Lewis and Kripke, Counterfactuals and Determinism, and Authors and Truth
April 29, 2025
Mark Stanley
Foreword
This is my final paper for Philosophy 516: Language and Meaning. It connects a few ideas relating to truth in fiction and the truth of counterfactuals. The first part of the discussion was initially based on a misinterpretation of how Lewis argued we should determine the truth of a counterfactual statement. Though that ended up not being the case, it can rather serve as an analogy as to why the consensus of possible worlds seemed appealing to him, and why we should not prefer it as a theory. I finally propose the appeal to the author of a fiction as how we should determine truth in fiction, and point to a problem that such a viewpoint may present.
Body
Kripke and Lewis promote varying arguments dealing with the truth in fiction. How can one make seemingly sensible claims, such as “Sherlock Holmes was a detective.” without the name Sherlock Holmes rigidly designating anything? To Kripke, that claim is strictly speaking false. However, to evaluate the statement, we consider the pretend world in which Sherlock Holmes exists: the Sherlock Holmes stories. What people often mean when they utter “Sherlock Holmes was a detective.” is something like “In the Sherlock Holmes stories, Sherlock Holmes was a detective.” This is an example of a pretense, which can differ from speaker. The latter is a true statement, the former are not even propositions. For the purposes of my discussion, my writing of the former sentence will mean the latter sentence.
Lewis builds upon this theory. Lewis specifically promotes that sentence, meaning I pointed out earlier. Lewis initially notes that we can make true assertions about things in a fictional work without it being expressly stated. It was never explicitly stated that Sherlock Holmes did not go to Jupiter; however, claiming that “Sherlock Holmes never went to Jupiter” is seemingly true. His solution is two theories that depend on, roughly, the similarities of our world to make assertions about the truths in the fiction. We are told to assume that the claim actually happened in a world similar to ours, and if we can imagine that as being true in most of these similar worlds, we are to evaluate the sentence as true. So, if we are to assume that Sherlock Holmes was indeed a real person at that point in history, most possible worlds we can imagine do not have him travelling to Jupiter. Thus, “Sherlock Holmes did not travel to Jupiter” is a true statement.
To contextualize my argument against Lewis’s use of possible worlds for evaluating a fiction's truth, I will start by dissenting from the consensus of possible worlds when considering counterfactuals: evaluating the truth of a counterfactual by considering if, in most possible worlds from that counterfactual, the claim is true. Though this seems like a common approach, I wish to argue that there is only a single possible world we need to consider. This discussion will assume determinism is true.
Consider regular counterfactuals, for example: “If I did not set an alarm, I would not have made it to class.” To predict whether this is true or not, we would make predictions from what would happen at that instant of that possible world. We would further consider multiple possible worlds after that initial antecedent to help us think of the possibilities. In our example, we may consider: where I kept sleeping in, where my roommate awakens me, where a later alarm goes off, etc., in an effort to figure out what could happen, and how this would affect the consequence. Importantly, this is also how we make inferences about what will happen in our regular world, not necessarily a counterfactual world. The reason we go through this thought process is that we cannot omnisciently predict how our world will go; rather, we need to consider likely future worlds to help us make good decisions against likely future outcomes. However, determinism claims there is only one way the world can go from the present. Thus, any future claim we make should be true if it will happen, and false if it won’t. It doesn’t matter what is most probable from possible worlds. “It will rain tomorrow.” should be evaluated as true if it does rain tomorrow, and false if it doesn’t. One ought not worry about what fraction of future possible worlds it rains during.
Analagously, it seems from a single instance of a counterfactual, there is only a possible world that can come from it. Sure, to attempt to predict the truth value of a claim in that possible world, we may consider further possible future worlds to make an inference, just like we would in our world. However, from that specific instance, there is only one way it can go. Thus, any single counterfactual claim can be evaluated solely on whether it is true or not in that single possible world. We don’t have Laplace’s Demon to figure out what would have happened from that instance, but because it is theoretically possible, that is what that truth should be evaluated to. Possible worlds only help us predict the truth of a counterfactual, they do not determine it.
All of this discussion is to say that I think Lewis is making this same mistake. Considering other possible worlds is useful to us when considering whether or not we think something should be in fiction. It may indeed be true in most similar possible worlds to ours, where the stories of Sherlock Holmes are assumed truthful, that Sherlock Holmes did not go to Jupiter. This theory allows us to make useful predictions about what we think is true in fiction, and correctly goes through our intuitions on how we predict whether something is true in fiction. This predictive utility does not imply that it grounds truth.
However, this needs a little more motivation. I have only pointed out how this is analogous to a system of inference, and not motivated why we should not prefer this theory. I think this is also quite clear: “most” is just too vague and not a useful way to determine truth. We should want truth to be dependent on something specific, not an amalgamation of different things. Furthermore, ‘most’ is hardly well defined. If something is true in 51% of worlds, should that be true, while 50% is untrue? Not to mention the efficacy of actually considering and determining what possible worlds qualify. It is practically convoluted.
I wish to promote a better account. Let us first consider the intuition behind this. First, I agree with Lewis that we cannot purely determine the truth of the fictional world exclusively based on the text. Consider the unreliable narrator trope. Most people would not claim that the way they describe their situation is what is true in that fictional world. I wish to promote the idea that it is the author’s intention that determines whether or not something is true in a fictional work. Roughly, “‘In fiction f, p’ is true iff the author of the fiction a would think it is true.”
For our intuition, imagine there is some ambiguous detail in a book that sparks intense debate in the reading community. Imagine some of the interpretations have controversial conclusions. So, the author, wanting to clear up the text, releases a statement saying that the controversial conclusion is not true. I would find their account moving, and read the book as described. Considering what an author meant should determine what is true in the fiction.
However, what about situations that the author hadn’t considered? Sir Arthur Conan Doyle never considered whether or not Sherlock Holmes owned a lightsaber. To respond to this, I say that the truth of these situations is how the author evaluates the statement “In fiction, Sherlock Holmes owned a lightsaber.” Though they may not have had the intention, “Sherlock Holmes does not own a lightsaber.” It is clear that when asked, Doyle would say this is false (after, of course, learning what a light saber is).
This presents another issue, that the author could change their mind when asked at different times. This is common practice in many works of fiction, that there is a name for this practice, retconning: “a piece of new information that imposes a different interpretation on previously described events, typically used to facilitate a dramatic plot shift or account for an inconsistency.” Oftentimes, the author intended for something to be the case, but in wanting to further the plot, adds a detail into the story that keeps the fictional world consistent, but goes against what they initially intended. Thus, the time at which this question is asked can change the answer.
I argue all that this does is add another stipulation to the theory: “‘In fiction f, p’ is true at time t1 is true iff the author of the fiction a would think it is true at time t2.” t1 here is the time in which the claim is made, and t2 is any specific time that the author existed. In most ordinary language, I think most people intend t2 to be just after the author finished writing the last story in the canon. This account is compelling because we may want to be able to say that “Before the retcon in Friday the 13th: Part II, Jason Voorhees died” and be able to say this is true. In this case, t2 is the time at which Sean S. Cunningham just finished writing the original Friday the 13th. However, just saying “Jason Voorhees dies in Friday the 13th” is not (strictly) true, as most people assume one is talking about the entire canon (t2 is the time at which Cunningham just finished writing the last movie).
One final issue presents itself: a work of fiction having multiple writers. I am not concerned with people like editors, who just look back over a book for grammatical errors; these people would not have a say over fictitious truth. It is not hard to imagine writers for a movie each individually writing their own portions. In many cases, we may be able to say that the creative director has the truth power, or the person who initially conceived the idea. However, we can imagine an improvisational show in which the fictitious world being painted is being equally contributed to by multiple participants. In this case, I would have to bite the bullet and say that the truth is what most of the creatives involved believe is true. I unfortunately cannot imagine a better way out of this.
Some dissenters may argue that the fictitious world in a text should exist independently of its author. I wish to promote a few ideas that I hope will guide our intuitions otherwise. First, I wish to say that a book should be interpreted independently of the author's intention; one can gather any interpretation and apply it in any way they see fit. However, what the text says is another story. Initially, consider some non-fiction work, like a philosophy paper. If Russell wrote some ambiguous statement, and someone asked him to clarify, he would respond with his clarification. It would be quite odd for someone, when hearing this, to say, “It doesn’t matter what he intended; it is not in the text, thus we must throw this knowledge out.” We understand that the intentionality of non-fiction works can be used to understand what the author meant.
Why would this be different from fiction? I think the intuition is that people want to be able to split the work from the author for practical purposes. It is better for colleagues to make inferences from the text when discussing truth in a fiction. A claim like “But the author actually intended p.” will typically be a baseless claim and not promote discussion. Furthermore, as I stated earlier, the interpretation of a text is most useful independent of what an author intended, so people may want truth in fiction to be equally independent. However, it is more useful to have an account that is determinate, not an account where we need to consider all possible worlds.
Overall, I hope to have presented an interesting discussion against the typical Lewis approach to evaluating counterfactuals. Assuming that conclusion, we have reason to question Lewis’s account of truth in fiction. Rather, having the author determine the truth in fiction presents us with a more consistent way to evaluate truth.