When I was in grad school, Wim Wenders came to talk about a movie of his, Der Stand der Dinge (the State of Things). I loved Wenders, and was glad that he was coming. After the movie I asked him what I thought was a very clever question about what was hidden under some stairs (iirc). He said he didn't know (which I knew he wouldn't), and I suggested that it was something from Johnny Guitar (Nicholas Ray is an extremely important precursor and mentor for Wenders). He looked at me as though I were batshit crazy, said no, it definitely wasn't that, and went on to the next questioner. I had imagined that he would get the important theoretical point that he was no more privileged as an interpreter of his own film than I was, and that what counted was the penetration of the reading (my reading), not the supposed authority of the reader (an authority that the author could pre-eminently claim). Literary theory, through its immense (and perennial) philosophical idealism had gone back round to treating fiction as though it were the representation of a true state of affairs, that anyone might be the first to see.
Truth in fiction didn't depend on what the fiction-maker meant. Its existence was independent of the fictionist's intention. Of course what made something true in a fiction was the interpretive aptness of the claim (like the notorious nineteenth century claim that Hamlet was a woman), such aptness measured by the insight it made possible. Insight into what? Well, into what was true in the fictional world. Such insight made, and could therefore find, the truth it claimed. Let's say it established truth. But that's what we do in the real world - we try to establish the truth.
Thus the only difference between the two - a difference which made possible the many-worlds interpretation of fictional interpretation - was the difference the article (the "the") suggests. In the real world we try to establish the truth, in fiction we try to establish truth.
Kendall Walton rightly argues that "truth in fiction," as D. K. Lewis called it, is a misnomer, since there's no requirement for logical consistency in a fictional world, on pain of deal-breaking incoherence. Deconstructive readings exploited the fact that most fictions are inconsistent, almost by their very nature, since fiction purports to know and to show things that cannot be known or showed: e.g. people alone with their thoughts, and the thoughts they're alone with (this particular inconsistency, rightly understood, is probably the one most central to deconstruction). Walton therefore prefers a technical use of the word "fictional": a proposition in a fiction would be called fictional if, as a stand-alone, it bore a relation to the fictional world it refers to analogous to the relation a true proposition bears to the real world. Fictional propositions don't have to appear in the fiction itself: they can be paraphrases or reasonable deductions or inductions from the propositions that appear there ("Hamlet dies at the end of the play"; and, probably, "Horatio lives on, with Fortinbras as King"). The reason for calling them fictional rather than "true in the fiction" is to suggest that not all their logical consequences are also true in the fiction. The dead Hermione's ghost appears to Antigonus... Hermione turns out not to have died. I think it's easier to say that both those statements are true in The Winter's Tale, rather than saying they're fictional in the play, but I've paused to rehearse Walton's argument because it brings out the difference between what I'm calling and will call fictional truth and the truth.
So we can tease out the implications of the difference the "the" makes by saying that our basic view of truth in the real world is Tractarian (i.e. conforms to the arguments of the early Wittgenstein): the consistency of the world will guarantee the consistency of the elementary propositions that picture it. Hence the world is all that is the case. Whereas our view of truth in fiction would be much more a coherence theory of truth: arguments about what happens in fiction require a reasonable amount of consistency among the various things that are true in that fiction, a consistency that makes it possible to handle the inconsistent parts that themselves contribute to the sense of coherence.
Still, at that time, in those days, the similarities seemed to us more important than the differences: the real world was coherent, and so was the fictional world. Ideal it may have been, but it shared with reality a presumption of completeness, and anything which made it complete could count as a live hypothesis about the fictional world, just as anything which explains away an apparent contradiction counts as a live hypothesis in the real world. In the real world, we are taught, we should always prefer the simplest possible account; in the fictional world we also used Occam's razor, but found that his straight edge didn't cut it and we had to plug in the electric one, which made possible all sorts of stylistic choices in the barbering of fictions hirsute with unexplained tufts of incident, character, or description. The simplest explanation is the best, but it's hard to define simplicity when in principle there's no reality check: it became a question of explaining all the fictional facts with a story supplementing the one we received. This of course was also what the New Testament did, and Midrash (where was Isaac after the Akedah?) and Kabbalah, and all manner of theologically inspired commentary and complement. Chandler might not know who killed Owen Taylor, but we could try to figure it out.
Now as the later Wittgenstein points out, there are an infinite number of sequences (of stories) that will explain any data (any fictional facts) that we are given. Since whatever sequence the author may have had in mind doesn't count more than any other, doesn't count more than the sequences readers may invent; since the logical inconsistencies, however trivial, show that even if we credit the author with authority over the meaning of her fiction, she nevertheless hasn't specified the whole sequence, item by item (any more than I have specified a whole sequence in my mind when I count 2, 4, 6, 8... that couldn't continue 1000, 1004, 1008, or - my favorite - 0, 1, 2, 720!, a number with 1,747 digits)* we deep readers felt entitled to our own penetrating, sequence producing fictional assertions about what happened offstage in the fictional world. Addition had no priority over quaddition, no matter what kind of real world type of pragmatism you inevitably evinced. There was no cash value to pragmatic truth in interpreting fiction - quite the reverse.
But to think this way is to lose the very thing that makes a fiction fiction, the universal literary genre we call fiction. It is to lose sight of the central law that the truth is what the author thinks it is (or what an authorial narrator, the last in the series, the narrator who has the author's full confidence, thinks it is). Narrating is one of the most basic forms of human interaction, of human sociability. "I've got a story": words which promise pleasure to both teller and told. The pleasures are different: the teller takes pleasure in promulgating, the listener or reader in learning (as Aristotle pointed out already in the Poetics). No stories without tellers is the moral of this one.
A moral more complicated than it might seem, it plays out differently according to the kind of story being told. A quick taxonomy would distinguish between true stories and fiction, but we have to add a third phylum: anonymous stories whose origin is lost in the mists of time (folk tales, myths, legends, etc.). When someone is telling a true story, we're entitled (rude though it might be) to second-guess her interpretations. Some people always do. I tell a story about a jerk cutting me off on the 405, but my skeptical listener suggests I might be in the wrong. He thinks the truth (the single truth) might be different from what my story is suggesting, that there is a fact of the matter and I'm misrepresenting it. This is true of third person stories as well: I say that Babe Ruth called his shot; she says, No, he was stretching prior to batting, and it just looked like he was pointing.
My skeptical chum doesn't have the same right to say that about a fictional narrative I originate. I get to say what my characters have known, have planned, have anticipated, have done. If Wenders denies that there's something under the stairs, if his denial is serious, his skepticism genuine, no one is entitled to gainsay him. It doesn't matter if the author is dead (you know, literally, biologically, dead). Our sense of her is that what she thought happened happened. We may not know what she thought happened, but we're still appealing to that category. Who killed Edwin Drood? We'll never know, but Dickens sure did.
The third phylum is the tale, which intersects the other two, and with their common ground. If you've heard a story, I can think you've heard it wrong, or that there's a way to tweak it to make it better. It's fiction, but it's like the truth in the sense that it's public property. No one has exclusive rights to it. Here the teller is more or less like a literary critic, or an actor: an interpreter of a story that comes from elsewhere. But her interpretation also gives her the authority that a witness has when it comes to telling true stories: she has a somewhat privileged, but defeasible relation to a public truth. More defeasible than an actual witnesses would have, since once I know the story I am as entitled to tell it in the way I think best as she was. There are no rules against hearsay in this phylum: indeed hearsay is obligatory, even or especially with all the hopeful mutation hearsay can introduce.
My interest, though, is in the authority the teller has over the tale, an authority most marked in the second phylum, the one where the fiction has an indentifiable author. Here the strangeness of fiction - that we care about what we know isn't true - and the importance of the teller are both at their maxima. And yet, the author is still governed by some coherence-producing restraints. 0, 1, 2, 720! will rarely do (though perhaps that's David Lynch's speciality). Chandler may not know who killed Owen Taylor, but he would have wanted to know, would have decided and established who did, had he realized that hadn't known. He doesn't know, and now there's nothing to know. There is something to know about who killed Edwin Drood, but we never will know it: ignorabimus.
That constraint, like poetic form, can be a goad and a spur to the fictionist. Lewis Carroll has to come up with the answer to random riddles he's posed - and he does (How is a raven like a writing desk?) The whole movie in the can, and being shown to test audiences, Hitchcock decides (the audience has a hand in this, as it should) that Cary Grant had better be innocent. Hitchcock comes up with an ingenious ending explaining away all the Suspicions. Javier Marías never returns to revise a page once he's done with it: he has to cope with the fictional truth of the fictional past, to explain the drop of blood or the behavior of young Pérez Nuix. Writers had to do this all the time in the age of serials: TV writers still do, though it's more interesting as in the case of Marías or (I think) Helen DeWitt when you have produced your own constraints. (DeWitt is endlessly inventive and then endlessly attentive to the implications and consequences of her inventions. DFW is sometimes like that too.)
Truth in fiction: there is a fact of the matter, but that means there are privileged relations to the facts, authoritative perspectives, certified fact-finders. (Otherwise no one would need to listen - everyone would already know that the tale referred to the fictional just as Frege says that true propositions refer to the true.) Even when we tweak anonymous tales, we usually think we're getting back to what they must originally have been, or at least we present them as the more authentic versions. (Or else we transmute them into something frankly our own, become their announced authors, even if we're anonymous: "by the author of Waverly, &c.") The risibly clever "fact" I offered Wenders showed that I got right that fiction trades in truths; what I got wrong though, was the way that such truth is in principle only available through the teller and her actions as a teller. We all trust our own judgment, and all human communication is a comparison of divergent judgment (no matter how small the divergence) - otherwise it's not communication.
But then how can we compare our own judgment about the world she has invented with that of the fictionist? And how can we know the truths she's left to our own inference - what prevents us from inferring the world we want to infer when we have a chance?
*In case you're curious, here are all 1,747 digits of the fourth number in the sequence whose first three numbers are 0, 1, 2 written out: