Five new papers and three new journals on the philosophy papers blog. One of the new papers, A Pragmatic Framework for Truth in Fiction by Andrea Bonomi and Alessandro Zucchi, was noted on semantics, etc yesterday. (There are even more papers there already. I can’t keep up!) Bonomi and Zucchi aim to, among other things, solve the problem of imaginative resistance. I don’t think their solution is going to work – as far as I can tell on their view a racist author writing in a racist community gets to make racist theses true in his story, but I may be misinterpreting them here. (See pages 7 to 8 of the text and see what you can make of it. I think their criticism of Lewis is right here, but I honestly can’t tell if they do better.)
They are also remarkably hesitant to allow that a narrator may be making a mistake. Their formal proposal is
(21) “In fiction x, p” is true iff p is true in every world w meeting (a)-(d):
a. w is compatible with the conventions of the fiction actually denoted by x,
b. the fiction is narrated in w by a teller who believes that the fiction is true in w,
c. w is as much as the teller believes it is as is required by the presumption of reliability,
d. among the worlds meeting conditions (a)-(c), w is closest to the set of worlds that represent the overt beliefs of the author in the community where the fiction originates.
And the presumption of reliability is
Presumption of reliability. Let w be a world in which fiction x is narrated and the narrator believes that x is true. If the narrator tells that q, q is true in w unless there are reasons intrinsic to the fiction that indicate otherwise.
Well, I simply don’t think this can be true. For one thing, it means that an author cannot signal that the narrator is unreliable by having the narrator make a mistake about the real world. But that’s clearly possible. It’s hard to find a perfect illustration of this, but for an imperfect illustration, consider Eveline.
The narrator of Eveline is quite unreliable, in many ways. Although the story has a third person narrator, we’re sort of told that the narrator is Eveline herself, at least until the last few lines of the story when suddenly it isn’t. But none of the hints that this narrator is unreliable are strictly speaking intrinsic. The most prominent such hint is when Eveline thinks that when she is married she will be treated with respect, unlike her mother. The story never says that her parents are married, we are meant to take that as given and infer that Eveline’s thinking (and hence presumably the narrator’s) is unreliable at best. This is backed up when she fails to notice that ‘stories about the terrible Patagonians’ are just stories. (Or perhaps she does realise this at the end of the story? I suspect not.)
According to Bonomi and Zucchi, we cannot use real world facts at this stage of working out what is true in the story – we find out the set of worlds compatible with the story by looking at what is intrinsic to the story and then among those pick the worlds closest to the (believed to be) actual world. So we can never use actual or believed facts about the world to form judgements about reliability. Since we can, and must, do this, their proposal needs at least some emendation.
By the end of last night I had decided that the argument for the JTB theory below was totally flawed. The worry was that if intuitions about Gettier cases are grounded in a tacit commitment to sensitivity, then we should expect all insensitive beliefs are not judged to be pieces of knowledge. But we don’t always all do this. I now think this worry can be overcome, provided we say that tacitly held theories can be inconsistent, so as long as some other tacit theories are operative, we can explain why some insensitive beliefs are judged to constitute knowledge, while still dismissing other sceptical intuitions (like Gettier intuitions) as being grounded in an utterly mistaken tacit theory.