April 2008

April 2008

THE SET-THEORETIC VIEW OF FICTION

by

Mark Barber Syracuse University for

Special Paper Requirement

1. Introduction1 Literary fictions are incomplete. No fiction specifies all of the possible details of the story—aside from being infinitely long, a work that did give every detail of the story would be unbearably boring. To compensate for this incompleteness, we must use our imaginations to fill in the gaps. Without our ability to imagine,2 fiction would simply not be understandable; imagination is indispensable to understanding fiction. We often fill in the gaps without being aware that we are doing so, but we do it nonetheless. Because of the close relation between fiction and imagination, imagination will play a role in any theory of fiction.3 The theory that I defend in this paper is a theory of fiction; the theory is called ‘The Set-Theoretic View of Fiction’ (or, ‘the Set View’ for short). The Set View provides an account of what kind of entity a fiction is, what is true in a fiction, and it outlines the role imagination plays in understanding fiction.4 In the Sections that follow, it should become clear how resourceful the Set View is; it not only stands up to philosophical objections but also is consistent with a

1

Strictly speaking, the Set View is a theory of literary fiction as it doesn’t address issues of visual fiction.

2

I use the term ‘imagination’, but I could use ‘make-believe’ or another similar term. The role to be played by imagination will not preclude any theory of imagination or make-believe.

3

A theory of fiction is not limited to a theory of truth in fiction. Nor is it limited to a theory of what fictions are. A theory of fiction will also include a theory of how we understand fictions—this is where imagination is indispensable.

4

I am not giving a theory of imagination. Instead I identify the role that imagination plays in understanding fiction and the reader is free to insert her or his favourite theory of imagination. I see no reason to think that the theory of fiction precludes any particular theory of imagination.

2

number of our strongest intuitions about fiction. In this paper I will both elucidate the Set View and respond to objections and answer questions about the theory.

2. Motivation for the Set View First, if ‘Holmes smokes a pipe’, is true in the Sherlock Holmes mysteries, then we need a theory to tell us why this is true. The Set View provides a needed alternative to the philosophical theories of fiction.5 Second, the Set View is not just a theory of what is true in a fiction, but also a theory of what fictions are. Too often philosophical accounts of truth in fiction are divorced from a theory of what fictions are—after all, if we are aiming to give a theory of truth in fiction, it might be good to know what the truths are supposed to be in. Third, since fictions are incomplete, a good theory of fiction will explain how we understand fictions. As presented in this paper the Set View outlines the first steps in a theory of understanding fiction. When developing a theory, it is essential that the theory accommodate the data. In the case of fictions, the data are our intuitions, manuscripts authors write, and how we interact with these manuscripts. This is a very unstable data set, but the Set View does a surprisingly good job of accommodating the data.6 That the Set View

5

I will not be arguing explicitly against other theories of fiction in this paper. However I do find all that I am aware of lacking.

6

With data such as this, we should not expect our theory explaining the data to be mathematically precise.

3

accommodates the data so well is good reason, says David Lewis, to think that it’s true.7

3. Imagination Every actual fiction is incomplete. No fiction gives us every detail of every event and character in the story and so we fill in the blanks with the use of our imagination. Some of the ways that we fill in the blanks are more reasonable than others; for example, it would be completely unreasonable to fill in the blanks about Hamlet’s past by saying that Hamlet played professional basketball. Whereas it would not be unreasonable to believe that Hamlet’s birth was widely announced. In either of these cases, we needed to do something in order to fill out the unspecified details of the story; what we did was use our imaginations. Imagination is fundamental to the proper functioning of fiction; for that matter, most of philosophy would not be possible if we didn’t have our imaginations. There are many theories of what imagination is and the Set View is neutral regarding which theory is true. Any theory of imagination8 is sufficient to play the role that the Set View calls it to play and I think that everyone understands what imagination is and how we use our imaginations well enough to understand the role that imagination plays in the Set View. So, I will not try to present any theory of imagination; instead I take a basic

7

Lewis 1986 (p. 3).

8

That is, any theory of imagination that I’m aware of or that I can reasonably imagine.

4

understanding for granted and let the reader fill in her or his favourite theory of imagination as appropriate. Just as imagination plays a fundamental role in our understanding fiction, it also plays a large explanatory role in the Set View. Some might think that a theory that relies so heavily on the imagination is flawed; but, since fiction relies so heavily on the imagination for its functioning, it would be wrong to leave imagination completely out of a theory of fiction.9 The primary role of imagination in the Set View is that it enables us to understand fictions, but exactly how important imagination is to the explanatory power of the Set View will become more apparent as this paper proceeds.

4. The Set View On Works of Fiction The first question that any theory of fiction must answer is: What is a fiction?10 The Set View gives a short answer to this question: Propositional Setism (henceforth, ‘Setism’). According to Setism a work of fiction is the set of propositions that are communicated by the sentences in the manuscript that is created by the author.11, 12, 13

9

I think it would even be wrong to minimize the role that imagination does in fact play.

10

I’m using the term ‘fiction’ to refer to things like A Study in Scarlet, The Pickwick Papers, and The Book of Job.

11

The communicates relation is explained in section 5.

12

One of the things that Setism doesn’t do—so far—is distinguish between fiction and non-fiction. I will not tackle this problem here; being a set of propositions is a necessary condition for being a fiction, I’ve given no sufficient conditions. If there are sufficient conditions for being a fiction, then we’ll have specified what the fiction/non-fiction distinction is; I have no reason to think that it could not be done but that is a project for another paper.

5

One of the things that we like about fictions is that they help us imagine a whole world of activity, characters in the most intricate detail, and they help us vicariously experience situations that we would not actually have the opportunity to experience. The fullness of fiction is one of the things that we like about it. According to the Set View, a fiction is a set of propositions; so one thing that the Set View must explain is how a set of propositions can create such a vivid image. The answer to this is relatively simple and it is best explained by an analogy. The set of propositions—which is the fiction—is like a blueprint for our imagination. When we read the sentences that communicate these propositions, we are able to fill in the gaps and thus create a vivid imagining. This is an everyday practice; take the following case, for example: The room is small with cement walls painted white. Two fluorescent lights are hung from the ceiling and there is a red door on the south side of the room that leads to the common area. These two sentences communicate propositions; if a reader understands the sentences, they might very well imagine a room with cement walls that are painted white, and so on. But this passage is far from a complete description of the room. For example, it doesn’t say how many walls there are, how clean the walls are, what shape the lights are, and so on. But most readers, if asked about what they imagined, would describe a particular type of room, a particular type of lights, and a particular type of door. One

13

There is a debate in the philosophy of literature about whether ‘the author’ refers to the actual author or a fictional author. The Set View is compatible with each of these positions. In either case, there will be a set of propositions that are communicated by the manuscript. Which set it is, however will depend on whether we are talking about the actual or fictional author.

6

reader might have imagined a four-walled room with new white paint and round hanging lights, while another might have imagined a round room with scribbling on the walls. In either case, the truth of my explanation of how a set of propositions can lead to such a vivid picture has been confirmed. There are objections to the Set View. I will consider objections in sections 7 and 8.2 of this paper. Before then, something must more precise must be said about the propositions that are members of the set— which is the fiction.

5. Propositions and the Communicates Relation One of the concepts that is vital to Setism, and thus the Set View, is the communicates relation. A sentence communicates a proposition and a proposition is communicated by a sentence. So the communicates relation obtains between sentences and propositions, but this does not tell us what the communicates relation is. Before trying to define the communicates relation, let us go on a small tangent. There is a distinction to be made between what is said and what is communicated. François Recanati14 argues that there are three steps from the sentence to what is communicated. We start with a sentence, say (1): (1) I went to the bank today. Sentence (1) has a meaning; this meaning combined with “contextual ingredients” gets us what is said. Sentence (1) contains at least three words that require contextual ingredients; they are: ‘I’, ‘today’, and ‘bank’. Both ‘I’ and ‘today’ are indexicals; some 14

Recanati 1989 (pp. 106-7). Grice famously argues for this as well.

7

indexicals are time-sensitive (like ‘today’, ‘yesterday’, and ‘now’), others are speakersensitive (like ‘I’, ‘you’, and ‘we’), and others still are location-sensitive (like ‘here’ and ‘there’). When I say that indexicals are ‘sensitive’, I mean that the reference of the indexical will change depending on who utters the indexical, when it is uttered, or where it is uttered. So, if (1) was uttered by Nixon on June 17th 1962, the sentence would mean something like Nixon went to the bank on June 17th 1962; whereas, if I uttered the sentence on July 1st 2004, the meaning would be different. The word ‘bank’ also requires contextual ingredients in order to get what is said. The reason is that ‘bank’ has more than one meaning. Let us say that there are only two possible meanings for ‘bank’ in sentence (1): ‘financial institution’ or ‘the sloped land at the edge of a river’. The contextual ingredients for (1) will fix the meaning of ‘bank’. So the contextual ingredients combined with the sentence meaning get us what is said. We get what is communicated from a combination of what is said with what is conversationally implicated. Sometimes there is nothing that is conversationally implicated; in these cases what is said is what is communicated; but, in other cases, there is conversational implicature, so what is said is not ‘what is communicated. Sentence (2) is uttered by a mother whose son has just cut his finger. (2) You are not going to die. 15 About this sentence Recanati says: This is interpreted in context as: You’re not going to die from that cut. The restriction ‘from that cut’ results from free enrichment. The sentence

15

Bach 1994 (p. 135) & Recanati 2002 (p. 301).

8

itself conveys no such thing—it simply says that the child is not going to die.16 The enriching of (2) is free because it is not linguistically controlled by the sentence in any way. So what is said is very often enriched such that what is communicated is different than what is said.17 We can now get back to defining the communicates relation. If we take a proposition to be what is communicated, then: the communicates relation=df the relation between sentences and propositions; sentences combined with contextual ingredients yield what is said, which is then enriched by conversational implicature to yield the proposition that is communicated. Given the heavy dependence on the distinction between what is said and what is communicated, the Set View must be supplemented with some rules governing the implicature. Grice’s maxims seem like the obvious candidate. However, there are other theories that will work; for example, those of Kent Bach and François Recanati. Given the aim of generality, I want to withhold commitment to a particular set of maxims governing implicature. The Set View can be supplemented with any number of maxims governing implicature. Depending on the choice of maxims, the Set View

16

Recanati 2002 (p. 301).

17

I will not argue for this further here. For objections to this account of the difference between what is said and what is communicated, see Bach 1994a and 1994b. Although Bach objects to Recanati’s explicit position, he also argues for a difference between what is said and what is communicated. He primarily argues against the existence of unarticulated constituents.

9

might yield slightly different answers in particular cases. What is true in a fiction, then, will depend on which version of the Set View is adopted.18 As defined, the communicates relation holds between sentences and propositions, however there is an extension of this relation that holds between collections of sentences (manuscripts) and sets of propositions (fictions). So, just as a proposition can contain an unarticulated constituent—which isn’t the content of any expression in the sentence—a set of propositions might contain an unarticulated proposition—which isn’t the content of any sentence in the manuscript—that is the result of a form of free enrichment.19

6. Ontological Commitments Most theories of what a fiction is make significant ontological commitments. For example, some commit us to the existence of a new ontological category of things— namely, objects of social discourse20—others commit us—if we weren’t already

18

There is also a debate in the philosophy of language about whether the contextual ingredients are parts of the proposition that is communicated. On the one hand, philosophers like Recanati 2002 and Perry 1998 say that there are unarticulated constituents that contain this information and are parts of the proposition. On the other hand, philosophers like Bach 1994a & 1994b say that the proposition communicated has no such constituent and that the conversational impliciture plays a role only in taking us from the literal meaning to what is communicated. The Set View also does not take sides on this debate, either view can be made to work.

19

I don’t intend to say anything about the nature of the unarticulated proposition. If this suggestion—of the unarticulated proposition—is palatable, then it may provide solutions to some problems with the Set View’s account of Truth in Fiction. See Section 7.

20

Thomasson 1999.

10

committed for other reasons—to possible worlds,21 and others still commit us to objects for which we do not have a good metaphysic.22 One way to avoid gratuitous additions to our ontology is to explain what a fiction is in terms of things that we already committed to. I assume that we are already committed to the existence of both sets and propositions and since the Set View explains fictions in terms of sets and propositions, it avoids additional ontological commitments—that is, commitments over and above the ones we already have to sets and propositions.

7. Questions and Objections for Setism We are now ready to answer some questions and respond to objections to Setism.

Objection: An author produces a manuscript; call it ‘manuscript A’. The author sends the manuscript off to a publisher who returns the manuscript with the message “We will publish this if you change the passage in Chapter Five where the protagonist runs into an oak tree; you should change it to a sycamore tree.”23 Since this request does not change the story in any significant way, the author complies with the request. After making the change, the author sends the publisher the modified version; call it ‘manuscript B’. Manuscript A and manuscript B are not identical, since they do not

21

Lewis 1978.

22

In Van Inwagen 2003 (pp. 149-50), Van Inwagen claims that although his argument implies that fictional characters exist, it implies little about what fictional characters are like.

23

Perhaps the reason for the change is that the primary audience of the fiction is more familiar with sycamore trees than with oak trees.

11

contain the same sentences. Also, the sentences in the manuscripts do not communicate the same propositions; so, according to the Set View, they are not the same fiction. This is a problem since anyone who reads both manuscript A and manuscript B will say that they are the same work.

Response: This might seem like a problem for the Set View; however, the intuition that manuscript A and B are the same is very easily explained. It is true that the Set View is committed to the claim that fiction communicated by manuscript A is not the same as the fiction communicated by manuscript B. Although this seems counterintuitive, most people will at least acknowledge that there is one difference between the fictions communicated by manuscripts A and B—namely, the type of tree that the protagonist runs into in Chapter Five is not the same. If the reader argues that the two works are exactly the same despite this change, then we should just modify the example to make it the case that two things are changed. If this still is not enough, then perhaps three changes will suffice. We have a typical Sorites paradox on our hands and, clearly, if we change every detail, then A and B aren’t the same work; but any point at which the reader wants to claim that “now the works are not the same” will simply be an arbitrary cut-off. So A ≠ B is the only non-arbitrary cutoff (otherwise A = B = C = … ).24 Hence there is a very good sense in which the two manuscripts with one minor difference communicate different fictions. 24

There is another non-arbitrary cut-off; namely, for any work A and any work B, A and B are the same work. However, this answer doesn’t even do justice to the intuition that The Illiad and The Idiot are different works. That’s absurd!

12

With this said, there is also a very good sense in which manuscript A and manuscript B do tell the same story. There really is something to this intuition. The Set View also has an explanation for why this is so. As we know, imagination is crucial to understanding fiction. Without it, we would read a number of sentences that communicate a certain set of propositions, but we would not be able to understand the story. The one vice—or perhaps virtue—of our imagination is that it sometimes individuates stories only coarsely. Because of this, various fictions fall into classes within which our imagination does not normally differentiate. This fact also explains why the scary campfire stories that I was told when I was a child are reasonably said to be the same as the stories that I can tell to children today. Even though the words are different, the blueprints are close enough to being the same that the imaginative picture that they create is, for all intents the same.

Objection: In his “Truth in Fiction,”25 David Lewis raises a case where there are two works, both named Don Quixote, one by Cervantes and the other by Menard. The manuscripts for each of the works are indistinguishable. Many people think that they are not the same work; but, if the manuscripts are the same, then isn’t the Set View committed to the claim that they are in fact the same work? If so, then this is very counterintuitive for some people.

25

Lewis 1978 (p. 40).

13

Response: According to the Set View, a manuscript is a collection of sentences and a work is a set of propositions. Sentences and propositions are related by the communicates relation; similarly, manuscripts and works are related by the communicates relation. Don Quixote as written by Cervantes would be the same work as Don Quixote as written by Menard, except for one fact: namely, sameness of sentences does not imply sameness of propositions. As was described in Section 5, sentences and contextual ingredients comprise what is said; and what is said is enriched by conversational implicature that yields what is communicated. There is a sentence s that is found in Don Quixote by Cervantes and Don Quixote by Menard. Perhaps what is said by s, call it ‘P1’, is the same in both Don Quixotes. However, what is relevant is whether what is communicated by s is the same. If what is communicated by s in Cervantes’s Don Quixote is P2; yet P3 is communicated in Menard’s Don Quixote, then Cervante’s and Menard’s Don Quixote’s are not identical since they communicate different propositions. The difference between the conversational implicatures is Menard’s Don Quixote was written as a commentary piece or perhaps as irony. The different intentions might find themselves in the conversational implicature and hence different propositions are communicated. Therefore Don Quixote by Cervantes is not the same set of propositions as Don Quixote by Menard; thus they are not the same works.

14

Objection: Some people26 claim that, if there is one intuition that a theory of fiction must conform to, then it must conform to the intuition that authors create works of fiction. If propositions are the type of thing that always exist, then sets of propositions have always existed; if fictions are sets of propositions, then have fictions always existed. So the Set View precludes the possibility of fictions being created; perhaps authors discover fictions, but they do not create them. Therefore, the Set View isn’t the correct theory of fiction.

Response: First, I acknowledge that many people do have the very strong intuition that authors create fictions, but what I do not understand is why our theory of fiction must align itself so strongly with this one intuition. It seems that, if we put enough effort into trying to explain how it is metaphysically possible that fictions can be created and we are still not able to do it, then we should abandon the intuition. Even though this might be the case, the Set View does have a response to the creatability objection. One reply would be to deny that propositions always exist. In this case, the argument for the conclusion that fictions have always existed would have a false premise. A second reply would be deny that there was the set of propositions before the author wrote the manuscript. Or, alternatively, that the propositions that are elements of the sets didn’t exist before the manuscript. If either of these denials is Metaphysically acceptable or the Set View can explain why we have the intuition, then the objection will have been answered. 26

Thomasson 1999.

15

Assuming that both of the premises are true, authors do not create sets of propositions. However, they do create something: namely, manuscripts. There is no disputing this. According to the Set View, the sentences in the manuscript communicate the propositions that comprise the set that is the fiction. Although there is no point in time when the set of propositions comes into existence, there is a point in time when the manuscript comes into existence (after all, a manuscript is a concrete object). At the moment when the manuscript comes into existence, the set of propositions—i.e. the fiction—gains a new property. The fiction gains the (extrinsic) property being communicated by the manuscript written by S. So, although this is not creation ex nihilo, there is a sense in which the actions of the author did affect the set of properties that is the fiction. But does this really count as creation? I don’t think so. After all, this is no more creation than “Cambridge change” is actually change. If this is what happens, then maybe it can explain why we have the creation intuition but, strictly speaking, it is not creation. Finally, another way of answering the question would follow Harry Deutsch in “The Creation Problem.” In this paper, Deutsch argues that there is a sense of ‘creation’ that does not entail coming into existence. Creation in this sense would be more like creative discovery. Although a proponent of the Set View could take Deutsch’s suggestion, I’m not so inclined given the objection presented in “The Destruction Problem” by Barber and Caplan.

16

Question: According to the Set View, what are fictions that are within other fictions?27 Answer: Since fictions are sets of propositions that are communicated by the manuscript, a fiction that is within a fiction is also a set of propositions that are communicated by the manuscript. So the fiction within a fiction is a subset of the original fiction.28 A similar objection will be discussed in section 7.2 but in relation to truth in fiction.

8. The Set View On Truth in Fiction 8.1. The Theory When we read a fiction, there are some things that are unquestionably true in the fiction. For example, it is true in Lord of the Rings that Frodo is from The Shire; it is also true in the fiction that Frodo is a hobbit, that Frodo has hairy feet, and that he is a friend of Sam Gamgee. Why are these things true in the fiction? Because Frodo is said to be from The Shire in the work, he is said to be a hobbit, the property of having hairy feet is attributed to him in the work, and there are many sentences about Frodo and Sam’s friendship throughout the work. So people agree that these things are true in the work, but what about claims like ‘Frodo prefers green beans to yellow beans’, ‘Frodo secretly desires to fly’, or ‘Frodo prefers trotting on a horse to cantering’? There are three answers that can be given to these questions; the propositions might be true, they might be false, or they might have no truth-value (i.e. they might be neither true 27

This objection is a very serious problem for some other theories of fiction, hence the inclusion.

28

This response purports to cause problems for what is true in a fiction; this problem is dealt with in Section 8.3.

17

nor false). A good theory of fiction ought to tell us which truth-value should be assigned to each proposition. The Set View has such an answer. According to the Set View, a proposition is true in a fiction iff the proposition is in the work. The claim that Gandalf is a wizard is true in the fiction because it is in the work of fiction—i.e. the proposition communicated by ‘Gandalf is a wizard’ is a member of the set that is the fiction. Before responding to objections, something should be said about what it would mean for a proposition to be either false in a fiction or have no truth value in a fiction. A claim is false in a fiction iff it is the negation of a proposition that is in the work. The following is false in Lord of the Rings: it is not the case that Frodo is the nephew of Bilbo Baggins. This is false precisely because the work claims that Frodo is the nephew of Bilbo. A proposition has no truth-value in a fiction iff (1) it is not in the work and (2) is not the negation of a proposition in the work. So, if P = a proposition and W = a work, then: P is true in W=df P ∈ W P is false in W=df ¬P ∈ W • P ∉ W P is neither true nor false in W=df P ∉ W • ¬P ∉ W.

8.2. Explicit vs. Derivative Truth in a Fiction

18

David Davies introduces a distinction between Explicit and Derivative truth in a fiction.29 The account of truth in a fiction outlined above is a theory of what is explicitly true in a fiction. According to Davies, there is also something called Derivative truth in fiction. We can define what it is for some proposition P to be derivatively true in a work of fiction W in the following way: P is derivatively true in W=df P is entailed by propositions that are explicitly true in W or P is entailed by propositions that are explicitly true in W and background assumptions made by the author. The primary reason for adopting derivative truth in a work is to accommodate some intuitions that we have about works of fiction. So, it is well motivated, but the adoption is not without some serious costs; in fact, the costs are so serious that we should reject the notion of derivative truth all together. So, the Set View rejects derivative truth. Nearly all readers of Hamlet think that Hamlet had a determinate number of hairs on his head—it is probably only philosophers that explicitly think about this, but everyone else would assent to the truth of the proposition. Furthermore, we’d think someone crazy who didn’t think that Hamlet had a determinate number of hairs on his head, that’s how strong the intuition is. That ‘Hamlet has a determinate number of hairs on his head’ is not explicitly true in the fiction but it is derivatively true, it follows from what is communicated by the sentences in the fiction and some very reasonable

29

Davies p.53.

19

background assumptions. So far so good, adopting derivative truth seems to have benefits. One problem with derivative truth becomes obvious when we consider inconsistent fictions. Let’s say that a fiction W is inconsistent iff either propositions (p and ¬p)30 or (p & ¬p) are communicated by the fiction. If there are derivatively true propositions in W, then in either case, p & ¬p is entailed by the propositions in the fiction. Then, it quickly follows that everything is entailed by the fiction and consequently, every possible proposition q is true in the fiction. However, that is not what happens with inconsistent fictions, we certainly don’t think that just because there is an inconsistency in a fiction that everything is true. This is the first problem with the notion of derivative truth. The second problem with derivative truth comes with the other disjunct in the definition—namely, ‘P is entailed by propositions that are explicitly true in W and background assumptions made by the author.’ Suppose that p is communicated by a work W. Suppose that there are two possible sets of background beliefs A and B. A and B are non-identical but are equally reasonable. A and p jointly entail q while B and p entail ¬q. Imagine that I am arguing with my wife over whether of not q is true in W. She says it is and I say it’s not. Who is right? Are both q and ¬q derivatively true? If so, then we have the problem noted above, namely that everything is true in the fiction. If not, then we need a reason for ruling either q or ¬q out, but by hypothesis, A

30

The ‘and’ in ‘p and ¬p’ is the English ‘and.’ I use it to make explicit that the contradictory propositions need not be conjoined in a work in order for that work to be contradictory, but they may be.

20

and B are equally good background assumptions and so there is no reason for ruling either q or ¬q out.31 Derivative truth is deeply problematic and so we should reject the notion. Despite being problematic, the notion of derivative truth is headed in the right direction. In the first objection of the next section, the Set View employs imagination to do, more or less, the work that derivative truth was aiming to do.

8.3. Objections and Questions Objection: If the only propositions that are true in a fiction are those that are in the work, then what about things that seem very obviously entailed by things in the fiction? For example, the Sherlock Holmes stories might say that Holmes is a man but not that he has some blood type. It seems to be the case that being a man implies having some blood type, but the work doesn’t contain the proposition that Sherlock Holmes has blood; so, according to the Set View, it is not true in the fiction that Sherlock Holmes has a blood type. This just seems wrong.32

Response: This is what the Set View claims. The Set View also claims that it is not true in the fiction that Sherlock Holmes has a brain—provided that the work doesn’t

31

There is a dispute in the literature over whether ‘the author’ refers to the actual author or to a fictional author. The given the nature of the objection, I do not need to take sides on this debate.

32

It could be that Holmes has some blood type is implicated by the sentences in the work, but I think that this is unlikely since blood types weren’t discovered until 1901.

21

contain a proposition that attributes this property to Holmes. Although this may seem strange, the Set View does have an account of why we have contrary intuitions. One reason this seems odd is that we often assume that, if something is not true, then it is false; but when it comes to fictions this is not the case—and in fact, it shouldn’t be the case given inconsistent fictions. If something is not true, then it is either false or it lacks a truth value. So just because it’s not true in the fiction that Holmes has a brain, it doesn’t follow that it’s false that Holmes has a brain. As we know, our imagination plays a vital role in our understanding fiction and, as I said in Section 5, this set of propositions—i.e. the fiction—provides us with a blueprint for an imaginative world. Since the propositions do not provide us with a complete description of the world that we imagine, we fill in the gaps in a way that allows us to best understand the blueprint. If this is true, we have the resources to respond to the objection. One of the claims made about Holmes in the fiction is that he is a man; but, as with all fictions, we are not given a complete description of him. One thing we might think is entailed by Holmes being a man is that he has a blood type and a brain but if the fiction does not say anything about whether or not Holmes has a blood brain, then according to the Set View it is neither true nor false that Holmes has either a blood type or a brain. When we read a work of fiction we read a number of sentences that communicate certain propositions. So again, these propositions form an incomplete blueprint; this blueprint can be filled out in many different ways. We generally fill the 22

gaps so that they fit with reasonable background beliefs and in a way that allows us to best understand or appreciate the work. In the case of Holmes, we are told that he is a man; and, since we normally think that all men have both blood types and brains, we fill out the blueprint so that Holmes has both of these. But this doesn’t seem necessary. To see why, suppose that Conan Doyle had added sentences that communicated the propositions that Holmes does not have a brain and Holmes does not have blood; instead he has a substitute called ‘blain’ that does the task of both. Since Conan Doyle could have done this and hardly changed our understanding of the work at all, we can infer that, just because the work claims that Holmes is a man, that doesn’t entail that Holmes also must have a blood type and a brain. So the reason that we have the strong intuition that it is true in the fiction that Holmes has both a brain and blood type is that it seems to be the most natural way to fill in the framework created by the set of propositions—i.e. the fiction. Although this is (likely) the most natural way to read the stories, there is nothing that says we must read the stories in that way. So, it is not true in the fiction.

Objection: If a proposition is not mentioned in a fiction and neither is its negation, then that proposition is neither true nor false in the fiction. This seems to deprive the Set View of the resources necessary for distinguishing the following cases suggested by André Gallois. Case 1: We agree that since Shakespeare conveys nothing about the exact number of lady Macbeth’s children, it is neither true nor false that she has exactly two. 23

Case 2: I write a follow up to Macbeth. In it I add this: ‘Lady Macbeth is an alien from a parallel universe. So alien is she that it is neither true nor false that she has exactly two children.’ According to the Set View, in case 1 it is neither true nor false that Lady Macbeth has exactly two children. In case 2, it is also neither true nor false that Lady Macbeth has exactly two children.

Response: The Set View can distinguish these cases. The objection turns on a scope ambiguity. In case 1 and 2, the Set View says: Case 1—It is neither true nor false that in the fiction Lady Macbeth has exactly two children. Case 2—It is true that in the fiction it is neither true nor false that Lady Macbeth has exactly two children. The difference between the cases should be clear. The Set View tells us whether or not some proposition is true, false, or neither true nor false in the fiction. It does not tell us whether or not some proposition is true, false, or neither true nor false, full stop. So, the Set View can distinguish these cases.

Objection: If we ask the author about some unmentioned detail in his work and he gives an answer, isn’t that then true in the fiction? For example, if someone asked Conan Doyle whether or not Holmes’s second toe is longer than his first and he said that it was, then isn’t that true in the fiction even though that proposition is not a member of the set?

24

Response: According to the Set View, the answer is no; it would not be true in the fiction that Sherlock Holmes’s second toe is longer than his first. Once again, I think that an explanation of the intuition is available on the Set View. Many people think that the author’s intentions are important to a work of fiction and the Set View takes this into account. As we know (from Section 3), what is included in the propositions that make up the set that is the fiction is partly determined by the sentences written by the author and partly a function of the context within which they are written. One thing that is sometimes a part of the context is the author’s intentions. If an author writes a sentence in her manuscript that says “Jennifer is flagging,” the proposition that is communicated is (partially) determined by which meaning of the word ‘flagging’ the author intended.33 So intentions are important to fiction. But they are important only inasmuch as they are instrumental to which propositions are actually communicated by the sentences. The author, just like the reader, takes her work and fills in the gaps in a certain way; but there is one respect in which the author has an advantage over the reader. The advantage is that (hopefully) the author knows what her intentions were when she wrote the sentences that compose the manuscript; because of this, she will know more accurately which propositions are communicated by those sentences. If, for example, the author intended to communicate that Jennifer was marking the trail, then she would likely know; and, if she intended to communicate that Jennifer was getting

33

There are contexts wherein all of the non-intentional contextual elements are insufficient to fix the intension of a term.

25

tired, then hopefully she would know this also. So the author does have this one advantage; but, when it is questions about details that are in no way part of the fiction, the author has no more say than the reader.34

Question: An inconsistent fiction is one that has inconsistent propositions. For example, if the Holmes stories claimed that Holmes always has an even number of hairs on his head and that he always has an odd number of hairs on his head, then the Holmes stories would be inconsistent. There are inconsistent fictions. Can the Set View accommodate and explain inconsistent fictions?

Answer: Yes it can. According to the Set View, what it means for a proposition to be true in the fiction is for that proposition to be a member of the set that is the fiction. Since these sets are not logically closed, there is no reason that a fiction cannot contain both the propositions P and ¬P. If the fiction contains both of these propositions as members, then both of those propositions are true in the fiction. Sometimes, inconsistencies happen because the author was careless in editing; and, other times, they happen by design. Regardless of the cause, we seem to be able to understand such fictions; because of this, our theory of fiction should give us some account of why this is the case. Again it is our imagination that provides us with the account. If the inconsistency is a result of careless editing, like Watson’s single wound 34

This advantage will only hold if we think it is the intentions of the actual author that count. If we think that it is a fictional author that matters, then the actual author will not be at an advantage in the way suggested.

26

that is apparently on both his shoulder and his leg, then any person is liable to build one of two imaginative images: one that accommodates each of the inconsistencies. Very often, designed inconsistencies work as clues to the reader that other sentences in the manuscript might not communicate the propositions that we initially thought they did. In either case, we do have a way to make sense of these inconsistencies; and the Set View can accommodate the inconsistent propositions all being true in the fiction. This is a major advantage of the Set View over theories like Lewis’s, which cannot account for inconsistencies. According to Lewis’s theory35, the way to determine what is true in a fiction is to look at the closest possible world to ours where the story is told as truth. To answer questions like “Does Sherlock Holmes have a heart?,” we should look to the closest world where there is a detective who lives at 221b Baker St., who smokes a pipe, who solves all of his cases, and so on, and see if that detective has a heart. If he does, then it is true in the fiction that Sherlock Holmes has a heart. Questions like “Where is Watson’s war wound?” are not so easy to answer on Lewis’s theory. If we employ Lewis’s suggested technique, then we will look to the closest world where the Holmes stories are told as truth. In the Holmes stories we are told that Watson’s war wound is located on his shoulder and another time we are told that it is located on his leg. The problem is that there is no world at which it is true that the war wound is multiply located. Since possible worlds cannot contain contradictions and fictions can contain contradictions, Lewis’s account of truth in 35

Lewis 1978 & 1983.

27

fiction is unable to account for this feature of fiction.36 That the Set View is able to account for contradictions is a major benefit of the theory.

Objection: According to the Set View, fictions within fictions are subsets of the original fiction; and a claim is true in a fiction iff that claim is a member of the set of propositions. Imagine a fiction that tells an elaborate story of a struggling soccer team that works hard but never succeeds in winning. After nearly a whole season of losing, the team finally works together and wins their first game. One sentence in the fiction is: (a) The team had finally won a game. When the fiction is nearly finished the narrator (who is also the main character in the fiction) says: (b) Then I woke up and discovered it was all a dream; my soccer team never won. When we read (b), we realize that everything that we just read was a fiction, so there is a fiction within a fiction. If the Set View is true, then (a) is true in the fiction because it is a member of the set of propositions that is the fiction. But we don’t think that (a) is true in the fiction because (a) was a dream and (b) alerts us to this fact. Since the Set View says that (a) is true in the fiction and we know that (a) is not true in the fiction, the Set View is false.

36

In Lewis 1983 (“Postscripts”) Lewis considers this as a major drawback of his theory; he does not have a satisfactory solution to the problem.

28

Response: It is true that, according to the Set View a claim is true in a fiction iff it is a member of the set of propositions that is the fiction. The reason (a) is not true in the fiction is that the proposition normally communicated by (a) is not a member of the set. Which propositions are communicated by the sentences in the manuscript depends on the context that the sentences fall within. The rest of the manuscript is part of the context that (a) falls within; so the proposition that (a) communicates depends (at least partially) on the rest of the manuscript. Initially we think (a) communicates the proposition (P1), (P1) The team won a game. but it doesn’t. When we read (b), we should realize that (a) doesn’t communicate (P1). Instead (a) communicates something like (P2), (P2) In the dream, the team won a game. which is true in the fiction. The Set View is not false, because the objection rested on a false assumption: namely, (a) communicates proposition (P1). This strategy for understanding fictions within fictions will accommodate fictions within fictions within fictions also.37

Question: We often participate in inter-fictional discourse. We make claims like “Sherlock Holmes is smarter than Inspector Clouseau.” Not only do we make these claims; we also think that they are true. Can the Set View account for this?

37

In fact, there is no reason to think that it would no work with any number of fictions within fictions. Thanks to Ben Caplan for a very helpful discussion of this objection.

29

Answer: The standard way of cashing out this claim—namely, Holmes is smarter than Clouseau—is to say, it is true in the fiction that Holmes is n degrees smart and it is true in the fiction that Clouseau is m degrees smart. Since n > m, we conclude that Holmes is smarter than Clouseau. The problem with the Set View using this response is that, unless it is true in the fiction that Clousseau is m degrees smart, then we don’t have a basis of comparison. Furthermore, even if it is true in the fiction that Clousseau is m degrees smart and it is true in the fiction that Holmes is n degrees smart, we don’t have any good basis for comparing m to n since there is no cross-fiction metric.38 The standard way of cashing these claims out is simply not available to the Set View; in fact, I doubt there is any account the Set View can give that has claims like these turning out literally true. However the Set View can provide an explanation for why we think that claims like these are true. When most people read the stories about Sherlock Holmes, they imagine that Holmes is a very smart detective who solves all of his cases; and when most people read the stories about Inspector Clouseau they imagine that Clouseau is rather dim witted. Once we imagine certain characters in certain ways, we are able to compare them to other characters that we have imagined. So in the case of Holmes and Clouseau, the comparison yields the claim that Holmes is smarter than Clouseau, even though it is just what we imagined.

38

I doubt that this is a problem only for the Set View. Lewis’s view, for example, might also face this objection. According to Lewis, something is true in a fiction iff it is true at a world where the fiction could be asserted truly. It might turn out that there is no world for some pairs of fictions in which case there is no answer to be given regarding the truth of these inter-fictional comparisons.

30

9. Conclusion Fictions are sets of propositions; this is the position taken by the Set View. A claim being true in a fiction depends on that claim being a member of the set that is the fiction. The success of any theory depends on its ability to explain the data. In the case of fiction, the data here are people’s intuitions and as I previously noted, this is not the most stable base, but we must do the best with what we have. Although the Set View might not fit with naïve intuitions about what a fiction is—which aren’t very strong intuitions anyway—, it provides us with explanations of the problems that arise from fiction. At the beginning of this paper, I said the reason we should think the Set View is true is that it is serviceable. By showing that the costs associated with the Set View are outweighed by the benefits, I have demonstrated that it is a serviceable theory. This is a good reason to think that it is true.

31

BIBLIOGRAPHY

Bach, Kent. 1994a. “Conversational Impliciture.” Mind & Language 9:2 (June) pp. 12462. ——. 1994b. “Semantic Slack: What is Said and More.” In Foundations of Speech Act Theory. S. Tsohatzidis (ed.). London: Routledge. pp. 267-91. Barber,

Mark

&

Ben

Caplan.

(Unpublished).

“The

Destruction

Problem.”

.

Davies, David. 2007. Aesthetics and Literature. New York: Continuum. Deutsch, Harry. 1991. “The Creation Problem.” Topoi 10: pp. 209-25. Grice, Paul. 1967. “Logic and Conversation” William James Lectures. Reprinted in A.P. Martinich The Philosophy of Language: 2nd Edition. pp. 149-160. (Page references are to the latter.) King, Jeffrey C. 2001. “Structured Propositions.” The Stanford Encyclopedia of Philosophy. http://plato.stanford.edu/entries/propositions-structured/. Lewis, David. 1991. Parts of Classes. Oxford: Basil Blackwell. ——. 1986. On the Plurality of Worlds. Oxford: Basil Blackwell. ——. 1983. Philosophical Papers Vol. I. New York: Oxford University Press ——. 1978. “Truth in Fiction.” American Philosophical Quarterly 15:1 pp. 37-46. Reprinted with “Postscripts” in Lewis 1983. pp. 261-80 Perry, John. 1998. “Indexicals, Contexts, and Unarticulated Constituents.” In 32

Computing Natural Language. D. Westerthal, et al. (eds.). Stanford: CSLI Publications. pp. 1-11. Recanatai, François. 2002. “Unarticulated Constituents.” Linguistics and Philosophy 25 pp. 299-345. ——. 1989. “The Pragmatics of What is Said.” Mind & Language 4 pp. 294-328. Reprinted in Davis, Steven (ed.). 1991. Pragmatics: A Reader. New York: Oxford University Press. pp. 97-120. All citations are from the later. Sider, Theodore. 1997. “The Paradox of the Question.” Analysis 57:2 (April) pp. 97101. Thomasson, Amie L. 1999. Fiction and Metaphysics. Cambridge: Cambridge University Press. Van Inwagen, Peter. 2003. “Existence, Ontological Commitment, and Fictional Entities.” In The Oxford Handbook of Metaphysics. Loux, Michael & Dean Zimmerman (eds.) New York: Oxford University Press. pp. 131-57. ——. 2001. Ontology, Identity, and Modality. Cambridge: Cambridge University Press. ——. 2000. “Quantification and Fictional Discourse.” In Empty Names, Fiction and the Puzzles of Non-Existence. Everett, Anthony & Thomas Hofweber (eds.). Stanford: CSLI Publications. pp. 235-47. ——. 1977. “Creatures of Fiction.” American Philosophical Quarterly 14:4 pp. 299308. Reprinted in Van Inwagen 2001.

33