Uppsala University, Sweden, autumn 2018
Claim: ‘Works of fiction are not capable of proving truths about the real world; therefore philosophers who claim that works of fiction can achieve “cognitive progress” are mistaken’.
When it comes to literary fiction, many people are cognitivists. They believe that novels and short stories can teach us truths about the real world. My native Russophone culture takes this point of view more or less for granted. From Pushkin and Gogol to Petrushevskaya and Grossman, the “great” authors are considered great at least partly because they are believed to give us insights into the human condition. As an illustration, here is an actual quote from a recent literature textbook for year 10 of Russian public school: “In your opinion, did Tolstoy discover a profound truth (istinu) about humankind? If yes, what is it? If no, explain why you think so” (Nelkin & Furayeva, 2013, my translation). Students are then provided 10 blank lines for their answers.
However, the role of fiction as a vehicle of truth has been called into question. This essay will discuss a competing view: that works of fiction “are not capable of proving truths about the real world”, and philosophers (and writers of Russian textbooks) who believe “that works of fiction can achieve ‘cognitive progress’ are mistaken”.
Here is a quick outline of the discussion that is about to follow. As is standard practice in analytical philosophy, I begin by defining the key terms. Next, I identify two versions of anti-cognitivism regarding literature, which I call the Strong Claim and the Weak Claim. Roughly, the Strong Claim is that fiction has no role whatsoever in cognitive progress, while the Weak Claim is that fiction has no unique role in cognitive progress. In section 3, I discuss and reject the Strong Claim. In section 4, I make a case for the Weak Claim, based on Stolnitz (1992). In section 5, I consider arguments that might undermine the Weak Claim.
Before I go on, it is also worth clarifying what I do not intend to do. I assume that the question whether fiction can convey truths is distinct from the question whether truth ought to be considered a literary value. It is conceivable that works of fiction qua fiction do tell us something new about the real world even if truth has nothing to do with their status as literature. For this reason, nothing I am about to say below is meant to bear on the issue of truth as a literary value (though of course it might). To paraphrase Peter Lamarque (2006, p. 133), the question is not “what value to attach” to fiction-induced learning, but “whether we can learn from fiction” in the first place.
Now for the definitions. I take a work of fiction to be a text whose author acknowledges, either explicitly or tacitly, that her writing contains propositions that are intentionally false, such as factual claims about non-existent people or descriptions of events that never took place. Needless to say, a work of nonfiction, such as Mein Kampf or any history textbook with a nationalist slant, may also contain a great number of false propositions. What ultimately matters is the authorial intent: an author of fiction will readily admit that she makes things up, either to tell a good story or to achieve some other goal. She will not normally expect her readers to take every claim she makes in her works at face value.
In a fairly traditional fashion, I take truth to be the way things are as opposed to any of the ways they are not. I believe we have no viable practical alternative to assuming both that there are truths (i.e. there is a certain way things are) and that some truths can be known by humans. To know a truth is to have a justified true belief (Gettier’s examples notwithstanding). A known truth can manifest itself as a linguistic utterance, e.g. “Bulgakov wrote Heart of a Dog”, or as some other action that successfully latches onto reality, e.g. when my fingers somehow hit the right keys when I’m typing Russian on a keyboard with no Cyrillic stickers.
I am not entirely sure how to understand proving a truth. “Prove that p” seems to imply that the truth of p is demonstrated once and for all, but I doubt that such demonstrations are possible outside logic or mathematics. This is why I take proving a truth to mean “providing a good reason to believe that things are a certain way”. With that in mind, cognitive progress can be understood as finding more and more good reasons to believe more and more truths.
Now that the key terms are defined, I can restate the view I am about to evaluate. There seem to be at least two distinct interpretations of the non-cognitivist take on fiction: a weaker one and a stronger one. The weaker version denies that fiction provides a unique method of cognitive progress: there are no truths that can only (or at least primarily) be grasped by writing and reading texts about made-up events and characters. Let us call this the Weak Claim. The stronger version denies that fiction provides any means of cognitive progress at all: texts about made-up events and characters cannot give us good reasons to believe a truth, full stop. Let us call this the Strong Claim.
It is easy to see that the Strong Claim entails the Weak Claim: if fiction cannot provide any good reasons for believing truths, it cannot possibly possess a unique method of cognitive progress. This is why I will consider the Strong Claim first.
3. The Strong Claim
It is difficult to assess the claim that fiction cannot give us good reasons to believe truths about the real world before we clarify what is meant by “us”. “We” may be individual human readers, each with their own story of personal cognitive progress. Alternatively, “we” may refer to humankind as a whole. In that case, cognitive progress would be a factor of the accumulated knowledge stored by our information technology, such as books and computers, and shared among the better informed members of our species.
If what we have in mind are individual readers, the Strong Claim lends itself to falsification in a pleasantly straightforward Popperian fashion. Here is an excerpt from a Goodreads review of the novel Life and Fate by Vasily Grossman (Beavers, 2013):
I have to use the “M” word for this panoramic portrayal of the Soviet experience of World War 2—masterpiece. I was moved and uplifted, enlightened and devastated, and ultimately made into a better person with more empathy and understanding of the human condition.
The reviewer claims to have gained more “understanding of the human condition”. As the rest of the review suggests, he also believes that he has learned truths about “the Soviet experience of World War 2”, both in terms of propositional knowledge (“knowing that”) and empathic knowledge (“knowing what it was like”) (the latter distinction comes from Lamarque 2009, p. 245). Taken at face value, this review alone is enough to falsify the Strong Claim for individual cognitive progress.
What makes the falsification definitive is the fact that the review I quoted is fairly typical: individual readers routinely claim to have learned things from works of fiction. I see no reason to doubt their claims, partly because I have learned things from fiction myself, and partly because it is not difficult to see how such learning is possible. To use Elgin’s synonym for creation (2007, p. 46), writing fiction is a process of reconfiguration in which features of the real world are reshuffled and combined with made-up persons, events and objects, and even the fictional is usually a more or less novel mixture of predicates that have been true of various real persons, events and objects. In Grossman’s Life and Fate, to stick with the same example, fictional characters are woven together from threads of actual lives and personalities. They fight the Battle of Stalingrad with Soviet weapons of the period; they die in the Holocaust; they talk to Stalin. Add to that the knowledge that the author experienced the Battle of Stalingrad first hand and later became the first reporter to enter a Nazi extermination camp, and the cognitive value of the novel for the average 21st-century reader seems uncontroversial.
Can works of fiction serve as a vehicle of cognitive progress for our civilization as a whole? This question also seems to lend itself to empirical investigation. Has fiction been used to discover or convey truths about the real world? Again, it seems almost trivially true that it has been. For example, one of the few surviving accounts of the Great Terror that were actually written during the Great Terror is Sofia Petrovna, a novella by Lydia Chukovskaya. It is a fictional narrative, populated by fictional characters. As such, it is arguably less reliable as a historical document than a straightforward memoir. However, its value for our understanding of the workings of the repressive totalitarian state in its Soviet incarnation is considerable. If we keep in mind that most non-fictional accounts of the pre-war Stalinist purges, such as Yevgenia Ginzburg’s powerful Journey into the Whirlwind, were written many years after the events and often use fictional narrative elements (Ginzburg’s memoir, for instance, contains a lot of dialogue), the importance of Sofia Petrovna for our knowledge of history becomes even harder to deny. Just as undeniable is the fact that fiction – or at least what we would call fiction today – has been used as a vehicle of philosophical reflection. Parmenides presents his thoughts in the form of a fictional encounter with a deity; Plato has his characters have fictional conversations about fictional lands in the middle of the ocean.
To sum up, the Strong Claim appears to be untenable. Both individual human readers and humankind as a whole can and do learn truths from works of fiction. However, the anti-cognitivist has an obvious rejoinder up her sleeve: sure, fiction can moonlight as a historical document or be press-ganged into expounding philosophy, but non-fiction can do the same job at least just as well and likely much better. It is understandable, the anti-cognitivist might say, that in the days of old, when the boundaries between functionally different kinds of writing were blurred or non-existent, fiction was employed for all sorts of purposes. Nowadays, however, we have news reporting, political debates, and academic journals dealing with all manner of subjects, including ethics, psychology, sociology and anthropology. Granted, works of fiction may still teach individual readers a thing or two about the human condition, but they have no truths to contribute to what is discovered by other means. Fiction, the anti-cognitivist will insist, provides no special access to the way things are.
In other words, it is time to consider what I have called the Weak Claim.
4. The Weak Claim
The Weak Claim, recall, denies that fiction has its own unique method of cognitive progress. On this view, any truth that can be known at all can be learned by means other than writing and reading narratives about made-up characters and events. A version of the Weak Claim is forcefully put forward by Jerome Stolnitz (1992). I will begin with a summary of his argument.
So far as I can tell, the variety of cognitivism Stolnitz attacks is the one that has been around at least since Aristotle. In chapter 9 of Poetics, Aristotle famously states that poetry, i.e. fiction, “is at once more like philosophy and more worth while than history”, i.e. non-fiction, because “poetry tends to make general statements” (or “express the universal”, in another translation of τὰ καθόλου). This view has proved resilient. Great works of fiction, Stolnitz observes, are still felt to reveal “psychological truths about people in the great world, truths universal” (1992, p. 193).
However, Stolnitz argues, whenever we try to cash out this view, the supposedly deep truths of literary fiction invariably turn out to be “distinctly banal” (1992, p. 200) and readily available elsewhere. (To be exact, Stolnitz talks about the “cognitive triviality of art” in general, but all his examples come from fiction or poetry.) As an illustration of Stolnitz’s approach, let us take a great novel and see what profound insights it can afford us. What, for example, can we learn from Grossman’s Life and Fate? Here are some candidate propositions:
War is bad.
Hitler’s totalitarianism and Stalin’s totalitarianism were fundamentally the same.
Even intelligent, freedom-loving people are flattered by the attention of those who hold absolute power.
And so on. Life and Fate is a wide-ranging work on a par with War and Peace, and it is possible to extract from it quite a few other general statements about human psychology and society. However, as Stolnitz would be quick to point out, the novel is still a fiction, i.e. a collection of untruths, and as such it “does not and cannot provide the evidence” (1992, p. 196) that any of these statements are in fact true of the real world. As a civilization, i.e. a community of cognizing subjects engaged in a collective pursuit of truths, we know that war is bad because we have had real wars, not because we have read great novels. We know that Nazi Germany and Stalin’s Soviet Union were in many ways similar because we have insightful nonfictional analyses of their history, such as Hannah Arendt’s The Origins of Totalitarianism or Raymond Aron’s Démocratie et Totalitarisme. We know that intellectuals can succumb to a dictator’s charm because nonfictional intellectuals have been lured into service by nonfictional dictators. And so on. To be sure, works of fiction may contain truths and even teach them to less informed individuals. Yet fiction has “no method of arriving at” truths, and it offers “no confirmation or possibility of confirmation” (Stolnitz, 1992, p. 200).
It is hard to disagree that Stolnitz has a point: the cognitive role of fiction has been exaggerated, at times greatly. I am tempted to add that some cultures have been more guilty of this than others, perhaps for understandable historical reasons. For instance, the traditional Russophone view of fiction as a natural medium of truth may well have something to do with the relative underdevelopment of journalism, public political discourse and academic philosophy in the 19th-century Russian Empire. Considering that Soviet political and ideological censorship of the public sphere was, if anything, even more intense, it is then little wonder that a radically cognitivist view of literary fiction did not go away. Many Russian and Soviet readers alike would read fiction hoping to glimpse the truth about their lives that was not available anywhere else.
From a modern Western point of view, such cognitive reliance on fiction seems abnormal, and efforts to cut fiction down to size as regards its truth claims are intuitively appealing. However, the Weak Claim as defended by Stolnitz does seem to have some blind spots. They are the subject of the next section.
5. Undermining the Weak Claim
Catherine Elgin has argued that much of the skepticism about the cognitive value of fiction presupposes an “information-transfer model of cognitive progress”. According to this model, knowledge consists of “discrete bits of information”. Our epistemic goal is then to “amass as many true bits and as few false bits as possible” (Elgin, 2007, p. 44).
The information-transfer model is evident in Stolnitz’s treatment of fiction. The discrete units of information he operates with are propositions. Consider his proposal for the universal truth supposedly conveyed by Pride and Prejudice (Stolnitz, 1992, p. 193):
[It is true that] “[s]tubborn pride and ignorant prejudice keep attractive people apart”.
If we want to condense the main theme of Austen’s novel in one proposition, preferably free of any time-specific or place-specific detail, Stolnitz’s summary is as good as any other. However, the question is whether this proposition – or, for that matter, any other proposition – offers the same cognitive pay-off as the novel itself, when read from cover to cover. What I mean by this is not that we might need many more propositions in order to do justice to Pride and Prejudice. Rather, my concern is what exactly we manage to communicate when we say “stubborn pride”, “ignorant prejudice” or “keep people apart” – or when we use any other noun phrases and predicates to summarise an accomplished novel as a list of propositions. In other words, I suspect that certain truths may be hard to communicate in non-narrative, stripped-down propositional form.
Yury Lotman once observed that a completely accurate transmission of any “sufficiently complex message” requires conditions that are impossible to meet “in a natural situation”. For a truly flawless transmission, we would need two people who encode and decode messages in exactly the same way. Besides speaking the same language (“English, Russian, Estonian etc.”), they would have to follow identical rules of reference and pragmatics. They would need to have had the same linguistic and cultural experiences. Basically, they would need an identical set of memories. This is why in actual human communication “a perfect match between the code of the transmitter and that of the receiver is only possible to some degree of approximation” (Lotman, 2016, p. 20, my translation).
What is a “sufficiently complex message”? I am not in a position to offer a theory of semantic complexity. Fortunately, I do not need one; for my current purpose it is enough to accept that the degree of approximation involved in transmitting some messages is simply greater than in others. For example, it seems intuitive that a message such as “Pride and prejudice keep attractive people apart” stands more chance of being construed in wildly different ways than the claim that it rains in England.
Now recall the proposition that I used earlier to sum up one of the key themes of Grossman’s Life and Fate:
[It is true that] war is bad.
The formulation is deceptively – should I say “seductively”? – simple. But what does it actually say? There is a multitude of ways in which a complex phenomenon such as war can be bad for different people, non-human animals, or the planet as a whole. Even those with first-hand experience of war may not be fully aware of many of its terrible consequences. Then there is the normative aspect of the proposition. “War is bad” sounds like a claim with straightforward normative implications, but it is unclear what kind of “ought” we are supposed to derive from it. Is war bad as in “we ought to do everything we can to avoid it”? Or is it bad as in “we only ought to wage war if we really want to expand our colonial holdings”? And so on.
All of this suggests that, despite its formal simplicity, “War is bad” carries a relatively complex message. If we want to communicate that message with a reasonable degree of approximation, we have two options. We could look for a receiver who is sufficiently like us and hope that they decode “War is bad” the way we want them to. Alternatively, we could try to encode the message in such a way that, to use a computer metaphor, it might gradually “self-extract” and “install” itself in the receiving mind by imitating some of the experiences necessary for its decoding. One thing the anti-cognitivist seems to overlook is that, when it comes to complex messages about the human condition, literary fiction might be particularly suited to such life-imitating message encoding. The key advantage of fiction here is the freedom of what Peter Swirski calls narrative modelling (Swirski 2000). An author of fiction is free to use whatever reconfiguration of real and imaginary features she finds necessary to provoke a response in the reader’s mind. She can verbalise her characters’ stream of consciousness; she can have them speak in catchy one-liners and follow a dramatic narrative arc. By contrast, nonfiction writers have the freedom of narrative modelling only to the extent they are willing to fictionalise their texts (e.g. by imagining people’s thoughts and feelings on particular occasions or “re-creating” unrecorded dialogue, not to mention fictionalising by omission, i.e. by leaving out potentially significant facts that interfere with the desired narrative).
To summarise the discussion so far, one potentially distinct cognitive role for fiction is that of bridging the experiential gap between the author and the reader, allowing complex messages about the human condition to be transmitted more efficiently. Earlier I have described the cognitive progress of our civilization as a factor of the accumulated knowledge stored by our technology and shared by individual humans. On the information-transfer model of cognitive progress, all knowledge can be stored and shared as lists of discrete propositions, ideally worded as single sentences of manageable length. However, as long as we have reason to suspect that certain kinds of knowledge cannot be stored and transmitted in that way, we need to look for cognitive tools beyond lists of propositions. Fiction may well be one such tool.
Another potential cognitive role for fiction is, in a sense, an extension of the one just discussed. Here the key idea is not that Life and Fate (the novel) and “War is bad” (the proposition) may transmit the same message with a different degree of accuracy. Instead, the focus is on the fact that the very concepts “war” and “bad” may come to acquire new meanings for someone who has read Grossman’s novel. As Eileen John puts it, fiction can challenge and refine our conceptual scheme; it can put into question “the very terms of our thought”, causing us to use them in novel ways (John, 1998). Since our cognitive progress must involve a feedback loop between empirical and conceptual investigation, such conceptual probing and stretching by means of fiction may eventually help us to learn new truths about the real world. John admits that this view assumes a continuity between the use of concepts in fictional contexts and their application in real life (1998, 341-342). This assumption, however, seems psychologically plausible – or at least more plausible than the competing claim that readers routinely distinguish between concepts as used in fiction and concepts as used elsewhere.
A particularly helpful way to understand how fiction can challenge and refine our conceptual habits is to think of certain works of fiction as extended thought experiments. Peter Swirski defends and develops this approach at book length in Swirski (2007), arguing that literary fictions can generate “nonfictional knowledge” thanks to “their capacity for doing what philosophy and science do – generating thought experiments” (p. 4). Just like thought experiments in philosophy and science, literary Gedankenexperimente tease out “consequences of events that by definition did not occur” (p. 6) and thus alert us to what can happen in the real world, given similar circumstances. As long as we are willing to grant cognitive value to an account of an imaginary event in a philosophy paper, we should be willing to do the same for an imaginary event in a work of fiction.
Catherine Elgin is another scholar who offers an account of the importance of literary thought experiments for cognitive progress (Elgin, 2007). Works of fiction, Elgin argues, can train us to notice the cognitively salient features of a situation. Quite often, our epistemological challenge is not that we have too few facts at our disposal; rather, it is that we have too many. This is why, to make sense of what is going on, we need to decide which facts to ignore and which to keep in focus; and we need to decide how to arrange the facts we keep so that they causally hang together (Elgin, 2007, p. 44). To use another computer-inspired metaphor, we need to impose a “select and crop” pattern on an unpatterned flow of reality. Now, imposing a causal pattern on a set of events could easily serve as a definition of building a narrative. It therefore stands to reason that fiction writing, as the narrative activity par excellence, might help us to produce new “select and crop” patterns to impose on the world. Equipped with these patterns, we might then “discern truths that we would otherwise not see or not see so clearly” (Elgin, 2007, p. 53).
To sum up, I have tried to show that there are at least two plausible roles fiction can play in the cognitive progress of our species. Firstly, we can use fiction to store and transmit complex messages whose efficient decoding requires a high degree of similarity between the transmitter and the receiver. Fiction may be particularly suited to this purpose because it can freely model and imitate various experiences in the reader’s mind. Secondly, we can view works of fiction as extended thought experiments. Like thought experiments in other domains, they can make us think about the way we use our concepts, and they can offer us conceptual models for making sense of the real world. In other words, an author of good fiction such as Tolstoy may not be able to give us an istina, i.e. profound truth, about humankind, but his narrative modelling might well help us on the way towards that cognitive goal. That this help comes in the form of conceptual, rather than empirical, reasons to believe new truths about the world need not detract from its contribution. As the second half of the well-worn Kantian adage has it (KrV, A 51, B 75), “Anschauungen ohne Begriffe sind blind”; we have to impose concepts on observations – our “Anschauungen … unter Begriffe zu bringen” – to see and make sense of the world, and we ought to keep working on the concepts we use for that purpose.
There are at least two obvious objections available to the anti-cognitivist, and both of them arise even if we accept the cognitive roles of fiction I have just described. As regards the first role (bridging experiential gaps), the anti-cognitivist might argue that storing or even spreading truths is not the same as arriving at them or confirming them. No one denies that fiction can serve a didactic purpose, but does that really count as a contribution to the cognitive progress of our species in the strict sense? As for thought experiments in literary fiction, it is not immediately clear what makes them distinct from thought experiments in philosophy or science. Can a Gedankenexperiment built into a novel do something that a strictly philosophical thought experiment cannot? And even if we grant that literary thought experiments are somehow distinct, one does wonder if “distinct” in this case is not synonymous with “distinctly inferior”. All those additional narrative and aesthetic dimensions one typically finds in fiction – do they not rather distract from whatever cognitive value the author has managed to produce?
Elgin (2007) and especially Swirski (2007) both go some way towards answering the second objection, and I believe there is a way to diffuse the first objection as well. Dealing with these objections, however, goes well beyond the scope of this essay. As they say, I hope to be able to return to them in the future. In the meantime, I will conclude with a farewell quote from Swirski (2000, p. 76), writing here about Stanisław Lem, an author responsible for much of my cognitive development in my early to mid-teens:
In [Lem’s] hands fiction becomes a flexible and versatile instrument of inquiry, while his sophisticated use of thought-experiments, counterfactual scenarios, or various semantic, semiotic, and mimetic games shows that there are almost no limits to its modelling power.
 Whether philosophy itself, however presented, has contributed to the cognitive progress of our civilization is a different matter. For the sake of this discussion, I simply assume that it has.
 With the possible exception of highly trained academics.
 Swirski’s treatment of fictional thought experiments is both comprehensive, insightful and somewhat dispiriting since a mere reference to his monograph would likely be a better answer to the question of this assignment than anything I might ever say.
In case anyone cares to cite this essay
Zarubin, K. (2018). Can we learn from fiction? Unpublished manuscript. Retrieved from https://kostia.me/english-deutsch/can-we-learn-from-fiction/
Beavers, W. (2013). Life and Fate by Vasily Grossman. Goodreads. Retrieved 9 December 2018, from https://www.goodreads.com/review/show/621152021
Elgin, C. Z. (2007). The laboratory of the mind. In W. Huemer, J. Gibson, & L. Pocci (Eds.), A Sense of the World: Essays on Fiction, Narrative and Knowledge (pp. 43–54). London: Routledge.
John, E. (1998). Reading fiction and conceptual knowledge: Philosophical thought in literary context. Journal of Aesthetics and Art Criticism, 56(4), 331–348.
Lamarque, P. (2006). Cognitive values in the arts: Marking the boundaries. In M. Kieran (Ed.), Contemporary debates in aesthetics and the philosophy of art (pp. 127—39). Blackwell.
Lamarque, P. (2009). The philosophy of literature. Oxford: Blackwell.
Lotman, Y. (2016). Vnutri myslyashchih mirov. St. Petersburg: Azbuka-Klassika. Non-Fiction.
Nelkin, A., & Furayeva, L. (2013). 10 klass. Literatura. Rabochaya tetrad’. St. Petersburg: SMIO-Press.
Stolnitz, J. (1992). On the cognitive triviality of art. British Journal of Aesthetics, 32(3), 191–200.
Swirski, P. (2000). Between literature and science: Poe, Lem and explorations in aesthetics, cognitive science, and literary knowledge. Montreal: McGill-Queen’s University Press.
Swirski, P. (2007). Of literature and knowledge: Explorations in narrative thought experiments, evolution and game theory. New York: Routledge.