Monday, 30 September 2013

A non-semantic-based argument for fictional realism

Hi everyone,

I just read an interesting paper by Alberto Voltolini, and I wanted to share the argument with you all.  Fictional realism, as many of you know, is the view that fictional characters exist.  Fictional realists, as Voltolini points out, typically rest their case on semantic arguments. The general move:  there are some sincerely asserted sentences that appear to make reference to fictional characters, and that appear to be true, so (given simple semantic assumptions) we should think fictional characters exist.  Voltolini thinks that it's possible for fictional anti-realists to continue to combat the traditional semantic argument by offering more and more complicated paraphrasing alternatives to the relevant sentences.  So, if that is true, it would be nice for the realist to offer a non-semantic-based argument in support of her view.  (You might even think that, regardless of the anti-realist's prospects for combating the semantic argument, it would be nice for realists to offer more arguments in favor of their view).  Voltolini's argument for realism is intended to be just that:

Voltolini’s ontological argument for fictional realism:

  1.  There are fictional texts.
  2. If (1), then there are sets of propositions that those texts non-connivingly express(call these sets ‘fictional worlds’).
  3. So, there are fictional worlds.
  4. If (3), then there are identity conditions for some existing entities that involve fictional characters.
  5. So, there are identity conditions for some existing entities that involve fictional characters.
  6. If (5), then fictional realism is true.
  7. So, fictional realism is true.

Justification for (1):  Texts are syntactical entities composed of sentences.  There are some texts that are used as props for games of make-believe.  If there are some texts used as props for games of make-believe, then there are some fictional texts.  So, there are some fictional texts.

Justification for (2):  A fictional text’s use in normal contexts is conniving, because this is what is required for them to be treated as props in games of make-believe (in fact, what Voltolini appears to mean by 'conniving' is just inviting to make-believe, rather than sincerely asserting; depending on your view of make-believe, such sentences may not even express propositions, but rather pretend to express propositions).  There is nothing inherent to the text itself, however, that prevents it from having a non-conniving dimension.  In fact, when we engage in acts of make-believe with the text, part of what that requires is for us to imagine that some speaker (the narrator, for instance) is sincerely asserting the sentences that make up the text.  If such a text has a non-conniving dimension, the sentences constituting that text must express a set of propositions (for, this is just what comes from non-connivingly employed sentences, ie. sincerely asserted sentences – they express propositions).  This set of propositions will include those explicitly expressed by the sentences, and those implicitly expressed by the sentences (this corresponds with the intuitive thought that fictional works contain both explicit and implicit truths about the fictional world - according to Voltolini, the implicit propositions will be entailed by the explicit propositions).  It’s natural to call this set of non-connivingly expressed propositions 'the fictional world.'

(3) follows from modus ponens.

Justification for (4):  It’s possible for there to be two syntactically identical fictional texts coincidentally written by different authors at the same time in roughly the same location (suppose the two authors are neighbors, but that neither is aware of the other or their writings).  It is intuitive to think that these two authors wrote distinct fictional works about correspondingly distinct fictional worlds.  Since a fictional world is just the set of explicit and implicit propositions a fictional text expresses, then the difference between the two must amount to a difference between the sets of propositions.  What difference, however, could there be in two sets of propositions expressed by syntactically identical texts?  There appear to be two options:  the two sets differ in virtue of the ‘real’ (non-fictional) constituents of the proposition-members, or the two sets differ in virtue of the fictional constituents of the proposition-members.  It cannot be the real constituents of those propositions (if there are any), since (if there are any) such entities would be the same in both works (under the assumption that a name like 'Napolean' in a fictional text could refer to Napolean Boneparte, then it would do so in any work in which the name was so used).  So, what makes the two worlds distinct must be the fictional entities constituting some of those propositions.  But, saying that fictional entities are what make two syntactically identical texts express distinct fictional worlds is just to say that those fictional entities are what constitute (at least in part) the identity conditions of those worlds.  So, fictional entities are needed to fix the identity conditions of fictional worlds.  If fictional worlds exist, then fictional entities are needed to fix the identity conditions of some existing entities.

(5) follows from modus ponens.

Justification for (6):  We should think (6) is true because for something to play a role in the identity conditions of an existing entity, then that thing must itself exist.  Fictional realism is the view that fictional entities exist, so, if fictional entities play a role in the identity conditions of an existing entity, then fictional realism is true.

(7) follows from modus ponens.

Voltolini considers some objections, and I have some objections of my own in mind, but for now I'll simply post the argument to see what y'all think.


  1. I take it that the place to push will be (4). The argument will be something along the lines of: ‘Sure, your intuitions say that P, but we have good independent reasons to think that ~P, so your intuitions are wrong’. I know it’s lame, but if there are independent reasons against fictional realism, you bite the bullet. This would be my response to the parallel version of this argument for mythical realism, since I am a mythical anti-realist.

    You can also deny (2). Looking like a sentence in English is not sufficient for expressing a proposition. A meteor might hit the Earth and have its bits randomly arranged in a way that looks indistinguishable from the sentence ‘Tillman is fat’. That does not make it the case that a proposition is expressed by that arrangement of meteor-bits. You might take this a bit further. Maybe you think that in order for a sentence to express a proposition, it must be written with the intention of asserting something, or at least without the explicit intention not to do so. But if you think that fictional works are written with the intention not to assert something (but rather to be used as a prop), then you can deny that there are any propositions expressed by fictional texts.

    Side note: It is not clear to me why this is not a semantic argument. van Inwagen’s argument says ‘claim x about fictional characters is true; if x is true, then fictional realism’. Voltolini’s argument (oversimplified) says ‘claim x about a fictional character is meaningful; if x is meaningful, then fictional realism.’ I don’t know why we’d want to call van Inwagen’s argument ‘semantic’, but suppose we do. Why wouldn’t we also want to call Voltolini’s argument ‘semantic’? Is appealing to the meaning of some sentences any less semantic than appealing to their truth and reference?

    This argument also seems weaker than van Inwagen’s, since Fregeans are forced to reject the justification for (4). After all, if Fregeanism is true, then the two fictional texts can vary in meaning without varying in reference. Van Inwagen’s argument, on the other hand, is neutral on the Fregeanism-Millianism debate, which I take to be a virtue of the argument.

    I guess that you can also do some ‘contingently gappy proposition’ move here, and say: ‘the reason why these fictional works are different is because of the modal character (not the content) of the propositions that make them up. Were the characters in one or both of the fictional works to exist, then the content of the propositions would be different.’ What grounds the modal character of these propositions, if not the content? I don’t know, ask Justin.

  2. Thank You Shawn for sharing this.
    I would like to deny the claim that in order for a sentence to express a proposition, it must be written without the explicit intention not to assert something. One of the conditions of assertion is inviting your audience to "believe" something. However, when Doyle writes "Holmes lives in Baker street", he has the explicit intention not to invite anyone to believe that Holmes lives in Baker street, i.e. making an assertion. He has the intention to invite the audience to make-believe, rather than believe, that Holmes lives in Baker street, and this is not assertion, maybe it is pretending to assert. But still his sentence express the proposition.

  3. Damian,

    Your comments are very useful. Thanks. I have a few things I'd like to say in response.

    First, I agree, premise (4) seems to be where most of the work is done. I also agree that one way to deny premise (4) would be to explain away the intuition that the two authors wrote distinct works (in the event that we have the intuition). However, I don't think the explanation necessarily involves independent reasons against fictional realism. The intuition in this case appears to be one we might have regardless of our beliefs about the existence of fictional characters. If the two authors sent me copies of their work, and I knew that the two authors were completely unaware of each other, then I might still have the intuition that they wrote distinct works, even though (by some miracle of chance) they ended up with syntactically identical texts. I can have that intuition even if I have no beliefs about fictional realism. So, I take it that the explanation for why we should think the two authors wrote the same work rather than two distinct works is possibly independent any concerns one might have about fictional realism.

    For what it's worth, I don't have very strong intuitions either way in cases like these, but I lean towards thinking they are distinct works. However, my suspicion is that what makes them distinct has nothing to do with what the terms refer to in some of the sentences expressing singular propositions, but rather something about the historical/intentional source of the work. Fortunately, Voltolini predicts this sort of objection (that the explanation for why the works are distinct has to do with intentions, not fictional characters), and offers a reply. Unfortunately, I have a hard time making sense of this section of his paper. Here's my best effort at a summary:

    Suppose you think that the fictional work is not simply the syntactico-semantical entity that Voltolini thinks it is ({text + fictional world}), but something more complex, say, {text + fictional world + author's intentions}. Voltolini denies that this helps the objection. According to Voltolini, even if the two authors employ type-different intentions to use the same expressions with the same meaning, such a difference would be "superfluous...[f]or the difference between syntactically identical fictional works may then be entirely accounted for by appealing to the meaning difference between the corresponding [entities]" (233). I have to be honest, I don't quite understand Voltolini's reply. As soon as you grant that a work's identity is partly constituted by the author's intentions, and grant that two works are different in this regard, then you have all you need to account for the difference between the works (without commitment to fictional characters). Whatever Voltolini's reply is, I don't understand how it changes any of that. ......

  4. I like your objection to (2). Voltolini's motivation for thinking that a text has its non-conniving dimension is similar to why it has its conniving dimension: a text gets to be conniving or non-conniving derivatively (through its use). He thinks fictional texts get used non-connivingly, and that "typically, this happens when people engage themselves in the practice that would intuitively be labeled as the telling-the-contents of a given story (consider for instance the sentences uttered by students sitting for an exam on literature)" (Voltolini 226). I think this is a bad reason to think fictional texts have non-conniving dimensions in the same way that they have conniving dimensions. A fictional text's conniving dimension is essential to it, while its non-conniving dimension is contingent at best. So, deny premise (2).

    What I find very funny about the paper is exactly what you touch upon in your side note. Voltolini wants a non-semantic argument for fictional realism, so he helps himself to an ontological argument concerning an entity that is essentially semantic. And I don't mean this in a "let's laugh at Voltolini" kind of way, I actually think it shows a kind of cheeky cleverness.

    Anyway, thanks so much for your comments. They are very insightful.

    For everyone, the paper I have been talking about is Voltolini's 2003 "How fictional works are related to fictional entities", Dialectica vol. 57 no. 2: 225-38.

  5. Shawn, I think that Voltolini's intuition is not just that the two books are different, but that they're different because they talk about different fictional worlds. That is, the two books are different because they talk about different things. If that's the case, then your response about author intentions doesn't quite get to the argument.