have to be advanced. As far as I know, no such evidence has been provided by theorists who take ontological disputes to be trivial.

evidence has been provided by theorists who take ontological disputes to be trivial.

If Schaffer is right about existence questions, then a lot of philosophical questions will turn out to have trivial answers. Perhaps this is no bad thing. However, I see no reason to think Schaffer is right about existence questions. The arguments he gives to show that existence questions are trivial are themselves question-begging. There is a further sort of problem with Schaffer’s arguments, however. It is not clear that the existence language in the truisms supports an inference to an existential conclusion. Compare Schaffer’s inferences with the following:

1*. There are job vacancies.

2*. Therefore there are vacancies.

3*. There are deficiencies that you and I share.

4* Therefore there are deficiencies.

5*. My body has a lack of ketones.

6*. There are lacks.

7*. Obama created a stir (kerfuffle, commotion, fuss) with his healthcare policy.

8* Therefore a stir (kerfuffle, commotion, fuss) exists.

9* Democratic legislation created an absence of jobs.

10* An absence exists.

In each of the starred arguments, we have an ordinary truism and an existential conclusion that follows. But, no one would seriously bloat their ontology by including such objects as vacancies, deficiencies, lacks, absences, stirs, kerfuffles, commotions, or fusses. These phenomena involve existents, but they are not themselves existents. A stir is a certain pattern of behavior, vacancies, lacks, and deficiencies are not themselves objects, but instead involve ways that objects are.

Perhaps Schaffer will object, allowing into his ontology things like vacancies, lacks, stirs, and so on, only claiming that these things are not fundamental. We needn’t go along with him here.

I want to consider one last direction from which a charge of epistemic defectiveness may be lodged. Neo-Fregean arguments for the existence of mathematical objects, if generalizable, threaten to trivialize ontology. The idea is something, roughly, like this. Consider Hume’s principle:

(H) The number of the Fs = the number of the Gs iff the Fs are equinumerous with the
Gs.

The sense of ‘equinumerous’ is given by the following:

F and G are equinumerous (or, F and G are in one-to-one correspondence) just in case there is a relation R such that: (1) every object falling under F is R-related to a unique object falling under G, and (2) every object falling under G is such that there is a unique object falling under F which is R-related to it.[238]

According to the neo-Fregean line, Hume’s principle “can be shown to endow all sentences in which number words occur with meaning in such a way that some such sentences which require for their truth that numbers exist…are true.”[239] This, taken together with the Context Principle:

CP: words have their meanings in virtue of the meaning of sentences in which
they occur as constituents.[240]

Given CP and the claim that a number of sentences that require for their truth that numbers exist, the neo-Fregean argues that

[a]ny further question about whether there really are numbers asks for the meaning or reference of number words taken in isolation, flouting the context principle. In this way, the context principle is regarded as having significant implications both regarding what exists (there exist abstract objects) and regarding the nature of the relevant existence claims (they can have the status of something like conceptual truths).[241]

So, the ontologist might want to ask seriously “Are there really numbers?” The neo-Fregean replies, “The answer to this question is trivial: given CP and some mathematical sentences that we regard as trivially true (e.g., ‘Five is odd’), it follows that number terms refer. So numbers exist. QED.”

The argument is (something like) this:

1. ‘Five is odd’ is true.

2. For any atomic sentence of the form ‘F(t)’, of any language, that sentence is true only
if the object refered to by ‘t’ is in the extension of the predicate ‘F’.

3. ‘Five is odd’ is true only if the object referred to by ‘five’ is in the extension of the
predicate ‘is odd’.

4. The object referred to by ‘five’ is in the extension of the predicate ‘is odd’.

5. If 4, then ‘five’ refers.

6. If ‘five’ refers, then numbers exist.

7. Numbers exist.[242]

If the above argument is sound, then it looks like ontological questions turn out to be way easier to answer than ontologists might have thought. How so? Let us frame the argument in terms of our Nihilist-Universalist dispute. Consider a region where both theorists will agree that there are some simples arranged car-wise. Suppose that the Universalist introduces the name ‘Herbie’ “to refer to the car, if any, in this location.”[243] Now the Nihilist and the Universalist get into a dispute about whether ‘Herbie’ refers.

Here is where things go badly for the Nihilist, according to the neo-Fregean.[244] This is because

[f]or her, the question of whether Herbie exists is a question of whether some sentences in which ‘herbie’ occurs in the right way are true; and that in turn is a question of whether; by ‘orindary criteria’ the truth-conditions of these sentences are satisfied. But it certainly seems that there can be a successful practice of assertively uttering such sentences. The thought would be this. Any inclination to think that the practice would have to be unsuccessful would seem to have to rely on the belief that there simply is nothing for the term ‘Herbie’ to refer to . But this consideration, putting reference before truth, so to speak, the neo-Fregean must rule inadmissible.[245]

So, if something like what the neo-Fregean says is correct, then there will be, given the Universalist’s introduction of the name ‘Herbie’, some true sentence of the form ‘Herbie is F’, e.g., ‘Herbie is white’, ‘Herbie is a Ford’. But, if this is the case, then we can run the following argument:

N1. ‘Herbie is a Ford’ is true.

N2. For any atomic sentence of the form ‘F(t)’, of any language, that sentence is true
only if the object refered to by ‘t’ is in the extension of the predicate ‘F’.

N3. ‘Herbie is a Ford’ is true only if the object referred to by ‘Herbie’ is in the extension
of the predicate ‘is a Ford’.

N4. The object referred to by ‘Herbie’ is in the extension of the predicate ‘is a Ford’.

N5. If 4, then ‘Herbie’ refers.

N6. If ‘Herbie’ refers, then there are cars.

N7. There are (contra nihilism) cars.

So, the Universalism-Nihilism dispute turned out to be absurdly easy; all we had to do was acknowledge some simple facts about singular terms and reference, and the ontological claims get sorted out as a matter of a simple argument.

Notice, on this proposal, no complicated reasoning about vagueness, theoretical simplicity, problem cases, or anything like that is required to settle ontological issues. There are a number of ways to respond to this style of argument.

First, the neo-Fregean’s arguments only go through if the logical form of the sentence mentioned in premise (1) and (N1) are subject-predicate sentences. Compare: “Nobody is immortal.” This seems true. However, its truth does not require that the object referred to by ‘Nobody’ be included in the extension of the predicate ‘is immortal’. This is because the sentence “Nobody is immortal”, while it may look like a subject-predicate sentence, is really a disguised quantified sentence: “There is no person such that that person is immortal.”

So, one way to resist the above arguments is to claim that the sentences mentioned in premise 1 and N1 have some hidden, complex structure. This is what some Nihilists maintain when they offer paraphrases of ordinary assertions about material composites. According to such Nihilists ‘Herbie is a Ford’ is true only if it has a deep structure that is (something like) ‘Some simples here are arranged Ford-wise’. On this strategy, the Nihilist could accept (N1) and (N2), but reject the instantiation of (N2), (N3), with the sentence ‘Herbie is a Ford’ on the grounds that this sentence is not of the logical form ‘F(t)’.

Presumably, this approach will be rejected by Eklund’s neo-Fregean, for it seems to put questions of reference before questions of truth. The only reason, apparently, for the Nihilist to insist on the paraphrase is that she thinks there is nothing in the vicinity a singular term like ‘Herbie’ could refer to, or at least, nothing mereologically complex that ‘Herbie’ could refer to.

So, to block the argument, it looks like the Nihilist will have to do one of the following: (i) reject premise (N1), (ii) block the instantiation of (N2) that occurs in premise (N3), (iii) reject (N6), or (iv) reject (N2). (N2) is fairly plausible, and it seems like the Nihilist should accept it, at least when ‘t’ takes as a value a term purporting to refer to a simple. Rejecting either (N1) or blocking the instantiation of (N2) seems to require that the Nihilist reject the Context Principle, thereby rendering her position incompatible with neo-Fregeanism. How might the Nihilist reject (N6)? One way to do this would be to adopt a suitable fictionalism about material objects.[246]

To see how this would work, let’s take a parallel argument.

S1. ‘Zeus is bearded’ is true.

S2. For any atomic sentence of the form ‘F(t)’, of any language, that sentence is true
only if the object refered to by ‘t’ is in the extension of the predicate ‘F’.

S3. ‘Zeus is bearded’ is true only if the object referred to be ‘t’ is in the extension
of the predicate ‘is bearded’.

S4. The object referred to by ‘Zeus’ is in the extension of the predicate ‘is
bearded’.

S5. If (S4), then ‘Zeus’ refers.

S6. If ‘Zeus’ refers, then a god exists.

S7. A god exists.

It is fairly easy to see where this argument goes wrong. S6 is clearly false. If ‘Zeus’ refers at all, then it refers to something that is a denizen of myth, a mythical creature.[247] The problem with the Herbie argument, the Nihilist can argue, is that (N6) is false. The truth of ‘Herbie is a Ford’ only establishes that some mythical object (the thing that would answer to ‘Herbie’ on the assumption that Nihilism is false and there are composite objects) exists. But, just as the mythical character Zeus is not a god (but ‘Zeus is a god’ is true according to the myths about Zeus), so too Herbie is not a car (but ‘Herbie is a car’ is true according to the myth about material composites).

Again, the neo-Fregean might complain that the only reason our Nihilist thinks that Herbie is a mythical character is because she believes that there is nothing for ‘Herbie’ to refer to. This is not quite right. What are our reasons for thinking that Zeus is a mythical figure (instead of a fictional character or a real person)? We think Zeus is a mythical creature because the name ‘Zeus’ was introduced as part of a false theory about the world: namely the theory that a moral agent was responsible for (phenomena like) thunder and lightning. Why not think instead that ‘Zeus’ referred to some natural phenomena or that ‘Zeus’ (like ‘Santa Claus’) referred to some fictional character dreamed up by the Greeks? The answer here is that the Greeks did not merely intend to create a story about the world; they meant to be giving an explanation. They did not intend the name ‘Zeus’ to refer to whatever it may be that is the cause of thunder and lightning; they intended the name to refer to the person (or sufficiently person-like creature) responsible. So, their use of the name refers to a mythical character. What’s the difference? Roughly this: mythical creatures are those that are referred to by folks who introduce a name in connection with some story that they believe to be true, while fictional characters are those things referred to by folks who introduce names in connection with some story that they do not believe to be true. This distinction can be found in Salmon:

Whatever good reason there is for acknowledging the real existence of…Holmes extends to Vulcan…Myths and fictions are both made up. The principle difference between mythical and fictional objects is that the myth is believed while the fiction is only make-believe. This difference does nothing to obliterate the reality of either fictional or mythical objects.[248]

The Nihilist does think that there are plenty of (nonmythical) things that ‘Herbie’ could have referred to: simples. ‘Herbie’ does not refer to any of these, however. The attempt to christen ‘Herbie’ as a name for a real entity presupposed a false theory: namely the theory that some objects are mereologically composite. According to this false theory, the world contains mereological composites; introducers of the name ‘Herbie’ nevertheless believed the theory to be true. Compare with the case of the name ‘Vulcan’; this name was introduced by believers in a false theory. The neo-Fregean needn’t hold that it is without reference, however, rather it refers to a mythical object. Generalizing, sortal kind terms like ‘table’, ‘mountain’, and ‘chair’ refer to mythical substances in the same way that ‘phlogiston’ refers to a mythical substance.

So, even if a Nihilist feels attracted to neo-Fregeanism, she can allow that singular terms that purport to refer to composites can genuinely refer without giving up her Nihilism. She can hold that these terms refer to mythical objects.

Perhaps this will seem too incredible. Rather than accept an ontology of concrete physical objects, the Nihilist is accepting a bloated ontology of mythical creatures. How is this better than just accepting that there are tables? According to this picture, having realized that terms which purport to pick out composite objects only refer to mythical objects would place us at an epistemic advantage similar to the one the Greeks would have been in had they come to realize that ‘Zeus’ referred to a mythical creature, and not to a person (or person-like creature) responsible for various natural phenomena.[249]

It turns out that a Nihilist can safely be a neo-Fregean. True, she has to give up a simpleminded characterization of her view as the view that there are no composite objects and instead redescribe her view as the view that composite material objects are mythical objects. But, the Nihilist who takes this line is no worse off than the atheist who admits that God, while he most certainly exists, is for all that a fictional character (or mythical character). To describe his view this way is not for the atheist to give up his atheism.[250]

The above mythicist-Nihilist view may not be very plausible. I do not intend to defend this view against possible objections, rather I merely put it forward as a consistent way for the Nihilist to resist the neo-Fregean arguments against her view. I want to consider one last argument in this connection. Recall that the neo-Fregean arguments only threaten to show that ontological disputes are epistemologically defective if they can show that ontological questions are absurdly easy to answer. Here I want to resist this claim. Perhaps it is the case that given the neo-Fregean’s view, certain ontological questions become absurdly easy (because they are on the nature of conceptual truths), however, this does not suffice to show that the ontological questions are absurdly easy simpliciter. Neo-Fregeanism itself requires motivation. The truth of neo-Fregean principles like CP is not obvious. Perhaps there are good arguments for such principles, but it hardly seems like settling ontological debates by motivating and then applying neo-Fregean semantics is something that is particularly easy, let alone absurdly easy. But, in order for the complaint that ontological questions are epistemically defective to have bite, it would have to be shown that these questions are absurdly easy, which, I maintain, has not been shown.

Where does this leave us? Barring any reason to think that the Nihilism-Universalism dispute is an absurdly easy dispute, it looks like Nihilists and Universalists can defend themselves from the charge that their dispute is epistemically defective. Further, even if the Nihilism-Universalism dispute turns out to be epistemically defective (a) that is no reason to stop pursuing the dispute seriously, (b) there is no reason to think ontologists are alone amongst philosophers in this regard, and (c) there is reason to think that many empirical researchers are in the same epistemic boat. So, I conclude that there is no reason to think that the Nihilism-Universalism dispute is epistemically defective in our sense, namely there is no reason to think it is a dispute that rational people ought not to be having.

Chapter 10

MERELY VERBAL

We come now to the question of whether there is any meaningful sense of ‘merely verbal’ according to which it is appropriate to say that the dispute between Nihilists and Universalists is merely verbal. In Part One we examined a number of different analyses of the expression ‘merely verbal dispute’ and found them all lacking. Even if proponents of these various analyses have not succeeded in giving a correct account of merely verbal disputes, it is still instructive to see whether their indictments of the ontological disputes come off successfully. Eli Hirsch and Alan Sidelle both argue persuasively that the Nihilist-Universalist dispute is merely verbal. In this chapter I consider and reject both of their arguments.

In this and the following sections I shall consider some problems with Hirsch’s argument. In this section, I shall attempt to show that his argument does not establish the conclusion that the Nihilists and Universalists are having a merely verbal dispute even in the Possible Communities sense.

Recall, that on the Possible Communities Model of merely verbal disputes that Hirsch provides, a dispute is merely verbal if, and only if, each disputant, given the correct rules for linguistic interpretation, ought (in the wide sense of ‘ought’) to agree that the other party speaks the truth in the imagined public language associated with her position.

Let N-lish be the imagined public language associated with the Nihilist’s position and U-lish be the one associated with the Universalist’s position. Hirsch’s master argument, is as follows:

1. Interpretations of N-lish and U-lish on which the disputed sentences (e.g. ‘there are merological sums” and “there are no composite objects”) came out false would violate charity to perception.

2. Interpretations of N-lish and U-lish on which disputed sentences came out false would violate charity to understanding.

3. Interpretations of N-lish and U-lish on which disputed sentences came out true would not violate charity to retraction.

4. No other principles of linguistic interpretation favor interpretations of N-lish and U-lish on which the disputed sentences are false.

5. If 1-4, then the most plausible interpretations of N-lish and U-lish respectively are ones on which each language’s disputed sentences come out true in that language.

6. The most plausible interpretations of N-lish and U-lish are ones on which each language’s disputed sentences come out true respectively in that language.

7. If (6), then the dispute between Nihilists and the Universalists is merely verbal (in the Possible Communities sense)

8. The dispute between Nihilists and Universalists is merely verbal (in the Possible Communities sense).

The key steps in this argument occur in premises (1)-(4). Accordingly, we shall examine these premises to see if there are good reasons for rejecting them.

Premise (1)

Suppose that a speaker of N-lish and a speaker of U-lish are both viewing the same scene before them. The scene contains a book, a pencil, and a cup. Hirsch argues that it would be uncharitable to interpret speakers of N-lish as uttering falsehoods when they affirm sentences like:

(N) There is nothing composed of the book, pencil, and the cup[251],

and that it would be uncharitable to interpret speakers of U-lish as uttering falsehoods when they affirm sentences like:

(U) There is an object whose parts are the book, the pencil, and the cup.

To interpret N-lish and U-lish in such a way would be uncharitable because to do so would involve attributing widespread perceptual errors to members of the N-community and the U-community. According to Charity to Perception, this is to be avoided if possible.

Does Charity to Perception require interpreting (N) such that it is true in N-lish and (U) such that it is true in U-lish? Perhaps not. Charity to Perception creates a presumption in favor of interpretations of a language that do not attribute widespread perceptual errors to speakers of that language. This means that Charity to Perception favors those interpretations on which one does not have to assume that speakers make mistakes about highly visible properties of objects perceived in good perceptual conditions. If one thinks that an interpretation violates Charity to Perception, then one must be in a position to identify which perceptible properties of which objects the interpretation suggests speakers of the target language are mistaken about.

Suppose that speakers of U-lish were to interpret N-lish disquotationally. They would then affirm (in U-lish) the following:

                  (UN)   “There is nothing composed of the book, the pencil, and the cup” is
                                true in N-lish just in case there nothing composed of the book, the
                               pencil, and the cup.

Given the truth conditions laid out in (UN) and the circumstances imagined above, U-lish speakers would regard the N-community’s utterance of N in those circumstances as false.

What perceptual mistake would speakers of U-lish be attributing to their N-lish speaking counterparts? What perceptual feature of the environment are N-lish speakers getting wrong on this interpretation? One candidate would be the number of physical objects present in the observed scene. But, according to both Nihilists and Universalists, this is not a visible feature of the scene.

According to theorists of both persuasions, perception alone does not reveal whether one object is numerically identical to another object or whether some xs compose a further thing, y. To determine how many objects are present in the above scenario requires some philosophical argumentation. If perception alone revealed the truth about how many material objects were there, then the Nihilist’s and the Universalist’s arguments would be unnecessary; we could just go and look.

So, premise (1) of Hirsch’s argument looks dubious. It is not clear that a disquotational interpretation of U-lish or N-lish would attribute perceptual mistakes to its speakers.

Premise (2)

The above considerations shift the weight of Hirsch’s argument to the claim that charity to understanding favors conciliatory interpretations of N-lish and U-lish. Even if disquotational interpretations do not violate charity to perception, Hirsch’s argument will still go through if they violate charity to understanding.

Hirsch presents a view of ontology on which

ontological disputes concern matters of a priori necessity. Perdurantists assert, and endurantists deny, that a tree (rock, table) is made up of a succession of temporal parts. Both sides are to be understood as defending their claims on grounds of a priori necessity (Hirsch 2009, 233).

According to Hirsch, on Universalism, (U) follows a priori from

(N¢) There are simples arranged book-wise, pencil-wise, and cup-wise here.

while, on Nihilism, (N) follows a priori from (N¢)

Since Nihilists and Universalists can agree about the truth of (N¢), speakers of N-lish and U-lish should each interpret the others as saying something true when uttering (U) and (N) respectively. Otherwise, each would be interpreting the other as making an a priori mistake.

On Hirsch’s view of the epistemology of ontology, it follows that an interpretation of N-lish on which (N¢) is regarded as true but (N) is false would be one that attributes an a priori mistake to N-lish speakers. Is this the correct view of the epistemology of ontology? There is some controversy over this matter. John Hawthorne writes,

…metaphysicians proposing this or that ontology often do not fit the profile that Hirsch presumes. Even if they regard their favored ontology as necessary, they often do not presume any special a priori access to its truth, being content rather to defend it on the grounds of broad theoretical virtues like simplicity, reasonable conformity with common sense, elegance, and so on. They thus regard such theses as that the physical facts fix the phenomenal facts, that there is some elite stock of fundamental properties, and that classical mereology is correct as quasi-empirical theses whose tenuous connection to experience is not different in kind to that of various bits of high-level physical theory (Hawthorne 2009, 217).

On Hawthorne’s view of the epistemology of ontology, disquotational interpretations of the disputed ontological sentences do not violate charity to understanding. If Hawthorne’s view is correct, then Hirsch’s premise (2) is false.

Hirsch disagrees with Hawthorne’s view on the epistemology of ontology. Hirsch writes

John Hawthorne…states that many recent ontologists—he mentions Lewis, in particular—do not appeal primarily to a priori arguments. He and I must be using “a priori” in different senses. In Lewis (1986) the main (or only) argument for perdurantism derives from an analysis of intrinsic change. His main (or only) argument for unrestricted mereology derives from an analysis of vagueness. These arguments are paradigmatically a priori in the sense of appealing not to sensory experience or empirical facts, but to reasoning, understanding, and intuitive insight (Hirsch 2009, 233 fn 3).

Hirsch goes on to cite several prominent ontologists and to claim that their most prominent arguments are also of an a priori nature.[252]

Hirsch intimates that his disagreement with Hawthorne over the epistemology of ontology may be merely verbal. I do not wish to try to adjudicate this dispute here. I shall therefore grant Hirsch his claim that the relevant ontological claims are a priori. I want to argue instead that this is irrelevant to an application of the Charity to Understanding principle.

As Hirsch characterizes the principle of Charity to Understanding, it is

the strong presumption that typical speakers of a language have a sufficiently adequate grasp of their linguistic and conceptual resources so that they don’t generally make a priori (conceptually) false assertions, especially when these assertions seem to be relatively simple, not ostensibly involving any complicated calculations or computations (Hirsch 2005, 72).

As an example of Charity to Understanding at work, Hirsch cites his dispute with a friend, A, over whether a glass is a kind of cup. Hirsch says that a glass is not a kind of cup, while A denies this. Suppose there were a community of people who spoke just like Hirsch’s friend A. According to Hirsch,

If “cup” were interpreted in the A-language to mean what it means in our language, the A-speakers would inexplicably make such a priori false assertions as “A glass is a kind of cup”. Worst of all, they would make a priori false perceptual assertions, such as, “There’s one cup here, and it’s a glass.” (2005, 72).

I think Hirsch is right about this. If we interpreted A-lish speakers as meaning what we do by ‘cup’, then it would look like they misunderstand their own concept of cuphood. For our concept is one that, according to Hirsch, excludes glasses. Is the matter similar with respect to the ontological disputes? If U-lish speakers interpreted N-lish speakers as meaning what they do by their utterance of (N) would they be attributing a conceptual confusion to speakers of N-lish who take this sentence to follow from (N¢)?

This is not so clear. For it is not apparent that the move from (N¢) to (N) is one that does not “ostensibly invol[ve] any complicated calculations or computations.” For the Nihilists think that the argument from (N¢) to (N) is fairly complicated. Seeing that (N) follows from (N¢) is not merely a matter of introspecting one’s concept of material composition. Compare the case of (N¢) and (N) with Hirsch’s example of cups and glasses. The A-community’s concept of cuphood is a concept that includes glasses. For them the inference from ‘Here’s a glass’ to ‘Here’s a glass, and it’s a cup’ is fairly simple. Indeed to think they go wrong in making this inference is to attribute to them an inexplicable a priori mistake. However, to think that the N-lish speakers go wrong in their inference from (N¢) to (N) is not to think they make an inexplicable a priori mistake. The Universalists and the Nihilists think that complicated considerations involving vagueness, simplicity, theoretical elegance, ability to solve complicated metaphysical puzzles, and other sorts of theoretical concerns are involved in the arguments from (N¢) to (N). If U-lish speakers attribute a mistake in this complicated reasoning, they are not attributing a mistake like the one that we would attribute to A-lish speakers if we took them to mean by ‘cup’ what we do. Similar considerations apply to the N-community’s interpretation of U-lish.

The purpose of the charity to understanding principle is to prohibit interpretations of a language that attribute obvious and widespread a priori mistakes to its speakers. While it seems true that homophonic translations from N-lish into U-lish would attribute widespread mistakes to members of the P-community, given the complexity of the arguments for perdurantism, it should be clear that such interpretations do not attribute obvious mistakes.

Premise (3)

In considering the state of play with respect to the perdurantism-endurantism dispute, Hirsh claims that charity to retraction does not deliver the verdict that the dispute is substantive. This is because neither side seems disposed to retract its favored ontological claims. Hirsch insists that neither side is disposed to retract on the grounds that debate over this issue in ontology has reached the critical ‘all is said and done stage’. At this stage in the development of a philosophical disagreement, all the positions have been staked out, all of the clever arguments have been heard, and all serious parties to the dispute are apprised of the evidence relevant to making a decision on the matter. About the state of affairs in the perdurantism-endurantism debate, Hirsch says,

[p]rior to this stage, if an endurantist, say, is disposed to change her mind in response to some perdurantist arguments, then charity to use may favor interpreting her language as P-[lish], so that the change of mind is deemed reasonable and her earlier judgment deemed mistaken. But after the “all is said and done” stage has been reached, there is nothing to be said but that each side speaks the truth in their own language (2009, 241).

Hirsch believes the state of the dispute over Nihilism and Universalism has similarly reached the ‘all is said and done’ stage. I am reluctant to assert, as Hirsch does, that this dispute has reached the ‘all is said and done’ stage. I think it is epistemically possible that some philosopher gives an argument that convinces everybody, or most everybody, of the truth of Nihilism, say. Quine convinced many to give up on the analytic-synthetic distinction; Kripke convinced many that there are necessary a posteriori truths; Gettier convinced just about everybody that justified true belief is not knowledge. Prior to each of these moves, some piece of philosophical ground may have seemed solid and unshakeable. What reason do we have to think that there is not a Gettier-like argument against Universalism (or Nihilism)?

I do not think charity to retraction should be taken off the table at this stage in the debate. I think all parties to the ontological disputes Hirsch mentions are disposed to give up their view in light of a knock-down argument against their position.[253] In light of this shared disposition, I think it is premature to conclude that their disagreements are merely verbal.

Indeed, since Hirsch published his 2009 indictment of ontology, a staunch supporter of Universalism, Ted Sider, has come to feel the theoretical pull in favor of Nihilism. In unpublished work, Sider advances arguments on behalf of Nihilism, concluding it to be the all-things-considered better theoretical option.[254] This shift in position shows that, if charity to retraction ought to be respected, that Sider’s dispute with Nihilists and Quasi-Nihilists (like Van Inwagen) was probably not verbal after all.

I think that charity to retraction poses a special, dialectical problem for Hirsch’s view which I shall sketch in the remainder of this section. Suppose that I am right and that it is premature to claim that the Nihilism-Universalism debate has reached the ‘all is said and done’ stage. If this is the case, then so long as theorists are disposed to retract their view in light of some argument, it seems like charity to retraction delivers the result that their dispute with rival theorists was substantive all along.

Take Hirsch’s own argument. If Hirsch’s argument is successful at persuading Nihilists (Universalists), then it seems that they will give up on Nihilism (Universalism). But, if this is the case, then charity to retraction kicks in and renders their dispute substantive. So, Hirsch’s view has a dialectical problem: it is true only if the ontologists he’s writing about refrain from accepting it. Surely this is an unstable position.

There is a wrinkle, of course. According to Hirsch, charity to retraction only applies if a party to the dispute is willing, in light of some evidence or argument, to retract her disputed claims in the following sense. The disputant must think that her disputed claims were false, and, in retracting them, she must mean to assert their denials. It is not so clear that the Nihilist (Universalist) who accepts Hirsch’s view is willing to do this. For, the Nihilist who becomes convinced that Hirsch is right will be willing to say, “While I was expressing true propositions before, I am now convinced that I was doing so in the wrong language.” This statement does not amount to a retraction of her Nihilism in the above sense of ‘retraction’ which Hirsch claims is operative in the charity principle.

Suppose, however, that the Nihilist (Universalist) in question also accepts the influential view of Tyler Burge according to which the language she speaks is the one spoken in her linguistic community. Hirsch argues that the language spoken in ordinary English-speaking communities is one in which the claims of so-called ‘commonsense ontology’ come out true. Call such a language C-lish. If our Nihilist (Universalist) accepts Burge’s view and becomes convinced that Hirsch is right, then it seems appropriate for her to say

(H) Prior to appreciating Hirsch’s arguments, I was
uttering false sentences (in C-lish).

If she comes to accept (H), then she is retracting her earlier assertions in the sense of ‘retraction’ relevant to the charity principle. So, if Hirsch’s remarks on charity to retraction are correct, then, so long as our imagined theorists all accept Burge’s view, their disputes with rival theorists are not merely verbal. This is a very untenable position for Hirsch: his view that the disputes are verbal is only acceptable if one rejects Burge’s view. But this was a result that Hirsch is at great pains to avoid.[255]

Premise (4)

Theodore Sider has forcefully argued that Hirsch’s target ontological disputes are not merely verbal in the Possible Communities sense[256]. Sider rejects premise (4) of Hirsch’s argument on the grounds that there is another principle of linguistic interpretation, charity to intrinsic eligibility, that renders the ontological disputes substantive.

Take the example of a disagreement between PVI (for Peter Van Inwagen) and DKL (for David Lewis) over whether there are tables. DKL says

(DKL1) Tables exist,

while PVI says

(PVI1) It is not the case that tables exist.

Hirsch argues that this dispute is merely verbal. This means that in the DKL-community (DKL1) expresses a true proposition, and in the PVI-community, (PVI1) expresses a true proposition. The only way this could be the case is if the different communities mean something different by ‘tables’, ‘it is not the case that’, or ‘exist’.

Sider maintains that the only reasonable choice for the ontological deflationist here is to claim that the dispute is verbal on the grounds that ‘exist’ and related expressions mean different things in the different languages. Sider argues, and Hirsch agrees, that Hirsch’s position requires the truth of Quantifier Variance.[257] In the DKL-PVI dispute, the quantifiers are to blame for the mere verbalness. This is because there are many candidate meanings for expressions that play the inferential role of unrestricted quantifiers. Sider formulates Quantifier variance as follows.

Quantifier variance: There is a class, C, containing many inferentially adequate candidate meanings, including two that we may call existence_PVIand existence_DKL. PVI’s claims are true when ‘exists’ means existence_PVI and DKL’s claims are true when ‘exists’ means existence_DKL…Further, no member of C carves the world at the joints better than the rest, and no other candidate meaning carves the world at the joints as well as any member of C—either because there is no such notion of carving at the joints that applies to candidate meanings, or because there is such a notion and C is maximal with respect to it.[258]

Sider denies Quantifier Variance. It is important to note that Sider does not think there couldn’t be multiple candidate meanings for unrestricted quantificational locutions. He admits that there can be. Let ‘Shmenglish’ name a possible language orthographically and phonetically identical to English, but in which the quantifiers range only over objects that so-called ‘commonsense ontologists’ think exist.[259] Sider says, “Shmenglish is possible, just not actual,” meaning that Shmenglish is a language English speakers could have spoken, but it is not English.[260] Sider denies Quantifier Variance because he claims there is a single candidate meaning that is intrinsically eligible to be meant by communities attempting to express unrestricted existence with their uses of quantifier-like expressions.[261]

Sider calls his unique, intrinsically eligible candidate meaning existence. On Sider’s preferred view of linguistic interpretation

meaning is determined by two vectors, one dependant on us, cal it use, another dependent on the world, call it the intrinsic eligibility of candidate meanings (reality’s joints).[262]

On this view, since “electronhood is a highly eligible meaning, a (natural, not logical) joint in reality,” it is more eligible to be meant by our use of the expression ‘is an electron’. Even if we believe false things about electrons, and thereby assert falsehoods of them, so that electronhood does not fit use as well as some, gerrymandered, unnatural property, we still mean electronhood by our use of the expression ‘is an electron’. On Sider’s view, “[e]legibility, not use, plays the dominant role in determining the meaning of ‘electron’”, while “our conventions, beliefs, or whatever, concerning ‘electron’ play a comparatively small role in determining the truth conditions of statements involving that term”.[263]

According to Sider’s view, ‘exists’ is similar to ‘is an electron’ in this respect. These terms differ, however, in that the former expresses a logical joint in reality whereas the latter expresses a natural joint in reality. What is the difference? Sider cashes out natural joints in reality in terms of similarity relations. The intuitive idea is that we tend to group things together in terms of their similarity, and that some possible groupings are better than others. So, while being an electron and being an electron or a dinner jacket or a dog are both possible ways to group things, the former is better than the latter in virtue of cleaving to the objective structure of reality. According to Sider, “[i]f all groupings are equally good, then the world is an amorphous collection of objects” (2009, 398). This view of the world neglects the fact that “the world has objective streaks in it; it has structure”.[264] Natural properties, then, correspond to ways of grouping things that reveal the distinguished structure of the world.

Sider extends the notion of similarity operative in cashing out natural joints in reality to facts. According to Sider’s broadly Lewisian account of property naturalness, natural properties are “properties whose sharing makes for similarity”.[265] This similarity criterion, Sider maintains, can be extended to the logical apparatus of a language in the following way. When each of the following

Ted is sitting

John is sitting

is true, “we have similarity between the facts: between the fact that Ted is sitting and the fact that John is sitting”.[266] Both are instances of something sitting. Similarly, when

Something is human

and

Something is located in North America

are both true there is similarity in the facts: both facts are cases of something’s being some way or other.[267]

On Sider’s view, the above similarity between the facts that something is human and that something is located in North America is a “genuine similarity”, i.e., one that holds in virtue of the joints in reality. To think the similarity between these two facts is not genuine is to think that it is like the “similarity” that holds between all the objects that are grue or bleen.[268] This latter view, Sider thinks, is what Quantifier Variantists believe. The Quantifier Variantist, according to Sider, is missing out on a fundamental piece of reality’s structure: its quantificational structure.

If Sider is correct, then Hirsch’s premise (4) is false. For, in interpreting Hirsch’s imagined communities, we must take charity to intrinsic eligibility into account. Intrinsically eligible candidate meanings are “reference magnets” (Sider 2009, 402). Let the PVI-community be a community of speakers who ‘talk just like’ Peter Van Inwagen, and let the DKL-community be a community of speakers who ‘talk just like’ David K. Lewis. So, given that (for example) the PVI-community and the DKL-community each intend to express unrestricted existence, it is irrelevant whether existence fits their respective uses better than some other candidate meaning. Since existence is intrinsically eligible to be meant in virtue of carving at the joints, speakers in both communities mean the same by ‘exists’ and related expressions: existence. Conciliatory interpretations are therefore implausible. Sider argues,

Suppose, for the sake of argument, that existence fits PVI’s use of ‘there exists’ perfectly, and therefore fails to fit DKL’s use. Does this mean that DKL does not mean existence by ‘there exists’ and rather means plural existence_DKL (say) instead? Surely not; surely existence’s superior naturalness outweighs its failure to fit DKL’s use of ‘existence’ perfectly…[269]

Supposing that existence fit DKL’s use perfectly, similar considerations could be adduced to show that PVI’s quantificational expressions mean existence.

Suppose, however, that the reference magnetism exerted by existence is too weak to trump charity to use in interpreting English quantifier expressions. Then it may appear that Hirsch is right: if revisionary ontologists are interpreted as speaking English, they are making false or ridiculous assertions. Sider claims that, even if this is the case, ontologists “can introduce a new language in which to conduct their debate”.[270] Sider claims that the following performance would suffice to introduce the new language:

Let’s give the speakers of ordinary English ‘there exists’; let us henceforth conduct our debate using ‘$’. We hereby stipulate that ‘$’ is to express an austere relative of the ordinary English notion of existence. We hereby stipulate that although the meaning of ‘$’ is to obey the core inferential role of English quantifiers, ordinary, casual use of disputed sentence involving ‘there exists’ (such as ‘Tables exist’) are not to affect at all what we mean by ‘$’. We hereby stipulate that if there is a highly natural meaning that satisfies these constraints, then that is what we mean by ‘$’. Perhaps the resulting ‘$’ has no synonym in English. Fine—we hereby dub our new language Ontologese.[271]

If this performance is successful, then the ontological debates will be saved from mere verbalness so long as all participants in the debates are speaking Ontologese.

Hirsch disagrees. He thinks that considerations of use in the form of the above charity principles trump property naturalness when it comes to interpreting the speech of natural language users and users of Ontologese. This difference of opinion, I suggest, is counted as a merely verbal difference by Hirsch’s own view. In the following section, I will argue for this claim. If my argument is sound, then Hirsch’s view faces a dialectical problem akin to the one I argued faces neo-Carnapians.

In this section I argue that Hirsch’s dispute with Sider over whether the target ontological disputes are verbal is, by Hirsch’s own lights, itself merely verbal.[272] Let ‘Siderish’ be the language spoken by members of a possible community who utter sentences orthographically similar to Sider’s. Let ‘Hirschese’ be the language spoken by members of a possible community who utter sentences orthographically similar to Hirsch’s. My argument proceeds in two stages. In the first stage, I shall provide plausible conciliatory interpretations of the disputed sentences. In the second stage, I shall argue that these interpretations are more charitable than non-conciliatory, disquotational interpretations.

Siderish speakers say

(S1) Quantifier Variance is false,

while Hirschese speakers say

(H1) Quantifier Variance is true.

Upon observing McA and McB arguing over mereological composition, Siderish speakers say

(S2) McA and McB are having a genuine dispute,

while Hirschese speakers say

(H2) McA and McB are having a merely verbal dispute.

If the Hirschese interpret (S1) and (S2) disquotationally as follows,

(S1D) ‘Quantifier Variance is false’ is true in Siderish just in case Quantifier Variance is false.

(S2D) ‘McA and McB are having a genuine dispute’ is true just in case McA and McB are having a genuine dispute,

then, by the Hirschese’s lights, these sentences are false. The same is true of the Siderish and (H1) and (H2). Is there a way for the Hirschese to interpret Siderish speakers such that they come out expressing truths by uttering (S1) and (S2) and vice versa? I think there is.

Hirsch’s dispute with Sider turns on their difference of opinion over what counts as an eligible candidate interpretation of a language. Specifically, the dispute turns on a difference of opinion over what counts as an eligible candidate interpretation of a language’s quantificational apparatus. Accordingly, I suggest that the Hirschese and the Siderish each take the other to mean something different by ‘interpretation’. Let a ‘Hirsch-interpretation’ be one on which charity to use trumps naturalness. Let a ‘Sider-interpretation be one on which naturalness trumps charity to use. The Hirschese can interpret Siderish utterances of (S1) as follows

(S1C) ‘Quantifier Variance is false’ is true in Siderish iff if there are logical kinds, then it is not the case that there are multiple existence-like candidate Sider-interpretations for expressions like ‘there are’ each of which is adequate for describing the world.[273]

on this interpretation, (S1) is true. The Siderish can interpret (H1) in a similar fashion:

(H1C) ‘Quantifier Variance is true’ is true in Hirschese iff there are multiple existence-like candidate Hirsch-interpretations for expressions like ‘there are’ each of which is adequate for describing the world.

The Hirschese can interpret (S2) as follows:

(S2C) ‘McA and McB are having a genuine dispute’ is true in Siderish iff it is not the case that on the best Sider-interpretation McA’s disputed sentences are true in A-lish and McB’s disputed sentences are true in B-lish.

The Siderish can interpret (H2) as follows:

(H2C) ‘McA and McB are having a merely verbal dispute’ is true in Hirschese iff on the best Hirsch-interpretation McA’s disputed sentences are true in A-lish and McB’s disputed sentences are true in B-lish.

On the above interpretations, each side comes out saying something true.[274] Now it remains to be shown that the above interpretations are more charitable than disquotational interpretations.

Let us now examine consideration of interpretive charity in interpreting Siderish and Hirschese. Charity to perception cannot determine whether the conciliatory interpretations I’ve offered above are the more charitable ones. For no perceptual experience seems relevant to deciding how to interpret (S1)-(S2) and (H1)-(H2).

Charity to understanding favors the conciliatory interpretations. Suppose the Hirschese interpreted the Siderish as meaning the same by ‘interpretation’ as they do. On such an interpretation, the Hirschese would attribute to the Siderish a number of a priori or conceptual errors. Where Iand I' are suggested interpretationsof language L, there are many values of I, I', and L such that Siderish speakers will say

(S4) Considerations of charity to use favor I over I', but since considerations of property naturalness favor I', I' is the better interpretation of L.

If the Siderish mean by ‘interpretation’ what the Hirschese do, then (S4) is not just false; asserting it displays a conceptual confusion like the one displayed by an English speaker who says,

(*) I know Smith is taller than Jones, but Jones’s height is greater than Smith’s.

Similar considerations apply to Siderish disquotational interpretations of Hirschese claims like

(H4) Considerations of property naturalness favor I over I', but since considerations of charity to use favor I', I' is the better interpretation of L.

On the conciliatory interpretations I suggest, neither community attributes such a priori or conceptual mistakes to the other. So, charity to understanding favors my interpretations over the disquotational ones.

Hirsch thinks that charity to retraction shows that most intuitively genuine disputes are not merely verbal in his sense (Hirsch 2005, 74). It might be argued that charity to retraction demands that either the Hirschese or the Siderish interpret the other disquotationally. Isn’t it prima facie plausible that there is some piece of evidence or some argument such that, were Sider (or Hirsch) to learn of it, he would retract the disputed sentences?

In order for charity to retraction to favor the disquotational interpretations over the conciliatory ones, there would have to be some evidential circumstances such that, in these circumstances, Sider (or Hirsch) would retract his disputed sentences in favor of his opponent’s. At the current stage in the dispute over the truth of Quantifier Variance and ontological deflationism, it seems unlikely that there is evidence that could convince Hirsch or Sider to accept the other’s position.

Still, that such evidence come to light is possible. It seems just as likely, however, that evidence will come to light that convinces Sider of the truth of Hirsch’s position (or vice versa) as it seems that some evidence will come to light that convinces Sider of the truth of Peter Van Inwagen’s position. For the purposes of my argument, then, I shall pose a dilemma. Either there is some evidence such that charity to retraction renders Hirsch’s metaontological disputes with Sider genuine or there is not. If there is some such evidence, then it is just as likely that there is some evidence such that charity to retraction renders Sider’s ontological dispute with Van Inwagen genuine. If there is no such evidence, then charity to retraction cannot favor the disquotational interpretations over my conciliatory ones.

So, either Hirsch’s charity principles favor my conciliatory interpretations, or they weigh against Hirsch’s claim that certain object-level ontological disputes are merely verbal.

In this section I shall consider a couple of objections to my argument. First, one might object that either ‘Hirsch-interpretation’ or ‘Sider-interpretation’ is a more eligible candidate meaning based on naturalness concerns. While neither meaning seems perfectly natural, one might still be more natural than the other. While considerations of relative naturalness may not induce the Hirschese to interpret Siderish disquotationally, considerations of naturalness might induce Siderish speakers to interpret Hirschese disquotationally. If ‘Sider-interpretation’ turns out to be a more natural candidate meaning, Siderish speakers ought to interpret the Hirschese disquotationally, so the dispute is not merely verbal.

It is unclear whether considerations of naturalness favor one interpretation over the other. This claim would have to be argued for. However, suppose that considerations of naturalness do favor one interpretation over the other. Still, given their views on the relative importance of charity to use, the Hirschese ought to interpret the Siderish conciliatorily. Even if naturalness considerations weigh in here, then, it seems as though the dispute involves at least a one-sided conciliation. While such a one-sided conciliation is not sufficient for full-blown mere verbalness on Hirsch’s view, it still seems to fall short of a genuine dispute.[275]

Second, it might be objected that the conciliatory Hirschese interpretations of Siderish will attribute to the Siderish strange meanings for a number of expressions. Let ‘Inwagese’ be the language spoken by an imagined community who accepts all the ontological assertions of Peter Van Inwagen. The Inwagese assert

(I1) There are no chairs.

Upon hearing (I1), the Siderish say

(SI) The Inwagese said something false,

and the Hirschese say

(HI) The Inwagese said something true.

If the dispute about the mere verbalness of the Sider-Van Inwagen dispute is merely verbal on Hirsch’s view, as I am suggesting, then the Hirschese should interpret (SI) such that it comes out true. But if they so interpret (SI), won’t they have to think that ‘true’ means something peculiar in Siderish? For if the Siderish mean by ‘true’ the same as the Hirschese, then a contradiction will be derivable in Hirschese.[276]

Serious trouble looms. If they are conciliatory, the Hirschese will interpret the Siderish as meaning by ‘true’ something like:

(ST) ‘Snow is white’ is true in Siderish iff ‘Snow is white’ comes out T on the best Sider-interpretation, if there are logical kinds.

where T indicates whatever property plays the role in Siderish semantics played by ‘true’ in Hirschese semantics. Given that Siderish speakers make Tarskian disquotational assertions and take those assertions to be meaning-constitutive and a priori, the Hirschese will have to treat sentences such as “Snow is white” as having a peculiar meaning in Siderish.[277]

I suggest that the Hirschese not interpret the Siderish as meaning something different by ‘true’ than they do. Rather, I suggest that the Hirschese interpret the Siderish as meaning something different by ‘expresses’ than they do. Call whatever relation that plays the role of expressing a proposition in the Siderish semantics ‘Sider-expressing’. Then the Hirschese can interpret (SI) as follows:

(SIH) ‘The Inwagese said something false’ is true in Siderish iff on the best Sider-interpretation of Inwagese, ‘There are no chairs’ Sider-expresses a false proposition.

That the Siderish mean something different by ‘expresses’ goes hand in hand with their meaning something different by ‘interpretation’.

At this point a further problem arises. The basic assumption behind charity in interpretation is that speakers mean to express truths, not Sider-express truths. Can an interpretation on which speakers are viewed as trying to Sider-express truths be the uniquely charitable one?[278] Perhaps not. I do not have a well-developed view of linguistic interpretation to offer here, but some remarks may suggest an answer. I think that giving a charitable interpretation of a language is motivated by the presupposition that speakers of that language are trying, in some sense, to capture the truth. Perhaps there are a number of relations between sentences and propositions in virtue of which sentences can be said to capture the truth. The most charitable interpretation of a language, then, may not be one on which speakers of that language are interpreted as trying to express true propositions, but rather one on which speakers are interpreted as trying to stand in some other relation to true propositions, some relation that supports locutions like ‘capturing the truth’.

If my argument is sound, then Hirsch’s metaontological dispute with Sider is, by his own lights, merely verbal. This result is surprising. I think that it suggests Hirsch’s possible communities model doesn’t capture the sense in which McA’s dispute with McB over cups and glasses is merely verbal. Sider’s dispute with Hirsch appears to have all the markings of a genuine dispute. There is room, however, for a view on which this is an illusion. I take it that neither Hirsch nor Sider would adopt such a view.

What should the rest of us believe? I’m inclined to think that the dispute is genuine, and it is Hirsch’s account of a merely verbal dispute that is in need of revision.[279]

In light of the above considerations I reject Hirsch’s view. I do not think the Possible Communities model of verbal disputes provides a sense of ‘merely verbal’ on which the ontological disputes in question are verbal. Hirsch’s view requires invariably giving more weight to use than naturalness in linguistic interpretation. This seems wrong. Further, accepting Hirsch’s view requires accepting the controversial thesis of Quantifier Variance. I have argued that, if Hirsch’s view is correct, then the question of whether Quantifier Variance is true is itself merely verbal. This generates instability in the position we’d do better to avoid.

Further, it is unclear to me that a dispute’s being merely verbal in Hirsch’s sense counts against that dispute. I find it easy to accept that philosophers should avoid arguing past one another or equivocating on the use of some term or making claims that do not express propositions. But Hirsch does not argue that ontologists are doing any of these things. It is hard to see what is wrong with their dispute. For all Hirsch says, it is the dispute that would go on between his imagined communities that is verbal, not the dispute that is really going on in our English-speaking community. That dispute looks substantive. To say that the dispute that’s going on in our community is verbal in virtue of the possible dispute is merely to stipulate a sense of ‘verbal dispute’ divorced from our ordinary understanding of that expression. Perhaps, my above arguments notwithstanding, the disputes are verbal in this sense. But why should we care whether the disputes are verbal in some philosopher’s stipulative sense of ‘merely verbal’? It is our sense of ‘merely verbal’ that counts when it comes to whether a merely verbal dispute should be dissolved or abandoned. Hirsch has not shown that the disputes are verbal in any ordinary sense of ‘merely verbal’.

Alan Sidelle argues that there are no facts of the matter that could, even in principle, settle the Nihilist-Universalist dispute.[280] Sidelle also argues that this dispute is merely verbal.[281] In this section, we shall examine his reasons for thinking that this dispute over composition is verbal. Unlike Hirsch, Sidelle does not offer a set of necessary and sufficient conditions for a dispute to be verbal. Instead, he provides a diagnostic test. It is symptomatic of verbal disputes that they be such that disputants “could agree on all the facts (even if they don’t), while still persisting in their apparently contradictory sentences.”[282]

This condition is pretty vague, for it is not entirely clear what ‘all the facts’ is supposed to refer to. In order to get a grip on which facts Sidelle has in mind, we need to appreciate his view of the world and our thought and talk as “layered”. According to Sidelle,

Objects (I use the term here very broadly to include actions and events, and almost anything that can be referred to with a singular term) have parts, both large and small, and lots of properties. They are similar to each other in lots of ways, and so can be grouped in many ways—either in a simple “red and round” sort of way, or a more complex, “similar enough in a variety of these respects” sort of way. I count as “layering” the way in which these more complex respects of similarity—or, if one likes, properties—depend on others.[283]

The world, and our representations of it, are “built up”, as it were. Take the property of being an instructor of philosophy. This property is pretty complex; it consists in an object’s having a number of properties: being a sufficiently personlike creature, teaching at least one class in philosophy, getting paid for teaching the class, and so on. Statues (if there are any) are complex things. Something’s being a statue is a matter of it being made up of some stuff that was arranged, or at least selected, by somebody with artistic intentions.[284] Suppose we disagree about whether a particular object is a statue. You and I are on a walk, and we see a wooden figure that looks like a person. You say, “There’s a statue of a person over there,” and I disagree. There are (at least) two things that could be going on here.

First, we might disagree about some fact at a level of description that does not involve the concept of statuehood or use of the word ‘statue’. So, suppose you think that the piece of wood became person-shaped by means of a bucolic artisan’s having chipped at it with a chisel, but I think that erosion or some natural process shaped the figure. In this case, we would be having a disagreement about the facts that we both take to constitute being a statue.

Suppose our disagreement were a bit different, though. Suppose that we both agree that an artist shaped the figure, but I think that a statue is something displayed in an art gallery, while you think that the location where something is displayed is irrelevant to determining whether it is a statue. Here we can agree on all of the base facts (i.e., the facts we can state without using the term ‘statue’ or a cognate expression), but we can still persist in our disagreement about whether the figure we glimpse in the woods is a statue.

The way we learn terms like ‘statue’, ‘table’, and the like typically involves “having the term[s] applied—used—in various real-life situations.” It is for this reason that “[t]o a certain extent…the base concept we form is “’enough like this (or these) in relevant respects.’”[285] It is fairly easy to see, then, how it is “possible for [people] to agree about all the ‘base’ facts while disagreeing about whether or not some term—whose application can only depend upon what the base facts are—applies in [some] case.”[286] We are each of us introduced to the terms that make up our public language through different patterns of use. Given this feature of our language learning environment, “it will hardly be surprising if people would not always apply [a] term the same way to new cases, cases which are similar in some ways, but different in others”.[287] Take a case in which there is disagreement about whether to apply some term F in a novel case. Users of the term have had exposure to cases where F was agreed to apply, and in this novel case there are many features that resemble the original cases of F-hood, but some differences. Some parties think F should apply, others think it should not. In many of these sorts of case, the richness of the world’s layered properties makes it very likely that, when people propose different meanings of a term in response to a novel case, the world will lack the right kind of structure to “determine one of them as the meaning” of the term.

Let’s take a case. Is Duchamp’s Fountain a work of art? Some agree, some disagree. The term ‘work of art’ is learned by seeing it applied (and applying it ourselves) to objects that share a cluster of features, say C. Duchamp’s piece, unlike many things to which we apply the term ‘work of art’, was not itself manufactured, created, or even altered in any significant way by the artist; it is a “ready made” piece. Is such so-called ‘found object art’ really art? There is disagreement. But is there a natural joint in reality here that could serve to fix the meaning of ‘work of art’? This may seem unlikely. In such event, “the stage is set for verbal disagreements’ about whether Duchamp’s piece is a work of art, and since the term ‘work of art’ has been learned by all parties to the disagreement “in sufficiently similar ways” and all parties to the dispute “have similar associations,” it is likely that “both sides suppose that their sentences contradict the others.’”[288] But, what we have in the Duchamp case is an instance of some people who (at least can) agree about all the base facts (e.g., the exhibit features a urinal, Duchamp selected the urinal, it is made of porcelain, Duchamp did not fashion it himself, etc.), but still continue to have their disagreement. This is a symptom, Sidelle thinks, of a merely verbal dispute.

It is fairly easy to see that this feature is symptomatic only, and not sufficient for a dispute to be verbal. [289] For take a dispute about Bill: I like him and you hate him, and we each try to change the others’ mind. It could be the case that we agree about all the facts, we just differ about how we feel about Bill. Also, we could have a non-verbal dispute about some moral issue in which we agree about all the (nonmoral) facts, but we still disagree about what is morally right.

So, the first step, according to the method for spotting verbal disputes that Sidelle sketches, is to see if the Nihilism-Universalism dispute has the symptom: can Nihilists and Universalists agree on all the “base facts” but persist in their dispute? Sidelle thinks they can. Let’s take a concrete example. We’ll use our (by now very tired) example of Peter V. and David L.’s dispute over the number of material objects in the vacuum chamber populated by two simples: a and b.

Peter V. and David L. purport to disagree about the truth of:

PARTS: There is a material object in the vacuum chamber that has a and b as
parts.

What “base facts” are relevant to determining the truth of PARTS? Well, here’s a (tentative and perhaps incomplete) list: a is in the chamber, b is in the chamber, a is a mereological simple, b is a mereological simple, a and b are n distance apart, a could exist without b, b could exist without a, and so on. It seems there could agreement on all of these facts. Where is the source of the disagreement, then? It seems like the only way to find a ground for the disagreement is to take account of Peter V.’s belief that nothing is a material object unless it is simple, and David L.’s belief that some things are material objects even though they are not simple.

But now it should be obvious that the disagreement is not about a base fact, like our above disagreement over whether the wooden figure was formed by an artist or not; instead, the disagreement is about one of the application conditions for ‘material object’. Insofar as Peter V. and David L. disagree about how to apply the term ‘material object’, in virtue of their disagreements about its correct application conditions, they do not disagree, or at least need not disagree, about the “base facts”.

But this does not entail that their dispute is merely verbal. According to Sidelle, however, agreement on all the base facts does suffice for a non-genuine dispute: “any dispute for which there is complete agreement at the ‘base’ level—or even, could be such agreement (without the disagreement going away)—is not a genuine dispute.”[290] According to this criterion, Peter V.’s dispute with David L. about how many material objects there are is non-genuine. However, this is not to say that Peter V. and David L. are having a merely verbal dispute, for they could still be having a genuine dispute over the correct application conditions for ‘material object’. This dispute could be perfectly genuine for their difference of opinion as to the correct application conditions for ‘material object’ could be construed as their holding different, and incompatible, theories about the nature of material objects. In order for their dispute to be verbal, then, something further would have to be added.

What would it take for the dispute to count as verbal? Here is one suggestion, the dispute is verbal “if we can plausibly assign different properties to the disputants’ uses of [‘material object’] such that if we took each party to be meaning, or referring to, these properties, each could be making a true claim that would not conflict with the other.”[291] This condition needs a little bit of tinkering if it is to deliver the right results.

First, I think we should drop the bit about both parties making a true claim on the supposed plausible assignment of properties. Why? I think that our criterion here allow for cases in which a dispute is verbal even if the most plausible interpretation of what both parties say is false or without truth value. Sidelle also wants to take this line, too, for he wants to allow that certain disputes over the application of vague predicates to borderline cases are verbal disputes. But, with a vague term in a borderline case, no plausible interpretation will yield a claim that evaluable for truth.

Also, the notion of ‘conflict’ here needs to be unpacked. What Sidelle has in mind here does not seem to be mere logical inconsistency. For we want to count even persons who are interpreted as uttering logical contradictions as sometimes saying things that do not conflict with each other in the relevant sense. I can have a merely verbal dispute with you if I say, “There are married men who not married” and you say “There are no married men who are not married” so long as I mean by “married” something different than what you mean. But what I say (no matter what I mean by ‘married’) will be a logical contradiction, so our views will conflict in the sense that it is not logically possible for them both to be true. But, our above criterion should rule our dispute verbal despite this.

It is not entirely clear how to make ‘conflict’ precise here. One suggestion would be the following:

METAP: S’s view conflicts with S*’s view iff the conjunction of S and S*’s view is not
metaphysically possible.

A correct account of merely verbal disputes should be able to rule as merely verbal the dispute between our two confused undergraduates about whether an omnipotent God could create a stone so heavy he couldn’t lift it.[292] However, if METAP is the right way to understand ‘conflict’ the two undergraduates’ views do conflict, because they both believe something metaphysically impossible.

I think that the reading of ‘conflict’ that is most charitable to Sidelle is one cashed out in terms of subject matter. When our views or claims fail to conflict, in this sense, our claims are about different subject matter: “[t]he idea is that disputes appear to involve some common subject matter, with each party saying something different, and conflicting, about that subject.”[293] The operative notion of subject matter is not too precise, but I think it will do. Let’s take our undergraduates:

Larry: An omnipotent God could create a stone so heavy he could not lift it.

Mary: No way! An omnipotent God could do no such thing.

There is an important sense in which the subject matter of Larrry’s claim is distinct from the subject matter of Mary’s. Recall, Larry is talking about a God that could do anything logically/metaphysically possible, while Mary is talking about a God that could do anything whatever, even what is logically/metaphysically impossible. In an important sense, then, Larry and Mary, even while they both purport to be talking about God, are really talking about different subjects altogether.[294] So, their remarks, while metaphysically incompatible in the sense that there is no possible world at which both claims are true, nevertheless do not conflict with each other. In what follows, we’ll understand ‘conflict’ in this way.

So, as far as I understand Sidelle’s argument, it is as follows:

If Peter V. and David L. can agree on all the base facts and if we can plausibly assign different meanings to ‘material object’ such that if we took each party to be meaning, or referring to, these properties, each could be making a claim that would not conflict with the other, then it is reasonable to believe that Peter V. and David L.’s dispute is verbal.
Peter V. and David L. can agree on all the base facts.
We can plausibly assign different meanings to ‘material object’ such that such that if we took each party to be meaning, or referring to, these properties, each could be making a claim that would not conflict with the other.
It is reasonable to believe that Peter V. and David L.’s dispute is verbal.[295]

A couple of things bear mention here. While I do not have a fully worked out view of what it takes to make an assignment of meaning plausible, I think a few remarks in this connection are in order. A plausible assignment is not just a possible one. After all, it is possible that Peter V. meant by ‘material object’ what most of us mean by ‘orangutan’. If we could plausibly assign this meaning to Peter V., then premise 3 would be true. However, as I hope should be obvious, this assignment is implausible. At the very least, a plausible assignment of a meaning to a speaker has to make sense out of their verbal and non-verbal behavior. Considerations of property-naturalness, however, may also get into the mix. It would be implausible to assign the meaning ‘the property of being a material object or an abstract object’ to David L., for this meaning is a non-natural, disjunctive property.

The key premises here are (2) and (3). I think we can plausibly reject both. Let’s start with (3). My reason for rejecting (3) is a bit more controversial than my reason for rejecting (2). What are the candidate meanings we could supply that would make (3) true? It seems they’d have to be (something like):

Peter V.: ‘material object’ means the same as ‘simple object’

David L.: ‘material object’ means the same as ‘simple object or mereologically
complex object’

There are two problems here. First, the meaning assigned to David L. is a disjunctive property. Isn’t there a non-disjunctive property he could mean instead? Sure there is: the property of being a material object. But if this is what he means, then it is perplexing why Peter V. doesn’t mean this as well, for if this candidate meaning is natural enough to be meant by David L., what could stop it from being meant by Peter V.? Perhaps Peter V.’s usage does not fit. But how would we argue that his usage does not fit? It would seem that we’d have to give an argument like this: Letting M be the property of being a material object, the extension of M is G, Peter V. rejects applications of ‘material object’ to sufficiently many things that are members of G, so the property of being a material object is not a good fit for his pattern of use. But any such argument would be very problematic for Sidelle. For, if we presuppose that we know the extension of the property of being a material object, it looks like we’re presupposing that there is a right answer here. But Sidelle argues that there is no fact of the matter in disputes over what counts as a material object.[296]

Of course it might be suggested that the problem term here is not ‘material object’, they mean the same by that, the problem is ‘There is’. This sort of reply invites a second argument against premise (3): the argument from logical joints. If something like Ted Sider’s position about how quantifier expressions get their meaning is true, then it would be implausible to assign to Peter V. and David L. different meanings to ‘There is’ and related expressions. For there is a unique, intrinsically eligible candidate meaning for ‘There is”: existence. Since existence is so eligible to be meant, both Peter V. and David L. mean it, even if their patterns of use do not quite match up with the extension of this candidate meaning.

Is there such a candidate meaning as existence? I think that Sidelle would not go along with this. In fact, he may take the view that the dispute over whether there is or is not such a candidate meaning is defective, either by because it is a no fact of the matter dispute, a verbal dispute, or both.[297]

The dispute over whether (3) is true, then, seems to come down to whether there is or is not a unique domain for evaluating the truth of quantified sentences. If there is no unique domain, then Sidelle wins and premise (3) is likely to be true; if there is a unique domain, then Sider wins and premise (3) is false. I do not see how to resolve the dispute here.

However, I do not need to. There is a good reason to reject Sidelle’s argument. First of all, I need to say something about premise (1). If the conditional in premise (1) is true, then I suggest that “can agree on all the base facts” must be interpreted as “can agree on all of what they take to be the base facts”. Let’s take an example. Suppose that I have a dispute with my neighbor about whether aliens kidnapped and probed him. He says, “Last night I was kidnapped and probed by aliens!” and I say, “Oh, get real, you were not.” But, I don’t disagree with my neighbor because I do not believe in aliens (let’s suppose I do) or because I do not believe he was abducted last night (let’s suppose that I do), I disagree with him because I mean by ‘probed’ (something like) interrogated, and my neighbor means by ‘probed’ something a little more invasive. We should say that our dispute suffers from Sidelle’s symptom. We have a verbal dispute in part because we can agree on all of what we take to be the facts. If Sidelle means by “can agree on all the facts” merely “can agree on all the facts that actually obtain, regardless of which ones we believe obtain”, then premise (1) is false.

This having been said, it seems clear to me that Peter V. and David L. cannot agree on all the base facts while persisting in their dispute. Why not? Because Peter V. believes that composition is metaphysically impossible and David L. thinks this is false. Whether composition is metaphysically possible is certainly a relevant base fact in determining whether the term ‘material object’ applies to anything with a and b as parts. It is a metaphysically necessary fact, Peter V. maintains, that composition never occurs. It is a metaphysically necessary fact, David L. argues, that composition always occurs when two things exist. Here we have our difference of opinion on a relevant matter of fact.

Notice that it is irrelevant whether Sidelle can present compelling arguments that show there are no such facts as the ones that Peter V. and David L. believe in. All that we need to secure the non-verbalness of their dispute is that our disputants believe there to be such a fact.

Sidelle is sensitive to this objection. If it is plausible to think that vying ontologists believe in such metaphysically necessary facts, then, Sidelle admits, it may not be plausible to think that their disputes are verbal. For suppose that, in addition to all the facts about the location of a and b, their simplicity, their qualitative features and spatiotemporal relations to each other, there is a “further fact” that it is metaphysically necessary that they compose something. And suppose that Peter V. thinks there is a further fact: the fact that it is metaphysically impossible for a and b to compose something. Sidelle says that “perhaps for such people” as Peter V. “it is not charitable to interpret ‘object,’ in their mouths, so that it applies” where there are some simples arranged in a certain way. If David L. honestly believes in the above further fact, “perhaps there is a genuine dispute between such philosophers and those who deny that there are composite objects: they disagree about whether there are such further facts.”

Sidelle still thinks their disputes are defective, for he thinks there are no such metaphysically necessary facts as the ones Peter V. and David L. believe in. We have already assessed his arguments for this claim and found them lacking.
Chapter 11

POSSIBLE WORLDS: AN INTRODUCTION

There were once two communities. Members of both communities believed the way things are to be just one of many, probably infinitely many, ways that things could have been. Further, both communities concurred that when they said “There are ways things could have been but in fact are not”, what they meant by ‘There are’ was the same as what they meant when they said things like “There are human beings”. They shared the belief that most things like human beings, donkeys, puddles, and stars have some of their features only accidentally and have others of their features non-accidentally. While there was some squabbling amongst both groups about which features things have non-accidentally, everyone agreed that the features (whichever they are) that something has non-accidentally are those that it couldn’t have failed to have no matter how things turned out. There was also some disagreement about whether there are any things that would have existed no matter how things turned out, but both communities agreed that if there are any such things, they would be necessary things.

One community called themselves the Wezekana. The Wezekana believed there to be infinitely many universes. These universes, according to the Wezekana, were not the sorts of places you could ever travel to. This, they said, was not because the universes were too far away. Rather, the Wezekana thought that these other universes were not related to each other in the ways that sticks, stones, and persons are. No part of our universe, they thought, stood in any of the familiar relationships like being close to, far from, on top of, sharing a part, or anything like that, with any part of any other universe. Each universe, according to the Wezekana, was totally isolated from each other universe. According to the Wezekana, some universes are very small, for example some universes have as parts just a donkey. Other universes, the Wezekana believed, are very large, for example some universes contain as parts duplicates of the entire universe.

The Wezekana believed that there are so many universes that for any possible way that a universe could be, there is some universe that is in fact that way. The Wezekana called these universes ‘werlds’. They referred to the universe of which they were all parts as ‘the actual werld’ and all the others as ‘merely possible werlds’. They believed that other possible werlds are occupied by persons, donkeys, and stars, some of which are exact duplicates of persons, donkeys and stars from their own universe. According to the Wezekana, no object (except for maybe numbers and other strange objects like numbers) is a part of more than one werld. Since werlds are isolated, they cannot overlap. But, the Wezekana believed that each other-werldly person would be right to say “My werld is the actual werld”. So, according to the Wezekana, ‘actual’ worked kind of like their word ‘here’.

The Wezekana believed that when they said, “Kanu could have gone fishing today, though he did not” what they said is true just in case there is some werld where a person relevantly similar to Kanu does in fact go fishing today. They further believed that when they said, “Kanu could not have been taller than himself” what they said is true just in case there is no werld where a person relevantly similar to Kanu is taller than himself.

The other community called themselves the Kweli. They had a different sort of view about things. The Kweli believed in infinitely many simple objects. These objects, according to the Kweli, have no parts or constituents. The Kweli called these objects ‘states of affairs’. They believed that there is the state of affairs of Neema’s being wise, the state of affairs of Jafari’s being foolish, and the state of affairs of seven’s being prime. According to the Kweli, there are some states of affairs that are actual and others that are not actual. The state of affairs of seven’s being prime, for example, the Kweli believed to be actual, whereas the state of affairs of Jafari’s being foolish they thought to be non-actual.

According to the Kweli, there is an interesting relationship that one state of affairs can stand in to another: one state of affairs, say S, can include another, say S*. This is the case if and only if were S actual, S* would also be actual. So, Neema’s climbing a mountain includes Neema’s climbing something or other. States of affairs, the Kweli thought, can preclude each other. S precludes S* just in case S and S* cannot both be actual. For example, Neema’s being the only one to climb Mt Wanakani precludes someone else’s climbing it. The Kweli thought that there are infinitely many states of affairs that are maximal. They called a state of affairs maximal when it is such that for any state of affairs S, it either includes or precludes S. Some states of affairs, according to the Kweli, are possible. That is, it could have been the case that these states of affairs were actual. The Kweli called the states of affairs that are both possible and maximal ‘wurlds’.

According to the Kweli, when they say “Neema could have climbed Wanakani”, what they say is true just in case there is a wurld that includes the state of affairs of Neema’s climbing Wanakani. And when they say “Seven could not have been even” what they say is true just in case every wurld precludes the state of affairs of seven’s being even.

The Kweli also held that an individual exists in a wurld just in case, were that wurld actual that individual would exist. So, according to the Kweli, one and the same individual, say Zuwena, exists in many different wurlds.

Throughout the early history of the communities, the Wezekana and the Kweli were largely isolated from one another. But one summer a drought forced the members of these two communities to become closer. Supplies of food and water were scarce, so the Wezekana and the Kweli had to trade with one another to survive. As the relationship between these two communities developed, they came to learn that they had much the same cultural practices and even seemed to share a language. Since they got along so famously, they observed what a great thing is was that they were forced by environmental conditions to get together. Indeed, it was popularly believed that, had things been otherwise and there were no drought, each community would have missed out tremendously. This shared thought occasioned discussion about just what was meant by saying that things might have been different had there been no drought. Each community claimed to believe in the existence of entities they called ‘ways things could have been’. They differed, however, with respect to what they thought these entities were like. The Wezekana and the Kweli bickered fiercely. Wezekana elders would say, “No individual exists at more than one werld!” while Kweli elders would forcefully claim, “Many individuals exist at more than one wurld!”

As each community carefully explained its views to the other, however, there grew a rising sense that there was something wrong about the bickering. Not everyone agreed what was wrong. Some felt that the Wezekana were using a slightly different language than the Kweli, and so that, at least when talk turned to matters modal, they were talking past each other. Others felt differently. They thought that the Wezekana had in mind a different sort of project than the Kweli. Proponents of this view pointed to the fact that the Wezekana claimed to provide an analysis of words like ‘possible’ and ‘necessary’, while the Kweli had no such pretensions. Proponents of this view claimed that the Wezekana were talking at cross purposes with the Kweli, and that properly understanding what each community intended to do with its fragment of modal discourse would resolve the bickering.

Still others thought that something else was going on. They maintained that the Wezekana and the Kweli had different and competing theories about what makes something possible. Among those who held this view, however, there were many who still thought the bickering was silly. Some of these folks maintained that there was just no way to know which theory was true, or if both were false. These folks despaired at ever finding any evidence that would make it rational to accept the Wezekana theory over the Kweli theory or vice versa.

Some took a different tack. There is just no fact of the matter as to which theory is correct. These folks took the bickering to be senseless because there was in principle no way the bickering could ever be resolved.

Still others thought that the Wezekana and the Kweli agreed on all the facts, but just opted to describe them differently. These folks held that the Wezekana and the Kweli spoke a shared language, but that each community used different, but equivalent, expressions to describe the same modal phenomena.

Let’s call people who thought there was something wrong with the Wezekana and the Kweli’s disagreement naysayers. While the naysayers were in the majority, there was a relatively small but strident group of people in both communities who held that the bickering was not merely petty bickering, but was instead a profound and deep disagreement about the nature of reality. These people believed that the naysayers had it wrong. These folks claimed that the Wezekana offered one theory about modal reality and the Kweli offered a competing theory, and that one at most of these theories is true. Moreover, they believed that there is evidence that would make belief in one theory more rational than belief in the other. Unlike competing theories about whether the number of atoms in the universe is odd or even, the dispute over werlds and wurlds is such that there could be some evidence in favor of the Wezekana’s picture and against the Kweli’s picture or vice versa.

Which explanation of the apparent disagreement between the Wezekana and the Kweli is correct? Unfortunately, we may never know, since historical record does not reveal whether the Wezekana and the Kweli share a common ancestry and hence whether it is reasonable to believe they spoke a common language. The issue remains largely unresolved, a problem for scholars to puzzle over and occasionally get into loud drunken arguments about.

The above is, of course, a fictional story. Readers who were raised on modal metaphysics will no doubt have guessed its origin in the famous dispute between Alvin Plantinga and David Lewis over the nature of possible worlds and the so-called Problem of Transworld Identity.[298] David Lewis presents us with his Pluriverse: the worlds in all their glory.[299] According to Lewis possible worlds are things just like the universe we inhabit: they are maximal mereological sums of spatiotemporally interrelated things. No possible individual is a part of more than one world, and every possible individual is part of one world. According to Plantinga, possible worlds are not like the universe we inhabit. Possible worlds, including the actual world, are maximal possible states of affairs. Each individual exists in countless such possible worlds. For to say that an individual exists in a world is to say no more than were that world to be actual, that individual would exist.

Consider the following thesis, which I will call the Thesis of Transworld Identity:

TI: Some concrete individuals exist in more than one possible world.

It is tempting to see Plantinga and Lewis as offering different and incompatible answers to the question of whether TI is true. David Lewis says (things like) “Worlds do not overlap in the sense of sharing parts. Each possible individual inhabits just one world.” Plantinga, on the other hand, says (things like) “Individuals exist in more than one world.” Call an individual that exists wholly in more than one world a transworld individual. At first blush, it looks like Lewis and Plantinga disagree about the claim:

TW: There are transworld individuals.

Lewis, apparently, denies TW and Plantinga seemingly affirms it.

Who is right? There are arguments on both sides. The main argument for Lewis’s position is the Contingent Intrinsics Argument; the main argument for Plantinga’s position is the Relevance Argument. Let’s consider each in turn.

First the argument from Contingent Intrinsics: how can one and the same individual exist in more than one world? The believer in transworld individuals says that Socrates, for example, exists in a number of worlds. Let’s consider two such worlds: one, call it w, in which he has some intrinsic feature, say snubnosedness, and other, call it w*, in which he lacks that feature. Is Socrates in w one and the same as Socrates in w*. Yes, the lover of transworld individuals would have us believe. But, so goes the argument, we should demur. For Socrates in w has a feature that Socrates in w* lacks: namely snubnosedness. The person who is snubnosed in w cannot be one and the same as the one who is not snubnosed in w* on pain of contradiction.

The only way out for the believer in transworld individuals is to claim that it is not after all one and the same property that Socrates has in w but lacks in w*. In fact, there are two different properties: being snubnosed in w and being snubnosed in w*. It is no contradiction to attribute both to Socrates. But now there is a problem. Surely if anything is an intrinsic property, snubnosedness is. But intrinsic properties are clearly not relations. My having some intrinsic property P is just a matter of how I am; it is not a relation between me and some other thing. I can just plain have P. But, the believer in transworld individuals cannot say this on pain of contradiction. So much the worse for transworld individuals.[300]

Now let us turn to the Relevance Argument. How could it be the case that individuals are world bound in the way Lewis says they are? Clearly it is possible for me to have gone to the track today instead of staying in my office writing. But if I only exist in one world, then there is no possible world in which I do go to the track. So, as it turns out, going to the track is not a possibility for me at all. This is too quick. The lover of world bound individuals says that there is some otherworldly person very similar to me, in fact more similar to me than anything else in his world, and this person goes to the track. Such a person, though he is not identical to me, is nevertheless a very close counterpart of me. His going to the track is sufficient for me to have the modal property possibly going to the track. What could be wrong with this?

According to the Relevance Argument, the behavior of this otherworldly person is irrelevant to my having the property of possibly going to the track. How could what this person does have anything to do with whether I myself could have gone to the track? Suppose that there is someone in Albuquerque who resembles me pretty closely, in fact suppose this man resembles me more closely than anything else in Albuquerque. How is his going to the track relevant to whether it is possible for me to have gone to the track? Clearly it is not. But this is what the advocate of world bound individuals (plus counterpart theory) asks us to believe.

At first blush it seems as though Lewis and Plantinga’s views are inconsistent with one another; either we each exist at more than one world or we do not. In what follows, I want to explore whether any case can be made for a deflationary view of their dispute. That is, can any of the attitudes of the naysayers from above be justified?

Chapter 12

MODAL REALISM: TWO VERSIONS

In the last chapter, I gave in broad brushstrokes Lewis and Plantinga’s respective views. Before evaluating the status of their famous dispute, we must spend some effort getting clear on the details of what each theorist commits himself to. We shall begin with Plantinga’s view.

1. Abstract Worlds: Plantinga

Plantinga calls his own view a version of modal realism. This is somewhat problematic, for Lewis claims that distinction for himself as well. As Plantinga construes modal realism, the modal realist is committed to three theses: (i) that things have both accidental and essential properties, (ii) that there are such things as possible worlds, and (iii) that things have properties in, or according to, worlds. If these are the commitments of modal realism, it is hard to see how Plantinga’s view is at odds with Lewis’s: they both seem to be modal realists on this account.

Not so says Plantinga. The commitment to (i), as Plantinga tells the story, requires commitment to the existence of propositions. According to Plantinga, “[p]ropositions are claims, or assertions; they attribute or predicate properties to or of objects; they represent reality or some part of it as having a certain character.”[301] For reasons that will become clear, Plantinga thinks Lewis does not acknowledge the existence of propositions or, for that matter, of properties either. According to Plantinga, the modal realist is committed to two varieties of propositions: “those that could have been false, and those that could not.”[302]

Plantinga also requires of the modal realist that she accept possible worlds. One might think that Lewis’s credentials are impeccable on this score. Not so, according to Plantinga. As Plantinga thinks of possible worlds, they are such that for any state of affairs or proposition P, P is possible if and only if there is a possible world that includes or entails P. But ‘includes’ and ‘entails’ are technical terms for Plantinga. The relation of inclusion is one that holds between states of affairs as follows:

INCLUSION: S includes S* iff were S to obtain, S* would obtain,

while entailment is a relation between propositions as follows:

ENTAILMENT: S entails S* iff were S true, then S* would be true.

According to Plantinga’s account of what a possible world is, it follows that only propositions or states of affairs can be possible worlds. For, since possible worlds must be able to stand in the inclusion relation to states of affairs or the entailment relation to propositions, and both terms of the former relation must be states of affairs while both terms of the latter relation must be propositions, a possible world can only be a state of affairs or a proposition. Lewis, according to Plantinga, does not believe in states of affairs or propositions, so he therefore cannot believe in possible worlds.

On Plantinga’s preferred view, possible worlds are maximal possible states of affairs. Plantinga officially distinguishes states of affairs from propositions, saying of the former that they have the property of either obtaining or not obtaining and of the latter that they have the property of being true or the property of being false.[303] On Plantinga’s view states of affairs are simples. They do not have parts or constituents. Socrates is, however, in some way involved in both the state of affairs of Socrates’ being wise and of Socrates’ being snubnosed. This is not to say that Socrates is a mereological part of both states of affairs or that he is a (non-mereological) constituent of both. Rather, to say that Socrates is involved in both of these states of affairs is to say that had either obtained Socrates would exist.

This brings us to the third commitment of modal realism. Objects have properties in worlds. According to Plantinga, for an object O to have a property P in a world W is for W to be such that, had it obtained or been actual O would exist and have P. Possible worlds, then, do not represent that individuals have properties by containing those individuals (or some surrogates of those individuals), rather worlds represent that individuals have properties by including states of affairs (or propositions) that represent those individuals as having those properties. For Plantinga, possible worlds, including the actual world, are by their very nature representational. Things are the case according to them.

Plantinga accepts an unanalyzable metaphysical property of being actual or obtaining. While every world represents of itself that it is actual, only one world is in fact actual: the world that we call ‘the actual world’. But we are not parts of this world. We are no more (mereological) parts of this world than we are (mereological) parts of any state of affairs. According to Plantinga states of affairs do not have parts. We are parts of the universe which is a maximal concrete particular; it is the mereological fusion of us and everything spatiotemporally related to us and nothing else. But this huge object, Plantinga says, is not the actual world. The actual world is a state of affairs and it, like all other possible worlds, is an abstract object.

Only one world is actual, according to Plantinga, but this is not to say that only one world exists. All the other worlds exist, in fact, they exist in the actual world. Indeed, each world exists according to each other world. Plantinga calls himself an actualist. He takes actualism to be the view that there are no non-existent, merely possible entities. Everything that exists exists at the actual world. How then, can Plantinga account for the seemingly obvious truth of the following:

CLAIM: There could have been an object distinct from every object that there is.

On Plantinga’s view, this claim is true, not in virtue of the existence of unactual, merely possible objects, but rather in virtue of the existence of unexemplified essences. An essence of an object O is a property P such that: (i) O has P essentially (i.e., O couldn’t have existed and not had P) and (ii) any object that has P is identical to O. On Plantinga’s view the property of being identical to Socrates is an essence of Socrates. To say that there might have been more objects is, then, just to say that there are some essences that, as things happen to be, are not instantiated but could possibly have been instantiated. That is, there is some possible world in which they are instantiated. Plantinga also claims strict actualism. Not only are there no individuals that do not exist, there could not have been any individuals that do not exist.

2. Concrete Worlds: Lewis

Lewis also claims to be a modal realist. Lewis’s version of modal realism (LMR) is the existential thesis that there exists a plurality of worlds. Stalnaker refers to this thesis as ‘extreme modal realism’, though David Lewis ponders in what dimension its extremity is supposed to lie. Unpacking this thesis requires saying what a Lewisian modal realist thinks a world is. What is the nature of a world? What kind of thing is it? Abstract? Concrete? Unpacking LMR also requires saying what the Lewisian modal realist means by a plurality. How many worlds are there? Two? Six? Seventeen? Infinitely many? Let us consider Lewis’s answers.

What sort of thing is a world? Most philosophers are fairly comfortable talking about what they call the actual world. This world that we inhabit and call actual has as parts human beings, planets, protons, puddles, stars, and donkeys. According to Lewis, our world consists of all of these things as parts, but also has as parts anything spatiotemporally related to these things. Nothing, Lewis says, is so far away from us in space or in time so as not to be a part of this world. All the distant galaxies, protons, and stars are parts of this world. Believers in immanent universals will also want to include universals instantiated by any part of the actual world as themselves parts of the actual world. Believers in abstract objects will also want to count such objects as parts of the actual world.

LMR is the claim that, in addition to the world we inhabit and call actual, there are a number of other worlds. These worlds, according to the Lewisian modal realist, are isolated from the actual world and each from one another. What does this claim amount to? The Isolation Thesis (IT) consists in several claims. First, worlds do not overlap. That is worlds do not share spatiotemporal parts.^{^[304]} Second, worlds are spatiotemporally unrelated. No part of one world stands in any spatiotemporal relations or analogically spatiotemporal relations to a part of another world. A system of relations is analogically spatiotemporal if, and only if, it consists of relations that are (i) natural, (ii) external, (iii) pervasive, and (iv) discriminating. Natural relations are, roughly, those relations the sharing of which makes for similarity. External relations are, roughly, those relations that needn’t be shared by intrinsic duplicates. A system of relations R is pervasive if, and only if, if X and Y stand in some chain of relations in R, then there is a more direct relation in R that holds between X and Y. A system of relations R is discriminating if, and only if, it is at least possible that there be a great number of things related by the relations in R such that no two of these things share exactly the same place in the system of relations. Third, worlds are causally isolated. No event at one world causes any event at any other world. For Lewis, this claim falls out of a counterfactual analysis of causation in terms of possible worlds.

So, worlds are isolated from one another. Worlds are also concrete individuals. Lewis observes that there is no one way of making the distinction between abstracta and concreta precise. He considers a number of different ways of making the distinction including, (i) the way of example, (ii) the way of conflation, (iii) the negative way, and (iv) the way of abstraction. According to each way, Lewis argues, worlds are more like paradigmatic concreta than they are like paradigmatic abstracta. Let us consider the ways briefly. According to the way of example, concrete objects are things like protons, puddles, donkeys, and stars, while abstracta are things like numbers. Lewis observes that, at least some of the parts of worlds literally are protons, donkeys, puddles, and stars, so at least some parts of worlds are paradigmatically concrete. Worlds, for Lewis, are just mereological fusions of some parts, so he is inclined to think that, in virtue of the fact that worlds have concrete parts, worlds are more like concrete objects than they are like abstract objects.

According to the way of conflation, the distinction between concrete and abstract is just the distinction between pure sets and individuals or between numbers and physical objects. Lewis takes worlds to be large individuals: mereological fusions. So, worlds end up looking more concrete than abstract. Again, worlds also seem to inherit concreteness from their concrete parts.

According to the negative way, abstracta are not located in time or space and they are not indiscernible one from another. Lewis thinks that, at least some parts of worlds lack these features. Parts of worlds like donkeys, for example, are located in space and time. Lewis also leaves open the epistemic possibility that there are indiscernible worlds. So, it looks like worlds are more like paradigmatically concrete things than they are like abstracta, according to the negative way.

According to the way of abstraction, abstract objects are abstractions from other objects. So, the abstract object, the property of manhood, say, is just an abstraction away from particular men. We get an abstract object from one that is not abstract by subtracting away some detail or specificity. According to Lewis, worlds are, in an important sense, complete. They are not lacking in any detail. So, even on the way of abstraction (which Lewis regards as the historical source of the concept of an abstract object) worlds are more concrete than abstract.

According to LMR, worlds are plenitudinous. This means, roughly, for any way a world could possibly be, there is a world that is that way. Also, for any way that a part of a world could be, there is a world that has a part that is that way. Lewis rejects these platitudes as genuine principles of plenitude, for he thinks they reduce to triviality. Lewis’s preferred principle of plenitude employs a Humean principle of recombination. According to Lewis a duplicate of any individual can coexist with a duplicate of any other individual, size and shape of spacetime permitting. This principle of recombination, Lewis thinks, suffices to populate his Pluriverse (all the worlds that exist according to LMR) with sufficiently many worlds to accommodate every possible way a world could be and every possible way an individual (or part of a world) could be.

Lewisian modal realism, then, is the view that there exist a plurality of worlds that fit the following characteristics: (i) they are isolated, (ii) they are concrete individuals, and (iii) they are plenitudinous.

In addition to these claims about the nature of worlds, the Lewisian modal realist claims that worlds are fit to play a number of theoretical roles in philosophy. Foremost, perhaps, it their fitness to aid in a reductive analysis of modal concepts like possibility and necessity. Roughly, according to the Lewisian realist’s analysis of these concepts, they are quantifications over possible worlds. So, ‘necessarily p’ is true if, and only if, p is true for every possible world (or true at, or true in, every possible world), and ‘possibly p’ is true, if, and only if, p is true for some possible world (or true at, or true in, some possible world). In addition to providing a reductive analysis of modal concepts, Lewis takes his possible worlds to be fit for a number of other theoretical purposes including, but not limited to, (i) providing analyses of propositional content or content more generally, (ii) providing analyses of causal notions, (iii) providing analyses of concepts like belief, and so on.

For Lewis, propositions are set theoretical constructs out of possible worlds. Roughly, a proposition is the set of worlds at which it is true. Properties, also, are for Lewis set theoretical constructions out of possibilia. The property of being a donkey, for example, is just the set of all the this-worldly and other-worldly donkeys.[305]

3. No Conflict?

Above we have what seem to be two radically different ways of thinking about modality. Lewis and Plantinga seem to have a substantive dispute over the nature of modal reality. Is this appearance deceiving in some way? In this section I will consider whether any reason can be given to think that Lewis and Plantinga are either (i) arguing at cross purposes, or (ii) are arguing past each other.

Recall that two parties are arguing at cross purposes when, due to different antecedent goals, plans, or projects, their dispute ends up being defective. Consider the case of Mary and Larry who are arguing about whether it would be best to take highway 50 or the 101. Mary thinks the plan is to go to the cinema first and then later have a stroll in the park, while Larry thinks the reverse of this. So long as this difference in plans or projects remains unknown to the two parties, any disagreement on the issue of which route is best to take will be defective.

Let us first take up the charge that Plantinga and Lewis are arguing at cross purposes. Lewis and Plantinga seem to disagree about a number of claims:

D1. Possible worlds (including the actual world) are maximal mereological fusions of
concreta.[306]

D2. No concrete individual exists at more than one possible world.

D3. Properties are sets of possible individuals.[307]

D4. Propositions are sets of possible worlds.[308]

D5. An individual, S, possibly has a property P iff S has a (non-identical) counterpart
that has P.

D6. An individual, S, essentially has a property P iff all of S’s (non-identical)
counterparts have P.

Lewis accepts each of D1-D6, while Plantinga rejects them.[309] Perhaps, however, like Mary and Larry from above, Lewis and Plantinga are engaging in different projects or have different goals. Perhaps this difference in goals undermines the legitimacy of their disagreement on D1-D6. What reason could there be for believing this?

A difference between Lewis’s and Plantinga’s projects is not far to seek. As Lewis understands his goal, it is to systematize philosophical theory. As he remarks,

We have only to believe in the vast realm of possibilia, and there we find what we need to advance our endeavors. We find the wherewithal to reduce the diversity of notions we must accept as primitive, and thereby to improve the unity and economy of the theory that is our professional concern—total theory, the whole of what we take to be true.[310]

Lewis is out to reduce the number of primitives in total theory. As part of this project, he seeks to eliminate primitive, unanalyzed modality. He complains, against what he calls “magical ersatzism” that the ersatzer of this stripe requires primitive modality: “Does [magical ersatzism] take modality as primitive?—Yes, overtly and abundantly.”[311] It counts as a black eye for magical ersatzism (Lewis’s unflattering name for Plantinga’s view) that the ersatzer must appeal to unanalyzed modality.

Yet, Plantinga unabashedly appeals to unanalyzed modality. He says,

[n]ote also that we begin with the notions of possibility and actuality for states of affairs. Given this explanation of possible worlds, we couldn’t sensibly go on to explain possibility as inclusion in some possible world, or actuality as inclusion in the actual world; the explanation must go the other way around.[312]

Clearly Plantinga does not regard it as an objection to his view that it requires such primitive modality. This is evidence that, whatever Plantinga takes his project to include, a reductive analysis of modality is not on the cards.

Lewis and Plantinga, then, have a deep and fundamental disagreement, it seems, about the truth of the following claim:

R: It is possible to give a complete and thoroughgoing reductive analysis of alethic
modality, an analysis the analysans of which does not contain any modal concepts.

It is because of their disagreement over the truth of R that Plantinga and Lewis disagree over D1-D6. It is tempting to think that Lewis and Plantinga, then, do not really disagree at all. For we can cast their dispute in a way that highlights their difference of goals or purposes. How might we do this? Let’s start with D1-D6. Even though Plantinga disagrees with Lewis on the truth of these claims, aren’t there related claims on which they can agree? Consider:

D1*: If we want to give a reductive analysis of modality, then it is more reasonable to take possible worlds as maximal mereological fusions than it is to take them as something else.

D2*: If we want to reduce modality to non-modal terms, then we’d be reasonable to consider individuals as being world-bound.

D3*: If we want to reduce unanalyzed primitives, we’d be reasonable to take properties to be sets.

D4*: If we want to reduce unanalyzed primitives, we’d be reasonable to take propositions to be sets.

D5*: Counterpart theory provides a successful reductive analysis of the modal concept of possibly having a property.

D6*: Counterpart theory provides a successful reductive analysis of the modal concept of essentially having a property.

Can’t Plantinga accept all of D1*-D6*?

There is a problem with taking Lewis and Plantinga’s dispute over D1-D6 to be a straightforward case of arguing at cross purposes. Recall that in typical cases of arguing at cross purposes, S and S* are arguing at cross purposes only if there is some project, plan, or goal, G, such that S ascribes to or accepts or has G and S* doesn’t and S and S* are unaware of this difference between them. Once it becomes clear to S and S* that there is a relevant difference in their antecedent plans or goals, they will, if rational, recast their dispute. Take Mary and Larry. Once they learn of their difference in plans, they would, if rational, stop arguing about which route is best to take and instead focus on settling the issue of what their mutual plan ought to be, whether they should indeed have a mutual plan, or whether they should just go their separate ways.

The problem with taking the Plantinga-Lewis dispute to be a case of arguing at cross purposes is that each theorist is aware that there is at least one goal with respect to which they differ. It is a well-known fact that Plantinga does not purport to reduce modality. So what’s going on? There are two explanations. First: Plantinga and Lewis are just being irrational. They are like the following two people:

Jake: We should buy a canoe.

Paula: No, we shouldn’t.

Jake: But if we don’t buy a canoe, how can we go canoeing this weekend?

Paula: Oh, I didn’t want to go canoeing at all this weekend; I thought we were going
hiking.

Jake: Oh, that explains our disagreement about the canoe. Still, we should buy one.

Paula: No, we shouldn’t.

There is something odd about what Jake and Paula are doing here. After acknowledging the disparity in their plans, they go on to renew the original disagreement. Shouldn’t they, if rational, instead resolve the issue of what their joint plan (if there is to be any) should be?

Interpreting the Lewis-Plantiga dispute this way would be grossly uncharitable to both parties. There is another way. Plantinga and Lewis have a deep disagreement about the sober, metaphysical truth of the matter. That is, Lewis really believes in the Pluriverse and Plantinga just cannot bring himself to believe in it. Compare with a revised version of the Jake-Paula dispute:

Jake: We should buy a canoe.

Paula: No, we shouldn’t.

Jake: But if we don’t buy a canoe, how can we go canoeing this weekend?

Paula: Oh, I didn’t want to go canoeing at all this weekend; I thought we were going
hiking.

Jake: Oh, that explains our disagreement about the canoe. Still, we should buy one,
because we like the outdoors, and everyone who likes the outdoors should own a
canoe.

Paula: No, I disagree. Just because we’re naturalists doesn’t mean we should own a
canoe.

Here Jake and Paula begin with an argument at cross purposes, but in the course of their conversation it is revealed that they have a deeper disagreement. Jake thinks naturalists should own canoes, and Paula does not think this.

Compare with Lewis and Plantinga. Plantinga and Lewis, though they may appear to begin with a disagreement at cross purposes, they have a deeper disagreement. Plantinga takes the nature of modal reality to be one way, while Lewis takes it to be another. What could be more fundamental than this disagreement?

Perhaps it is uncharitable to take the Lewis-Plantinga dispute as a simple case of arguing at cross purposes. But it does seem that, to a large extent, they are simply talking past one another. We find Plantinga saying: “Possible worlds are abstract” and Lewis saying, “Possible worlds are not (or at least are not clearly) abstract”. But Plantinga uses the expression ‘possible world’ in the following way: “A possible world, then, is a possible state of affairs,”[313] while Lewis uses it to refer to maximal spatiotemporally interconnected mereological fusions.

At first blush, then, the proposition that Lewis takes himself to be denying when he claims worlds are not abstract is not the same as the proposition that Plantinga takes himself to be affirming when he says worlds are abstract. Similarly, the proposition that Lewis affirms when he says “Individuals are world-bound” is not the same as the proposition that Plantinga denies when he says “Individuals are not world-bound”. There couldn’t be a clearer case of two people arguing past each other than this. So, isn’t their dispute defective?

This is too quick. Consider two fictional philosophers: Al P. and Dave L. Al P. introduces the term ‘possible world’ by making the following speech:

Let me stipulate a meaning for the term ‘possible world’. This meaning is to be divorced from any entity that we take to play a role in our non-philosophical thought and talk. By ‘possible world’, then I have in mind a highly technical term. As I plan to use the term, it will refer to a maximal possible state of affairs.

Now let’s suppose that Dave L. introduces the term ‘possible world’ as follows:

Allow me to make a theoretical neologism. I’m going to coin the term ‘possible world’. As I understand this term, it refers to a maximal spatiotemporally interconnected mereological fusion.

Now imagine Dave L. and Al P. getting into a dispute over the nature of possible worlds. Dave says, “Worlds are not abstract” and Al says, “They are too”. This would be an instance of two philosophers arguing past one another.

Things are different with respect to the state of play between Plantinga and Lewis. They do not take themselves to be introducing a technical vocabulary when they use the term ‘possible world’. Take the following passage from David Lewis:

I believe that there are possible worlds other than the one we happen to inhabit. If an argument is wanted, it is this. It is uncontroversially true that things might have been otherwise than they are. I believe, and so do you, that things could have been different in countless ways. But what does this mean? Ordinary language permits the paraphrase: there are many ways thing could have been besides the way they actually are. On the face of it, this sentence is an existential quantification. It says that there exist many entities of a certain description, to wit ‘ways things could have been’. I believe that things could have been different in countless ways; I believe permissible paraphrases of what I believe; taking the paraphrase at its face value, I therefore believe in the existence of entities that might be called ‘ways things could have been’. I prefer to call them ‘possible worlds’.[314]

This passage makes it clear that Lewis does not take himself to be introducing a piece of technical vocabulary. Rather, Lewis observes that ordinary thought and talk is committed to the existence of some entities that play a certain role in our cognitive economy. He means to use the expression ‘possible worlds’ to refer to those things; things he takes us all to believe in.

Compare with Plantinga who says,

In exploring and explaining the nature of necessity, Leibniz turns to the idea of possible worlds; we can do no better. So we must ask initially what sort of a thing a possible world is. The first and rough answer is that it is a way things could have been; it is a way the world could have been; it is a possible state of affairs of some kind.[315]

Plantinga too acknowledges the currency of the ordinary language expression ‘way things could have been’ and he too chooses to call entities of this sort ‘possible worlds’. However, he thinks the nature of these entities is abstract; they are possible states of affairs.

Both theorists, then, are attempting to characterize the nature of a type of entity that (they claim) we acknowledge in our pretheoretical thought and talk. Compare with talk of natural kinds. Suppose that two biologists want to talk about a certain kind of animal. They both agree that there is an ordinary language expression that serves to pick out such creatures, say ‘tiger’. Theorist A, however, takes these creatures to have a certain nature, say, being feline, while theorist B takes them to have some other nature, say, being bovine. In such a case, it seems that the thing to say about the nature of their disagreement is the following. It is not the case that theorist A and theorist B succeed in specifying different meanings for a pair of technical terms (tiger_a and tiger_b resepectively), rather theorist A and theorist B propose different views about the nature of one and the same creatures.

Something similar seems to be going on with Plantinga and Lewis. Both seek to explain the nature of a class of entities that we mark out in our pretheoretic thought and talk by means of expressions like ‘way things could have been’. Now, perhaps there is an important disanalogy between the case of our biologists and the case of our possible worlds theorists. In the case of the biologists, there is a relevant biological kind that we mean to pick out when we prethoretically use the term ‘tiger’. We mean, roughly, things of that kind. And it seems like nature answers us by supplying a joint in reality, namely a natural kind of things: tigers (whatever their intrinsic nature happens to be). Is the same thing going on with ways things could have been? Is there some joint in reality that our expression can glom onto? If not, then it looks like all Lewis and Plantinga accomplish is to specify two different senses of the term ‘possible world’. When we talk pretheoretically about ways the world could be, are we, in effect, talking about ‘whatever entities make for the truth about our talk of possibility’? This may seem like a stretch.

The extent to which one is inclined to think of our pretheortical use of ‘ways the world might be’ as an attempt to pick out a kind of thing or phenomenon that explains possibility talk will determine the extent to which one is inclined to think that Lewis and Plantinga are theorizing about these entities rather than merely stipulating a meaning for the expression ‘possible world’. If Lewis and Plantinga are doing the latter, then their dispute seems to be straightforwardly a case of arguing past each other; if they are doing the former, then this is not the case.

But there is something further that can be said. Suppose that there is no joint in reality corresponding to our pretheoretical talk of ways things could have been. There are no things such that we mean to be talking about them when we use this expression and its cognates. So, both Plantinga and Lewis are wrong in taking there to be such entities. Ought we to conclude that Plantinga and Lewis are thereby arguing past each other? This would be too hasty. For even if there are no things that we pick out by means of the expression ‘ways things could have been’, Lewis and Plantinga think that there are such things. Compare their dispute with a fictional dispute between McA and McB about the nature of lurches. McA says, and McB agrees, that ordinary thought and talk is committed to the existence of some entities which might be called ‘lurches’ owing to our commitment to the occasional truth of claims like the following:

LURCH: Jones left me in the lurch.

McA thinks that lurches are sets of space-time points while McB thinks that lurches are concreta of some kind.

Clearly McA and McB are doing something silly; they have hypostatized or reified lurches when all the evidence from ordinary usage points to the fact that we do not pretheoretically believe in such things. But, they are not arguing past one another, for they really do take there to be entities of this sort, and they take themselves to be giving a theory about their nature. Perhaps this does not convince. Perhaps McA and McB are better described as just talking nonsense when they say, respectively, “Lurches are abstract” and “Lurches are non-abstract”. After all, they each mean to be using the term ‘lurch’ in the way it is used in ordinary thought and talk. And according to this usage, both the claim that lurches are abstract and the claim that lurches are concrete are nonsense.

Perhaps a different example will be more illuminating. Suppose that two scientists, Dr. A and Dr. B, observe perturbations in the orbit of some planet, X. Dr. A and Dr. B have different views about the nature of the cause of these perturbations. Dr. A takes the cause to be a hitherto unobserved planet, while Dr. B believes the cause is a hitherto unobserved belt of highly magnetic particles. Now, suppose that the observation of the perturbations was a mistake; it was due to faulty equipment and bad calculations. So, there is no phenomenon that explains the perturbations, for there are no perturbations to be explained. Still, Dr. A and Dr. B are not just arguing past each other when each proposes a different theory.

Compare with Lewis and Plantinga. Suppose their observations are fraudulent: we just do not take there to be entities called ‘ways things could have been’. So, there is no need to explain what these things are like. Still, however, Plantinga and Lewis are not arguing past one another when they present their different theories about the nature of these things they believe in.

I have argued that the Plantinga-Lewis dispute is not an instance of an argument at cross purposes and it is not an instance of persons arguing past one another. What does this mean? That the dispute is not a case of arguing at cross purposes entails that the dispute is not rendered defective in virtue of Lewis and Plantinga having different goals or projects. That their dispute is not a case of arguing past each other entails that it is not the case that when Plantinga says ‘Socrates exists at more than one world’ he is affirming a proposition different from the one that Lewis denies when he says “Socrates is world bound”. Haven argued for these two claims, I do not take myself to have shown that the Lewis-Plantinga dispute is not merely verbal. Nor do I take myself to have shown that it is not defective in some other way. In what follows we shall take up the question of whether there is any reason to think Lewis’s dispute with Plantinga is defective in some other way.

4. No Way to Know?

There are good reasons to think that Lewis and Plantinga are not arguing past one another or at cross purposes. But some might claim to detect an air of unresolvability about the transworld identity issue. How could we ever know which (if either) theory is correct? What’s the evidence?

The evidence, as I see it, consists of a number of arguments. Are any of these arguments knock down? Probably not. But are the evidentiary scales tilted even slightly in favor of transworld individuals or against them? This remains to be seen.

What are the arguments? On behalf of world-bound individuals Lewis gives one main argument: The Argument from Contingent Intrinsics. On behalf of transworld individuals Plantinga gives two main arguments: The Relevance Problem and The Argument from Essentialism. Let us look at these in turn.

4.1 The Argument from Contingent Intrinsics

This argument is made very forcefully and compellingly by David Lewis. In short, the argument is this. Suppose, for reductio, that Humphrey exists in more than one world. Now, consider one of Humphrey’s intrinsic properties, say the shape of his left hand the way it is at the actual world. Let’s assume his left hand has five fingers. It is clearly possible that his left hand had six fingers, however. According to the believer in transworld individuals, then, there are two worlds: the actual world where Humphrey has five fingers and some other world, say W, where he has six. According to Lewis:

He could have had six fingers on his left hand. There is some other world that so represents him. We are supposeing now that representation de re works by trans-world identity. So Humphrey, who is part of this world and here has five fingers on the left hand, is also part of some other world and there has six fingers on his left hand.[316]

But this is problematic. For, according to transworld identity, Humphrey’s hand has both five fingers and six fingers, but “[t]hat is double talk and contradiction. Here is the hand. Never mind what else it is part of. How many fingers does it have? What shape is it?”[317]

Schematically, here is the structure of the argument:

1. If the thesis of transworld identity is true, then there are is a world that has Humphrey as a part and where he has five fingers on his left hand and there is a world that has Humphrey as a part and where he has six fingers on his left hand.

2. If there is a world that has Humphrey as a part and where he has five fingers on his left hand and there is a world that has Humphrey as a part and where he has six fingers on his left hand, then it is the case that Humphrey has six fingers on his left hand and it is not the case that Humphrey has six fingers on his left hand.

3. The thesis of transworld identity is false.

Lewis anticipates that the lover of transworld identity will reject 2. Lewis anticipates that the transworld identity theorists will say “though it would be a contradiction to say, simply, that Humphrey had five fingers on the left hand and also that he had not five but six, that is not what was said. He has five at this world; he has six at that world.”[318]

Lewis convincingly argues that introducing the modifiers ‘at this world’ and ‘at that world’ does not help.[319] But rejecting 2 is not Plantinga’s preferred method of reply to the above argument. Plantinga denies 1. Plantinga denies that there is any world that has Humphrey as a part. Worlds, for Plantinga, are abstract simples. They do not have any mereological parts. Lewis’s mistake in giving the above argument, according to Plantinga, is his accepting a certain picture of what worlds are like. Worlds, according to Lewis, are mereological fusions of concrete particulars. So, on this way of thinking about worlds, if there were transworld identity, there would be mereological overlap between worlds. But this is not the correct way to think about worlds. According to Plantinga, to say that Humphrey exists at a world is not to say he is part of it, rather it is to say that were that world actual, Humphrey would have existed. No concrete individual, however, is part of a world. A fortiori, no concrete individual (including Humphrey) is part of two worlds. So, according to Plantinga, premise 1 is false.

4.2 The Relevance Argument

Socrates might have been a carpenter. Had he eschewed philosophy and taken up carpentry instead, then Socrates—the very person who we know and love and use endlessly in examples—would have been a carpenter. But, Lewis thinks that Socrates exists in only one world. There is no world in which Socrates, our Socrates, is a carpenter. Sure there are a number of Socrates-simulcra who are carpenters, but our Socrates is nowhere a carpenter. How is the existence of these Socrates stand-ins even remotely relevant to the truth of whether Socrates himself could have been a carpenter? Surely if Socrates had been a carpenter it would have been Socrates, and not some imposter, who would have been a carpenter. So, Lewis’s view is mistaken in that it supplies as a truth condition for ‘Socrates might have been a carpenter’ some state of affairs that is irrelevant to whether Socrates could have been a carpenter.

Here is the structure of the argument.

1. Socrates could have been a carpenter.

2. If David Lewis’s view is correct, then Socrates has a counterpart that is a carpenter,
but there is no world where Socrates himself is a carpenter.

3. Whether Socrates might have been a carpenter is a matter of how Socrates might have
been, not a matter of how someone else in fact is.

4. David Lewis’s view is incorrect.[320]

Lewis can respond by claiming that this argument begs the question. Premise 3 includes the claim that whether Socrates might have been a carpenter is not a matter of how someone else is. But this is just to deny Lewis his counterpart theoretic analysis of modality de re. According to Lewis, Socrates’ having the modal property possibly being a carpenter just is a matter of his having a counterpart who is a carpenter. So, 3 begs the question.

4.3 The Argument from Essentialism

Plantinga claims that Lewis’s view entails a rather absurd version of essentialism. If Lewis’s view is correct, then each concrete individual exists in one and only one world. Let’s take such an individual: Socrates. Socrates exists in only one world: the actual world. So consider any seemingly contingent property he has: being wise, being a philosopher, corrupting the youth, etc. Since there are no worlds where Socrates lacks any of these properties (for, if there were, then he would have to exist at those worlds), then each of these properties is essential to Socrates. But this is highly counterintuitive.

Here is the argument.

1. If Lewis’s view is correct, then Socrates exists only at the actual world.

2. If Socrates exists only at the actual world, then there is no world where he exists and has different properties from the ones he as at the actual world.

3. If there is no world where he exists and has different properties from the ones he as at the actual world, then Socrates has all of his properties essentially.

4. Socrates does not have all of his properties essentially.

5. Lewis’s view is incorrect.[321]

Lewis’s response is to deny premise 3. To have a property essentially is to be such that all of one’s counterparts have that property. But Socrates is not such that for any property p he has at the actual world, every one of his counterparts has p. So, it is not the case that Socrates has all of his properties essentially. So, 3 is false.

4.4 Where Does this Leave Us?

It looks like we are at a stalemate. There is no argument for which there is not a ready response. Lewis argues that Plantinga’s view of modality, if it makes sense at all, smuggles in primitive modality.[322] But we have seen that this argument does not move Plantinga at all. So, what evidence could there be that could make it more rational to accept world-bound individuals over transworld individuals or vice versa? There doesn’t appear to be any. Both theorists give us complete and consistent packages. Both engage in “difference minimizing”, Lewis playing down his ontology of possibilia by means of the machinery of quantifier domain restriction, and Plantinga playing up the expressive power of his system by insisting on unexemplified essences.[323] So, the dispute over the issue looks epistemically defective; there’s just no evidence that could uniquely determine rational belief here.

There are a few things to say at this point. Some philosophers will take the ontological extravagance of Lewis’s view as a reason to reject it. After all, Lewis holds that there are (really are) talking donkeys. Isn’t this just plain absurd? Perhaps. But this consideration does not tell against Lewis and in favor of Plantings. For Plantinga holds that there are mereologically simple, abstract entities that represent (somehow, we know not how) that there are talking donkeys. Isn’t this just as absurd?

Isn’t Lewis’s theory overall more elegant? His view has fewer unanalyzed primitives, after all. Isn’t this a reason to prefer Lewis’s view? Lewis certainly thought it was. It is very hard to determine how to trade off simplicity and theoretical elegance against other theoretical virtues like parsimony. I do not feel like I have a grip on whether Lewis’s theory wins out due to its having a better balance of theoretical virtues than Plantinga’s or vice versa. But there is something to be said to this line of thinking. Suppose that Lewis succeeds in reducing primitives: he analyzes properties and propositions away. Plantinga objects that the analysans in each case is unfit to do the job. Properties cannot be sets of individuals, Plantinga argues, because sets cannot be exemplified and properties can. Propositions cannot be sets of worlds (or sets of anything else for that matter), for propositions can be true or false and sets cannot.[324] These considerations are compelling. So maybe Lewis cannot claim the theoretical virtues in question.

I am inclined to make a tentative and qualified concession to the argument that Lewis and Plantinga’s dispute is epistemically defective. My concession is tentative for the following reason. I do not have sufficient reason to believe that there is not an argument on behalf of either view that its opponent could not answer. These issues are very complicated., and perhaps there is an argument that no one has yet thought of that tells decisively against one or the other view. I am not sure there isn’t such an argument. Also, I’m not sure that I’ve correctly weighed all the reasons. Lewis argues at great length that so-called ‘magical ersatzism’ is untenable. His arguments are subtle. Perhaps they go wrong or can be answered easily by a supporter of Plantinga’s view; perhaps they cannot.

My concession is qualified for the following reason. Suppose that it is correct that Plantinga’s dispute with Lewis is epistemically defective. This means that there is a correct answer about which theory (if either) is correct, but we cannot ever be in a position to know it. Who cares? Why should this mean that we ought to abandon the dispute? The dispute has born fruit. We’ve gotten clearer about the nature of and relationships between various of our modal concepts. Continuing the dispute and teaching its intricacies to our students is a worthwhile endeavor.

There is an old joke. A guy goes to a psychiatrist and says, 'Doc, my brother's crazy, he thinks he's a chicken,' and the doctor says, 'well why don't you turn him in?' And the guy says, 'I would, but I need the eggs.'[325] Compare the situation of philosophers with the situation of the brother: we have some reason to think Plantinga’s dispute with Lewis is epistemically defective. But if we do, then why don’t we just give up on it? Because we need the eggs. Many philosophical disputes seem epistemically irresolvable. Is this a reason to stop doing philosophy? No more than it’s a reason to stop doing high level mathematics or physics or history or literary criticism. Why do we keep having disputes that, epistemically, seem to go nowhere? Because we need the eggs. Having these disputes can inform us about our world, about our concepts, about interesting distinctions and important arguments. Having the disputes sharpens our minds, hones our acumen, and gives us information we couldn’t get in other ways. So, even if the disputes are defective because they are not resolvable by the available evidence, or even by any possible evidence, this in itself is no reason to abandon them.

Contrast the Plantinga-Lewis dispute with a dispute over whether the number of atoms in the universe is odd or even. In the latter case, there is clearly a fact of the matter, but it seems that there is no way to determine what that fact is. Should theorists hotly dispute the issue of whether the number is odd or even? Clearly not; such a dispute would be silly. But why would it be silly? Not merely because there is no way to know who has it right, but because it is hard to see what good could possibly come out of such a dispute. What kinds of arguments could be given for believing one way rather than another? What sorts of interesting and useful distinctions would need to be made in order to assess these arguments? How could having the dispute inform us about or world, or concepts, or about what makes for good reasoning? How could having this dispute force us to evaluate our theory selection procedures? It seems clear that a dispute over the number of atoms in the universe could do none of these things.

Consider the Plantinga-Lewis dispute. There are a number of interesting arguments that can be made on both sides. Assessing these arguments requires us to make important and interesting distinctions between ways that a world could represent something to be the case and between qualitative and quantitative parsimony. Having this dispute informs us about which possibilities we are pre-theoretically inclined to accept, it informs us about which inferences in modal reasoning we have reasons to endorse, it forces us to evaluate principles of modal plenitude and evaluate what makes for possibility and necessity. Having this dispute and learning about it makes salient the need to establish methods for weighing and trading off theoretical virtues like simplicity, elegance, number of primitives, and parsimony. This dispute, unlike a dispute over whether the number of atoms in the universe is odd or even or how many angels could dance on the head of a pin or how many hairs were on Socrates’ head the moment he drank the hemlock, has borne fruit. That is a reason to continue discussing it, teaching it, and even looking for new arguments to add to it.

Maybe we could get the benefits cheaper, without having this epistemically defective dispute. Perhaps this is true, but the burden of proof lies with the philosopher who thinks we ought to abandon the dispute.

5. No Fact of the Matter?
Perhaps we can never know whether Plantinga’s theory is correct or whether Lewis’s is or whether neither is. A darker counsel of despair tells that we cannot know because there is nothing to be known; there is no fact of the matter. What reasons might be given for thinking that the Plantinga-Lewis dispute over transworld individuals is defective in this way?

One argument mirrors Amy Thomasson’s indictment of physical object ontology.[326] We have already discussed Thomasson’s argument in some detail above, so I will only sketch a version of it leveled against the Plantinga-Lewis dispute.

Both Plantinga and Lewis aim to systematize our pretheoretical thought and talk about ways things could have been. There is a unique answer as to which theorist gets it right only if our linguistic practice with the expression ‘ways things could have been’ is sufficient to determine a unique referent. To figure out the nature of ways things might have been, we need to go to the source. What frame-level application conditions does the use of this term determine? If the usage is indeterminate with respect to frame-level application conditions, then questions like, “What is the nature of a way that things could have been” or “Can one and the same individual exist according to different ways the world might have been?” will be unanswerable questions in the sense that no answer to them given in terms of the question will be truth evaluable. We have no reason to think that our usage of the expression in question suffices to determine unique frame-level application conditions. So, Plantinga and Lewis are trying to answer unanswerable questions.

Where, if anywhere, does this argument go wrong? There are two responses. If it turns out that there are highly natural candidate referents for the expression ‘way things could have been’, then our expression may serve to pick these out independently of our associating clear frame-level application conditions with that term. Our term would glom onto these referents by reference magnetism. If there were Lewisian possible worlds, then these would be highly natural candidate referents for our term ‘way things could have been’.

Now for the second reply. David Lewis’s view requires the truth of

DONKEY: There are (quantifier unrestricted) talking donkeys.

If Lewis has things right, then DONKEY is true. If Plantinga has things right, then DONKEY is false. I am more convinced of the truth of:

EITHERDONKEY: Either there are talking donkeys or there are not talking donkeys,

than I am of any of Thomasson’s claims about frame-level application conditions. Here’s my argument:

EITHERDONKEY is true.
If Lewis’s view is true, then there are talking donkeys.
If Plantinga’s view is true, then there are not talking donkeys.
Either Lewis’s view is false or Plantinga’s view is false.

If one is trying to prove there to be a fact of the matter as to which view, if any, is correct, one could not hope for a conclusion more amenable to this claim than 4. 4 clearly follows validly from 1-3, and the credentials of each of 1-3 are thoroughly unimpeachable. I conclude, therefore, that the above Thomasson-style argument goes wrong somewhere.[327]

It seems clear to me that any argument designed to show that there is not a fact of the matter in the Plantinga-Lewis dispute will fall victim to my counterargument. The no fact of the matter theorist, then, must reject my argument. Perhaps she can claim that Lewis and Plantinga mean different things by ‘there are’, so that my argument equivocates on the use of this phrase. Since this would be to take the line that the Plantinga-Lewis dispute is verbal in some way, I will put off discussion of this reply until the next section. Can the no fact of the matter theorist plausibly reject a premise? I do not see how. 2 and 3 are just straightforward consequences of Lewis’s and Plantinga’s views respectively. 1 is very hard to deny. Any argument that 1 is false, it seems to me, would have to appeal to some premise that is more questionable than 1 itself.

There is one other thing that may be said on behalf of the Plantinga-Lewis dispute. Suppose that we grant there is no fact of the matter. Does this entail that we should give up on the dispute? Yes and no. If there is no fact of the matter, then perhaps we ought to shift to having a debate about the issue rather than a dispute. There being no fact of the matter is a reason to abandon a debate only if that debate does not bear fruit. But, as I’ve argued above, the Plantinga-Lewis debate is fruitful. Contrast the Plantinga-Lewis debate with a no fact of the matter debate in literary criticism. There just seems to be no fact of the matter as to whether or not Sherlock Holmes had a heart-shaped birthmark on his left buttock. Two literary critics who hotly debated this issue would be doing something silly. Why? It is not clear that debating this issue would lead to any theoretical gains that could not be gotten some other, more reasonable way. Perhaps having the debate would motivate a close reading of the Holmes stories which might uncover some interesting facts about those stories, but it is clear that such a benefit could be had without the debate in question. The Plantinga-Lewis debate is different. Engaging in this debate has taught us things about our modal concepts that we could not easily have learned in another context.

Barring, then, the claim that my argument equivocates on the use of ‘there are’, I can see no plausible objection to it. Let us now turn to the charge of equivocation.

6. Merely Verbal?

We saw above that it is not plausible to think of Lewis and Plantinga as arguing past each other. However, perhaps there is some other sense in which their dispute is merely verbal. Eli Hirsch provides a sense of ‘merely verbal’ in which it might be plausible to think that Lewis and Plantinga have a merely verbal dispute. Take our two fictional communities from above: the Wezekana and the Kweli. Let’s suppose that the Wezekana utter sentences orthographically and phonetically identical to the ones that Lewis would utter, while the Kweli utter sentences orthographically and phonetically identical to the ones that Plantinga would utter.[328] How should each community charitably interpret the other?

The Wezekana say,

W1: A world is a mereological fusion.

If the Kweli interpret them disquotationally, not only would the Wezekana come out saying something false, it would be obviously and a priori false. For according to the Kweli, ‘world’ just means ‘maximal possible state of affairs’, and states of affairs, by definition, do not have mereological parts. So, in order to be charitable, the Kweli would have to interpret W1 as meaning, in the Wezekana dialect, something like:

W1*: A physical universe is a mereological fusion.

But now we run into trouble. For the Wezekana also say

W2: There is more than one world.

In order to maintain a charitable interpretation, the Kweli will have to interpret ‘world’ in W2 as meaning something different than it means in W1. Perhaps this is not too wildly implausible, for the Kweli may just think that the Wezekana employ two different senses of the word ‘world’. So, the Kweli might take W2 to mean something like

W2*: There is more than one maximal possible state of affairs.

How might the Wezekana go about interpreting the linguistic behavior of the Kweli? The Kweli say things like:

K1: Worlds are abstract objects.

In order for the Wezekana to interpret K1 so that it comes out true, they will have to interpret the Kweli non-disquotationally as meaning something like:

K1*: Maximal consistent propositions are abstract objects.

K1* comes out true in the Wezekana dialect, for they take propositions to be sets. The Wezekana can interpret the Kweli as meaning by ‘world’ the same that they mean by ‘maximal consistent proposition’. The Kweli say,

K2: One and the same individual exists at more than one world.

The Wezekana can interpret them as meaning something like:

K2*: The existence of one and the same individual is entailed by more than one
maximal consistent proposition.

How about the Wezekana’s claim that:

W3: There are talking donkeys.

The Kweli can interpret this claim as meaning something like:

W3*: There is a maximal possible state of affairs S such that were S actual a talking
donkey would exist.

The Wezekana can interpret the Kweli’s apparent denial of W3 as involving a quantifier restricted to only those things that exist at the actual world.

It seems as though any sentence the Wezekana care to utter can be charitably interpreted by the Kweli, and any sentence the Kweli care to utter can be charitably interpreted by the Wezekana. So, given Hirsch’s criterion of mere verbalness, it follows that Lewis’s dispute with Plantinga is merely verbal.

I have criticized Hirsh’s criterion of mere verbalness above. I think his criterion fails to capture an ordinary language sense of ‘merely verbal’. It is not clear to me that being merely verbal in Hirsch’s sense is a mark against a dispute. Hirschean merely verbal disputes do not strike me as obviously defective.

According to my preferred analysis of merely verbal disputes, it is not plausible to think that Lewis’s dispute with Plantinga is merely verbal. Why not? Recall my preferred analysis:

Plantinga and Lewis satisfy the first two conditions, but fail to satisfy the third. Why? While it is true that Lewis uses the term ‘there are’ in such a way that talking donkeys fall under its extension and Plantinga does not, this fact alone does not explain their dispute. What does explain their dispute? Their dispute arises out of a difference of opinion about whether modality is an analyzable primitive or not. Plantinga thinks the notion of possibility cannot be analyzed in non-modal terms, while Lewis thinks it can be. Why not, then, just take Plantinga to be talking about one sense of ‘possible’ and Lewis to be talking about a different sense of ‘possible’? Given the way that they introduce their views, it is not plausible to think that they diverge on the sense of the word ‘possible’. After all, there is broad agreement between Lewis and Plantinga about what is and what is not possible, they take themselves to be talking about what we pretheoretically mean by ‘possible’, and they both ground their discussion of possibility in the use of ordinary expressions like ‘ways things could have been’.

I therefore conclude that it is not plausible to interpret Lewis and Plantinga as being engaged in a merely verbal dispute.

In light of the above arguments, I see no clear sense in which Plantinga’s dispute with Lewis is silly or otherwise defective. Unlike clear cases of obviously defective disputes, Lewis and Plantinga’s dispute bears theoretical fruit. I have argued that it is implausible to think there is no fact of the matter in the dispute, so there is no need to shift from having a dispute over the issue to having a mere debate. But, even if there were a reason to shift to debating rather than disputing, the resultant debate would still be fruitful. Therefore, I conclude that the Plantinga-Lewis dispute is one that is worth having.

Concluding Remarks: Whither Disputes in Philosophy?

This project has been about philosophical disputes in general. We have looked at two test cases of philosophical disputes: the Universalist-Nihilist dispute and the Plantinga-Lewis dispute. It is fairly easy to lose sight of the forest for the trees, however. In looking at these disputes, we hoped to learn about what makes for a valuable, as opposed to silly or defective, dispute in philosophy.

Many non-philosophers share the opinion that philosophy, to be blunt, is bullshit. According to this way of thinking, there are facts and there are opinions. Intellectual fields that traffic in facts contribute to our understanding of the world, generate meaningful distinctions, new technologies, and make our lives better. Intellectual fields that traffic solely in opinions do not contribute to our understanding of the world, generate new technologies, or make our lives better. Considering opinions can be fun, so goes this popular belief, but at the end of the day those fields that consist entirely in opinion slinging are not productive. You have your opinion; I have mine, what else is there to say? Philosophy is one of those fields that traffics in opinions, not facts, according to this way of thinking.

This attitude about the activity of philosophers is dangerous. It is dangerous for a number of reasons. First, it depends on a flimsy and poorly understood distinction between facts and opinions. Ways of making this distinction precise are non-equivalent and lead to conflicting judgments about what to count as facts and what to count as opinions. The distinction is plainly not helpful.

Secondly, this attitude towards philosophy is dangerous because it misrepresents what philosophy is about. Philosophers are engaged in the practice of inventing and deploying intellectual technologies: distinctions, arguments, inference patterns, practices of theory selection. These things do make our lives better. The activity of philosophers has made us better reasoners, advanced thinking in the sciences, and contributed to moral development.

Another popular belief among non-philosophers is that philosophical questions are questions that do not have answers or do not have knowable answers. Many take this as a reason to think the activity of philosophers is mere word-play, wheels spinning idly, the pretentions of overeducated and underskilled savants. Perhaps this belief about the nature of philosophical questions is true. However, its truth does not support the contention that doing philosophy is without value.

Does God exist? This is a paradigm philosophical question. According to popular belief however, the answer is unknowable.[329] Is this a reason to abandon the question? No. Why not? Think of all the intellectual technology consideration of this question has produced. We have a distinction between two kinds of omnipotence: the power to do anything whatsoever on the one hand, and on the other the power to do only that which is logically possible. Considering Paley’s teleological argument for the existence of God forces us to think about the properties that good arguments by analogy must have and the properties that cogent inferences to the best explanation must have. These are immensely important pieces of intellectual machinery, for we employ such arguments in the sciences where most people believe there are answerable questions.

When ought a philosophical dispute to be abandoned because it is, in some principled way, irresolvable or defective? My answer is when it fails to bear fruit, or when the fruit can be gotten more cheaply some other way. To the best of my knowledge, no opponent of a particular dispute in philosophy has succeeded in making the case that the dispute in question is fruitless.

What is the upshot? Philosophers should be on their guard against defective disputes. We should avoid wasting time in having silly or pointless disputes with each other. Such disputes do not advance our goals and they make us look bad. In times where serious research in the humanities in general and philosophy in particular is being scrutinized by those in a position to fund it, philosophers cannot afford to make themselves look unnecessary. I contend that the best case philosophers can make in favor of having a particular dispute is to argue, not that there is a knowable answer, not that there is a principled fact of the matter, but instead to argue that the dispute can foster the development of useful intellectual technologies.

This way of recasting the question of whether a philosophical dispute is worth perusing, to my mind, avoids a number of difficulties. If first-order disputes in philosophy are verbal, epistemically irresolvable, or metaphysically underdetermined, I find it hard to see how meta-disputes can avoid being defective as well. On my preferred way of gauging the importance and meaningfulness of a philosophical dispute, there needn’t be any facts of the matter as to which meta-philosophical views are correct. My proposed method is a cost-benefit analysis. Can we reap the benefits of having this dispute without wasting our time in its intricacies? If the answer is yes, then the dispute—if defective in any of the above ways—should be abandoned. If the answer is no, then it is a dispute worth having.

REFERENCES

[Insert Your Text Here]

[2] For a good survey of this literature see Richard Feldman and Ted A. Warfield, Disagreement (Oxford University Press, 2010).

[3] For an interesting discussion of arguments relevant to this issue see John MacFarlane, “Relativism and Disagreement,” Philosophical Studies 132, 1 (2007): 17-31.

[4] For a good survey of this so-called “meta-ontological” literature see David Chalmers, David Manley, and Ryan Wasserman, Metametaphysics: New Essays on the Foundations of Ontology (Oxford: Clarendon Press, 2009).

[5] A notable exception is John MacFarlane. In his 2007 and his unpublished manuscript, he takes up the question of exactly what disagreement is. Another notable exception, though not from recent literature, is Charles Stevenson, who takes up the question of disagreement in his Facts and Values (New Haven and London: Yale University Press, 1963).

[6] Not having made up one’s mind is not the same thing as being agnostic, as we will see later.

[7] For a discussion of speaker meaning and sentence meaning, see H. P. Grice, 1975, “Logic and conversation,” in P. Cole & J. Morgan (ed.), Syntax and Semantics, 3: Speech Acts (New York: Academic Press, 1975): 41-58. Reprinted in H. P. Grice (ed.), Studies in the Way of Words (Cambridge, MA: Harvard University Press,1989): 22-40.

[8] As opposed to degrees of belief. Below I will discuss differences in degrees of belief as a kind of disagreement.

[9] Indifference to p, like not having made up one’s mind about whether p, is not the same as suspension of belief with respect to p

[10] Cappelen and Hawthorne 2009, 60-1. John MacFarlane cites Cappelen and Hawthorne in connection with this observation in his unpublished manuscript “Varieties of Disagreement” given as a talk at University College Dublin and the Institute of Philosophy in London in May 2009 and available at http://johnmacfarlane.net/varieties.pdf.

[12] Perhaps my writing these words is sufficient for my disagreeing with all of the conspiracy theorists, taken together, in the activity sense of ‘to disagree’, but it seems plausible to me that I’d have to do more, e.g., single out a particular theorist, in order to count as disagreeing in the activity sense with that person.

[13] Indeed, if they are rational, it seems like after they make second of the above speeches, they should stop activity-disagreeing until they’ve sorted out whether there is anything at all about which they state-disagree.

[14] Intuitively, actors do not count as activity-disagreeing with one another because neither believes that the other is asserting what she is saying. But, whether assertions are being made in performance is a controversial topic in the philosophy of fiction and performance. I hope to skirt it for the moment, though I will return to the issue of whether one can activity disagree with someone who is not making assertions below.

[16] Some might want to call this “pretending” to hold the position in question; this sounds wrong to my ear.

[18] Even people who are just plain disagreeable have to be such that they are disposed to disagree with someone or something.

[19] We’ll talk about the sorts of commitment that are involved here, but I mean ‘commitment’ in a very loose sense here. In the sense I have in mind, to be regimented later, one can commit to a content, say the proposition that p, even if they merely suggest that p or assume that p. One needn’t assert that p in order to be committed to p in this very loose sense.

[20] MacFarlane makes this point in his 2009 as well as Stevenson in his 1963.

[21] Stevenson calls such disagreements “disagreements in interest” (Stevenson 1963, 26).

[22] I here say ‘otherwise inconsistent’ so as to include the possibility of propositions that do not entail a logical contradiction, but are nevertheless incompatible, perhaps in the following sense. Practical incompatibility: p and q are practically incompatible iff it would be impractical (imprudent) for a person to act on the belief that (p & q).

[23] Though they might not know that they disagree.

[24] The restriction to persons here is crucial. Suppose I vociferously object to the proposition that gruyere is tasty. It would be comical to remark “Gerald is having a disagreement with the proposition that gruyere is tasty.” Why would this be comical? When ‘disagrees’ is taken in the activity sense, “S disagrees with x’ implies ‘S is having a disagreement with x’ only if x is a person or suitable person-like creature.

[25] Why is this incorrect? I take it that McCain fails to count as having a disagreement with Bill because McCain’s comments are not appropriately directed at Bill. This is a sticky issue, for McCain might intend his remarks to be directed as whoever is listening to his speech. We will explore this issue in some detail in what follows, but I am inclined to think this is not enough for McCain to count as having a disagreement with Bill. McCain must single Bill out in some way in order to count as having a disagreement with him.

[26] Compare with the locution ‘are fighting’. My spouse and I can be such that “My spouse and I are fighting” is true of us even though she is not currently throwing plates, curses, or anything else at me. In fact, she can be in the next room (perhaps plotting what deprecation she plans to assail me with next) and we can count as being engaged in the activity of fighting, albeit in a dispositional, non-occurrent sense.

[27] And this is not because it is always obvious what the disagreement is about.

[30] For an account of the sort of rejection of a proposition that Neg has in mind, see Mark Richard’s 2008 When Truth Gives Out (Oxford: Oxford University Press, 2008).

[31] Doubtless, some philosophers will probably try to find such a proposition. Let them propose one, then we’ll see if we can recast the case so that Peggy and Neggy still disagree, but they do not disagree about that proposition.

[32] Both MacFarlane and Stevenson also hold this view.

[33] Thanks to Margaret Walker for helpful discussion here.

[35] I do not mean to suggest that this is what MacFarlane means by ‘reject’.

[36] Or, at least, they are not obviously directed at propositions.

[41] Though, of course, she may like Bill in one respect and not like him in another without inconsistence. What she cannot do consistently, however, is like him all things considered and not like him all things considered.

[42] This example is MacFarlane’s. MacFarlane also worries about the fact that there are some who would say Amy and Ari have distinct attitudes: Amy’s being wanting that Amy eat the cupcake and Ari’s being wanting that Ari eat the cupcake. A discussion of the content of desires would bring me too far afield here. If something like this view of desire content turns out to be right, I will need a different example, but for our purposes the above should suffice.

[45] I am here allowing that there can be disagreements between people that do not involve some attitude towards a proposition, but instead involve different and incompatible ways of representing the world in thought. Theorists who do not take there to be such non-propositional ways of representing can ignore this sort of accuracy.

[49] I have in mind here so-called cognitivists about emotions like Robert Solomon and others.

[50] These are very similar to Austin’s A and B conditions and G conditions.

[51] Austin identified one use of the verb ‘to challenge’ as what he called a Behabitive illocutionary act.

[52] I say ‘this type’ because I want to leave it open that a dispute between A who thinks that ‘Bill is bald’ has no truth value and B who think it does have a truth value may not be defective. Disputes are individuated in terms of the attitude types of the disputants. In a token dispute where one disputant affirms p and another affirms not-p, the dispute type is picked out in terms of the propositional attitude types belief that p and belief that not-p.

[53] Carnap apparently thought this, and Chalmers (2009) defends this view as well.

[54] It might turn out that the dispute over Holmes’ sexuality is epistemically defective and not a no fact of the matter dispute. I take it that the Holmes-mole dispute is, however, a no fact of the matter dispute and not an epistemically defective dispute.

[55] Karen Bennet (2009) argues for such a view about certain disputes in ontology.

[56] Provided that we keep fixed the amount and distribution of hair on Bill’s head.

[57] Whether they are wrong because what they have asserted is false (as a Russellian about definite descriptions would have it) or because what they say is not truth apt (as a Strawsonian about definite descriptions would say) I do not want to attempt to settle here.

[58] I am here rejecting the view that emptiness is an all or nothing affair and allowing for partial emptiness.

[59] Eli Hirsch holds a version of this view, claiming that a merely verbal dispute is one in which both sides speak the truth in their respective languages, “Each side can plausibly interpret the other side as speaking a language in which the latter’s asserted sentences are true.” (Hirsch 2009, 231).

[61] William James “What Pragmatism Means” in The Writing of William James: A Comprehensive Edition, ed. John J. McDermott (New York: Random House, 1968): 376-390, 376.

[63] David Hume Dialogues Concerning Natural Religion in Introduction to Philosophy: Classical and Contemporary Readings fourth edition, ed. John Perry, Michael Bratman, and John Martin Fischer (Oxford: Oxford University Press, 2007) 91-124, 122

[64] Carrie Jenkins also makes this observation in her unpublished work on verbal disputes.

[65] David Chalmers, unpublished manuscript “Verbal Disputes and Philosophical Progress”.

[66] Theodore Sider “Quantifiers and Temporal Ontology” Mind 115 (2006): 75-97, 76. I do not mean to attribute the Equivocation Model to Sider, for he prefaces his comments by saying that an apparent dispute is merely verbal when it meets the conditions specified by the Equivocation Model. Perhaps Sider, then, means the Equivocation Model only to apply to apparent disputes and not disputes simpliciter.

[67] Chalmers “Verbal Disputes and Philosophical Progress” unpublished manuscript.

[69] Indeed, given the S5 modal system, what McA says is also necessarily false, given that round squares (in his sense) are possible.

[70] Is it? Some philosophers might find it a welcome result that merely verbal disputes are less widespread than they might appear.

[71] This is, perhaps, a bit too crude. Suppose I think that ‘murder’ means ‘unlawful killing’ and it turns out that Smith unlawfully and immorally killed Jones and I know this. Even if I had different metalinguistic beliefs, say I thought that ‘murder’ meant ‘immoral killing’, presumably I would not change my mind about Smith’s being a murderer. What we need is some finer-grained way to assess a change in mind. For now let’s say that a metalinguistic belief influences my position just in case were I not to have that belief, I’d make some sort of different judgments in some possible cases. Perhaps this is too general to help us, but I trust that some intuitive sense of ‘influence’ can be pressed into service here.

[72] Carrie Jenkins “Merely Verbal Disputes” unpublished manuscript.

[81] Hirsch 2002, 2004, 2005, 2007, 2008, 2009.

[82] This example is a modified version of an example form Hirsch’s 2005 paper.

[83] Each principle should be read as having an implicit ceteris paribus clause.

[84] The original case comes from Hirsch 2005, 69.

[87] This reply was suggested to me by Eli Hirsch in conversation (though he is not responsible for any distortion of the reply engendered by my statement of it).

[88] Hirsch has indicated in conversation that he holds something like this view.

[89] Or, if you prefer, each party believes something that, in the imagined public language associated with their position, would be contradictory. It is not clear to me, however, that the Burge point applies to ‘omnipotent’, so I am perfectly happy to say that each party here really does believe a contradiction. Not too much will turn on this, however, for wide scope rationality prohibits believing contradictions and the proposition that the other side speaks the truth in their imagined public language is a contradiction, so wide scope rationality would prohibit believing it.

[90] Hirsch counts such disputes as merely verbal (e.g., disputes over whether a borderline-bald person is bald or not), but such disputes do not seem merely verbal to me.

[91] I do not wish to represent this view as Chalmers’ considered view of verbal disputes, but only as one that he considers and tries to motivate in his unpublished manuscript, “Verbal Disputes and Philosophical Progress”.

[100] Thomasson, Hirsch, Sidelle, and Bennett give such arguments, just to name a few.

[101] Peter Van Inwagen popularized this term in his 1990 book Material Beings.

[103] This is, of course, on the assumption that human persons, if they existed, would be mereologically complex things, an assumption that is pretty hard to deny. Van Inwagen’s own view is what has been called Organicism, the view that the only simples that compose something are those simples whose collective activity constitutes a life. So, on Van Inwagen’s view, while there are no mountains, chairs, tables, stones, or the like, still there are people, badgers, plants, and other living organisms. Since I am going to restrict my examples to alleged composites that are inanimate, I will ignore this complexity and treat Van Inwagen as though he accepts Nihilism.

[104] The ‘simples arranged F-wise’ locution is Van Inwagen’s.

[105] This is a very Quinean approach to ontological commitment which some theorists would reject.

[106] The ‘and only these’ qualification is important, for on some non-Universalist views, there is an object that has all these things as parts, namely the world. The Universalist acknowledges that there is a world that has everything as a proper part, but holds the view that there is another thing that has just these scattered objects as parts.

[107] Most noteworthy amongst those views left out here are: mereological essentialism (defended by Roderick Chisholm), sortal dominance theory (defended by Michael Burke), composition as identity (defended by Donald Baxter), monism (a version of which is defended by Terry Horgan and Matjaz Potrc), and four dimensionalism (defended by Ted Sider and David Lewis). For an excellent discussion of some views on composition, see Ted Sider’s chapter “Constitution” from Sider and Conee’s 2007 book Riddles of Existence: a Guided Tour of Metaphysics.

[108] Karen Bennett argues that certain ontological questions are epistemically defective in this way in her 2009 paper “Composition, Colocation, and Metaontology”.

[109] Provided these rival answers do not place the number of licks ridiculously low, like at, say just one.

[110] Those not convinced of this can compare the SCQ to the Special Atomic Question: Is the number of atoms in the Universe odd or even?

[111] This example comes from Amie Thomasson’s 2009 paper “Answerable and Unanswerable Questions”.

[112] Thomasson 2009, 445. She also argues for a related thesis, that Nihilism is an attempt to answer unanswerable questions, in her 2007.

[114] By ‘truth evaluable’ I am taking Thomasson to mean what some other philosophers have meant by ‘truth apt’, that is, having a truth value (either true or false).

[115] Thomasson intends her conclusion to apply with greater generality; she purports to show that disputes among theorists who hold any view on mereological composition arise out of attempts to answer unanswerable questions.

[119] Devitt 1981, 62, emphasis in original.

[129] See Devitt 1981 for more on multiple groundings or ‘anchorings’ of a nominative term.

[132] Peter V. can formulate his claim as: EPV: $x$y[x ≠ y & "z(z = x Ú z = y)], while David L. can formulate his claim as:EDL: $x$y$z[x ≠ y & x ≠ z & y ≠ z & "u(u = x Ú u = y Ú u = z)]

[133] She responds to this objection in section 6 of her 2009 paper.

[135] Sider makes this argument in his 2009 paper, “Ontological Realism”.

[137] This way of talking may seem to commit one to a view that reifies facts. One can rephrase the above in such a way as to avoid this commitment: When Jones is sitting, it is like when Smith is sitting, where the ‘it’ here is like the ‘it’ in “It is raining”. Sider makes this point in fn. 43 of his 2009.

[138] Sider 2009, 415. Sider gives explicit instructions on how to affect the shift to speaking Ontologese.

[142] Or, if it does, then Thomasson’s methodology renders disputes over the question either merely verbal or non-philosophical.

[143] Ted Sider 2010 (unpublished manuscript) “Against Parthood”.

[145] Carnap famously argued that something like this is the only way to make good sense out of ontological disputes, see his “Empiricism, Semantics, and Ontology”.

[146] For some explanation of the phenomenon of higher order vagueness, see Michael Tye 1994. and 1990..

[148] If you think this is implausible, adjust the case so that the disputed claim is “Holmes had a hairy mole on his back”.

[150] That is, assuming that ‘half empty’ is not an incoherent phrase.

[151] For discussion see Armstrong’s 2004 Truth and Truthmakers.

[152] There are divergent views about whether truthmakers must entail or necessitate the truths they make true. I hope to avoid taking a stance on this issue.

[153] I do not mean to suggest that Quine himself would have been committed to such a view.

[157] These formulations are taken from Sidelle 2002, 119.

[159] At least, once the package views are thoroughly worked out.

[160] For the sorts of problem cases that are relevant, see Sider’s 2001 Four Dimensionalism.

[162] Sidelle2002, 134. I have quoted this passage at such length because I think that Sidelle’s remarks here are very clear and are representative of the views of a lot of philosophers who think that disputes over material object ontology is misguided in some way.

[166] Chalmers unpublished discusses the phenomenon of vocabulary exhaustion.

[171] These rules of use are very similar to what Thomasson has in mind by “frame-level application and coapplication conditions”.

[177] There is some good reason to think that Sider himself does not suffer from the misunderstanding I attribute to him, for he says, “I do not pretend to refute Carnap, only to show that the opposing view, that meaningful ontological disagreement is possible, is coherent.” Perhaps this remark is Sider’s way of acknowledging that Carnap’s view is merely a proposal, but I cannot be sure. To be safe, imagine a fictional theorist named, Ted Scmider, who takes his objection to be a refutation of Carnap.

[178] To be fair, Chalmers does not explicitly endorse a neo-Carnapian view; rather, he defends such a view against objections and tries to make it seem plausible. I do not know whether Yablo also demurs when it comes to endorsing the neo-Carnapian view that he sets out to defend. In what follows, we can imagine every occurrence of ‘Chalmers’ and ‘Yablo’ replaced by names for fictional philosophers who wholeheartedly endorse the neo-Carnapian views defended by Chalmers and Yablo.

[181] Yablo 259-260. It bears mention here that Yablo’s Quine is the “’popular’ pre-late 1960’s Quine: the one who wrote ‘A logistical approach to the ontological problem’, ‘On what there is’ (ignoring the ontological relativism), ‘Two dogmas of empiricism’, ‘On Carnap’s views on ontology’, and Word and Object (ignoring the ontological relativity).” Yablo 1998, fn 3.

[192] I am talking pretty fast and loose here; there are, perhaps, important distinctions to make between pretense and make believe, but I gloss over these for they do not affect the point I wish to make.

[198] Timothy Williamson convincingly argues that knowledge is “the norm of assertion” in his 2002.

[199] For a good survey of some possible views see Hawthorne 2004.

[200] For further articulation and defense of this position with respect to knowledge attributions see Stewart Cohen, 1988, 1998, 2001 (and many others).

[201] There are a number of ways to make this view precise; I do not here wish to commit to a particular view.

[202] Or for accepting the view that the proposition expressed by an existence statement is sensitive to features of the conversational context.

[203] Chalmers allows for a view he calls lightweight realism that is not committed to the intelligibility of absolute quantification. According to lightweight realists, ontological questions have answers, but these answers are fairly trivial.

[205] Chalmers 2009, 91. Chalmers is open to the idea that there is a meta-concept or a pseudo-concept in the vicinity (Chalmers 2009, 91-92).

[209] I do not mean to claim that Chalmers himself endorses such an argument.

[220] It is, for example, an epistemic possibility that the world does not contain any simples, that matter is ‘gunky’: composed of ever smaller particles ad infinitum. If this is the case, then it would be a consideration that could tell against certain forms of Nihilism.

[222] Russell from The Problems of Philosophy.

[238] Edward Zalta 2010 Stanford Encyclopedia of Philosophy “Frege's Logic, Theorem, and Foundations for Arithmetic”.

[240] Zoltán Gendler Szabó 2007, Stanford Encyclopedia of Philosophy “Compositionality”.

[242] This version of the argument is borrowed from Eklund 2008, 388.

[244] At least, according to Eklund’s neo-Fregean.

[246] As we will see below, ‘mythiscism’ is probably a better name.

[247] One might think that if this is the case, then such a character cannot be in the extension of the predicate ‘is bearded’, for only flesh and blood persons (and other suitably barbate creatures). If you think this, then introduce a predicate ‘mbearded’ that a mythical creature has just in case, according to the relevant myth or myths they have a beard.

[249] There is worry that, as an empirical matter of fact the name ‘Zeus’ was not introduced in the way I’ve suggested, as a name for whatever person causes thunder and lightning. If this turns out to be the case, then imagine a civilization of people called Greecks who did introduce the name ‘Schmeus’ in the way I suggest and interpret the above to be about this community and their use of ‘Schmeus’.

[250] Eklund makes this point about recasting the Theism-Atheism debate in this way. Eklund 2006, 114-115.

[251] Indeed, the Nihilist would go on to say, “and this is so in part because there is no book, no pencil, and no cup.”

[252] Hirsch mentions Sider, Chisholm, Cartwright, Thomson, Lewis, Shoemaker, Unger, Sosa, Van Inwagen, and Van Cleve in this connection (2009, 233 fn 3).

[253] In making this claim I am assuming a certain measure of intellectual honesty is shared by all parties to the dispute.

[259] Hirsch uses such an example in his 2004 comments on Sider’s Four Dimensionalism.

[261] An expression is quantifier-like just in case it obeys the standard inference rules for ‘$’ in the predicate calculus.

[267] Sider 2009, 405. Even though the similarity is cast in terms of facts, “the similarity judgments in question don’t really require reifying facts. As Jason Turner pointed out, one might express such judgments thus: ‘when something is human, it’s like when something is located in North America’; the ‘it’ here is like the ‘it’ in ‘It is raining’” (Sider 2009, 206 fn 43).

[268] See Nelson Goodman 1979, Fact, Fiction, and Forecast.

[272] Much of this section repeats arguments I make in my forthcoming.

[273] It might be argued that Hirsch should reject the notion of a logical kind as unintelligible. I would say rather that the concept is, to date, under-described and under-motivated. In his 2008, Hirsch claims that disputes about logical kinds are related to, and as merely verbal as, the ontologists’ disputes about existence. In light of this claim, I see no reason why the Hirschese could not interpret the Siderish in terms of the notion of a logical kind. Thanks to Eli Hirsch here.

[274] In my above conciliatory interpretations, I have assumed that, if there are logical kinds, there is only one joint-carving, EXISTENCE-like logical kind. This assumption might be rejected, however. Those who reject this assumption may read ‘if there are logical kinds’ as ‘if there is a unique EXISTENCE-like logical kind’ throughout.

[275] Thanks to Eli Hirsch and an anonymous referee for the Australasian Journal of Philosophy for raising this issue.

[276] Thanks to an anonymous referee for making this point.

[279] For comments on earlier versions of this section, many thanks to John Devlin, Eli Hirsch, Ted Sider, Angel Pinillos, Richard Creath, Michelle Saint, and Michael Gifford.

[281] It is not clear whether Sidelle thinks there are no facts of the matter because the dispute is verbal.

[283] Sidelle 2007, 87. Sidelle points out that he realizes “that some people attaché a stronger metaphysical content to the notion of “layers” than” he intends. It is worth noting at least this, however: even given this limited metaphysical content, Sidelle’s view of reality as layered entails that Nihilism is false, at least, if we take his claim “Objects…have parts, both large and small” seriously. I do not intend to quibble about this point, however.

[284] I borrow this description of what it takes to be a statue from Thomasson 2007. Much else is probably required, but this seems a good enough description for our purposes.

[289] Sidelle explicitly states that the above “is not meant as a sufficient condition” on verbal disputes (2007, n. 15)

[291] Sidelle 2007, 91. Sidelle is somewhat cagey here, so a couple of points need mentioning. First: this quotation comes from a discussion of an example of an allegedly verbal dispute over whether bats are birds (here I’ve replaced “bird” with “material object’ accordingly). Secondly, Sidelle does not explicitly avow or disavow the view that the above is a sufficient condition on verbal disputes. It is fairly natural to read him as though he would endorse such a condition as sufficient, but he does not commit himself. Once can imagine a fictional philosopher, Albert Slidelle, who does think this is a sufficient condition. Those who wish to be extra careful, then, can interpret my above comments as being about Slidelle’s view.

[292] Recall: one student says “God couldn’t do it”, but means by ‘omnipotent’ (something like) ‘could do anything, even what is logically impossible’, whereas the other student says, “God could do it’ and means by ‘omnipotent ‘could do anything logically possible’.

[294] The subjects are related, of course, and there may be an interesting common subject matter that lies at the heart of their dispute. For example, Mary and Larry might really disagree about which kind of omnipotence is more interesting or important, or about which kind of God it would make more sense to worship (why worship a creature who is bound by what is logically possible? That would be a weak God not fit to be an object of adulation), etc. Thanks to Alan Sidelle for this point.

[295] Since Sidelle nowhere explicitly commits to the view that the two conditions in (1) are sufficient for a dispute to be verbal, I represent him as arguing for the claim that it is reasonable to believe that the dispute is verbal. I take it, however, that even if this weaker conclusion goes through, it will be a bitter pill for the Nihilists and the Universalists to swallow.

[296] Sidelle 2002. Of course, this is only a dialectical problem for Sidelle, other theorists not committed to the no fact of the matter view would not have to deal with this objection.

[297] Hirsch suggests this view and sympathizes with it, Hirsch 2009.

[298] See especially Plantinga’s The Nature of Necessity, “Two Concepts of Modality: Modal Realism and Modal Reductionism” and Lewis’s On the Plurality of Worlds and “Counterpart Theory and Quantified Modal Logic”.

[303] Plantinga is open, however, to the view that propositions and states of affairs are one and the same (Plantinga 1974, p. 45).

[304] This claim is consistent with the claim that some abstracta, e.g., numbers, may be said to exist at more than one world.

[305] Lewis does have the resources to provide other set theoretical constructions to capture various roles that properties and propositions are said to play (e.g., structured propositions and structured properties) (Lewis 1986, chapter 1, esp. sections 1.4 and 1.5).

[306] With the notable exception of sets, immanent universals, and other abstracta, such there be.

[307] Or some other suitable set theoretic construction out of possible individuals; perhaps sets of individual-world pairs.

[308] Or some other suitable set theoretic construction out of possible individuals.

[309] This list of disputed claims is not intended as exhaustive.

[320] Kripke gave a similar objection in his Naming and Necessity, the famous ‘Humphrey objection’.

[321] Plantinga gives a cluster of arguments of this sort. Each is designed to show that if Lewis is right, then some absurdity follows. Among the absurdities Plantinga claims follow from Lewis’s view are that ‘Socrates is foolish’ is necessarily false and that ‘Socrates exists’ entails every true proposition (Plantinga 1974, 115).

[322] See Lewis 1986, section 3.4 “Magical Ersatzism”.

[323] Recall Karen Bennett’s arguments from her 2009 that focus on difference minimization.

[324] See “World and Essence” for these arguments.

[327] Thomasson nowhere gives such an argument.

[328] This requires a slight modification of my above story, for according to my story the Wezekana use a word ‘werld’ that is orthographically distinct from the one Lewis employs, while the Kweli use a word, ‘wurld’ that is orthographically distinct from the one that Plantinga uses.

[329] Or at least, not knowable by reason alone unaided by faith in scripture or spiritual revelation.