Fictional realists claim that fictional characters like Spiderman exist in reality. Against this view, Anthony Everett (2005; 2013) argues that fictional realists cannot determine whether characters α and β are identical if the relevant fiction states that α and β are identical and distinct at the same time. Some fictional realists, such as Ross Cameron (2013) and Richard Woodward (2017), respond to this objection by saying that the sense in which α and β are identical differs from the sense in which they are distinct. In this paper, I argue that Cameron and Woodward cannot handle all cases without undermining the theoretical foundation of their approach, namely, the thesis that the identity of fictional characters must be determined by the content of the relevant fiction.
*A typo in (E2) has been corrected as of Jan. 2025: https://doi.org/10.3998/ergo.7251
1. Everett’s Puzzle
Fictional realism is the view that fictional characters like Spiderman, Sherlock Holmes, and Raskolnikov exist in reality. The idea may appear nonsensical initially, for we tend to think that fictional characters are merely fictional and thus not real. Even if there is some sense in which fictional characters are not real, however, they might still be existent. Some fictional realists hold, for example, that although Spiderman is not a real person, it exists as a cultural artifact created by Stan Lee.1
Furthermore, fictional realism allows us to take seemingly truthful statements about fictional entities at face value. That is, fictional realists can readily grant that sentences like “Spiderman is a fictional character” are as true as they appear. On the realist view, for example, this sentence is true because Spiderman really exists as a fictional character. For this and other reasons, many philosophers accept fictional realism: van Inwagen (1977), Searle (1979), Salmon (1998), Thomasson (1999), and Kripke (2013), to name just a few.
Nonetheless, fictional realism faces numerous challenges.2 Among others, Anthony Everett (2005; 2013) offers an influential argument against fictional realism. First, he presents a set of principles consisting of an existence criterion, (E1), and two identity criteria, (E2) and (E3), for fictional characters.
(E1) If α appears in fiction and α is not a real object, then α is a fictional character.3
(E2) If α and β are fictional characters, then [F(α=β) iff α=β].4
(E3) If α and β are fictional characters, then [F(α≠β) iff α≠β].5
Definition of F(φ)
F(φ)=df.It is true in the relevant fiction that φ.6
According to Everett, the fictional realist is committed to the above principles, which he takes as “having a near platitudinous status” (2005: 627).
Subsequently, Everett argues that (E1)–(E3) generate a metaphysical problem when applied to fictional works containing logical impossibilities. To illustrate this point, he presents a hypothetical work of fiction he created:
Dialetheialand
When she arrived in Dialetheialand, Jane met Jules and Jim. This confused Jane since Jules and Jim both were, and were not, distinct people. And this made it hard to know how to interact with them. For example, since Jules both was and was not Jim, if Jim came to tea Jules both would and wouldn’t come too. This made it hard for Jane to determine how many biscuits to serve. Then Jane realized what to do. She needed both to buy and not to buy extra biscuits whenever Jim came. After that everything was better. (Everett 2005: 633–34)7
To see why Everett sees the case of Dialetheialand as problematic, let us consider the results of applying (E1)–(E3) to the story.
(E1D) If Jules and Jim appear in Dialetheialand, and Jules and Jim are not real objects, then Jules and Jim are fictional characters.
(E2D) If Jules and Jim are fictional characters, then F(Jules=Jim) iff Jules=Jim.
(E3D) If Jules and Jim are fictional characters, then F(Jules≠Jim) iff Jules≠Jim.
Everett maintains that (E1D)–(E3D) lead to a contradiction via the following argument:
P1. Jules and Jim appear in Dialetheialand.
P2. Jules and Jim are not real objects.
P3. (E1D)
C1. Jules and Jim are fictional characters. (from P1–P3)
P4. F(Jules=Jim).
P5. (E2D)
C2. Jules=Jim. (from C1, P4, and P5)
P6. F(Jules≠Jim).
P7. (E3D)
C3. Jules≠Jim. (from C1, P6, and P7)
C4. C2&C3 (Contradiction)
If this argument is sound, then the fictional realist is committed to a logical contradiction. In this regard, Everett holds that “various impossibilities within the world of a story may infect the fictional characters that occur in that story” (2005: 633). Let us call this problem for fictional realism “Everett’s puzzle.”
This paper is a critical review of a well-recognized approach, which I call the “precisificational approach,” to resolving Everett’s puzzle. Our discussion will take the following steps. First, I lay out the basic strategies of the approach in the following section. Then I present the correct way of applying its core assumption in Section 3. In Section 4, I show that the assumption, when applied correctly, yields a powerful counterexample to the precisificational approach. Finally, in Section 5, I briefly examine an existing objection to the precisificational approach.
2. The Precisificational Approach to Fictional Inconsistencies
The precisificational approach attempts to resolve Everett’s puzzle by positing multiple standards of evaluation. Its core strategy appeals to the intuition that two seemingly contradictory verdicts can be mutually compatible if they result from distinct evaluative standards. To illustrate this point, I offer an analogy of serial stories.
Sometimes we regard a series of fictional stories—say, the Harry Potter series—as a single story with a continuous plot and a uniform worldview. At other times, however, we think of it as comprising multiple stories that share a single worldview.8 Yet we rarely view ourselves as inconsistent in having multiple conceptions in this way. Instead, we tend to think in the following way. First, although our use of “story” seems satisfactorily precise in ordinary contexts, its ambiguity stands out when we think of unusual cases such as a series of stories. Second, by precisifying our concept of a story—that is, by having two distinct conceptions of a story—we may resolve such ambiguity.
The precisificational approach attempts to resolve Everett’s puzzle by appeal to this type of intuition. First, although our concept of a fictional character seems sufficiently precise under ordinary contexts, odd cases like Dialetheialand may reveal its ambiguity, resulting in seeming inconsistencies. Second, by precisifying the concept of a fictional character, we may resolve the ambiguity. In this case, however, disagreements exist about the locus of ambiguity among the proponents of the approach: some say it consists in the ambiguous concept of a fictional character, yet others say it does in the ambiguous plot of the relevant story. Let me outline each view, starting from the latter.
According to Ross Cameron (2013), Dialetheialand has an ambiguous plot, and thus it has two interpretations. On one interpretation, which we may call Dialetheialand1, Jules and Jim are one and the same individual inhabiting Dialetheialand, while on the other interpretation, Dialetheialand2, they are distinct individuals. If so, then P4 in Everett’s argumentum ad absurdum would read “It is true in Dialetheialand1 that Jules and Jim are identical” and P6, “It is true in Dialetheialand2 that Jules and Jim are distinct.” Accordingly, there will be determinately three characters involved here: Jules(=Jim) of Dialetheialand1, Jules of Dialetheialand2, and Jim of Dialetheialand2. As a result, “Jules” and “Jim” in C2 will take distinct referents from those in C3. Along these lines, Cameron holds that Everett’s argument does not commit fictional realists to a contradiction.
In contrast, Richard Woodward (2017) suggests that we have two distinct conceptions of a fictional character. Under one conception, which we may call “character1,” there is one and the same character α(=β), just in case the relevant story describes α and β as identical. Conversely, there are two distinct characters α and β, just in case the relevant story does not describe α and β as identical.
Note that not describing α and β as identical does not amount to describing α and β as distinct; if the story leaves the matter of identity open, then it still counts as not describing α and β as identical. Thus, under the first conception, there are two distinct character1s α and β, even if the relevant story provides no information about whether α is identical with β.
Now, there is another logical option. When a story leaves open whether α is identical with β, we may say that there is one and the same character α(=β) instead of two distinct characters α and β. Under this second conception, character2, characters α and β are identical just in case the relevant story does not describe them as distinct.
In a sense, character1-type conceptions result in a strong identity criterion for fictional characters, whereas character2-type conceptions result in a weak criterion. According to the strong criterion, (W1), only definite fictional identity amounts to the real identity of fictional characters. But according to the weak criterion, (W2), anything but fictional distinctness amounts to the real identity of fictional characters. In summary:
Woodward’s Criteria of Identity
(W1) F(α=β) iff α1=β1.
(W2) F(α≠β) iff α2≠β2.
Along these lines, Woodward argues that the number of characters in Dialetheialand varies with the conception of a fictional character we adopt. For instance, if we adopted (W1), there would be two character1s, Jane1 and Jules1(=Jim1) since Jules and Jim are described as identical in Dialetheialand. On the other hand, if we adopted (W2), there would be three character2s, Jane2, Jules2, and Jim2, since Jules and Jim are described as distinct in Dialetheialand. If so, “Jules” and “Jim” occurring in P5 and C2 in Everett’s argument would have to be replaced with “Jules1” and “Jim1,” and those occurring in P7 and C3 with “Jules2” and “Jim2.” Consequently, the seeming contradiction in C4 would be resolved.
For the following reason, I believe Woodward’s approach is more advanced than Cameron’s, and thus I will focus mostly on his account in this paper. As Woodward (2017: 663–64) notes, Cameron’s approach suggests that there are definitely five characters—Jane and Jules(=Jim) of Dialetheialand1 and Jane, Jules, and Jim of Dialetheialand2—associated with Dialetheialand. It is plausible, however, that at most three characters—Jane, Jules, and Jim—appear in the story. Thus, “Cameron’s approach fails to do justice to our literary practices” (2017: 663).9
In contrast, Woodward’s account does not seem to carry the problematic implication. For unlike interpretations, conceptions do not generate multiple fictions, but multiple perspectives from which we conduct counting. To illustrate this point, let us briefly return to the analogy of serial stories. Previously, I pointed out that we have two distinct conceptions of a story. For example, since there are seven Harry Potter books, we could say either that there are seven Harry Potter stories or just one. But we do not think that this commits us to eight stories in total. Likewise, when we count fictional characters, the reasonable thing to do is not to count character1s and character2s together. Therefore, on Woodward’s account, it is false that five characters appear in Dialetheialand.
3. The Grounding Thesis
In this section, I specify the core assumption of the precisificational approach and argue for the correct way of applying it to determine the number of fictional characters. In Section 4, this will serve as the basis for my main argument against the approach.
Consider the following thesis.
Grounding Thesis (GT)
The existence, identity, and number of fictional characters appearing and originating10 in a story are determined by fictional truths concerning those things.
According to GT, if F(α exists), α exists as a fictional character in reality. Likewise, if F(α and β are identical), α and β are one and the same character in reality. In a sense, then, GT grounds the ontological status of fictional characters upon the content of (or equivalently, truths in) the relevant fiction.
Everett argues that (E2) and (E3) imply GT.11 And that seems correct. Look at the biconditional phrases in (E2) and (E3). The left-hand side states that some fictional characters are described as identical/distinct in fiction, while the right-hand side states that they are identical/distinct in reality. Hence, (E2) and (E3) suggest that the number of characters in reality depends on how many objects exist in fiction. Woodward’s criteria depend on GT in the same way, as they are just subscripted variants of (E2) and (E3). Thus, GT seems to play a central role in the precisificational approach.
GT states that certain ontological facts about fictional characters derive from the relevant truths in fiction. However, the thesis remains silent as to what counts as true in fiction. Thus, there can be multiple strategies for applying GT depending on the view one takes on the latter issue. In this section, I will compare two such strategies and show that one of them is superior to the other. In Section 4, I will argue that the precisificational approach cannot determine the exact number of characters in certain impossible fiction in line with the correct strategy for applying GT.
3.1. The Simple Strategy
It is best to compare the two strategies within a context in which their difference stands out most prominently. The following work of fiction taken from Everett (2005) will provide that context.
Asymmetryville
As soon as he got up in the morning Cicero knew that something was wrong. It was not that he was distinct from Tully. On the contrary, just as always he was identical to Tully. It was rather that while he was identical to Tully, Tully was distinct from him. In other words, some time during the night (he could not tell exactly when) the symmetry of identity failed. (Everett 2005: 634)
The simple strategy treats the identity relation in Asymmetryville the same way it would treat the ordinary, symmetrical identity relation. Since Cicero is said to be identical with Tully in the story, it seems natural to conclude that F(Cicero=Tully). So, by (W1), Cicero1=Tully1. In addition, since Tully is described to be distinct from Cicero, it is tempting to think that F(Tully≠Cicero). Therefore, by (W2), Cicero2≠Tully2. Thus, on the simple strategy, one fictional character1 and two fictional character2s appear in Asymmetryville.
3.2. The Immersive Strategy<xref ref-type="fn" rid="n12">12</xref>
Though the above suggestion comes off as intuitively plausible, it disregards the fact that the identity relation is described as asymmetrical in Asymmetryville. That is, it takes “F(Cicero=Tully)” as a correct description of what goes on in Asymmetryville, whereas the “=” relation is necessarily symmetrical.
Alternatively, we could think of another Woodwardian strategy that respects the asymmetrical nature of the identity relation in Asymmetryville. To do so, we must introduce a new identity symbol, such as “≃.” One way of specifying its logical properties is to employ a contextual definition like the following.
Definition of the “≃” Relation13
I. Identity
α=β≡(α≃β&β≃α).
II. Asymmetry
α≃β⊭β≃α.
According to Asymmetryville, Cicero is asymmetrically identical with Tully, but the converse does not hold. So F(Cicero≃Tully) and F(Tully≄Cicero). Moreover, due to Asymmetry, that Cicero≃Tully does not imply that Tully≃Cicero. Therefore, it is determinately the case that F(Tully≄Cicero).
Now, (W1) and (W2) can be restated as follows using Identity:
(W1*) F(α≃β&β≃α) iff α1=β1.
(W2*) F(~(α≃β&β≃α)) iff α2≠β2.
Adopting (W1*) and (W2*) results in a new set of verdicts via the following reasoning. First, since F(Tully≄Cicero), F(~(Cicero≃Tully&Tully≃Cicero)). So, (W2*) implies that Cicero2≠Tully2. Second, since F(~(Cicero≃Tully&Tully≃Cicero)), ~F(Cicero≃Tully&Tully≃Cicero). So, (W1*) implies that Cicero1≠Tully1. Thus, determinately two characters appear in Asymmetryville.
Should we dismiss the immersive strategy for disregarding the possibility that Cicero and Tully are identical? Not necessarily. Against such accusation, the fictional realist could respond that it is natural to conclude that Cicero and Tully are not identical in the ordinary sense, given that the identity relation holds only in one direction in Asymmetryville. Thus, further consideration is required to determine which of the two strategies is superior.
3.3. Simple versus Immersive: Which Strategy Is Correct?
The two strategies draw different verdicts of identity from distinct descriptions of the plot of Asymmetryville. While the simple strategy takes (DS) as the correct description, the immersive strategy advocates (DI).
(DS) F(Cicero=Tully)&F(Tully≠Cicero).
(DI) F(Cicero≃Tully)&F(Tully≄Cicero).
In other words, while the simple strategy denies that the content of a story is influenced by logical glitches in the world of the story, the immersive strategy takes this factor into account by employing a new identity symbol.
So, which strategy is correct? To answer this question, we must decide which of (DS) and (DI) represents the content of Asymmetryville more accurately. For the following reason, I believe (DI) is the more accurate description of the two, and thus the immersive strategy is the correct way to apply GT.
Strange as it may seem, Asymmetryville is not a logically inconsistent fiction—at least not in the sense that Daniel Nolan (2021) has in mind. According to Nolan, logical inconsistency is just one kind of many logical impossibilities. He says that a fiction is logically inconsistent only if “According to F, A and also According to F, not-A are both correct” (2021: 6, italics in original). But there is a good reason to think that Asymmetryville does not fall under this class of impossible fiction.
Observe that “φ&ψ” can be logically inconsistent in Nolan’s sense only if that ψ logically entails that ~φ, and vice versa. For example, “α=β&β≠α” is logically inconsistent because that α=β logically entails that ~β≠α. But it is only because the “=” relation is symmetrical that that α=β entails that ~β≠α. Consider “α<β&β≮α,” for instance. The sentence is not logically inconsistent, because that α<β is not equivalent to that β<α and thus fails to entail that ~β≮α.
Asymmetryville clearly states that the identity relation governing Asymmetryville is not symmetrical (“the symmetry of identity failed”).14 So, that α is identical with β does not entail that ~(β is distinct from α). Accordingly, the statement that α is identical with β and β is distinct from α does not commit us to the kind of contradiction that stories like Dialetheialand involve.15 By contrast, (DS) seems to imply that Asymmetryville is a logically inconsistent fiction since what is inside the F-operators constitutes a contradiction given the symmetry of the “=” relation. Therefore, (DS) is not an accurate description of what goes on in Asymmetryville.
Hence, I conclude that (DI) represents the content of Asymmetryville more accurately than (DS) does. In addition, GT states that it is the content of a story that determines the existence, identity, and number of fictional characters. Therefore, if GT is true, one must apply it in line with (DI) instead of (DS). To put it differently, advocates of GT must respect various logical peculiarities present in the world of a story.16
4. The Trilemma of Tridentity
The immersive strategy requires us to employ a new identity sign when the identity relation bears bizarre logical properties in the world of fiction. In the case of Asymmetryville, we were able to determine the number of fictional characters in line with the correct strategy. This was possible because Identity and Asymmetry served as principles that allow us to translate fictional identity statements invoking the “≃” relation into real ones invoking the “=” relation.
In general, however, there is no guarantee that suitable translation principles exist for every fictional identity relation. In the absence of such principles, there is no nonarbitrary way of determining how many characters appear in a story. To elucidate my point, I present a fictional story I wrote:
Tridentityland
When John arrived in Tridentityland, he met Jules, Jim, and Jane. This confused John, since the identity relation in Tridentityland seemed dramatically different from the familiar one. In Tridentityland, the identity relation was not a dyadic relation but a triadic one. For example, Jules, Jim, and Jane were identical, while Jules, Jim, and John were distinct. The relation was an exclusively triadic relation, so that no identity statement could make sense with a left-out term.17 In addition, the relation was commutative (i.e., insensitive to the order of its relata). For instance, because Jules, Jim, and Jane were identical, Jane, Jules, and Jim were also identical. As time went by, John got used to this strange situation little by little. Unfortunately, he was struck with greater confusion when he discovered that Jane, Jules, and John were identical and distinct at the same time.
In this section, I argue that the precisificational approach cannot determine the number of fictional characters in Tridentityland in line with the immersive strategy. My argument takes the form of a trilemma. Its horns trifurcate according to how the weak identity criterion determines the number of characters in Tridentityland.
A couple of preliminary remarks about notation: “J” will denote the set {Jane, Jules, John}, and “n(J)” the number of the referents of its members. For example, if Jane, Jules, and John are one and the same character, then n(J)=1.
4.1. The First Horn: According to the Weak Criterion, n(J) Is Indeterminate.
To implement the immersive strategy, we need an identity symbol that reflects the triadic nature of the identity relation in Tridentityland. So let us introduce a three-place predicate “I3” to represent the triadic identity relation.
In addition, we need a set of identity criteria to account for different senses in which Jane, Jules, and John are identical and distinct. Thus, we will construct a strong and a weak criterion using the “I3” relation. In doing this, we will write “I3{α, β, γ}” instead of “I3αβγ,” to articulate that the relation at issue is commutative. The resulting criteria will look like the following:
(W3) F(I3{α, β, γ}) iff α=β=γ.
(W4) ~F(~I3{α, β, γ}) iff α=β=γ.18
For the following reason, however, the precisificationist cannot depend solely on (W3) and (W4) to determine the exact number of characters in Tridentityland.
Since F(~I3{Jane, Jules, John}), (W4) implies that ~(Jane=Jules&Jules=John). This is compatible not only with Jane, Jules, and John being three fictional character4s but also with their being two. However, the number of things existent in the real world must be determinate. Furthermore, fictional realism holds that fictional characters reside in the real world. Therefore, if (W3) and (W4) are intended—as they are supposed to be—as principles telling us exactly how many fictional characters exist vis-à-vis particular fictional works, they conflict with fictional realism.
4.2. The Second Horn: According to the Weak Criterion, n(J) Is Determinate.
So, what we need is a set of criteria that can determine the exact number of characters in Tridentityland. Thus, we might resort to something like the following:
(W5) F(∃xI3{α, β, x}) iff α=β.
(W6) ~F(∃x~I3{α, β, x}) iff α=β.
To demonstrate how this idea is supposed to work: first, (W5) accounts for the sense in which Jane, Jules, and John are identical. Since F(I3{Jane, Jules, John}), by existential generalization, F(∃xI3{Jane, Jules, x}). Therefore, from (W5) it follows that Jane=Jules. In addition, since F(I3{Jane, John, Jules}), F(∃xI3{Jane, John, x}). So (W5) implies that Jane=John. Thus, according to (W5), Jane, Jules, and John are all identical character5s.
Now for distinctness: since F(~I3{Jane, Jules, John}), F(∃x~I3{Jane, Jules, x}). So, (W6) implies that Jane≠Jules. Moreover, since F(~I3{Jane, John, Jules}), F(∃x~I3{Jane, John, x}), and thus Jane≠John. Because these two verdicts accord with what follows from F(~I3{John, Jane, Jules}) and the like, (W6) issues a consistent verdict that Jane, Jules, and John are all distinct character6s and that determinately three character6s appear in Tridentityland.
Still, this is not satisfactory. All we know from the text is that Jane, Jules, and John collectively fail to stand in the triadic identity relation. Thus, we must allow for the possibility that n(J)=2. Nonetheless, (W6) seems to preempt this scenario since it entails that Jane, Jules, and John are all distinct characters. In this sense, (W6) achieves a complete verdict only to become ad hoc.19 Then, the remaining question is whether there is a way to represent the fact that n(J) can take on values of 1, 2, and 3 under different conceptions.
4.3. The Third Horn: Multiple Weak Criteria Cover Every Logical Possibility.
This requires us to consider both possibilities that F(~I3{α, β, γ}) iff n({α, β, γ})=2 and F(~I3{α, β, γ}) iff n({α, β, γ})=3. Moreover, we must employ multiple identity criteria to cover such possibilities since each criterion must yield a determinate value of n(J).20 To this end, we may disassemble (W4) into something like the following:21
(W4a) F(~I3{α, β, γ}) iff α=β&β≠γ.
(W4b) F(~I3{α, β, γ}) iff β=γ&γ≠α.
(W4c) F(~I3{α, β, γ}) iff γ=α&α≠β.
(W4d) F(~I3{α, β, γ}) iff α≠β≠γ.
This strategy may appear compelling because (i) each of (W3) and (W4a)–(W4d) yields a determinate value of n(J), and (ii) all possible values of n(J) are covered by the five criteria.
Unfortunately, this attempt is incompatible with GT. GT states that the content of a story determines the ontological status of the fictional characters that appear in the story. If so, then equivalent facts in a fictional world must result in equivalent verdicts about the identity of fictional characters in the real world. For the following reason, however, (W3) and (W4a)–(W4d) fail to meet this requirement.
The tridentity relation has been stipulated to be commutative. Thus, “~I3{Jane, Jules, John}” is equivalent to “~I3{Jules, Jane, John},” and so on. However, (W4a)–(W4c) do not treat these statements equivalently. For example, suppose (W4b) is true. Then, that F(~I3{Jane, Jules, John}) entails that Jules=John, while that F(~I3{Jules, Jane, John}) entails that Jules≠John. In other words, (W4b) draws distinct verdicts in reality from the same truth in fiction. Thus, it goes against a natural consequence of GT.
To summarize, the precisificational approach cannot simultaneously meet all the following requirements.
Determinacy: Only determinately many fictional characters can exist in reality.
Comprehensiveness: If a story makes it genuinely indeterminate whether there are n1, n2, . . ., or nk characters, at least k identity criteria must exist to account for the senses in which there are n1, n2, . . ., or nk characters.
Grounding: GT is true.
By Determinacy, each criterion must yield a determinate number of characters. By Comprehensiveness, those numbers must include 2 because the story makes it genuinely indeterminate whether there are one, two, or three characters (except Jim).22 So, it must be possible that exactly two characters among Jane, Jules, and John are identical in reality. However, it is impossible to account for that possibility without violating Grounding because there is no way to pick out two names in a tridentity statement without relying on the order in which they occur in the statement.23
5. Cohen on the Precisificational Approach
According to Wouter Cohen (2017), the precisificational approach obscures the distinction between semantic and metaphysical indeterminacies. Semantic indeterminacy occurs when a name refers to determinately many objects. For example, “Utrecht” is semantically indeterminate because we can use it to refer either to a Dutch province Utrecht or its capital city Utrecht. Metaphysical indeterminacy, on the other hand, is defined as one that remains even after all the relevant names are precisified to the extent that no semantic indeterminacy lingers.24
Cohen notes that Dialetheialand describes, or at least could have described, the identity between Jim and Jules as metaphysically indeterminate. For it does not seem that “Jim” or “Jules” take multiple distinct referents in Dialetheialand. Instead, they refer to some metaphysically indistinct object(s) whose number cannot be specified using any logical or linguistic apparatus.
As Cohen Correctly points out, Cameron and Woodward attempt to resolve this metaphysical impossibility by regarding fictional names like “Jim” and “Jules” as semantically ambiguous. In particular, Cameron holds that the former is ambiguous between “Jim of Dialetheialand1” and “Jim of Dialetheialand2,” while Woodward maintains that it is ambiguous between “Jim1” and “Jim2.” Thus, Cohen concludes that such semantic measures cannot be applied to a metaphysical problem like Everett’s puzzle.
I believe Cohen’s argument blurs the distinction between two levels of discourse: fictional and real. The distinction between semantic and metaphysical indeterminacy must be respected only insofar as it stays on one level of discourse. For example, if Woodward held that Jim1 and Jim2 are determinately distinct inhabitants of Dialetheialand, he would indeed be confounding semantic and metaphysical indeterminacies. However, that is not what he claims. His view is that we can individuate fictional characters in reality through linguistic means if they are described as metaphysically indeterminately identical in fiction.
To put the point differently, recall the fictional realist’s task. She wants to determine the exact number of characters based on the relevant fiction. In other words, she hopes to find an adequate function ϕ which maps fictional identity statements onto verdicts about identity in the real world. The problem with Cohen’s objection is that he seems to pose an overly strong restriction on this ϕ.
Cohen assumes that ϕ must preserve metaphysical indeterminacy such that ϕ(F(Im(α=β)))=Im(α=β). However, there are no prima facie grounds for such a strong restriction. As long as each verdict about identity is uniquely and consistently determined by fictional content, the verdict need not match exactly with the content.25 For example, we cannot preclude the Woodwardian formulation, ϕ(F(Im(α=β)))=(α1=β1&α2≠β2), just because “Im” does not occur on both sides of the equation. Thus, to refute the precisificational approach, one must show that there can be no unique and consistent ϕ that maps fictional identity statements onto verdicts about identity in reality. And that was the aim of this paper.
6. Concluding Remarks
This paper can be viewed as a general response to the type of fictional realism that upholds a content-based approach to the identity criteria for fictional characters.26 What the approach overlooks is that there are many types of impossible fiction with varying levels of tamability. Plausible as the approach may sound vis-à-vis Everett’s relatively “simple” fictions, it fails to maintain its initial appeal when more complex cases like Tridentityland are brought into the picture.
Tridentityland is problematic in that the binarity of the identity relation fails in it. Unlike other necessary properties that the identity relation bears, its binarity plays a seminal role in constructing a proper criterion of identity. This is because we can only translate a fictional identity statement in terms of our own when we can recognize which two characters are identical in the fictional world.
Note that I could further write stories like Quadradentitiyland, Pentadentityland, or even ones featuring more “unruly” identity relations. The moral here is that there are only minimal restrictions on the ways in which the identity relation can be twisted in fiction27 and it seems hopeless trying to come up with identity criteria corresponding to all the myriad impossibilities. Admittedly, there might be some original content-based identity criteria that I have not considered. However, until and unless content-based realism proposes compelling evidence for their existence, I believe that we are justified in rejecting the view.
While all fictional realists agree that fictional entities exist, some deny that fiction-makers create them: see Lewis (1986), for example. Although Parsons (1980) and Zalta (1983) deny that authors create fictional characters, they do not count as fictional realists as defined (somewhat narrowly) in this paper since they view fictional characters as nonexistent Meinongian objects. I thank an anonymous referee for pressing me to add this clarificatory note.
See Yagisawa (2001), Brock (2002; 2016), and Caplan and Muller (2014), for example.
See Everett (2013: 226).
See Everett (2013: 205).
See Everett (2013: 205).
There is wide-ranging disagreement over what truth in fiction is. Philosophers like D’Alessandro (2016) hold that only explicit statements made by the author are true in fiction. Others like Lewis (1978), Currie (1990), and Byrne (1993) say that implicit statements reasonably inferable from the text should also count as fictionally true. I remain neutral on this issue in this paper. Such neutrality can be maintained because all fictional statements this paper concerns are explicit statements that both explicitists and implicitists would take to be true in fiction. Thanks to an anonymous referee for pressing me to clarify this point.
Henceforth, I italicize the titles of fictional works, but not other names of the same spelling. For example, “Dialetheialand” refers to a fictional work, while “Dialetheialand” refers (or pretends to refer) to a place in a fictional world.
Motoarca (2017: 381) makes a similar observation.
For my original argument against Cameron’s account, see footnote 23.
The ontological status of fictional characters that occur in derivative works of fiction, such as sequels and parodies, is often thought to be determined by the truths in the original story in which they appear. Thus, GT shall be construed as concerning not just any story in which characters appear but the story in which they originate (I thank an anonymous referee for pointing out this issue). For related discussions, see Everett (2013: 199–200).
“[B]y [(E2)] and [(E3)], what exists in the world of a story determines which fictional characters occur in that story” (Everett 2005: 633).
The immersive strategy views the identity relation as asymmetrical—in the way the inhabitants of Asymmetryville, like Cicero, would experience it. Hence the label “immersive.”
Introducing the “≃” relation need not make Asymmetryville a story about some unfamiliar relation. Instead, we should take the situation as one in which the author describes the ordinary identity relation as having bizarre logical properties. Here is an analogy to further elaborate this point. In War and Peace, Napoleon is described as having done deeds that the real Napoleon did not do. To distinguish two Napoleons, we may (temporarily) call the real Napoleon “Napoleona” and the fictional one “Napoleonb.” However, that need not make War and Peace a story about some unfamiliar figure we have never heard of. As “Napoleonb” just means Napoleon as described in War and Peace, so “≃” indicates the identity relation as described in Asymmetryville.
Both explicitists and implicitists would regard this statement as fictionally true unless it results from the author’s mistakes or erroneous beliefs (I discuss these exceptional cases further in footnote 16). Thus, my point that GT should not treat it differently from other fictional truths can be taken as independent of the debate between explicitism and implicitism about truth in fiction. I thank an anonymous referee for making me consider this issue seriously.
Hanley (2004) maintains that all fictional stories must be understood as having a consistent plot. If his view is correct, then there will be no significant difference between the senses in which Asymmetryville and Dialetheialand are inconsistent. However, as Priest (1997) and Nolan (2015) point out, some stories are best understood when interpreted as having inconsistent plots. Dialetheialand is a case in point: Jane’s solution to the biscuit problem can be deemed reasonable as it seems only if we concede that Jules and Jim are identical and distinct at the same time. For other compelling instances of inconsistent fictions, see Priest (1997), Gendler (2000), and Chiang (2002).
The simple strategy might be helpful in some unusual cases. For example, if Everett described the identity relation as asymmetrical in Asymmetryville by mistake or as a result of erroneous beliefs, some might want to exclude statements about asymmetry from the content of Asymmetryville. At any rate, it is possible (and even plausible) that Everett had a firm and clear intention to describe the identity relation as asymmetrical. And my argument will remain equally sound as long as there are some cases where the immersive strategy is superior to the simple strategy. Thanks to an anonymous referee for pointing out this issue.
A non-fictional example of an exclusively triadic relation is found in Euclidean geometry. On a Euclidean plane, any three points define a particular circle. So, for any circle C on the plane, the C-defining relation is an exclusively triadic relation binding any three points on C.
Strictly speaking, the α, β, and γ on the right-hand side of (W3) and (W4) should be subscripted, respectively, with “3” and “4”. Henceforth, however, I will omit such a practice for conciseness when doing so is unlikely to introduce confusion.
According to Caplan and Muller (2015), the identity and distinctness of fictional characters are, in some sense, brutal. This view might be consistent with fictional realism, but neither Cameron nor Woodward can subscribe to it because it conflicts with GT, which states that the identity and distinctness of fictional characters derive from the content of the relevant story. I thank an anonymous referee for informing me of Caplan and Muller’s article.
Here, each criterion can count no other characters than John, Jules, and Jane. The core idea behind the Woodwardian strategy is to manipulate the identity sign in an intra-fictional identity statement featuring each fictional name to determine the exact number of characters. In this regard, as mentioned in Section 2, Woodward cogently argues that characters other than Jane, Jules, and Jim cannot appear in Dialetheialand. To be consistent, he must not allow each of (W3) and (W4a)–(W4d) to count characters other than John, Jules, and Jane. I thank an anonymous referee for giving me a chance to clarify this point.
I am assuming that a proper identity criterion must not only determine the number of things whose identity with other things is evaluated but also provide an accurate verdict as to which thing is identical with or distinct from which thing. This is a guaranteed consequence of settling all facts about identity.
As pointed out in the last paragraph of Section 3.2, one can plausibly argue that Asymmetryville does not make it genuinely indeterminate how many characters appear in it by appeal to the immersive strategy. In contrast, the trilemma presented in Section 4 shows that Tridentityland displays a genuine indeterminacy that remains even when the immersive strategy is employed.
Cameron’s strategy falls prey to a similar objection. First, an interpretation must not allow for n(J) to have an indeterminate value since the number of characters in reality must be determinate. However, n(J) must not be determinately 2 (or determinately 3), since that would be objectionably ad hoc given that the story only says that Jules, Jim, and John collectively fail to be identical.
This brings us to our last option, where we have five different interpretations to cover every logical possibility. Unfortunately, this option is also ad hoc in the following sense. The purpose of an interpretation is to make sense of the ambiguous plot of a story. Thus, for something to count as an interpretation, it must have sufficient textual grounds. Now consider the three interpretations saying that exactly two characters appear in Tridentityland. Since the story tells us that three individuals collectively fail to be identical, there seem to be no textual grounds to identify any two of them as identical. Therefore, Cameron’s approach fares no better than Woodward’s in determining the number of characters in Tridentityland.
Cohen borrows this formulation of metaphysical indeterminacy from Barnes (2010).
In this respect, one may regard Cohen’s argument as assuming a principle stronger than GT that fictional content not only determines but also is equivalent to the identity statement in reality.
My argument leaves intact other forms of fictional realism. For example, Friedell’s (2016) and Lee’s (2022) intention-based approaches are immune to my objection to Cameron and Woodward. However, there could be arguments independent of mine that commit intention-based realism to ontic indeterminacy. See Friedell (2020: especially 225–27) for a related discussion. (Thanks to an anonymous referee for informing me of this article.)
This is not to say that anything can be true in fiction. For example, Xhignesse (2021) claims that there can be no “universal fiction”, in which everything is true. This means that the conjunction of all propositions (but for itself) cannot be true in fiction. This view is clearly compatible with my claim that a certain logical relation can have bizarre properties in fiction.
Acknowledgements
I sincerely thank Yeolim Bae, Jake Beardsley, Daehwi Jeong, Eunhong Lee, Nikolaj Jang Lee Linding Pedersen, and two anonymous referees for Ergo for their helpful comments and suggestions. I am especially grateful to Jeonggyu Lee and Seong Soo Park, whose detailed feedback led to significant improvements in this paper.
Barnes, Elizabeth (2010). Ontic Vagueness: A Guide for the Perplexed. Noûs, 44, 601–27.Brock, Stuart (2002). Fictionalism about Fictional Characters. Noûs, 36(1), 1–21.Brock, Stuart (2016). Fictionalism about Fictional Characters Revisited. Res Philosophica, 93(2), 377–403.Byrne, Alex (1993). Truth in Fiction: The Story Continued. Australasian Journal of Philosophy, 71, 24–35.Cameron, Ross (2013). How to Be a Nominalist and a Fictional Realist. In Christy MagUidhir (Ed.), Art and Abstract Objects (179–96). Oxford University Press.Caplan, Ben and CathleenMuller (2014). Against a Defense of Fictional Realism. Philosophical Quarterly, 64(255), 211–24.Caplan, Ben and CathleenMuller (2015). Brutal Identity. In StuartBrock and AnthonyEverett (Eds.), Fictional Objects (174–207). Oxford University Press.Chiang, Ted (2002). Division by Zero. In Stories of Your Life and Others (71–90). Tor Books.Cohen, Wouter (2017). Fictional Realism and Metaphysically Indeterminate Identity. Analysis, 77, 511–19.Currie, Gregory (1990). The Nature of Fiction. Cambridge University Press.D’Alessandro, William (2016). Explicitism about Truth in Fiction. The British Journal of Aesthetics, 56(1), 53–65.Everett, Anthony (2005). Against Fictional Realism. The Journal of Philosophy, 102(12), 624–49.Everett, Anthony (2013). The Nonexistent. Oxford University Press.Friedell, David (2016). Abstract Creationism and Authorial Intention. Journal of Aesthetics and Art Criticism, 74(2), 129–37.Friedell, David (2020). Fiction and Indeterminate Identity. Analysis, 80(2), 221–29.Gendler, Tamar S. (2000). The Puzzle of Imaginative Resistance. The Journal of Philosophy, 97(2), 55–81.Hanley, Richard (2004). As Good As It Gets: Lewis on Truth in Fiction. Australasian Journal of Philosophy, 82(1), 112–28.Kripke, Saul (2013). Reference and Existence. Oxford University Press.Lee, Jeonggyu (2022). Authorial Intention, Readers’ Creation, and Reference Shift. Philosophical Quarterly, 72(2), 381–401.Lewis, David (1978). Truth in Fiction. American Journal of Philosophy, 15, 37–46. Reprinted with an added postscript in Philosophical Papers 1 (261–80), Oxford University Press.Lewis, David (1986). On the Plurality of Worlds. Blackwell.Motoarca, Ioan-Radu (2017). A Bad Theory of Truth in Fiction. British Journal of Aesthetics, 57(4), 379–87.Nolan, Daniel (2015). Personification and Impossible Fictions. British Journal of Aesthetics, 55(1), 57–69.Nolan, Daniel (2021). Impossible Fictions Part I: Lessons for Fiction. Philosophy Compass, 16(2), 1–12.Parsons, Terrence (1980). Nonexistent Objects. Yale University Press.Priest, Graham (1997). Sylvan’s Box: A Short Story and Ten Morals. Notre Dame Journal of Formal Logic, 38(4), 573–82.Salmon, Nathan (1998). Nonexistence. Noûs, 32, 277–319.Searle, John (1979). The Logical Status of Fictional Discourse. In his Expression and Meaning (58–75). Cambridge University Press.Thomasson, Amie (1999). Fiction and Metaphysics. Cambridge University Press.Woodward, Richard (2017). Identity in Fiction. Philosophy and Phenomenological Research, 94(3), 626–71.Van Inwagen, Peter (1977). Creature of Fiction. American Philosophical Quarterly, 14, 299–308.Xhignesse, Michel-Antoine (2021). Exploding Stories and the Limits of Fiction. Philosophical Studies, 178(3), 675–92.Yagisawa, Takashi (2001). Against Creationism in Fiction. Noûs, 35(s15), 153–72.Zalta, Edward N. (1983). Abstract Objects: An Introduction to Axiomatic Metaphysics. D.Reidel.