2. Part I. Some observations on reasoning through narrative

Let's explore some cases in order to illuminate the phenomenon I am interested in.

GRADUATE STUDENT

Imagine with me that I'm talking with a discouraged graduate student.

In this example, I am asking the graduate student (and you, dear reader) to reason through narrative.Footnote ⁴ It is something that the graduate student does effortlessly. The graduate student drew at least some of these inferences:

1. (a) I must be courageous, strong, and wise.

2. (b) I must not quit, even though I am weary.

3. (c) Once I finish this dissertation, I will have my life back.

There are other inferences that the graduate student does not draw.

1. (d) I must find armor

2. (e) The task is to behead the dissertation.

3. (f) If I complete the task, I get to marry a princess and become a ruler.

How is the graduate student (and you, my reader) able to successfully complete this reasoning? You are not George. The grad student is not George. There is no George. This story is a fairy tale. It does not even purport to describe reality. There are no dragons, and it is a rare person who would get excited about marrying a stranger and running a mediaeval village. And yet, despite the fact that I've painted an unappealing fantasy prize, the grad student is able to reason to conclusions that guide and even inspire his actions. But how is it that something that is not remotely about him, or reality, can lead to reasoning that results in practical reasoning and commitment to action?

Reasoning involves accepting (at least conditionally) the truth of one's premises.Footnote ⁵ In the case of a reductio, we might reason to the conclusion that a premise is false. When we reason through narratives like in GRADUATE STUDENT, though, there is no pretense that anything in the narrative is factual. It might be an exercise in imagination or fancy. But in order to identify it as reasoning at all we need to give an account of why a fanciful story can guide inferences of the sort the graduate student makes. At first blush, reasoning through narrative doesn't look like reasoning. An account of reasoning through narrative must account for how narrative can play a role in reasoning, one that leads to judgments, beliefs, practical commitments, even though nothing in the narrative purports to be about a matter of fact.

Reasoning through narrative involves imaginative engagement, and there is epistemic value in such engagement. Recent work in the epistemology of imagination has discussed how we can learn from imagination. Philosophers have argued that imaginative engagement with fiction can expand our concepts (Peacocke Reference Peacocke2020), give us knowledge of counterfactuals and conditionals (Williamson Reference Williamson2005, Reference Williamson2007), give us perceptual or quasi perceptual experiences (Jackson Reference Jackson, Macpherson and Dorsch2018), or serve as guides to metaphysical possibility (Yablo Reference Yablo1993). Classic texts highlight the way narratives illuminate causal structure (Carroll Reference Carroll2007) and salient counterfactuals (Beatty Reference Beatty2017). Velleman (Reference Velleman2003) points out that not all narratives are about causality. Some are about emotional cadencing, which trains us how to feel. Fraser (Reference Fraser2021) likewise points out that narratives can ground not just our judgments but also our reactive attitudes. All of this work shows that narratives can provide epistemic benefits. Notice, that these accounts argue that narratives reveal something about modal structure. They reveal something about what is possible, probable, plausible, permissible, required, relevant, desirable, and good.

Philosophical reasoning often makes use of narrative, in the form of thought experiments. When Mary exits the black and white room, or when the Chinese room spits out a sentence, or when our radically different counterpart utters “gavagai!” these narratives are supposed to tell us something about what is possible, or plausible, or impossible, or implausible. When I ask you whether you would flip the switch to redirect a train, or whether an atom-for-atom facsimile of a Van Gogh is equally valuable, I am asking you to reason in a narrative. Such reasoning tests the limits of our concepts, and depending on your metaphysics, tells us how to carve nature at its joints.

But there is an additional puzzle to reasoning through narrative. It's not just that the story tells us an abstract lesson about modal structure. When we reason through narrative, we take the story to bear on a particular situation. And we take it to bear on that situation directly. It is possible, of course, that the graduate student sees that George could slay the dragon, and so judges it's possible for them to finish their dissertation. But the connection is not very tight. Why think fictional-George slaying fictional-dragon provides evidential (or even analogical) support for real-graduate student finishing real-dissertation? In order to draw that inference, the graduate student needs to think that finishing the dissertation is slaying a dragon. And that is the very move that is puzzling.

The distinctive feature of reasoning through narrative involves taking a particular story (that may have no surface-level similarity to a real-life situation) and taking it as a guide to a particular situation. While it is exceedingly plausible that stories can teach us things, it's unclear what would ground the application of those lessons to our particular situation. Here's one way to unpack this.

WOLF

One way of reconstructing Steve's reasoning is that he's reasoning from a particular story (Boy who Cried Wolf) to the Moral of the Story (if you are untrustworthy you lose standing in your community) and then applies the Moral of the Story to his situation (refrain from posting to keep social capital). While I think this reconstruction is sometimes what we are doing, it doesn't quite make sense of the immediacy of Steve's reasoning. He doesn't move from particulars to abstraction to particulars again. He transposes the perspective of the story directly onto his life. He thinks, “Don't be the boy who cried wolf!” and “would this be crying wolf?” This is similar to the graduate student thinking “I am George.”

Elizabeth Camp explores how engaging with fiction can have “lingering cognitive effects” (Reference Camp2017: 74). On Camp's view, when we engage with fiction, we construct a perspective: a range of open ended dispositions to think and feel in certain ways in response to various inputs. When given a characterization, a way of seeing things, a perspective will update in various cognitive and conative ways. Camp's work helps us articulate just what is so puzzling about reasoning through narrative. When we reason through narrative, we take a perspective in response to a characterization. It is somewhere between believing and imagining. It is a way of perceiving the world. Camp writes, “Trying on a perspective requires more than just imagining that a set of propositions is true, or even imagining experiencing something. Rather it involves actually structuring one thinking in certain ways, so that certain sorts of properties stick out as especially notable and explanatorily central in one's intuitive thinking. Because it involves actual patterns of attention and response, adopting a perspective is partly but not entirely under one's voluntary control.” (Reference Camp2017: 74)

We can turn from a fiction and apply the fictional perspective to a particular object, person, or event. Camp notes that we have a tendency to take reasoning that occurs within narrative and transplant it into our actual lives. In this sense, narratives are stickier than mere pretense. Consider. If you and I are role playing doctor and patient, and you stab me with your finger, simulating a shot, I should react as “ouch! That hurt!” All of this is perfectly intelligible within our pretense. But it would not make sense for me to then take myself to actually be hurt, or experience actual side effects from the shot. Within the pretense, this makes sense. But there is a gap between that and the world itself. When we reason through narrative, we are not engaging in mere pretense, we are seeing the world from the perspective of the narrative.

Camp does not frame things as reasoning through narrative. But she does discuss the phenomenon. She considers, after reading Anna Karenina, she might think “I am Anna Karenina.”Footnote ⁷ From this identification, she casts her children as Anna's son, and her career ambition as Vronsky. Under this characterization, she experientially applies the perspective of what it would be for her to be Anna. Camp thinks there are abstract theoretical gains that can be made through this imaginative enterprise.

But what I am interested in is the way such perspective taking might transform our first order predictive and prescriptive judgments about our own personal situations. If Camp were to reason through Anna Karenina, she might come to see herself as a tragic heroine. She might predict that the end is only suffering and unhappiness. From this, she might do two things. First, she might abandon the perspective, she might reject it as a false narrative (at least when applied to herself). Or she might draw the judgments that the Anna Karenina perspective recommends. The second case would involve reasoning through narrative. One of the things I am interested in is when reasoning through a narrative goes well and has epistemic payoff, and when narratives can be rejected as false.

And I can already imagine, some of my readers think I have not done well to offer the GRAD STUDENT narrative to my student. It reifies gendered and sexist views about the role of women. We should not equate a beautiful princess with a post-doctoral future. That is demeaning to the woman in this story, and perpetuates the stereotype that women are not characters, but objects or rewards. And so while my pep talk might have worked wonders, it is a morally objectionable piece of reasoning because the standing in of a postdoctoral future for a woman is demeaning to women. Here is another feature of reasoning through narrative to be explained. Even when narratives do not purport to be about anything factual, we can still reject them on moral grounds.

Consider this retelling, with the same narrative but different reasoning.

DORM ROOM

From this framing of the narrative, the dorm resident was primed to draw a number of conclusions:

1. (g) I am weak and helpless, awaiting rescue.

2. (h) When my handsome, strong knight arrives, he will free me so we can go off and live a happy life.

3. (i) My role will be to marry my rescuer and submit to his rule in our community and our marriage.

And she is not primed to draw conclusions like

1. (j) I am being arbitrarily restrained from doing things I would otherwise enjoy

2. (k) I have no choice in who I marry because my father promised me to whoever slays the dragon

3. (l) My family is in harm's way

We can reason through the same narrative to vastly different sets of conclusions. Narratives are flexible, and can provide different frameworks for reasoning. The grad student interpretation of this narrative did not rely crucially on anything gendered. If the princess had been an especially desirable chocolate cake and George had been Georgi, it would not have changed the inferences the grad student draws. But DORM ROOM is gendered. The princess is awaiting a handsome, male rescuer. For her, the story bolsters her judgment that she is to be rescued, and she is the reward to her protector and rescuer. In DORM ROOM, the reasoner isn't primed to see anything particular as the dragon, and she is primed to see herself in the background, with the rescuer as the main character. This is supported by background cultural scripts about the role of women. What is notable is that she can reason through the same narrative to different conclusions, and she doesn't draw conclusions that might seem obvious to some of my readers.

So a single narrative can frame different reasoning. The narrative itself underdetermines how the agent will reason through it. I will speak now of a narrative as a general story, and a narrative frame as a particular version of that story that primes particular inferences. Narrative frames are generated by a number of contributing factors. Social scripts and schemas will prompt us to fill out the story in particular ways. Who is offering the narrative, and in what context, will shape which narrative frame is salient. And the individual reasoner has imaginative power to construct the interpretation. In contrast to DORM ROOM, we can imagine a person who puts up the George and the Dragon poster but casts herself as George and the dragon as patriarchal marriage norms.

One conclusion to draw from the contrast between GRAD STUDENT and DORM ROOM is that some reasoning through narrative seems to result in unobjectionable conclusions, and some results in objectionable conclusions. One view of reasoning through narrative could be that it is all bad reasoning and a mark of irrationality. Indeed, a common response to political discourse and folk reasoning is that it is irrational because it goes through narrative rather than robust statistical models. I will address this later. But even if that is the conclusion we wish to draw, it still seems that not all reasoning through narrative is equally bad. GRAD STUDENT leads to inferences that seem worthy of acceptance. DORM ROOM does not. Or at least, DORM ROOM is worse. But both cases of reasoning utilize the same narrative. Both are equally non-factual. So we need some explanation for why one narrative frame is better than another, given they arise from the same narrative.

So far, reasoning through narrative has led to predictive and normative conclusions.

1. (b) I must not quit, even though I am weary (normative)

2. (c) Once I finish this dissertation, I will have my life back (predictive)

And

1. (h) When my handsome, strong knight arrives, he will free me so we can go off and live a happy life. (predictive)

2. (i) My role will be to marry my rescuer and submit to his rule in our community and our marriage. (normative)

Reasoning through narrative yields these judgments in the same breath. It does not require extra content to be added. It also structures our affective responses and reactive attitudes. This is a widely remarked feature of narratives.Footnote ⁸ They are not collections of atomic sentences. But they are organized into an internally coherent perspective that calls for a response. As Rachel Fraser explains, they ask us to engage in the Strawsonian interpersonal stance. We engage interpersonally, not as scientific observers. By its very structure, narrative frames guide us to draw normative and predictive judgments about the actual world.

Reasoning through narrative seems to be a fundamental mode of human reasoning.Footnote ⁹ Psychologists bemoan and pundits giggle, but in the end, people are mostly moved by a good story, rather than a cogent argument or an extensive dataset. Human beings struggle to reason through statistical models. They struggle to respond to data, even when the conclusions are clearly laid out. But if given a compelling story – even a just so story with no argument! – human persons tend to endorse the conclusion. In his 2021 book, Michael Lewis writes about Carter Mecher's attempts to convince epidemiologists in 2007 that social distancing would be an effective way to slow the spread of airborne disease. He reported that it wasn't the data that convinced the epidemiologists but asking them to engage in narrative reasoning about whether they would put their children at risk with the disease floating in the air. According to Lewis, the exercise shifted the judgment.Footnote ¹⁰

So far, I have discussed non-referential narratives. These are narratives that do not refer to agents in our world, and do not purport to be actual. I started with these cases because they are the most puzzling. They play a robust role in shaping our sense of what to expect (descriptively and normatively). They shape our patterns of attention and guide our judgments. And yet it is hard to capture what they are doing within our standard models of reasoning.

However, many of the narratives we reason through have more contact with the actual world. Journalists will seek to shift public opinion on some issue by telling a vibrant story based in an actual occurrence. Scientific models gain sway through a case study of a significant case. As I see it, stories come on a continuum between referential and non-referential. On one end, every aspect of the story refers to some actual object, person, or event in our world. And on the other, all of it is fictional. Even fully referential stories contain choice points. They are structured to make some elements salient and downplay others. They simplify causal strands and prime us to draw a range of conclusions. If an actual event is told as a narrative, an interpretive lens has been applied.

Many of the stories that we reason through are not nearly as referential as we presume. Consider the death of Kitty Genovese. Kitty was brutally raped and murdered in 1964 in Queens. According to a New York Times article published after her death, there were thirty eight witnesses to her murder, no one intervened or called the police, and she was left to die alone in the alley.Footnote ¹¹ This is the paradigmatic story used to illustrate the so-called “Bystander Effect,” where onlookers will let something terrible take place and do nothing to stop it. But, it turns out, the story is part fiction. Neighbors did attempt to frighten off Kitty's assailant, and she was comforted by her neighbor until the ambulance came to take her to the hospital (she died on the way).Footnote ¹² Suppose we were to reason through the narrative of Kitty Genovese to draw conclusions about who would likely intervene in a public place. Does it matter that the story of Kitty is inaccurate? I think the answer depends on whether the bystander effect is real, more than whether what happened to Kitty Genovese is accurate. But learning that the Kitty Genovese murder does not exemplify the Bystander Effect does seem to cast at least some doubt on the Bystander Effect.

We find narrative reasoning at the heart of partisan national attempts to understand and respond to galvanizing events. In the wake of George Floyd's murder in 2020, protestors from around the globe protested police violence against black Americans. They were not protesting one instance of police brutality. They were protesting a particularly egregious instance that supported a narrative that policing in America is systematically racist. Counter protesters soon arrived. They agreed the George Floyd killing was disproportionate and unjust, but refused to see it as emblematic of a bigger problem because they accepted a narrative according to which George Floyd's treatment was an aberration from how the system works. In this case, everyone agreed on what had happened to Floyd, and that it was egregious. But what they disagreed about was where this fits within a more general understanding of policing in the US. For some, the narrative of George Floyd is taken up as a paradigmatic case of how policing plays out in cities across the country. For others, it is an aberration against a backdrop where policing looks more like COPS, Law and Order, or NCIS. In these shows, the prevailing narrative is good guys with guns fighting to keep the neighborhoods safe from bad guys who wish to do harm. I take it that one of these narratives is better than the other when it comes to reasoning about the state of policing in the US.

But it's important to note that the counter-protestors's narrative is not worse in virtue of it being fictional. In the lead up to the American civil war, many people were influenced by Harriet Beecher Stowe's Uncle Tom's Cabin, a fictional account of the atrocities of slavery. This narrative, though fictional, shifted public sentiment against the institution of slavery by asking voters to reason through the story and think about building a better world.Footnote ¹³ The problem with the counter-protestors narrative is not that it is fictional, but that it is that it is false. In the next section I will give an account of what it means for a narrative to be false.

And it's not as though protestors couldn't go astray in their inferences. If they reason that the police only kill black Americans, or that their local town must have police brutality as cruel as in Floyd's case, they would be incorrect. Indeed, it is exceedingly difficult to put Floyd's case in context since there was no national database of police misconduct at the time (and now three years later, it is still in the process of being established).Footnote ¹⁴ If they think that such murders occur every day, or that every cop is as cruel as Floyd's killer, they are mistaken. But they can infer, I think, that police act with impunity, and that policing involves racial profiling of Black Americans. For the average white protestor with no special background knowledge of policing in the US, who absorbed the narrative through Tiktok, they are primed to draw some correct inferences about policing in the US. But they are also in a position to draw some incorrect inferences.

So far, I have explored a phenomenon (reasoning through narrative). Reasoning through narrative involves using a story helps us make sense of something. Narratives involve persons, places, events. They involve, at minimum, an event, a change from one state to another. As a result, narratives have a temporality built into their very nature. I have argued that it is a psychologically important part of human reasoning. I have outlined some puzzling features of this form of reasoning. Sometimes, we reason through a fictional narrative to conclusions in the actual world. Narratives stably prime us draw a range of inferences. But narratives are also flexible and can be distinguished into different narrative frames that prime different inferences. Some narratives are better than others, and the factualness of the narrative does not seem to be the determining mark. In the next section, I will develop an account of reasoning through narrative that clarifies these puzzling features.

3. Part II. The modal model account of reasoning through narrative

Narratives are representational. They represent their patients as good, bad, possible, permissible, impermissible, necessary, contingent – in short, a narrative is a story that illuminates modal relations between its patients. When we reason through narrative, we apply a modal framework to a particular person, object, or event in our experience in order to better understand it. Or so I will now argue.

First, I asked how it is that a narrative can lead us to conclude things about the actual world even though it does not purport to be about the actual world. We can expand this problem by asking how it is that we can reason with one of the premises “I am George”, when we know this is not true. It cannot be treated as mere supposition, since we somehow reach a conclusion without discharging the supposition. So how can I reason to an unconditional conclusion, through something I do not take to be representing as true, with a premise I know to be false?

One suggestion might be that reasoning through narrative is like reasoning by analogy.Footnote ¹⁵ We make a comparison between two things. Once we notice that they share some significant traits in common, we project that they (probably) share other traits. Famously, Plato thought that by formulating the ideal city we could come to understand the nature of a well-ordered soul. This is because a city is analogous to a soul. Similarly, the graduate student is analogous to George and the dragon. I do think that reasoning by analogy is a recognized form of reasoning. And as far as it goes, thinking of reasoning through analogy as similar to reasoning by analogy may be helpful. But it doesn't illuminate much. In what sense is a dissertation a dragon? Does the fact that a dissertation bears some similarities to a dragon increase the chance that a dissertation can be beaten? In the case of the city and the soul, the connection is quite close (it turns out). But what grounds the validity of an analogy? Appealing to analogical reasoning does not help us tell a story on which reasoning through a narrative generates epistemic value.

Here is what I propose. We should begin by looking at the function narratives plays in reasoning. Narratives provide rich relational informational structures. They can be disambiguated a myriad of ways, into different narrative frames. The frames will highlight which modal sentences the narrative uses, and whether they take bound variables or proper names for their constituents. The narrative frame can highlight any number of modal claims, and one narrative can be disambiguated into many frames (As with George and the Dragon above). The modal structures represented in the narrative frame emerges from the narrative. This captures a key representational component of narratives. Narratives are not literal, atomic claims meant to be taken in conjunction. They are content rich structures that must be given a narrative frame and an interpretation in order to have a truth value. It would be a mistake to try to evaluate the truth of a narrative by giving each claim within it a truth value. That is not its function. Inferences from narrative are not straightforward inferences from a set of propositions. Instead, narratives guide reasoning by illuminating what is possible, plausible, probably, required, forbidden, implausible, improbably, impossible for the object being reasoned about.

What determines the narrative frame? How is it that GRAD STUDENT and DORM ROOM have such different narrative frames? Social scripts and schemas, background assumptions, and interpersonal cues determine the particular narrative frame being used by the reasoned. Additionally, the reasoned has some control over the narrative frame.

There are distinct forms of narratives, but here I will focus on just one: the archetypal narrative. The archetypal narrative presents a particular instance (or generalized fable) that is the paradigmatic exemplar for any other object of the same type. This is how a compelling story can be repurposed to shape our understanding of other events. It's not that one boy cried wolf, but that anyone who is repeatedly untrustworthy will be left to be eaten by a wolf (or equivalent).

Once the narrative has been translated into schematic modal sentences, we still cannot evaluate those statements for a truth value until we have given them an interpretation. An interpretation is formed by substituting salient objects from the discourse in for bound variables. This generates modal statements that can be evaluated as true/false. And so an interpretation of a narrative that represents X must be done when it is in fact true that X must be done is representing a truth.Footnote ¹⁶

Here is how I am understanding this: a narrative is a rich relational informational structure. It can be given a narrative frame, or a schematic representation (bound variables and names) that lay out various modal relations, deontic, epistemic, and metaphysical. The narrative can then be given an interpretation when its schematic sentences are filled with objects from the discourse (“you are George, your dissertation is the dragon”).

As I mentioned above, sometimes narratives contain bound variables, and sometimes they also contain names. As a result, narratives can be referential or non-referential. Referential narratives lay claim to persons, objects, events, as proper names. Historical narratives purport to be actual, even if their narrative explanations go beyond known facts. The narrative of Braveheart, for a historian, is bound by certain facts that are known about William Wallace. By contrast, non-referential narratives do not reference existing persons, objects, and events. Sometimes it is vague whether the narrative is referential or non-referential. The narrative of Braveheart, as portrayed by Mel Gibson, is not bound by the facts that are known about William Wallace. But there is an element of Braveheart that is referential to William Wallace. It just doesn't really care about the facts about William Wallace. In that sense, it might be “about” him, but only as a fantasy about a real person, not as laying claim to explaining realistic features of him. In her book Jesus and John Wayne, Kristin Kobes Du Mez (Reference Kobes du Mez2020) argues that American Evangelicals take Mel Gibson's William Wallace narrative and reason through it to draw conclusions about American Society and their response to it. What is important for Evangelicals, Du Mez argues, is not that William Wallace lived and died the way Mel Gibson portrayed him. But that the fictional depiction of Wallace animates the way they view themselves in relation to American culture. It would be in effective to disrupt evangelical narratives by pointing out the historical inaccuracies of the film.

So, I am suggesting, when we reason through narrative, the function of a narrative is to represent modal space. But it cannot have a truth value until we give it an interpretation. This means that a narrative is not something we can believe, at least not until it has an interpretation, or a range of interpretations. Instead, I think we should view narratives as representational aides. They frame the modal space. They make certain features of a situation salient, and reduce others to background noise. They tell us what is good, bad, desirable, undesirable, and using them we are able to reason to conclusions about our own situation.

As a representational aide, or tool of reasoning, narratives aren't proper objects of belief. But once a narrative frame is given an interpretation, they are truth evaluable, and they could be believed. Rather than talking about believing a narrative, we should talk about accepting a narrative frame.

Someone could accept a narrative and use it only one piece of reasoning. But often, our narrative frames become stable, so that we revert to the narrative in particular circumstances without any effort. This aids our reasoning for assessing the credibility of women asserting sexual assault, or for men who abuse their power, or for those who virtue signal online. Furthermore, social groups can converge on a set of narrative frames, making for swift and convenient convergence on shared meaning.

What makes some reasoning through narrative good and some less good? What we've said so far makes a narrative frame something like a scientific theory. A narrative frame helps us make sense of the world. It directs our inferences. It is a guide to prediction and normative expectation. It tells us which alternatives are relevant. It helps us navigate the world. But if narrative frames are to be evaluated as scientific theories, then they are bad scientific theories. They do not answer to empirical constraints or theoretical virtues. They are not refined and critiqued by institutions of scientific knowledge. Narratives are spun by advertisers who want to sell deodorant, or pundits who want to downplay a politician's betrayal. If pundits, Instagram influencers, and Heather Cox Richardson can all give us narratives, we must be able to distinguish between them.

As I said at the outset, narratives are unmoored representations. They cannot have truth values until they are given narrative frames and interpretations. As a result, typical tools for epistemic evaluation do not straightforwardly apply to them. But as I have argued, not all reasoning through narrative is on a par. So how can we distinguish the good from the bad?

What we need is an epistemology of narrative. Here is what I proposeFootnote ¹⁷

1. i. If the modal representation is accurate, then we will call the interpretation of the narrative frame correct or true Footnote ¹⁸

2. ii. If S accepts a narrative frame and she draws mostly accurate conclusions in the contexts she is disposed to reason through it, then we can say the narrative frame is reliable. In this case, S is defeasibly justified in her conclusions.

3. iii. If the interpretation of the narrative frame is true and all the conclusions that S is disposed to draw are true, then her reasoning is safe, and constitutes knowledge.Footnote ¹⁹

We might also reflect on the source of the narrative as a potential source of epistemic merit. If I tell my graduate student the tale of George and the Dragon, we might think the fact that I am telling it adds extra weight. After all, I have experience observing, finishing, and mentoring others to finish dissertations. And so the conclusions are reasonable because I am offering them to the graduate student. Some justification for reasoning through a narrative might come from its source.Footnote ²⁰ But not all narratives come from trusted sources. On TikTok they spread rapidly and no one knows where they came from. And even if the advisor is a trusted source, advisors can pass along banalities without much thought.

We can now give a rough characterization of what it is for a narrative frame to be a false narrative. It can't be strictly speaking false until it's been given an interpretation. But we could imagine that someone accepts a narrative and the range of situations in which they are disposed to reason through it leads them to false conclusions. The result is that it leads them astray. And this is what we call a false narrative. It is a misleading representation of modal space, when applied to a range of situations. With this in hand, we can now say the difference between the partisan narratives of police brutality against black Americans. If the narrative tends to produce incorrect judgments across the range of cases it is used, then we can say it is a false narrative. While I think the evidence supports the George Floyd protestor narrative, it's important to note that most people reason purely on the level of narrative and are not aware of actual frequencies, and would misjudge if asked.

Earlier, I argued that narratives seemingly violated a common constraint on reasoning. Seemingly, they used fictional premises in order to draw conclusions about the actual world. There is a common assumption that in order for one's reasoning to justify the conclusion, there must be some epistemic constraint on premises.Footnote ²¹ Narrative frames, as I have set them out here, cannot play this role because they are not truth evaluable. Until we have an interpretation, we cannot say whether it meets the epistemic constraint. And so reasoning through narrative seems to violate a basic constraint on reasoning. But why does this matter? The thought is that a narrative is an arbitrarily selected just-so story, and that another story could have easily been picked which would have generated a different conclusion. There is no constraint one which narratives one might use. And so what entitles one to believe (or judge, or act on) the output of reasoning through narrative?

I have said above that an epistemology of narrative could help us settle whether a narrative frame is reliable within a particular context. And I have suggested that if it is a reliable narrative frame, then she is justified in the conclusions she draws. However, she may not have any awareness about how reliable a narrative frame is. She might be unable to justify her reasoning to herself or someone else. The thought is that the narrative is a tool, like a calculator or a surveyor's tool, or an astrolabe. If it is reliable, then it grounds one's judgments. But one can use it without knowing its precise accuracy. Since it is a tool, not a premise, it doesn't violate the epistemic constraint on reasoning. And the things we actually use in reasoning can be justified.

My account does show that reasoning through narrative is fragile. Narratives can clash, and we may have no good reason to prefer one to another. In the case of social narratives about how police treat black Americans, or how trauma victims respond under duress, we can look to academic research. But it can also be that the truth is hard to find amidst the noise, and we are left without persuasive narratives. We need knowledgeable persons to be better story tellers.

Reasoning through narrative is ubiquitous. It is just how our brains are wired. Even advanced statisticians make use of narratives to explain their views to others. If you submit an NSF grant proposal, you will be asked to write a narrative. But on my account, a narrative is a compact, powerful way of communicating complex, nuanced modal claims. The fairy tale is true, not because it tells us dragons exist, but because it tells us dragons can be beaten. And narratives manage to convey this information in simple, easily graspable terms.Footnote ²² What would be the alternative for finite, social creatures with time constraints? A narrative allows us to simply communicate complex information. Narratives are easily transmitted and shared. Most of the kinds of claims that narratives support couldn't be empirically grounded, either because it is about modal dispositions, or because it is about moral permissibility, or it is about what ones current situation warrants.