1. Introduction
The debate about the potential value of art goes at least to Plato’s banning of artists from his ideal Republic for misrepresenting the world and thereby corrupting its citizens (Plato 1997). In doing so, Plato also set the terms of the debate which has since mostly revolved around whether art can be a source of epistemological and/or ethical knowledge (Aristotle 1987; Horace 1990; Boileau 1674; Schiller 1794; Spivak 2012). Put differently, the value of art has traditionally been construed in terms of the potential, on the one hand, for conveying truths about the world (Walsh 1969; Lamarque and Olsen 1994) and, on the other, in promoting ethical development (Shelley 1891; Levinson 1998). While in its broader sense, aesthetic cognitivism may be said to cover both these claims, in this paper I want to investigate only the former—I will refer to this as aesthetic cognitivism narrowly understood.
In what remains, I will argue that, while narrow aesthetic cognitivism is certainly a possibility, as it stands philosophers subscribing to it have not produced convincing arguments in favor of it. That is because they have primarily engaged in theoretical debates rather than in discussions of existing empirical literature on the subject. To make my case, I will first distinguish between narrow and broad aesthetic cognitivism. I will then propose that the scope of the narrow claim needs to be quantified with more care than has been hitherto. Having done so, I will survey different types of arguments in favor of the narrow claim and demonstrate that they fail to accomplish what they set out to do. I will conclude with a discussion of what would count as evidence in favor of aesthetic cognitivism.
2. Weak, broad, and narrow aesthetic cognitivism
Here is a sample of explicit articulations of aesthetic cognitivism.
Aesthetic cognitivism may be said to be the view that art at its best is a form of understanding and as such, though it differs greatly in other respects, is to be accorded the same evaluative status as science, a status which its undoubted capacity to entertain and give us pleasure could not justify. (Graham 1996, 1)
Aesthetic cognitivism (…) is best thought of as conjunction of two claims: first, that art can give us (non-trivial) knowledge, and second, that the capacity of art to give us (non-trivial) knowledge (partly) determines its value qua art, i.e. its aesthetic value. (Gaut 2005, 437)
Aesthetic cognitivism and aesthetic anti-cognitivism can be defined as opposing responses to two questions: (1) Can art provide knowledge? And, if it can, (2) how is this aesthetically relevant? (Thomson-Jones 2005, 376)
[T]he claim that narrative art is a source of truth, capable of transferring knowledge and other cognitively relevant states (aesthetic cognitivism). (Vidmar Jovanović 2019a, 18)
One possibility is that art can promote new knowledge and understanding, a notion referred to as aesthetic cognitivism. (Christensen, Cardillo, and Chatterjee 2023)
As we can see the claims are both quite broad and quite weak. In the former case, Gaut and Thomson-Jones point out that the key component of aesthetic cognitivism is not only the epistemic import of art but that this epistemic payoff is also a key part of the aesthetic value of art. Concerning the latter, their proponents explicitly state than only some art conveys knowledge or that art can on occasions be of epistemic value by conveying knowledge.
Here, I will be focusing exclusively on the question of epistemic value without exploring whether or how it relates to the problem of aesthetic value. But I want to be even narrower than that. I also want to delimit “knowledge” that I will be addressing from those more broadly understood “form[s] of understanding” and “other cognitively relevant states”. Here I am discounting potential wider cognitive benefits such as the positive impact of art on imagination, emotions, curiosity, reasoning, etc. I will also not be addressing the debate on whether art conveys moral knowledge (for discussion, see Nussbaum 1990; Gaut 2005; Plantinga 2018). For one, a recent research project led by Vidmar Jovanović explicitly treats moral knowledge under a banner distinct from aesthetic cognitivism which they refer to as aesthetic education: “the exploration of the transformative power of aesthetic experience with respect to one’s ethical development” (2020).1 The other reason is that skepticism of moral knowledge is far more prevalent than skepticism about traditional knowledge making the debate potentially moot (Campbell 2019). Instead, I am zeroing in on less controversial forms of knowledge usually captured under the banners of know that, know how, and know what it is like that I will outline in the next section.
The second point I want to address pertains to the “weakness” of the claim. In the forms expressed above, it is hard to deny that art can and does convey knowledge. Religious art regularly conveys religious truths. Michelangelo’s Sistine Chapel Ceiling (1508-12) imparts, among other things, the knowledge that in the Christian tradition God created Adam. Calvary paintings standardly inform the viewer that in the same tradition Jesus was crucified and died on the cross. Portraiture usually provides information about how the subject looked like. Jacques-Louis David’s painting Portrait of Pope Pius VII (1805) tells us a lot about the facial features of the eponymous sitter at the time the painting was made. Eddie Adams’ photograph Execution of Nguyễn Văn Lém (1968) informs us not only about the appearance of both the victim and his executioner but also a number of things about the act of killing (type of gun, place, its summary nature, etc.) as well as of the emotions of the two people at the fateful moment. Grizzly Bear (2005), a documentary by Werner Herzog, imparts next to, again, a range of visual information (about how its protagonist Timothy Treadwell looked like or the places he visited), also the information about Treadwell’s life, such as his enthusiasm for grizzlies and eventually death at the hands of one. Truman Capote’s 1966 non-fiction novel In Cold Blood teaches the readers about the particularly gruesome murders that took place in Holcomb, Kansas in 1959. Under this analysis, the weak version of aesthetic cognitivism is trivial, and the underlying claim must be stronger.
That this is the case and that the quoted general formulations are somewhat disingenuous is evinced by the fact that the discussions that follow regularly turn to fictional narrative arts such as literature and film. Graham (1996) immediately turns to literary fiction as does Gaut (2005) and Vidmar Jovanović (2019a) is explicit that she is concerned with narrative art and in practice tackles fictional narratives.2 In other words, the non-trivial discussions about whether art imparts traditional forms of knowledge are really about whether fictional works of art or art fiction does so. In the first instance, therefore, I take (non-trivial) narrow aesthetic cognitivism to mean something like art fiction conveys knowledge.3
3. Narrow aesthetic cognitivism
In the previous section the problem of quantification already arose, but there I discussed it in terms of the claims’ “weakness”. While modals and quantifiers are not the same thing, the way in which proponents of narrow aesthetic cognitivism articulated in terms of “can” proceeded to evince their position was to find examples of at least some art fictions which convey knowledge. This is why I also framed my discussion as a list of examples vindicating the weak thesis. And while I have avoided modals in my formulation of narrow aesthetic cognitivism the problem remains. Would one fictional work of art suffice to evince narrow aesthetic cognitivism? Would one person learning something be sufficient? Who is this person who acquires knowledge? How much knowledge and, or, what is the relevant type of knowledge that should be acquired here? Let us address these questions in turn.
3.1 Art fiction
Proponents of narrow aesthetic cognitivism do not usually explicitly address what subset of art fiction is relevant. One option is that the discussion here is really about high art fiction. The examples that reappear in the discussions such as Tolstoy and Dickens are certainly consistent with this version. However, when narrow aesthetic cognitivists address other media more middle brow example start appearing including Mission Impossible and Alien series as well as Game of Thrones (HBO 2011-2019). This suggests that the implicit relevant subset is at least something like high and middle brow art fiction. Importantly, we are talking about art fiction stories. When in the rest of the paper I write something like art fiction imparts knowledge I take this to mean a significant subset of art fiction stories of at least high and middle brow nature imparts knowledge.
3.2 Benefactors
The statement that art fiction confers knowledge does not specify who the beneficiaries are. I suggest that we should be interested here more in the audiences than in the makers of art fiction because it is uncontroversial that making an art fiction involves a more sustained engagement than reading, watching, or listening to it. For instance, writers regularly learn about the subject of their writing. It is hard to deny that in writing three books about Thomas Cromwell, Hillary Mantel acquired substantial knowledge about her protagonist and 16th-century England. The same holds for directors and screenwriters. In making Oppenheimer (2023), Christopher Nolan must have learned a lot about the eponymous physicist’s life. In bringing characters to life through performance, actors learn about the people they portray. Through filming The Iron Lady (Phyllida Lloyd 2011), Meryl Streep acquired, among other things, knowledge on both how Margaret Thatcher’s movements look like and how her voice sounds like as well as how to imitate them.
It might be objected that it is not really art fiction making that confers knowledge here but the preparation for the production of an artwork i.e., the research that precedes it is the relevant process. On this account Mantel learns about Cromwell and his times not by writing proper but in reading the books about the man and the period which she only later applies to writing. The same holds for Nolan and other biopic makers. And for Streep, similarly, it is watching and listening to the tapes of Thatcher that secures the actor’s knowledge about her subject, not the performance itself.
The objection holds only under an improperly limited understanding of what it means to make art fiction. This cannot exclusively refer to the acts of putting certain words on paper, instructing the film crew what to do, or making certain bodily movements in front of the camera. Making art fiction also entails researching one’s subject, thinking about how to execute one’s craft, drafting, and redrafting what has been done with reference to previous research and attempts, etc. This is no different from writing an academic paper—in writing this I have learned a lot about aesthetic cognitivism. But on the limited view of academic writing which entails only the typing of the words and leaves out the preceding research I would have acquired no knowledge. Moreover, even this limited view which considers writing only to be the act of putting words on paper, must admit that this act of writing is not merely the movement of hands but also involves a lot of mental processing such as organizing of thoughts and articulation of those thoughts in words. Similarly, the very act of executing one’s craft, be it novelists’ and scriptwriters’ writing, instructing the film crew, or moving for the camera, also brings clarity to the ideas that one has about the subject of that craft. So even in putting the words on paper and formulating her sentences Mantel must have formed new connections and gained a clearer picture of Cromwell and early modern England. In instructing Cillian Murphy on how to perform, Nolan certainly arrived at a firmer grasp of Oppenheimer and his motives. In performing for the camera through multiple takes Streep developed her knowledge of how to sound and move more like Thatcher.4
Therefore, it is hard to deny that art fiction regularly confers knowledge to its makers. The more controversial question—and certainly the one that has garnered more attention but remains underspecified in the definitions of aesthetic cognitivism—is whether art fiction conveys knowledge to its audiences.
Once this is determined, we also need to ask what the size of the effect is. In other words, what subsets of audiences should gain knowledge for narrow aesthetic cognitivism to hold? There appears to be agreement that the audiences who will profit are those who are sufficiently attentive. Gregory Currie, for instance, talks about “suitably prepared readers” (2020, 158). Iris Vidmar Jovanović has in mind “an active, reflective reader” (2019b, 366). Stacie Friend, similarly, argues that all audiences are expert audiences for at least some genre (2014, 244). At the same time, narrow aesthetic cognitivists would also probably allow that art fiction imparts knowledge of the aesthetic world more easily than knowledge of the non-aesthetic one. Put differently, while these attentive audiences will regularly gain knowledge about the story-world this will be less frequent for knowledge about the actual world. Put in yet another way, under this view learning from art fiction is something like learning through classroom lectures—some students will be able to say what the lecture was about and others not, and a subset of the first will also have learned about some claims made in the lecture while another subset of the same will not. When I write that art fiction imparts knowledge from hereon, then, this is shorthand for saying a subset of art fiction stories of at least high and middle brow nature imparts knowledge to a significant subset of audiences.
3.3 Knowledge
It is also uncontroversial to say that all art fiction conveys knowledge, if we do not specify what kind of knowledge we are interested in. Put otherwise, if we take knowledge simply to mean some form of justified true belief, then art fictions regularly convey knowledge about what is true in the world of fiction. Upon reading Anna Karenina (1878) audiences learn, among many other things, that Tolstoy’s heroine killed herself by jumping in front of a train. Furthermore, audiences regularly acquire knowledge of genre conventions through engaging its members. After a certain number of contemporary zombie horror films, audiences recognize that in these films humans are oftentimes a greater danger to protagonists than zombies. Having seen a sufficient string of whodunnits, audiences will know that the convention of the genre is an elaborate revelation of how the crime was committed and by whom in the last act by the protagonist. By engaging with an artform, the audiences will also gain knowledge of the stylistic features and narrative structure of that artform such as that the events depicted through parallel editing usually take place at the same time or that there is a certain class of narrators who are not to be trusted. While the knowledge about the fictional story might be trivial, the ensuing examples become less and less so, essentially moving into some basic, but crucial, territory of formal analysis and narratology. Yet, I suggest, because the acquisition of such knowledge is not controversial and/or because all these types of knowledge are about a specific work of art fiction or art fiction in general, discussion of narrow aesthetic cognitivism have mostly skirted the subject.5 Put differently, narrow aesthetic cognitivism is dominantly interested in whether art fiction conveys non-aesthetic knowledge to its audiences, i.e. the knowledge of the world (minus the world of art).
But I should be more precise and list at least some of the relevant types of knowledge that narrow aesthetic cognitivism is interested in within that non-aesthetic brand. These are: perceptual, propositional, experiential, social, and how to knowledge.
Perceptual knowledge relates to how something appears to the senses: how it looks, sounds, smells, feels or tastes. At present, art fiction standardly conveys only visual and sound information. As such, we may, for instance, be interested in whether Jurassic Park (Steven Spielberg 1993) conveys knowledge about how Tyrannosaurus Rex looked and sounded, or, to use the above example, how Margaret Thatcher appeared, moved, and spoke. But there are some relevant examples of knowledge import which are difficult to deny. For instance, Vertigo (Alfred Hitchcock 1958) confers, among other things, knowledge of how James Stewart and Kim Novak looked at the time of the shooting, as well as how many of the places in San Francisco appeared at the same time. More generally, every photographic fiction film conveys a plethora of historical information on the visual aspects of various entities including the actors, locations, and whatever was in front of the camera at that moment.6
Does this not mean that narrow aesthetic cognitivism is vindicated for there is a significant subset of art fiction—namely photographic film fiction—which provides nontrivial non-aesthetic knowledge to its audiences? I suggest this depends on what aspect of art fiction narrow aesthetic cognitivism takes to be important for conferring knowledge. The absence of discussions about the above point suggests that narrow aesthetic cognitivism is not really interested in this type of knowledge. I assume this is because the reason behind the imparting of knowledge in this case are medium properties—the photographic nature of traditional fiction film—rather than art fiction qua art fiction. The perceptual knowledge is conferred because conventionally photographic fiction film hinges on verisimilar photography. Except for, perhaps, radio drama there is no equivalent of imparting perceptual knowledge among other art fictions.7 Were the acquisition of knowledge due to aspects different than basic medium properties, I assume those debating narrow aesthetic cognitivism would have taken more interest. Instead, the implicit demands seem to be that the imparting of the knowledge should come through art fiction qua fictional representation.
What certainly commands considerable if not the most interest in these debates is propositional knowledge of the form knowing that p (Stolnitz 1992; Gaut 2005; Gibson, Huemer, and Pocci 2007; Vidmar Jovanović 2019b; McGregor 2021a; Slugan 2021a). Here the discussion is most often about statements about the world as it is. The question here is, do audiences, by reading Mantel’s trilogy about Cromwell and watching Nolan’s film about Oppenheimer acquire knowledge about these and related historical persons. When reading the sentence from the second instalment Bring Up the Bodies (2012)—“‘Cromwell has the skin of a lily’, the king pronounces”—do audiences learn something propositional about Cromwell or Henry VIII such as the shade of Cromwell’s skin and/or his king’s opinions about it? When watching the opening of Oppenheimer and seeing the titular character read the statement at a hearing do audiences learn something specific about the theoretical physicist such as the exact words that he said at the time in question or how he felt when saying them? Propositions can, of course, be broader and relate to general social and human condition such as the opening sentences of Anna Karenina—“All happy families are alike; each unhappy family is unhappy in its own way”—raising the question of whether novels impart such lessons. And propositions need not be explicit but may be implicitly conveyed through art fiction such as “drug crime is a structural problem” that is generally agreed to be articulated in The Wire (HBO 2002-2005). In the next section, when we discuss strategies for answering these questions, we will see that the crucial thing to remember here is that audiences need to learn from art fiction unlike makers who learn from the process of making art fiction based on research.8
An important subclass of propositional knowledge is the knowledge of possibilities, hypotheticals, and counterfactuals. In this case, it has been debated whether art fiction confers knowledge of what might happen and what could have happened (Putnam 1978; Gaut 2005; McGregor 2021a). Does, for instance, Dora Lessing’s The Golden Notebook (1966) account of her heroine impart knowledge of a possible person in a specific historical period? Does Marlon James’ A Brief History of Seven Killings (2014) confer CIA as a possible cause of crime and social harm in 1970s Jamaica?
Another type of knowledge relevant for narrow aesthetic cognitivism is experiential or phenomenal knowledge. This is the knowledge of what it is to have experiences more complex than perceptual ones such as what is to be in love, what is to be in mourning, as well as, less dramatically, how is to ride a horse or go deep diving (Walsh 1969; Putnam 1978; Gaut 2005; McGregor 2021a; Slugan 2021a). This type of knowledge also covers how a particularly valued knowledge among aesthetic cognitivists of how it is to be in somebody’s else’s shoes (Smith 2017). One question, for instance, might be does Memento (Nolan 2000) teach audiences how it is to have anterograde amnesia. Another: does Requiem for a Dream (Darren Aronofsky 2000) confer how the high of using controlled substances feels like?
The last type of knowledge falling under narrow aesthetic cognitivism I will address is how to or practical knowledge. This knowledge type involves non-propositional knowledge of how to undertake certain activities such as how to type with all ten fingers, how to play the guitar, how to tie a tie, how to score a penalty or how to dance. While it is not impossible that this type of knowledge could be articulated in propositional terms, it is often the case that the people who possess it are unable to do so. From the perspective of this knowledge, perhaps one can learn how to fight by watching kung fu films or how to dance and sing by watching musicals.
There is, admittedly, at least one other form of knowledge which bridges the propositional and practical type and generally refers to the domain of social relations. These include knowledge about human motivation and folk psychology more generally, knowledge of societal norms, how to behave in certain situations, and consequences of not doing so, knowledge about what it means to be a good parent or spouse, etc. It has been argued among those interested in evolutionary aspects of narrative and fiction that these are the reasons why the capacity to tell narratives and convey fictions evolved in the first place (Hutto 2008; Gottschall 2012; Boyd 2018).
While there are undeniably other forms of knowledge such as embodied knowledge, I will focus on the acquisition of propositional knowledge as this type is most often debated in literature and arguably the easiest to demonstrate.9 To do so, I will discuss the general structure of the argument in favor of narrow aesthetic cognitivism and argue that what they present as evidence of the claim is invalid. Many only argue that art fiction expresses truth. This fails because for knowledge to be obtained it also needs to be justified. Some also address the issue of justification. But even they regularly assume that justified truth will be automatically believed in by the audiences. There is an even smaller group who addresses potential mechanisms of belief acquisition. However, the empirical evidence they cite is selective and leaves out the most recent meta-analysis of the problem which contradicts their argument. In other words, to argue that art fiction confers knowledge on its audiences, one needs to argue that it expresses truth, justifies the truth it conveys, and instils belief in this justified truth.
4. Art fiction expresses truth
In a recent contribution to aesthetic cognitivism Rafe McGregor (2021a) has argued that fiction imparts at least three types of knowledge about causes of crime and harm—phenomenological, mimetic, and counterfactual. In my typology, McGregor’s phenomenological knowledge (e.g., what is it like to live in a war zone through a graphic novel The Sheriff of Babylon [Tom King and Mitch Gerard 2018]) corresponds to experiential knowledge, mimetic or knowledge of everyday reality (e.g., the resources at the disposal of criminal cartels and their appearance via Miami Vice [Michael Mann 2005]) to a combination of perceptual and propositional knowledge, and counterfactual knowledge (e.g. CIA as a possible counterfactual cause of crime in Jamaica in the 1970s through A Brief History of Seven Killings) to a subtype of propositional knowledge.10 While it is certainly possible that all these fictional representations are true, McGregor is generally uninterested in demonstrating the way in which fiction justifies its truth. Miami Vice, undeniably, depicts in a rich perceptual way a range of material resources at the cartel’s disposal—from motorcades transporting new recruits through mobile jammers to hundred meters killing fields. But given that the audiences are watching a fiction film, what is to justify them in believing that what is depicted is the case in the real world? McGregor does not tackle the question. Yet, if knowledge is some form of justified true belief, proponents of narrow aesthetic cognitivism need to address the question of how fiction justifies its truths.11
A critic of the idea that the discussion of propositional knowledge necessarily needs to involve a discussion of justification, might point out that conceptual knowledge also falls under the banner of propositional knowledge. The critic will point out, this type of knowledge requires no justification through empirical investigation of reality. Gaut (2005), for instance, argues that the knowledge of a definition of a term is conceptual knowledge. If we have been acquainted with the articulation of, say, the notion of “crystallisation” that Stendhal proposes and elaborates upon in his On Love (1822) we get to know what crystallisation is (Gaut 2005). While this is true in non-aesthetic settings, the case is different in art fiction.
First, it remains the case that the question of whether crystallisation—that love essentially rests on attribution of imaginary qualities to the object of love, qualities which are not actually possessed by it—is true or not still has not been settled. Only if it is true, do we gain knowledge of the non-aesthetic world through acquaintance with the term. And Stendhal is offering the concept in a fictional setting meaning that the main examples on which he articulates it are fictional. What we really have here is the knowledge of the fictional world in which crystallisation is the essential feature of love. And this is in fact not much different from the question of whether Tolstoy’s pronouncement about happy and unhappy families is true. In Anna Karenina it is, but it is an open question whether it is true in the actual world. The main difference is that Stendhal has come up with a word for his general claim about some aspects of human nature—"crystallisation".
Gaut might retort that I have missed the point. The relevant knowledge acquired here is simply what the term means, not whether the proposition the term expresses is true or not. But for the knowledge not to be only about the fictional world, the same concept needs also to be about the non-aesthetic world. And if this is so, then there is no need to draw out such a specific example as Stendhal’s. Any instance in which an art fiction explains what a word means should under Gaut’s view count as a case of acquiring conceptual knowledge. Pulp Fiction (Quentin Tarantino 1994) opens with a dictionary style definition of “pulp”. The Big Short (Adam McKay 2015) has Margot Robbie explain what “subprime mortgages” are and what “shorting” means. However, that Gaut (2005, 441) continues his discussion of the defense of conceptual knowledge by saying that the defense may go too far suggests that the truth of the matter is relevant after all. He talks about how the defense fails when applied to know how and instances in which it turns out that a detailed fictional account of plumbing is completely inaccurate and causes a flood when “lessons” learned are applied in one’s real home. From this perspective, conceptual knowledge can also fail when applied to real world. It is true it does not fail in Stendhal, Tarantino, and McKay, but that is only accidental. The definitions provided could well have been invented. And given that they occur in fiction, it is a reasonable possibility that they are. This is not different from biopics or historical fiction in which it is never quite clear what piece of information provided is only fictional and what is also true.
Gaut might again object that I have read too much into his later discussion of the failure of defense of conceptual knowledge. Truth of the claim is beside the point. And he could say there is a difference from Pulp Fiction and The Big Short insofar they do not invent new concepts while On Love introduces a term to the non-aesthetic world, regardless of whether the proposition it means is true or not. But even in that case there is a need for justification. What needs to be justified is not whether crystallisation accurately describes key features of love in the non-aesthetic world but whether crystallisation has the same meaning in the fictional world and in the non-aesthetic world. To Gaut this is obvious because he knows the history of the term. But to a less astute reader this is far from evident.12And from this perspective precisely because it is found in art fiction and developed in an art fictional context, there is no guarantee that this novel concept also circulates in the non-aesthetic world and that it circulates with the same meaning. After all, Stanisław Lem also develops a term “solaristics”—a scientific discipline focusing on the planet Solaris—in his eponymous novel, but the concept has no non-aesthetic equivalent.
But even if Gaut is correct that there is no need for justification of conceptual knowledge the question whether audiences reading On Love acquire knowledge of crystallisation and whether audiences engaging fictions acquire any knowledge in general is empirical. And rarely has it been tackled by philosophers in a sustained manner as we will see in the section after next.
Before proceeding, it is necessary to say a word or two about knowledge of possibilities as a subset of propositional knowledge. Like conceptual knowledge, it could be said that a knowledge of possibility only requires acquaintance with the possibility given that the space of possibilities is much wider than what is actually the case. And art fiction excels in conveying detailed stories which are possible, be it very probable historical accounts, plausible gangster and romantic narratives, or improbable but not impossible horror, fantasy, and science fiction tales. On this account, by reading A History of Seven Killings audiences become acquainted with a possible history in which CIA has supplied weapons for the assassination attempt on Bob Marley. Two points should be made here.
First, there is much art fiction which conveys impossible worlds, be it logically or against the laws of physics. Science fiction often involves travel at or higher than the speed of light which is not allowed by laws of physics. Other art fiction involves stories which are logically inconsistent either by mistake or design. Acquaintance with such stories could bring no knowledge of possibilities because what is depicted is impossible. Moreover, if the idea is that acquaintance with fiction in general leads to knowledge of possibilities, then there will be many art fictions which form inaccurate beliefs about what is possible.
Second and more importantly, knowledge of possibilities is arguably considerably less weighty than that of what actually is the case and is in a sense trivial. Certainly, upon reading James’ novel audiences might arrive at the knowledge that it is possible that the CIA was involved in Marley’s assassination attempt. But so could have KGB, MI5, or any other intelligence service for that matter. If an art fiction about a game of dice specifies that the sequence was 3-4-3-2 this does not seem to impart any special knowledge because audiences generally know that any sequence is possible in such a game. It might be retorted that the value is in a level of detail with which this sequence has been represented, how much the players bet, how they felt when betting, what the win or loss meant for them. But that does not change the fact that what is reported is a space of possibilities of human action and sentiment and that this space of possibilities is in a good sense already known. To put it in Jerome Stolnitz’s terms, these are trivial truths which “resemble garden variety truths” (1992, 192).
Lastly, proponents of narrow aesthetic cognitivism also usually focus on propositional cases which are plausibly true but there is another tradition, running from Plato, which argues that art regularly expresses falsities. In screen studies, for instance, there is a body of work usually dubbed screen theory which argues that films excel in misrepresentations especially when it comes to underrepresented groups and minorities (women, ethnic, and sexual minorities, etc.) (see Slugan 2019b). In other words, proponents of narrow aesthetic cognitivism who are uninterested in the way art fiction justifies its truth would end up with art fiction which, arguably, imparts inaccuracies more often than it does knowledge for, on average, fiction certainly involves more inventions and misinformation about the non-aesthetic world than truths.
5. Art fiction justifies its truth
Stolnitz (1992) has produced one of the strongest critiques of narrow aesthetic cognitivism by arguing that art fiction, unlike science and history, does not have established procedures for justifying what it conveys is true. Art fiction regularly mixes truth with other elements, be it inventions or inaccuracies, but for the audiences who consume it, it does not provide any tools for disentangling the two. In other words, while it is the case that art fiction can convey truth it cannot convey any knowledge because whatever belief audiences acquire through their engagement with art fiction, this belief is not justified. There have, of course, been responses to Stolnitz’s proposal arguing that art fiction does justify its truth.
One line of defense has been to argue that there are subsections of art fiction which audiences can assume are truthful. Proposed candidates have included realist fiction (Gaut 2005; Vidmar Jovanović 2019b) and sub-genres such as works by Dickens (Friend 2014). Vidmar Jovanović argues, for instance, that writers committed to realism such as Austen will accurately depict social and economic relations of their time. For Friend, we can assume Dickens to be a truthful guide on social facts but less so on scientific ones (given his recourse to spontaneous self-combustion in one of his novels). In other words, membership of a specific (sub)genre is what justifies truths in these fictions. From Stolnitz’s perspective, however, the problem remains because even the most realist (sub)genres involve admixtures of other elements. Despite what the opening of Dickens’ Our Mutual Friend (1865) suggests, for instance, there was no such a thing as collector of dead bodies from the banks of Thames. Put differently, for Stolnitz no art-fiction (sub)genre justifies its truth because it is a typical feature of all art fiction (sub)genres to involve inventions and inaccuracies.
Gaut has provided perhaps the strongest response to Stolnitz by arguing that while there is an institutional difference between non-fiction and art fiction in terms of immediate justification, in the end the deeper justification of both non-fiction and art fiction lies in one’s experience. Gaut acknowledges that, unlike for fiction, non-fiction which standardly makes a claim to imparting knowledge is vetted through some kind of expert process. In academia, for instance, this is through peer-review while in journalism it is through editorial fact-checking. However, Gaut points, there are still non-fiction works which have escaped vetting either by design (frauds like Hitler Diaries) or mistake (arguably, even the best history books involve some inaccuracies). And just by looking at such non-fictional work audiences cannot know whether the vetting has been successful. To ascertain that audiences need to go outside the text, examine the conditions of its making, check its claims against other sources, i.e. appeal to experience. In that sense there is no substantial difference with fiction because in both cases the ultimate justification lies in the appeal to experience.
While justification does ultimately lie with experience, the institutional guarantee does play an important role that is not addressed in Gaut’s retort. Justification pertains to ascertaining whether something is true or not. But knowledge also involves belief and just because a truth is justified it does not mean a belief about it will be formed. In other words, narrow aesthetic cognitivists also need to explain how audiences come to acquire beliefs and whether they do acquire them in the first place. Institutional guarantee that comes with non-fiction but is absent from art fiction is arguably part of a mechanism for acquiring beliefs. Put in yet another way, the underlying contract for consuming non-fiction is that one is supposed to believe in it. There is no similar underlying contract for art fiction not even in its most realist subsets.
6. Art fiction imparts beliefs
A minority of proponents of narrow aesthetic cognitivism have recognized that it is also necessary to investigate how and whether art fictions influence beliefs. They usually cite a selection of empirical studies and propose certain mechanisms of belief acquisition (see Friend 2014; Carl Plantinga 2018). The work that is usually put forward to support the idea that art fictions change beliefs comes from a few psychologists including frequent collaborators Deborah Gerrig and Richard Prentice (Gerrig 1991; Prentice et al. 1997; Prentice and Gerrig 1999), Melanie Green and Timothy Brock (2000), and Elizabeth Marsh (Marsh et al. 2003; Marsh and Fazio 2006). These are usually studies of literary art fiction where a general knowledge test follows the exposure to a fictional text set against a control group.
In one string of studies one group reads fictional stories with general statements about the non-aesthetic world which were either true (e.g., mental illness is not contagious) or false (e.g., mental illness is contagious). The control group reads either unrelated stories or no stories at all. Those exposed to fictional texts containing the above general statements (the test group) were statistically more likely to agree with the claims in the follow-up questionnaire so long as they were consistent with the information in the text, even if the information was false (Gerrig 1991; Prentice et al. 1997; Prentice and Gerrig 1999). When participants read one and the same story, but one group was told the story is fictional and the other nonfictional those in the fictional group were more likely to agree with the statements they have read (Prentice and Gerrig 1999). In another set of experiments, the general statements about the non-aesthetic world which could be either true (e.g., extant is used for navigation via stars) or false (e.g., compass is used for navigation via stars) were only peripheral or in the background of the fictional story. Much like in the previous studies, those exposed to the general statements did significantly better (in case of true statements) or worse (in case of false statements) than the control group on a follow-up general knowledge test (Marsh et al. 2003; Marsh and Fazio 2006). In other words, a significant subset of audiences acquires propositional beliefs from fiction. From this perspective, fiction is even more persuasive that non-fiction in imparting beliefs about propositional claims. Interestingly, even those more skeptical of knowledge acquisition like Gregory Currie have argued that “[i]t is an obvious truth that fictions affect the beliefs of those who come into contact with them” (2020, 161).
This, however, is belied by the currently most comprehensive meta-analysis of persuasive effects of fiction (Braddock and Dillard 2016). In the last four decades meta-analysis has become a key method of summarizing research on a given subject in various disciplines including psychology and an important way to tackle replication crisis in those disciplines (Sharpe and Poets 2020). Meta-analysis as a method identifies different studies focusing on a similar question and statistically combines the results of these studies to compute whether there is an effect and the size of the effect (Borenstein et al. 2011). Given that results of single studies are statistical analyses themselves based on sample sizes and quantified responses which provide values for whether there is an effect and what its size is, with right parameters and weightings these analyses can be combined into an overarching statistical analysis to ascertain whether the effect holds more broadly or not and its size if it does. As such meta-analyses are not competing studies which look at the same question and find the effect (size) or not. They are analyses of the whole body of research on a topic of interest and their results hold more generally than any single study.
Braddock and Dillard have produced a meta-analysis of 74 empirical studies investigating effects of fiction across media (visual, audio, theatre, literary) on four different aspects of persuasion—attitudes, desires, behaviors, and beliefs. For each of these aspects of persuasion they have made a separate analysis to see whether there is a difference in how fiction and nonfiction influences a given aspect. For beliefs, they have found that “nonfictional narrative stimuli significantly affect belief change, but fictional narrative stimuli do not” (2016, 457).13 In other words, while about dozen different studies that Friend et al. cite evince the sought-out effect—that art fictions change beliefs—the meta-analysis of 74 studies on the subject demonstrates that there is in fact no such effect—art fiction does not change beliefs. From a statistical perspective, the reason is that the effect sizes and sample sizes in studies reported by Friend are outweighed by the lack of effect and sample sizes in the studies she does not cite. In fact, the most recent non-statistical literature review of the subject arrives at the same conclusion (Dubourg and Baumard 2023).14
Moreover, Braddock and Dillard have also considered whether each of the four aspects of persuasion effects—attitudes, desires, behaviors, and beliefs—depends on the medium. They have found that “existing data support the conclusion that medium of presentation is unimportant for inducing persuasion via narrative” (Braddock and Dillard 2016, 462, italics in the original). Put differently, that fictional narratives do not change beliefs cannot be explained by the claim that one type of fictional narratives, say the textual ones put forward by Friend, change beliefs whereas the others do not. In general, in the fictional case no medium type changes beliefs (in the nonfictional case they all do).
Braddock and Dillard’s findings also problematize some of the proposed mechanisms of belief acquisition. Building on the work of Prentice and Gerrig (1999) and Gilbert (1991), Friend (2014) has advocated for scrutiny lowering—an idea that the very label “fictional” lowers a more critical outlook on what is consumed leading to easier persuasion than in non-fiction. This however is contradicted by the above meta-analysis which has the opposite results of what the scrutiny lowering theory predicts.15
Other mechanisms such as Currie’s (2020) narrative value have different issues. While Currie is skeptical of a general belief acquisition mechanism, he still proposes a typical way in which audiences infer beliefs about the non-aesthetic world from beliefs about fiction: “Deviations in fiction from truth are expected to be justified by their narrative payoff” (2020, 161). In other words, if information is simply in the background, then the tacit assumption is that it is true in real life. Because French interpolations by Russian aristocracy in Tolstoy’s War and Peace (1869) are not of particular narrative importance (they could have been in, say, Italian and nothing much would change narratively speaking) but happen in the background, audiences generally infer that 19th century Russian aristocracy spoke French. By contrast, because Dickens’ Our Mutual Friend opens in a particularly dramatic fashion with a collector of dead bodies from the Thames audiences can assume that the collector is not a historical profession. Counterexamples, however, abound. In the case of background, in historical fiction films characters regularly speak in contemporary languages rather than languages of the epoch depicted yet a diet of Hollywood epics on Rome will not lead to an inference that Romans spoke English. Similarly, irrespective of their occupation characters in American TV shows regularly look like models yet this hardly leads to inferences that a typical policeman, doctor, or a lawyer conforms to contemporary standards of beauty much more than an average person. When it comes to the foreground, Currie himself admits that when he first read Our Mutual Friend, he came to mistakenly believe that there was such a profession as collector of bodies from the Thames. This means that the difference between foreground and background is much less clear than he proposes leading to problems in determining what to believe in. There is also a matter of middle ground: what do we do with fictional truths which are neither crucial for the storyline but are at the same time not unimportant such as that queen Anne was a lesbian (The Favourite 2018, Yorgos Lanthimos) or that first inoculation took place on the Danish Court (A Royal Affair 2012, Nikolaj Arcel)?
In short then, existing studies disprove the claim that art fiction imparts beliefs making general proposals about belief acquisition moot. It is true that virtually all the studies in the meta-analysis are short term studies in the sense that they only look for effects immediately after the exposure to art fictions. But this arguably makes things even more difficult for narrow aesthetic cognitivists because it is unlikely there would be long term effects on beliefs without there being short term ones. In other words, the biggest hurdle for proponents of narrow aesthetic cognitivism is that empirical findings contradict the view that art fictions impart beliefs.16
It is possible, however, that proponents of narrow aesthetic cognitivism are not persuaded by the empirical findings I cite. One objection could be that psychological studies of the type that Friend et al. refer to and that are included in the meta-analysis are simply not a good way of exploring the question that interests us—whether art fiction changes beliefs. The objection broadly claims that what the cited experiments measure is not belief but something else. Therefore, it is irrelevant for the debate either way. If we think about beliefs as cognitive states which are relatively stable over time, then short-term studies deploying questionnaires only immediately after the exposure to art fiction to gage beliefs stemming from that text are lacking. Such studies are not really measuring belief but some kind of acceptance or subconscious recall. What is required is a long-term study where the questionnaires are repeated after a prolonged period of several weeks or more.
In another variant of the objection what is actually studied is not beliefs derived from fiction, but beliefs formed on the basis of who presented them. When participants are provided with information in experimental conditions because the information is provided by experimenters whom participants intuitively trust to convey truths, it does not matter whether information is labelled fiction or nonfiction. What matters is the implicit trust in the presenter of information. The conclusion of both variants is the same—short-term studies do not capture (relevant) beliefs, only long-term ones do.
Tackling the second variant first, if participants generally got their beliefs from the fact that experimenters provided implicitly trustworthy information, then all the experiments in which the control group engages unrelated or no work and the test group is presented with art fiction would need to show effects because questionnaires would necessarily include references to statements in the art fiction provided by the experimenters. Yet clearly there are such studies which do not find an effect. Moreover, as mentioned earlier there are studies which present one and the same text to the control and test group but label it as nonfiction and fiction, respectively (Prentice and Gerrig 1999). According to the second variant the result of this experiment should be no difference between the two groups (both groups are getting relevant information from the trustworthy experimenters), but the experiment reports a significant difference. Most importantly, there is also the matter of long-term studies which invalidates both variants of the objection.
While there is only a handful of such long-term studies, they do exist (see Strange and Leung 1999; Brodie et al. 2001; Marsh et al. 2003; Howell 2011; Schneider-Mayerson et al. 2020). Crucially, all but Marsh et al. (2003) report disappearance of the effect while even Marsh et al. report significant decrease in the effect size. In other words, even if we accept that short-term studies do not actually measure beliefs, long-term studies evince that art fiction does not establish imparting or change of beliefs construed as temporally stable cognitive states.17
7. Future directions for narrow aesthetic cognitivism
While the above invalidates current defenses of narrow aesthetic cognitivism that does not mean that narrow aesthetic cognitivism does not hold. One obvious response is that the empirical findings about belief relate only to propositional belief and knowledge and even that only in part. There are other (sub)types of belief and knowledge that could be acquired including experiential belief and knowledge. This is certainly one avenue of research that should be pursued both experimentally and philosophically.
Defenders of narrow aesthetic cognitivism should also not neglect long-term studies of a somewhat different type. It is possible that beliefs and knowledge accrues over time through a repeated exposure to art fiction (sub)categories. Audiences might not pick up that the Roman Republic was run by the Senate after watching Cleopatra (Joseph L. Mankiewicz 1963) once or if they do the effect might not be long-term, but they might acquire a robust belief and knowledge of this fact after a more extensive engagement with the genre of Hollywood epics about Rome. In other words, there is much space for long-term studies investigating repeated exposure to certain types of art fiction.
Focusing more on belief and knowledge of social relations which bridges propositional and how to knowledge, it has been recognized that it is difficult to evidence this type of belief and knowledge acquisition in adults because they might have already hit the ceiling having been embedded in a storytelling society from their birth (Boyd 2018). That is why we need to turn to children and adolescents to investigate in more detail whether art fiction imparts belief and knowledge of social relations.
However, proponents of narrow aesthetic cognitivism also need studies that address knowledge acquisition in addition to belief acquisition, because it is necessary to ascertain that, if belief acquisition is an empirical reality, it has an overall positive epistemic effect. In other words, narrow aesthetic cognitivists need to accept the fact that demonstrating belief acquisition might also lead to concede that, while art fiction can certainly have positive epistemic outcomes, in reality it may bring more epistemic harm than good.
Acknowledgments
I wish to thank the three anonymous reviewers for EuJAP for providing useful and constructive comments on a previous version of this paper. This work has been supported in part by the Croatian Science Foundation (project AETNA, grant UIP-2020–02–1309).