Wolfgang Schwarz

Impossible Fictions

Brian still believes that impossibilities are true in some fictions. I still disagree. But I wonder if the disagreement is substantial. Of course I agree that in some fictions one can find impossible statements. The question is whether these statements are true in the fiction. Not everything that is explicitly said in a fiction is true in the fiction. I'm now inclined to believe that our conception of truth in fiction may be ambiguous: on the standard conception, impossible fictions are ruled out; but on a different conception, they are tolerated. Let's call the former conception 'intensional' and the latter 'hyperintensional'.

There are several reasons why I think of the intensional conception as being standard. Here are five of them. First, this conception is simple and theoretically well understood. That's because it is modeled upon the simple and well understood Stalnaker-Lewis account of intentional content. Second, it is psychologically plausible: entertaining a fiction works and feels very much like pretending that a certain kind of world is actual. Third, the intensional conception is built into the best available semantics of fictional statements. This semantics regards utterances about what happens in a fiction as (implicitly) embedded in an intensional operator. The utterance is true iff it is true in the worlds selected by the operator. There are some difficulties in spelling out just what the relevant worlds are, but otherwise the approach works very well. Fourth, the intensional conception can nicely explain the phenomenon of Imaginative Resistance. Fifth, it matches my intuitive judgements about what's true in a fiction better than any alternative conception.

Let me briefly comment on Brian's current examples of impossible fictions before I talk about the non-standard, hyperintensional conception.

The restaurant at the end of the universe. Brian says it's impossible to observe the end of the universe. I don't think so. There would be a problem if by "the end of the universe" we meant the very last moment of the universe. Even then I think it would be possible to observe it, e.g. by means of backwards-causation powered projections to earlier times. But anyway, what people observe in Douglas Adams' restaurant is not the very last moment: they observe the collapsing of stars and galaxies and so on, which takes more than just a moment. It's more like watching the slow destruction of a building from within the building, and there is nothing impossible about that. Moreover, even if some claims in the text are really impossible, this doesn't imply that the impossibilities are true in the fiction. It's not true in the Holmes stories that Watson has his war wound at different body parts at different times. The fictional narrator is not the ultimate authority on what is true in a story, even less so if the narrator is as eccentric as the hitchhiker narrator.

Sylvan's box. I find nothing impossible about an empty box with a statue of a small elephant in the far corner. After all, "empty" doesn't mean "devoid of absolutely any content". Otherwise empty boxes would be physically impossible. All empty boxes contain at least space-time, gases and various fields. Most also contain bits of dust and dirt, and many even contain packaging material. I see no reason why an empty box shouldn't contain a statue of a small elephant.

That said, I think the examples can be modified in such a way that it sounds plausible to describe them as impossible fictions.

Sylvan's Box (Revised Edition)
Richard Sylvan never used to believe in true contradictions. Then one day shopping in a flea market in Indonesia, he came across a small box containg no statue of a small elephant but with a statue of a small elephant in the far corner. All of a sudden he saw that a contradiction was true. And if one was true, how many more must be true? He bought the box, and of course the statue inside it, and took them back to New South Wales.

To me, this story resembles the Tower of Goldbach in that I find it simply unintelligible. If I came across that paragraph in an ordinary story, I would say that the narrator must have been very confused at this point. Moreover, I wouldn't believe him, just as I don't believe Conan Doyle (or his fictional narrator) when he confusedly makes incredible claims about Watson's war wound. Our ordinary rules for interpreting fiction probably handle such cases by fragmentation: the contradictions or absurdities themselves aren't true in the fiction; what's true is some coherent revision of the narrative, though presumably it remains somewhat indeterminate exactly which revision that is (see Lewis, "Truth in Fiction", the postscript to it, and "Logic for Equivocators".)

Still, as I said, there is a sense in which one can reasonably say that contradictions are true in stories like the revised Sylvan's Box, and even in the Sherlock Holmes story. That's what I call the hyperintensional conception of truth in fiction. On this conception, truth in fiction is not closed under strict or logical implication, nor under any kind of relevant implication. It might not even be closed under synonymous paraphrase and translation. Here is a story that illustrates all these failures of closure:

The Holy Letter
"One day", the monk said, "you will find the holy letter. All words beginning with the holy letter are divine, and all sentences beginning with it are true."
"But", said Fred, "for every letter x and sentence S, there is a sentence S' that begins with x and that is logically equivalent to S. So if there is a holy letter, then every sentence must be true. Is that so?"
"No", the monk replied, "of course not. Don't be lured by classical logic. And don't be lured by paraconsistent logics either."
One day, the monk's prediction came true. On his way to the bathroom Fred suddenly realized that "O" is the holy letter: All words beginning with "O" were divine, and all sentences beginning with "O" were true. But even though it was true that octopuses are both animals and not animals -- for "octopuses are both animals and not animals" begins with an "O" --, it wasn't true that some things are both animals and not animals. As the monk had warned, classical and paraconsistent logics were both misguided.

These failures of closure show that the hyperintensional conception of truth in fiction can't be our ordinary conception. On the ordinary conception, truth in fiction is undoubtedly closed under synonymous paraphrase and translation. To me it also seems that using the hyperintensional conception works and feels very different from using the ordinary conception. In this case I don't interpret the story as describing an alternative reality. I don't engage in pretence that it's really true that octopuses are both animals and not animals, or that some box both contains and doesn't contain a statue. I don't know how I could pretend that (unless "pretend" just means "suppose", which is not what it means if it's supposed to describe what we do when we engage in fiction). Using the hyperintensional conception works more like saying what somebody said who uttered a contradiction. Suppose I said to you:

Octopuses are animals and grass is green, but grass is not green.

What would you say if someone asked you what I said to you? The only safe option is probably to repeat the very words I used. For instance, it is not even clear that I said that grass is green.

So I believe the ordinary conception of truth in fiction is the intensional conception. But I have a fallback position: Even if it turns out that the ordinary conception is somewhat hyperintensional, or (more likely) is ambiguous between various levels of hyperintensionality, I still say that the core notion of truth in fiction is intensional. In this case I would treat attributions of content to fictions like attributions of content to intentional states. The latter is a horribly messy and multifarious business, despite the fact that on a very attractive ("core") conception, intentional content simply is a probability distribution over the class of centered worlds. In fact, I take the horrible mess of hyperintensional content attribution to be one of the best argument for individuating intentional content intensionally. Likewise, if judgements about truth in fiction also turn out to be a hyperintensional mess (which I don't think they will), I wouldn't say this undermines the intensional account of truth in fiction.