Fictional entities

2. Deflationary theories

The intuition that there are no such things as fictional entities, so that any apparent ontological commitment to them must be removed, is powerful among philosophers. The logical analysis of Russell and Quine provides a striking example of how such a commitment can be avoided. The eliminative strategy applied to logical fictions can be applied to more familiar literary fictions as well. When we describe the exploits of Sherlock Holmes or the properties of Pegasus we appear to be referring to, or speaking about, entities which have some kind of existence. But this is an illusion, according to deflationary theories. In fact we are not referring to any such things.

Russell’s Theory of Descriptions sets the standard for deflationary theories. Suppose, regarding a character in a novel, we want to assert, ‘The invisible man could see but not be seen.’ For Russell, a logical analysis of our assertion would be somewhat as follows: ‘There is one and only one thing that is a man, is invisible, and can see without being seen’. The latter sentence removes the apparent naming expression ‘the invisible man’ and asserts only that something or other satisfies a collection of predicates. The sentence turns out to be false (for there is no such thing) but is perfectly meaningful.

Quine showed that the analysis could be generalized to eliminate all singular terms, proper names included. He suggested that, from a logical point of view, names can be turned into predicates purely formally. Thus he analyses the seemingly true sentence ‘Pegasus does not exist’ as ‘Nothing pegasizes’. By losing the troublesome singular term ‘Pegasus’ and speaking only of the instantiation (or lack of it) of a complex, albeit artificial, predicate, the analysis neatly shows how it is possible to deny the existence of a fictional entity without incurring any ontological commitment to that entity. It is a lesson, for example, that atheists can usefully apply in their denials of the existence of God (for it might seem that even to use the name ‘God’ is to admit that there is such a being).

To the extent that the logical problem of fictional entities is that of how to discriminate apparent reference from genuine reference the Russell/Quine deflationary theory is highly effective. But it also has serious drawbacks in accounting for other intuitions concerning ordinary talk ‘about fictional characters’. For one thing it has the consequence that all sentences containing singular terms for fictional characters turn out to be false (because they involve false existence claims). Yet is there not a sense in which the assertion ‘Sherlock Holmes is a detective’ is true, in contrast, say, to ‘Sherlock Holmes is a ballet dancer’, which is clearly false? On the Russell/Quine view they are both false. What needs to be captured is the fact that the claims are about the Holmes stories, not about persons in the real world. What is meant is something like: ‘Within the Holmes stories, as told by Conan Doyle, Holmes is a detective (and not a ballet dancer)’. It is a further question how assertions of that kind are to be analysed. In some deflationary theories they are said to refer only to novels and their component sentences. That has the advantage that it too removes unwelcome ontological commitments, but is surprisingly difficult to sustain in practice (not least because truths about the content of fictional stories extend beyond the explicit content of the stories’ sentences: for example, it is surely true that Holmes did not travel in a rocket, though that is never made explicit by Conan Doyle).

Another drawback of the Russell/Quine view is that it fails to distinguish storytelling discourse from discourse about stories. If the sentence ‘Holmes returned to London’ occurs in a Holmes story it seems inappropriate to analyse it as a false assertion about the real world. Conan Doyle, the author, is not asserting, or attempting to assert, facts about the world; he is making up a story. If on the other hand readers report the content of the story by using the same sentence they, in contrast to the author, are making an assertion (true or false), but one about the story, and again not about the world.

There are other deflationary theories, which attempt to remove commitment to fictional entities. Two deserve particular mention. The first is that of Nelson Goodman (§2), who defends a strict nominalist ontology. Given that there are no unicorns and no centaurs, the predicates ‘is a unicorn’ and ‘is a centaur’ have the same extension, according to Goodman, namely, a null extension: the set of unicorns is identical to the set of centaurs because it is the empty set. Yet surely unicorns and centaurs are different? Goodman seeks to explain the difference, not by invoking fictional entities, but by appeal to further predicates that do differ in extension. For example, a unicorn-picture, as he puts it, is different from a centaur-picture and can be recognized as such. While normally ‘X is a picture of Y’ is a two-place predicate, relating a picture and an object (for example, a portrait and the Duke of Wellington), in ‘fictional’ cases, such as ‘This is a picture of Sherlock Holmes’, there is only a one-place predicate involved, namely ‘is a Holmes-picture’. Pictures can be sorted into ‘Holmes-pictures’, ‘unicorn-pictures’ and so on, without commitment to a separate class of ‘fictional entities’. Pictures, after all, are unproblematically real.

Another deflationary strategy comes from Kendall Walton, for whom works of fiction are ‘props in games of make-believe’. Readers of Don Quixote seem to be introduced to a dreamy knight, who tilts at windmills, crusades against Evil, and so forth. Furthermore, they seem to think about him, refer to him and even feel sympathy for his predicament. However, for Walton it is only make-believe that they do such things. It is all pretence; they are playing a game with the novel as prop. In Walton’s alternative idiom, ‘it is fictional that’ they describe him, refer to him and respond emotionally to him. There is not even any such proposition that Don Quixote tilted at windmills. Walton offers a systematic theory which locates all ‘relations with fictional characters’ in imaginative games; to speak ‘about characters’ is just an elliptical way of speaking about the relevant games.