What follows is a guest post by Chris Tillman. Chris is an associate professor of philosophy at the University of Manitoba. His main interest is in metaphysics, but he considers practically everything to be an issue in metaphysics. He is originally from Missouri, where his first major was in painting and he spent his free time in bands, including a country/rap band (hick-hop, if you will). These days his free time is more likely to be consumed by curing meats, genre fiction, and making Korean farmer hooch (makgeolli).
Serial fictions pose special problems for accounts of truth in fiction. What is true according to a fiction at one time can appear to change as a story develops. Sometimes these changes are dramatic. According to A New Hope, and even according to early drafts of The Empire Strikes Back, it seems it was not true in the Star Wars fiction that Darth Vader is Luke Skywalker’s father. So in 1977 it seems wrong to say that Vader was Luke’s father, according to the relevant fiction. But in 1980, it seems right to say that Vader was Luke’s father, according to the relevant fiction.
Roy T. Cook mentions another famous example of this phenomenon in his post from July 24. Doyle intended to kill off Holmes in “The Final Problem” in 1893, but succumbed to external pressures and brought Holmes back in “The Adventure of the Empty House” in 1903.
Now the claim that Holmes is alive according to the relevant fiction seems defective in 1895 in a way that the same claim does not seem defective in 1905. The standard way of accounting for the defectiveness of the first claim is to say that it is false according to the relevant fiction. And the standard way of accounting for the non-defectiveness of the second claim is to say that it is true according to the relevant fiction. But this seems to lead to a problem.
There are a number of ways of bringing out the problem. One way would involve exploring subtleties or challenging assumptions glossed over in the set-up above. But I propose to ignore the subtleties and grant the assumptions for now.
Another way to bring out the problem is via what Ben Caplan (2014) calls “The Contradiction Problem”:
Each of the following seems true:
1. “The Final Problem” is part of the Holmes canon.
2. The Holmes canon does not contradict itself.
3. “The Final Problem” contradicts the Holmes canon.
But that seems strange. It would be nice to have an account of truth in fiction in which (1-3) are not all true.
So why think (1-3) are true?
The first claim seems true since it seems the Holmes canon is an extended fiction that begins with A Study in Scarlet and includes “The Final Problem” and “The Adventure of the Empty House”.
The second claim seems true since, ignoring the issue of the location of Watson’s wound, it does not appear that any proposition and its negation is true according to the Holmes canon. The claim that Holmes died at Reichenbach Falls is false in the canon, given that “The Adventure of the Empty House” is part of the canon and it is false according to that story that Holmes died at Reichenbach Falls. So it is false according to the Holmes canon that Holmes died at Reichenbach Falls. So it is true according to the Holmes canon that Holmes did not die at Reichenbach Falls. So, it seems—a few hiccups aside—that the Holmes canon does not contradict itself.
The third claim seems true since the claim that Holmes died at Reichenbach Falls is true according to “The Final Problem”. But its negation is true according to the Holmes canon.
In “Serial Fiction, Continued” Ben Caplan criticizes linguistic solutions to the Contradiction problem due to Ross Cameron (2012) and Andrew McGonigal (2013). Cameron’s view is a version of contextualism: to grossly oversimplify, there is no proposition that Holmes died at Reichenbach falls; rather, there is the proposition that Holmes* died at Reichenbach falls, which is what is expressed in 1895, and the proposition that Holmes** died at Reichenbach Falls, which is what is expressed in 1905. ‘Holmes’ refers to different characters in different contexts. As a result, “The Final Problem” does not contradict the Holmes canon (and isn’t really part of the canon since it’s about a different character—one that dies at Reichenbach Falls!)
McGonigal’s view is a version of relativism: to grossly oversimplify, there is one proposition expressed by ‘Holmes died at Reichenbach Falls’. And that proposition is true relative to a body of work that includes “The Final Problem” but excludes subsequent Holmes stories, and is false relative to a body of work that includes all of the Holmes stories. As a result, “The Final Problem” does not contradict the Holmes canon.
Caplan (2014) lodges a number of objections to Cameron’s contextualism and McGonigal’s relativism that I won’t rehearse here. Caplan himself prefers a different solution to the Contradiction problem—one that is more “metaphysical”. Caplan endorses what he calls “Work Contextualism”. Serial fictions have contents, in much the same way sentences do. That is, they express contents relative to contexts. To grossly oversimplify, just as a context-sensitive sentence expresses different contents relative to different contexts, serial fictions express different contents relative to different contexts. In 1895, the Holmes canon expressed some propositions that included the proposition that Holmes died at Reichenbach Falls. But in 1905, the Holmes canon expressed some propositions that included the proposition that Holmes did not die at Reichenbach Falls. As a result, “The Final Problem” does not contradict the Holmes canon.
I agree with Caplan that the problems arising from serial fiction are best addressed by a more “metaphysical” solution than by a more “linguistic” solution. But I doubt Caplan goes far enough in this regard. As Caplan notes, there seems to be an important disanalogy between context-sensitive expressions and serial fictions. Context-sensitive expressions are associated with Kaplanian characters. These can be thought of as functions from contexts to contents, and are associated with a sort of rule that “tells” us which content gets associated in which context. For example, in a context in which I am the speaker, the content of ‘I’ is me, whereas in a context in which you are the speaker, the content of ‘I’ is you. The associated rule for ‘I’ tells us to assign the speaker of a given context as the content of ‘I’ in that context. But there does not seem to be a correspondingly simple character that assigns propositions to the Holmes canon relative to contexts. This problem becomes especially difficult if we try to factor in retconning. Given sufficient artistic license, almost any serial fiction could have practically any propositions as its content at any given time.
One explanation of this apparent fact may be that serial fictions aren’t really analogous to context-sensitive expressions. But if that’s so, it seems we’re left with no solution to the Contradiction problem.
I propose that, when considering the nature of serial fiction, we should move even farther away from linguistic solutions. Let’s start by considering a case that has nothing to do with serial fiction: the Height problem. In 1977, I was less than 6 feet tall. In 2014, I am over 6 feet tall. Call ‘My Life’ the collection of facts about me. Call ‘1977-me’ the collection of facts about me that are restricted to what’s going on in 1977. Now we can present a problem that parallels the Contradiction problem:
4. 1977-me is a part of My Life.
5. My Life doesn’t contradict itself.
6. 1977-me contradicts My Life.
The first claim seems true since My Life is a collection of facts that begins with 1976-me and includes 2014-me.
The second claim seems true since it does not appear that any proposition and its negation is true according to My Life. The claim that I am less than 6 feet tall is false in My Life, given that 2014-me is part of My Life and it is false according to 2014-me that I am less than 6 feet tall. So it is false according to My Life that I am less than 6 feet tall. So it is true according to My Life that I am not less than 6 feet tall. So, it seems that My Life does not contradict itself.
The third claim seems true since the claim that I am less than 6 feet tall is true according to 1977-me. But its negation is true according to My Life.
The Height problem parallels the Contradiction problem. But it also parallels a familiar problem of change over time. My proposal is to treat the parallel problems in a parallel fashion.
According to the solution to the Height problem I prefer, I am distinct from my “matter”. Plausibly, I am the sort of thing that has cells as matter. And I change over time in part by having different matter at different times. My having considerably less matter in 1977 explains why it was true then that I was less than 6 feet tall. My having considerably more matter in 2014 partly explains why I am now over 6 feet tall. So (6) is false—that 1977-me was less than 6 feet tall does not contradict My Life since the former fact concerns the matter I had in 1977, which, as it happens, is not the matter I have today.
Note that this does not mean in any sense that I myself am a sort of context-sensitive being expressing different matter at different time. What’s going on here is just a matter of an ordinary object enduring ordinary change.
In the case of serial fictions, we can tell a parallel story. Serial fictions have “matter” as well, though they’re not the sorts of things that have cells as their matter. Rather, they have propositions as their matter. And what matter they have can change over time. In 1977, my matter determined that I was less than 6 feet tall. In 1895, the Holmes canon’s matter included the proposition that Holmes died at Reichenbach Falls. In 2014, my matter determined that I was not less than 6 feet tall. In 1905, the Holmes canon’s matter included the proposition that Holmes did not die at Reichenbach Falls. Thus (3) is false, but not because of the linguistic properties of utterances concerning the Holmes canon. Nor because the canon itself is a context-sensitive entity. Rather, (3) is false because the problem, at bottom, is the familiar problem of change over time. And the Holmes canon changed over time in such a way as to make (3) false.
This is, of course, just a sketch of my preferred solution to the problems posed by serial fictions. But I hope I have said enough to make clear what the problem is supposed to be and why it might be preferable to view the case as merely a case of change over time.
If successful, I think the account could be extended to deal in a fairly straightforward fashion with other works that are intended to change over time, such as Jason deCaires Taylor’s underwater sculptures or songs by Shellac. The matter for sculptures and songs is plausibly different from the matter for me or the Holmes canon. But I think the basic idea could be applied in the same way to these other cases.
Finally, I think the proposed solution has an advantage over Roy T. Cook’s probabilistic proposal. My proposal preserves flat-out truth and falsity with respect to fictions. I consider this an advantage because I think we have robust intuitions that some claims are flat-out true in certain fictions (Holmes is a consulting detective) while other claims are flat-out false in certain fictions (Holmes is a porcupine). And I think these intuitions are worth preserving if we can do so in a reasonable way. And I think we can.
(Thanks to Christy Mag Uidhir for the invitation to guest post and to Ben Caplan for comments.)