I’ve argued that there is an imaginative speech act involved in fictional discourse, but that is neither necessary nor sufficient for creating fiction. The imaginative speech act is analogous to assertion but is an expression of content from a particular imaginative model rather than one of content from a belief model. Since imaginative models represent possible objects and states of affairs with no commitment to their actuality, imaginative utterances are not true or false the way assertions are. The standard of correctness for an imaginative utterance depends on the context of utterance, the purpose of the model, and the practice of which the instance of model-building is a part.
In Chapter 2, I argued that it’s a norm of imagination that it models possibilities. Therefore, the intentional objects of imaginings are possible objects and states of affairs. The names and terms we use in our imaginative utterances refer to the intentional objects of the mental state that we express. So, when we make an imaginative utterance, we normatively refer to
possibilia.242 Additionally, fiction is the kind of social practice that I call an institution; it is a large-scale, collaborative practice that creates social entities and artifacts. Some of these artifacts are fictional objects. When we make assertions about fictional objects, such as “Paul Atreides was created by Frank Herbert,” we refer to these actual abstract artifacts. This means that the reference of ‘Paul Atreides’ is ambiguous. The ambiguity of reference in the case of fictional names is a well-established view, at least as far back as Kripke. Often, however, it is ambiguous between referring to nothing at all and referring to an abstract artifact. On my view, it either refers to a cluster of possible objects or an abstract artifact. This is a consequence of my imaginative practice view of fiction, as I just
described. Another, more familiar way to frame the view is as a modification of Currie’s ontology. Remember from Chapter 1 that Currie has a three-part semantic analysis for fictional names. His fictive and metafictive uses are based on his analysis of fictive
utterance versus assertions made under the fictional operator. On my view, however, there is no difference between so-called fictive and metafictive uses – they are both imaginative utterances.
The part of Currie’s analysis that is of primary interest here is that of transfictive uses. When we appear to talk about fictional characters outside of the fictional operator, we refer to roles that pick out possible people. This is a broadly similar picture as the one that results from my imaginative practice view of fiction. Rather than merely specifying work-bound roles, however, creators of fiction bring dynamic abstract artifacts into existence: artifacts that transcend their originating works and whose properties and relations can change over time. The relationship between the abstract artifacts and the possible candidates on my view is not as straightforward as the function-and-value relation in Currie’s theory. The abstract artifacts that we tend to call ‘fictional characters’ are created by the practice and their relation to possibilia is determined by the practice. When we imagine with a fictional work, we represent some possible objects and states of affairs and we refer to those possibilia with our intrafictional talk. Intrafictional talk is a practice-based species of 242 I stress that this is normative rather than descriptive, as things can go ‘wrong,’ such as in the case of a
posteriori identity failure mentioned in Chapter 2 and the connection between our imaginative acts and
imaginative utterance. When we use the name ‘Paul Atreides’ in an intrafictional claim, we refer to all those possible people that are candidates for the character in the story. We use the name ‘Paul Atreides’ to predicate of those possibilia that they are such candidates. The practice is responsible not just for establishing the relationship between its abstract artifacts and the possibilia represented in the imaginings of each participant, but also establishing the relation between the possible Paul Atreides candidates imagined by each participant and those imagined by each other participant.
This semantic view and the specifics of the resulting metaphysics is what I will develop in the coming sections. However, there’s one major hump left to get over before we entertain dualism. On Currie’s view, we never refer to the possible candidates picked out by the role, only to the role itself conceived as a theoretical entity. He explains:
Someone is Holmes in a world w if he is the value of our function fₕ at w. If it is possible for me to have been and done all the things Holmes is said to be and do, then there is a world in which I am and do those things and in that world I occupy the Holmes roles; I am Holmes in that world. But that does not make me or the countless others about whom we could say the same thing the subject of the Holmes story. Doyle’s use of Holmes is not a reference to me – no more than the
policeman’s use of “Smith’s murderer” is a reference to me just because I might have murdered Smith.243
What he’s showing here is that talk about a role is not talk about each or any of the role’s possible values. However, the force in the Smith case comes from the fact that, in this example, Smith was murdered in the actual world, the detective believes the role of Smith’s murderer is occupied by an actual person, and he makes assertions about that person via his use of ‘Smith’s murderer.’ Of course, there is no person he refers to merely because that person is possibly the murderer. But surely he does refer to some person by his use of ‘Smith’s murderer,’ because that person occupies that role in the actual world – the world about which the detective is making his assertions. This, I hasten to add, comports
perfectly well with Currie’s account, according to which our transfictive utterances are assertions about roles in the actual world that just happen to have no actual-world value. However, we shouldn’t ignore the disanalogies in the Smith’s murderer case. First, when the detective talks about Smith’s murderer, he is never talking about the role, but always its actual-world value. The detective does not need to know the identity of the murderer to talk about him. It’s difficult to conjure up an example in which the detective would be referring to a role, because ‘Smith’s murderer’ isn’t the kind of socially-defined office that counts as a theoretical entity of the type that fictional characters are on Currie’s account. To make things clearer, let’s use the office of the president of the United States instead. When we say that the president was born in the 1940s, or that the president is a man, or that the president gambles, we are referring to the actual-world value of the role. When we say that the president has never been a woman, we are saying something about the role rather than any one of its occupants. Most sentences about the office are explicit about this: “The office of the president was established in 1776,” “This behavior is beneath the office of the president,” etc. The detective never uses ‘Smith’s murderer’ this way; this ‘role’ can only have one actual value because it picks out the perpetrator of a singular, unrepeatable crime. There is no reason to talk about Smith’s murderer unless you are referring to the actual person who murdered Smith. The interest for this role is entirely in its actual-world value.
Fictional characters are not like this. We can distinguish between types of roles based on the possibilities for their values. Presumably, some roles have no actual values and no possible values. Others will have exactly one actual value and many possible values. A third kind may have many actual values and even more possible values. A fourth kind will have no actual values but many possible values. ‘Smith’s murderer’ is a role with only one actual value, but many possible values. ‘The president of the United States’ is a role that has many actual values and many possible values. Our primary interest in roles that have actual values is in their actual values. We care about the role ‘Smith’s murderer’ only because we want to catch and imprison the actual value of the role. We care about who is, has been, and will be the actual president of the United States. Only secondarily might we
imagine about the non-actual possible values of that role. Merely fictional characters have no actual value; we have no beliefs and make no assertions about the actual value, other than the trivial belief that it doesn’t exist. Our primary interest is in their possible values. It is the possible candidates for those roles that we represent in our imaginative models. When Currie uses the murderer analogy, we think “Of course the detective couldn’t be referring to him just because he possibly fills that role!” This is because the way we treat actually-filled roles is importantly different from how we treat fictional characters.
However, even without the analogy, it’s plausible enough that the Sherlock Holmes stories are not about Greg Currie. I think there’s also a sleight of hand in this move. It does not follow that a fiction is about an actual person if that actual person’s possible counterpart is a candidate for a fictional character. There’s a difference between imagining of an actual person that they did all the things Sherlock Holmes has done and imagining about possible Holmes candidates, of which one is an actual person’s counterpart. Even if we accept that names are rigid designators, this does not mean that we have no way at all of referring to some of an actual person’s possible counterparts without also referring to that actual person.
What is of interest to us in the Holmes stories is not which actual people may have Holmes counterparts, and our ‘picking out’ of possibilia in our imaginary representations need not be transparent or determinate. For me to imagine the possible blond-haired version of myself and make imaginative utterances about him, I need not consciously know exactly how many blond-haired counterparts I’m mentally representing, or what their other features are, or anything else about them to nevertheless represent them and refer to them with singular language: ‘the possible blond-haired Ash.’ I pick these possible people out by an identity relation, and predicate of them that they are me with blond hair. In the fiction case, we pick out the cluster of possible people not with an identity relation but with a Holmes-candidate relation, or an Atreides-candidate relation, and predicate of them that they are candidates for subjects of the story. We need not know whether or how many actual people might have an identity-relation counterpart among these possible people in
order to mentally represent and refer to them. The picking out is interest-relative, and the interest is determined by the practice.244