Long time reader, first time commenter. I love this argument. I would add that people who select into writing fiction aren't a representative sample of the population. They might be particularly likely to have certain personality traits--to be more like Albert Camus than Ernest Shackleton. The way they describe their characters' inner lives may reflect their own inner lives. That doesn't mean fiction isn't interesting or valuable. But it suggests that fiction might not give us a great sense of the inner lives of the kinds of people who don't write fiction.
Excellent read as always. I agree with the main points of the piece. As an aside, I’ve thought for a while that the fact people feel compelled to defend the “value” of reading in these sorts of ways is precisely the problem; we should value great books (and other works of art) for what they are, and the enjoyment they give us, not some marginal benefit that can be measured in a controlled experiment. (Something similar goes on with defenders of the value of philosophy that cite earning potential etc. Even if the stats are true they don’t capture all or most of the reasons we think philosophy is valuable.)
Anyway I wonder what you think of the following alternative defence of the epistemic or cognitive value of literature. Good books (and I certainly don’t just mean the classics, but I mean to include them) are valuable because they often contain deep insights, sometimes insights it is hard to get elsewhere (who would be a better source of insights into a peculiar section of English society at a particular period in time than Jane Austen?).
Perhaps the people who wrote these books *could* have put those insights in another form: an academic treatise, a work of commercial non-fiction, or whatever. But they didn’t so if you want them you need to read the book. Of course, sometimes you can get much the same ideas and perspectives in lots of places—but what makes really good books good is that they are a bit more original, a bit more idiosyncratic.
I would combine this with another claim, which is that some ideas, visions, perspectives might be better communicated in a work of fiction than in another form, certainly than in a less literary form. I imagine George Eliot could have written a dry academic treatise on social change in 19th century England. But she wrote Middlemarch instead and—no disrespect to whatever academic traits Eliot may have possessed—it was probably better that way.
I love this. Here's an example in that vein. Not too long ago I read "The Good Earth", by Pearl Buck. It tells the story of an early 20th century Chinese peasant, who at the beginning of the book lives in crushing poverty. His family is always on the brink of starvation, he loses a child to famine, etc. By the end of the book, through a combination of hard work, luck, and theft, he's rich. And yet, he doesn't at all seem to appreciate his situation. Rather, he's frustrated that his wife is old and ugly, and he wants a young concubine like the other rich folks have. And now he's worried that his children--who are literate, and have no need to fear starvation, both of which he would never have dreamed of ten years earlier--won't get prestigious government jobs. In reading the book you want to scream at him that he should appreciate how great his life is. But then, my life is a lot more comfortable than his even at the end of the book, so maybe I should be screaming at myself...
One could imagine just explicitly stating various truths about the hedonic treadmill, and the importance of gratitude, perspective, etc. But for me at least, I think seeing a vivid fictional example of exactly how *not* to navigate those issues helped those truths stick for me in a way that I don't think just reading them in treatise form would have. I know people who think if only they could publish in this journal, or get that job, then all would be well. And then other people who've published in this journal many times, and not only have the job, but have named chairs, but still are comparing themselves to Chomsky and coming up short. I tend to think reading the Good Earth was, for me, a kind of (partial) inoculation against that very strong psychological tendency.
One of the great joys of reading children’s historical fiction (to children) is that it makes history accessible and interesting in a way few other formats can match. As a homeschooler, I love using good historical fiction for kids to draw their interest into a specific time period!
As a fellow homeschooler, I'd welcome any specific suggestions (for older pre-teens) that you're willing to share. (Or just the link if you happen to have blogged such recommendations before!)
- By the Great Horned Spoon, Sid Fleischman (audio version is amazing)
- Great Illustrated Classics Treasure Island (annotated Treasure Island story; don’t recommend the real one until late middle school or high school due to how challenging the language is). Many of the Great Illustrated Classics are awesome including Sherlock Holmes.
- Anything by Avi. The “cross of lead” series is a classic for a reason but you can’t go wrong with him.
- We just read “The First State of Being” about Y2K and it was great. Prepare to feel old.
- I’m reading “Sweep: The Story of a Girl and Her Monster.” It’s set in Victorian England. It’s fabulous.
This reminds me of the thesis of a Facebook friend of mine (he published an abridged version of the claim at Aeon: https://aeon.co/essays/why-poetry-is-a-variety-of-mathematical-experience ), arguing that the machine learning concept of an "autoencoder" helps us understand how a genre of art can define a low-dimensional subspace of interesting conditions within a massively high-dimensional space of experiences, and that by "training on" this art, we can get a better understanding of the relevant dimensions than by trying to express it explicitly in a language we already have (the way a philosophical work might try).
Fictional characters are simplified, but learning to recognize something in simplified contexts is often how we train ourselves to eventually get the sensitivity to identify it in more complex contexts. This is the key to many sorts of education, ranging from physics (where we start with frictionless planes and two body systems in the vacuum of space, and eventually understand things well enough to think usefully and precisely about more complex systems) to wine tasting (where we can start with sniffing samples of individual molecules to build a smell vocabulary, and then learn to identify hints of these smells in wines where it is canonical, and then eventually in wines where it is an unexpected note).
I like this thought. Here's a reason for some pessimism about how well it works in the case of fiction. With most complex skills, you need a lot of feedback to make sure you're learning them right. It's a trope in the Brazillian jiu jitsu community that traditional martial arts, where sparring is a small to non-existent part of the training--mostly you're practicing moves against compliant partners--just doesn't prepare you all that well for real-world fights. The reason being that to actually learn a move in a way that's effective, you need a lot of experience trying out versions of it against people who are trying not to let you do it, and seeing what works and what doesn't. (So sport martial arts--boxing, wresting, judo--end up seeing much more use in MMA than traditional martial arts like karate.)
So I think the ideal "practice regime" for interpreting other people's mental lives using basics picked up from fiction would involve lots of opportunities for learning when you're doing it wrong. E.g., maybe a bartender, or a therapist--someone who has unusually deep conversations with a wide range of different sorts of people--has opportunities to form conjectures about what people's inner mental lives are like, and then to learn what sorts of mistakes they're prone to, and how to adjust their conjectures in the future so as to avoid those mistakes. Just like the wine taster learns when they correctly identified the molecule, and later the wine, or the physics student learns whether they got the question on the test right. But most of us don't get those sorts of opportunities to learn whether we're correctly imagining other people's inner mental lives.
It could be that staying within fiction is like staying with traditional martial arts or smell kits or frictionless planes, and you need to move beyond that. Or it could be that modernist and postmodernist literature contains writing that tries to push towards complexity, the way that graduate training in mechanical engineering goes beyond the frictionless plane.
The Philistines (who settled in Canaan around 12th century BCE) were about as well cultured as the peoples who were not Philistines. Of course their script and language have not survived due to assimilation. But we should cut them some slack. Likewise for non-Greek "barbarians", whose reputation the Greeks have done for :)
I recently stumbled upon this interesting YouTube video discussing Ursula LeGuin’s explanation for the convention of writing fiction in the past tense: https://youtu.be/4l23-1yWMPA
One of the points that was discussed is that written fiction gives an opportunity to discuss interiority in a way that other narrative media don’t - cinema sticks you in the flow of time, and gives you a better sense of experience but less of thought; video games give you the agency and the feel of motivation, but not of how someone else does it.
Whether or not there is a practical or epistemic benefit to this affordance of the medium, this is a kind of thing that one might be able to have an aesthetic appreciation of, and it is good to have aesthetic media that enable us to appreciate different things aesthetically! (I personally am much more motivated by the aesthetic appreciation of epistemology that I get from puzzles than I am of the aesthetic appreciations made possible in written fiction, but it’s valuable to be aware of each.)
For several centuries, engagement with literary fiction was one of the principle means by which people (not universally, but quite broadly) understood *their own* psychology. As recently as a decade ago, young people were still learning to interpret their inchoate experiences and shape their inchoate selves through engagement with J. K. Rowling's archetypical characters and plots. If you want to develop empathy for people who have constituted themselves as subjects in this way, reading novels is probably going to be helpful, perhaps even essential.
But self-fashioning through fiction is not an unchanging fact of human biology. It's a contingent product of particular historical circumstances. Perhaps the young people of 2035 will come to constitute themselves as subjects through their engagement with ChatGPT and memes. If that's how things turn out, developing empathy for this new species of human will require study of ChatGPT and memes, not novels.
I found this really engaging—especially because I share some of the skepticism about whether fiction helps us “understand” people in any reliable or transferable sense. But rather than settling the question, I’ve often been more interested in what kinds of effects fiction does reliably produce—especially at the level of cognition and affect.
For me, one of the most concrete effects is what I’d call the accumulation of reactive experience.
A sentence in fiction can act as a neural-emotional stimulus, and if it resonates, the brain doesn’t necessarily distinguish whether that resonance came from “diluted reality” or “symbolically processed construction.” In that sense, fiction isn’t just narrative—it's a field where various intensities of experience get stored as response samples.
We live within a limited pool of people and situations.
But fiction—and more broadly, created worlds—lets us expose our nervous system to a much wider range of human and situational contact than real life can afford.
And yes, fiction is narratively biased.
But I think this bias often mirrors how we ourselves process emotionally significant real-life events:
What happens is sometimes objectively trivial, but the subjective interpretation—often an unrefined form of narrative distortion—can feel enormous.
In that way, the structure of fiction echoes our own unstructured narrativizations.
Especially with certain impressions—vivid or affectively charged experiences—they tend to loop in memory,
and over time, we find that repetition itself starts to narrativize the event.
What was once just a reaction becomes something like a micro-story.
In that light, I’m less inclined to ask whether fiction teaches us “what it’s like” to be other people,
and more drawn to the idea that we inevitably reach for some form of story, however distorted,
to even begin imagining the motives of others.
And if that’s true, then fiction offers us patterns of structured misinterpretation—some of which might be livable.
Because no one can major in psychology just to get through a friendship.
And honestly, the more seriously one studies human cognition,
the stronger the conviction becomes that “humans are not fully knowable.”
Even so, in real life, if someone misunderstands me—but I can offer some plausible or self-defensible reason behind my action—
the relationship doesn't always break down.
So maybe what matters isn't whether fiction provides “truth,”
but whether it gives us forms of motivated explanation that function interpersonally.
If I had to put it flatly:
Fiction may not offer a comprehensive understanding of human beings,
but when narrative structures derived from fiction solidify into the only lens we use for interpreting real life,
then interpretive distortion becomes nearly unavoidable.
(I can probably get you a link to the full text of the article if you’re interested in reading it.)
A pretty low-powered study, but I think the claim that reading and writing fiction makes you more empathic is B.S. I think it is just that, for whatever reason (the authors of the study hypothesize that it’s the halo effect), reading and writing fiction increases the *perception* that you’re more empathic. Sigh.
Well, (1) fiction writers are avid readers themselves, so if you don’t see an effect among them, then it throws that whole idea that reading improves empathic abilities into doubt and (2) if fiction writers themselves don’t have greater empathic abilities (depending on how you define that term) than the average person, then why would you ever expect a novel to improve the empathic abilities of the person reading it?
Some studies failed to replicate the Kidd and Castano finding that reading literature improves theory of mind reasoning, but many more did replicate it. A meta-analysis, which includes the non-replicating study that you cite, finds that the effect is real. Here is the meta-analysis: https://labsites.rochester.edu/scplab/wp-content/uploads/2018/06/Dodell-Feder-JEPG-inpress.pdf
Jared Peterson has a post on stories as open-ended scientific models with more flexibility and context built in, which might be a nice complement to your own take:
I think the ways in which fiction is epistemically valuable is simply not capturable by your opening question, as formulated. After all, by definition it's not about anyone real. So it's not only a problem of accuracy; there's also the question of how to identify *which* real person in the world it should apply to. Fictional characters are either archetypes for a broad class of people you can only make generalizations about, or idiosyncratic examples that might remind you of someone you know. But it's not actually based on anyone you know, so any understanding it appears to yield would be your own projection after the fact. (To be clear, I do think fiction offers something epistemic, but that accuracy is an incoherent metric since there's no target).
I definitely like the analogy between stories and models. There's a strand in the literature on modeling in philosophy of science which starts with models (rather than stories) and analogizes them to certain sorts of stories (fables, parables):
The reason I'm pessimistic about how far this thought gets you towards the idea that a healthy diet of fiction makes you better at interpreting other *actual* people came up in a thread below with Kenny Easwaran. In the case of scientific models, it's a non-trivial thing to be good at fitting the right model to the right real-world scenario. I like Dani Rodrik's book "Economics Rules" a lot, and he emphasizes how much art rather than science goes into decisions about which models to use to think about which real-world economic problems. But at least in econ, you get to fit models to data, and you get feedback about which models perform poorly in which settings. So there's some reason for optimism about economists collectively converging over time on better models. (This will come up in a future post, as it turns out.) If we think what fiction gives you is a bunch of models that you can use for interpreting people, there's still the highly non-trivial skill of being good at picking which model to use for interpreting which people, and I tend to most of us don't get enough feedback about how well we're doing at fitting interpretive models to people to get all that good at it.
"If we think what fiction gives you is a bunch of models that you can use for interpreting people, there's still the highly non-trivial skill of being good at picking which model to use for interpreting which people."
This is exactly what I was getting at, in less precise language. The only connection the fiction model originally had to the real world in terms of data was the writer's own personal experiences and observations; meanwhile, the reader is getting some of that data secondhand and greatly transformed through the story, while simultaneously interpreting this firsthand through the very different data of their *own* prior experiences and observations. And then on top of that the reader must apply this to a specific situation back out in the world. So, many sources of distortion.
There is actually some empirical support for the premise that fiction reading increases social cognition and activates the default mode (aka mentalizing) network in the brain-- but you are totally right, this is a non-random sample.
Long time reader, first time commenter. I love this argument. I would add that people who select into writing fiction aren't a representative sample of the population. They might be particularly likely to have certain personality traits--to be more like Albert Camus than Ernest Shackleton. The way they describe their characters' inner lives may reflect their own inner lives. That doesn't mean fiction isn't interesting or valuable. But it suggests that fiction might not give us a great sense of the inner lives of the kinds of people who don't write fiction.
Excellent read as always. I agree with the main points of the piece. As an aside, I’ve thought for a while that the fact people feel compelled to defend the “value” of reading in these sorts of ways is precisely the problem; we should value great books (and other works of art) for what they are, and the enjoyment they give us, not some marginal benefit that can be measured in a controlled experiment. (Something similar goes on with defenders of the value of philosophy that cite earning potential etc. Even if the stats are true they don’t capture all or most of the reasons we think philosophy is valuable.)
Anyway I wonder what you think of the following alternative defence of the epistemic or cognitive value of literature. Good books (and I certainly don’t just mean the classics, but I mean to include them) are valuable because they often contain deep insights, sometimes insights it is hard to get elsewhere (who would be a better source of insights into a peculiar section of English society at a particular period in time than Jane Austen?).
Perhaps the people who wrote these books *could* have put those insights in another form: an academic treatise, a work of commercial non-fiction, or whatever. But they didn’t so if you want them you need to read the book. Of course, sometimes you can get much the same ideas and perspectives in lots of places—but what makes really good books good is that they are a bit more original, a bit more idiosyncratic.
I would combine this with another claim, which is that some ideas, visions, perspectives might be better communicated in a work of fiction than in another form, certainly than in a less literary form. I imagine George Eliot could have written a dry academic treatise on social change in 19th century England. But she wrote Middlemarch instead and—no disrespect to whatever academic traits Eliot may have possessed—it was probably better that way.
I love this. Here's an example in that vein. Not too long ago I read "The Good Earth", by Pearl Buck. It tells the story of an early 20th century Chinese peasant, who at the beginning of the book lives in crushing poverty. His family is always on the brink of starvation, he loses a child to famine, etc. By the end of the book, through a combination of hard work, luck, and theft, he's rich. And yet, he doesn't at all seem to appreciate his situation. Rather, he's frustrated that his wife is old and ugly, and he wants a young concubine like the other rich folks have. And now he's worried that his children--who are literate, and have no need to fear starvation, both of which he would never have dreamed of ten years earlier--won't get prestigious government jobs. In reading the book you want to scream at him that he should appreciate how great his life is. But then, my life is a lot more comfortable than his even at the end of the book, so maybe I should be screaming at myself...
One could imagine just explicitly stating various truths about the hedonic treadmill, and the importance of gratitude, perspective, etc. But for me at least, I think seeing a vivid fictional example of exactly how *not* to navigate those issues helped those truths stick for me in a way that I don't think just reading them in treatise form would have. I know people who think if only they could publish in this journal, or get that job, then all would be well. And then other people who've published in this journal many times, and not only have the job, but have named chairs, but still are comparing themselves to Chomsky and coming up short. I tend to think reading the Good Earth was, for me, a kind of (partial) inoculation against that very strong psychological tendency.
Yes this is an excellent example. Also hadn't heard of the book--it sounds worth checking out.
One of the great joys of reading children’s historical fiction (to children) is that it makes history accessible and interesting in a way few other formats can match. As a homeschooler, I love using good historical fiction for kids to draw their interest into a specific time period!
As a fellow homeschooler, I'd welcome any specific suggestions (for older pre-teens) that you're willing to share. (Or just the link if you happen to have blogged such recommendations before!)
One of my favorite questions!!!
Some of my favorites for that age:
- By the Great Horned Spoon, Sid Fleischman (audio version is amazing)
- Great Illustrated Classics Treasure Island (annotated Treasure Island story; don’t recommend the real one until late middle school or high school due to how challenging the language is). Many of the Great Illustrated Classics are awesome including Sherlock Holmes.
- Anything by Avi. The “cross of lead” series is a classic for a reason but you can’t go wrong with him.
- We just read “The First State of Being” about Y2K and it was great. Prepare to feel old.
- I’m reading “Sweep: The Story of a Girl and Her Monster.” It’s set in Victorian England. It’s fabulous.
Thanks so much! Excited to check these out :-)
This reminds me of the thesis of a Facebook friend of mine (he published an abridged version of the claim at Aeon: https://aeon.co/essays/why-poetry-is-a-variety-of-mathematical-experience ), arguing that the machine learning concept of an "autoencoder" helps us understand how a genre of art can define a low-dimensional subspace of interesting conditions within a massively high-dimensional space of experiences, and that by "training on" this art, we can get a better understanding of the relevant dimensions than by trying to express it explicitly in a language we already have (the way a philosophical work might try).
So many interesting themes to pick up on here!
One response:
Fictional characters are simplified, but learning to recognize something in simplified contexts is often how we train ourselves to eventually get the sensitivity to identify it in more complex contexts. This is the key to many sorts of education, ranging from physics (where we start with frictionless planes and two body systems in the vacuum of space, and eventually understand things well enough to think usefully and precisely about more complex systems) to wine tasting (where we can start with sniffing samples of individual molecules to build a smell vocabulary, and then learn to identify hints of these smells in wines where it is canonical, and then eventually in wines where it is an unexpected note).
I did not know that's how wine tasting worked!
I like this thought. Here's a reason for some pessimism about how well it works in the case of fiction. With most complex skills, you need a lot of feedback to make sure you're learning them right. It's a trope in the Brazillian jiu jitsu community that traditional martial arts, where sparring is a small to non-existent part of the training--mostly you're practicing moves against compliant partners--just doesn't prepare you all that well for real-world fights. The reason being that to actually learn a move in a way that's effective, you need a lot of experience trying out versions of it against people who are trying not to let you do it, and seeing what works and what doesn't. (So sport martial arts--boxing, wresting, judo--end up seeing much more use in MMA than traditional martial arts like karate.)
So I think the ideal "practice regime" for interpreting other people's mental lives using basics picked up from fiction would involve lots of opportunities for learning when you're doing it wrong. E.g., maybe a bartender, or a therapist--someone who has unusually deep conversations with a wide range of different sorts of people--has opportunities to form conjectures about what people's inner mental lives are like, and then to learn what sorts of mistakes they're prone to, and how to adjust their conjectures in the future so as to avoid those mistakes. Just like the wine taster learns when they correctly identified the molecule, and later the wine, or the physics student learns whether they got the question on the test right. But most of us don't get those sorts of opportunities to learn whether we're correctly imagining other people's inner mental lives.
I should probably hedge what I say about wine tasting - I don't know if people actually train that way! But a bit of searching does turn up these scent kits: https://napavalleywineacademy.com/collections/le-nez-du-vin-aroma-kits
It could be that staying within fiction is like staying with traditional martial arts or smell kits or frictionless planes, and you need to move beyond that. Or it could be that modernist and postmodernist literature contains writing that tries to push towards complexity, the way that graduate training in mechanical engineering goes beyond the frictionless plane.
I had the same thought, and you expressed it well!
This is an aside. Nevertheless!
The Philistines (who settled in Canaan around 12th century BCE) were about as well cultured as the peoples who were not Philistines. Of course their script and language have not survived due to assimilation. But we should cut them some slack. Likewise for non-Greek "barbarians", whose reputation the Greeks have done for :)
It never occurred to me to wonder about the Philistines. I really don't know the first thing about them. Thanks!
Another point:
I recently stumbled upon this interesting YouTube video discussing Ursula LeGuin’s explanation for the convention of writing fiction in the past tense: https://youtu.be/4l23-1yWMPA
(Incidentally, I was so struck by the philosophical significance of many of her quoted observations about the experience, metaphysics, and cognition of time that I want to read the whole book: this review gives a taste: https://slate.com/culture/2015/09/ursula-le-guins-writing-guide-steering-the-craft-reviewed.html )
One of the points that was discussed is that written fiction gives an opportunity to discuss interiority in a way that other narrative media don’t - cinema sticks you in the flow of time, and gives you a better sense of experience but less of thought; video games give you the agency and the feel of motivation, but not of how someone else does it.
Whether or not there is a practical or epistemic benefit to this affordance of the medium, this is a kind of thing that one might be able to have an aesthetic appreciation of, and it is good to have aesthetic media that enable us to appreciate different things aesthetically! (I personally am much more motivated by the aesthetic appreciation of epistemology that I get from puzzles than I am of the aesthetic appreciations made possible in written fiction, but it’s valuable to be aware of each.)
That reminds me of this interesting post on interiority in recent fiction: https://theculturewedeserve.substack.com/p/culture-digested-what-was-woke-art
For several centuries, engagement with literary fiction was one of the principle means by which people (not universally, but quite broadly) understood *their own* psychology. As recently as a decade ago, young people were still learning to interpret their inchoate experiences and shape their inchoate selves through engagement with J. K. Rowling's archetypical characters and plots. If you want to develop empathy for people who have constituted themselves as subjects in this way, reading novels is probably going to be helpful, perhaps even essential.
But self-fashioning through fiction is not an unchanging fact of human biology. It's a contingent product of particular historical circumstances. Perhaps the young people of 2035 will come to constitute themselves as subjects through their engagement with ChatGPT and memes. If that's how things turn out, developing empathy for this new species of human will require study of ChatGPT and memes, not novels.
Nice point! So to the extent that life imitates art, you want to know the art life is imitating to understand the life.
I found this really engaging—especially because I share some of the skepticism about whether fiction helps us “understand” people in any reliable or transferable sense. But rather than settling the question, I’ve often been more interested in what kinds of effects fiction does reliably produce—especially at the level of cognition and affect.
For me, one of the most concrete effects is what I’d call the accumulation of reactive experience.
A sentence in fiction can act as a neural-emotional stimulus, and if it resonates, the brain doesn’t necessarily distinguish whether that resonance came from “diluted reality” or “symbolically processed construction.” In that sense, fiction isn’t just narrative—it's a field where various intensities of experience get stored as response samples.
We live within a limited pool of people and situations.
But fiction—and more broadly, created worlds—lets us expose our nervous system to a much wider range of human and situational contact than real life can afford.
And yes, fiction is narratively biased.
But I think this bias often mirrors how we ourselves process emotionally significant real-life events:
What happens is sometimes objectively trivial, but the subjective interpretation—often an unrefined form of narrative distortion—can feel enormous.
In that way, the structure of fiction echoes our own unstructured narrativizations.
Especially with certain impressions—vivid or affectively charged experiences—they tend to loop in memory,
and over time, we find that repetition itself starts to narrativize the event.
What was once just a reaction becomes something like a micro-story.
In that light, I’m less inclined to ask whether fiction teaches us “what it’s like” to be other people,
and more drawn to the idea that we inevitably reach for some form of story, however distorted,
to even begin imagining the motives of others.
And if that’s true, then fiction offers us patterns of structured misinterpretation—some of which might be livable.
Because no one can major in psychology just to get through a friendship.
And honestly, the more seriously one studies human cognition,
the stronger the conviction becomes that “humans are not fully knowable.”
Even so, in real life, if someone misunderstands me—but I can offer some plausible or self-defensible reason behind my action—
the relationship doesn't always break down.
So maybe what matters isn't whether fiction provides “truth,”
but whether it gives us forms of motivated explanation that function interpersonally.
If I had to put it flatly:
Fiction may not offer a comprehensive understanding of human beings,
but when narrative structures derived from fiction solidify into the only lens we use for interpreting real life,
then interpretive distortion becomes nearly unavoidable.
Thanks for the piece—it left a lot resonating.
Empirical research that shows that experienced fiction writers do *not*, on objective measures, perceive the mental and emotional states of other people any more accurately than non-writers: https://www.jbe-platform.com/content/journals/10.1075/ssol.4.2.01bis
(I can probably get you a link to the full text of the article if you’re interested in reading it.)
A pretty low-powered study, but I think the claim that reading and writing fiction makes you more empathic is B.S. I think it is just that, for whatever reason (the authors of the study hypothesize that it’s the halo effect), reading and writing fiction increases the *perception* that you’re more empathic. Sigh.
Huh, I hadn't thought of testing the hypothesis by looking at how *writers* differ from everybody else. Cool idea.
Well, (1) fiction writers are avid readers themselves, so if you don’t see an effect among them, then it throws that whole idea that reading improves empathic abilities into doubt and (2) if fiction writers themselves don’t have greater empathic abilities (depending on how you define that term) than the average person, then why would you ever expect a novel to improve the empathic abilities of the person reading it?
Some studies failed to replicate the Kidd and Castano finding that reading literature improves theory of mind reasoning, but many more did replicate it. A meta-analysis, which includes the non-replicating study that you cite, finds that the effect is real. Here is the meta-analysis: https://labsites.rochester.edu/scplab/wp-content/uploads/2018/06/Dodell-Feder-JEPG-inpress.pdf
And here is an article that I wrote about the issue: https://open.substack.com/pub/eclecticinquiries/p/why-do-literary-people-hate-science?r=4952v2&utm_campaign=post&utm_medium=web&showWelcomeOnShare=false
Jared Peterson has a post on stories as open-ended scientific models with more flexibility and context built in, which might be a nice complement to your own take:
https://open.substack.com/pub/jtpeterson/p/stories-as-context-sensitive-models
I think the ways in which fiction is epistemically valuable is simply not capturable by your opening question, as formulated. After all, by definition it's not about anyone real. So it's not only a problem of accuracy; there's also the question of how to identify *which* real person in the world it should apply to. Fictional characters are either archetypes for a broad class of people you can only make generalizations about, or idiosyncratic examples that might remind you of someone you know. But it's not actually based on anyone you know, so any understanding it appears to yield would be your own projection after the fact. (To be clear, I do think fiction offers something epistemic, but that accuracy is an incoherent metric since there's no target).
I definitely like the analogy between stories and models. There's a strand in the literature on modeling in philosophy of science which starts with models (rather than stories) and analogizes them to certain sorts of stories (fables, parables):
https://www.iasdurham.org/wp-content/uploads/2021/03/Cartwright_Models-Parables-v-Fables.pdf
The reason I'm pessimistic about how far this thought gets you towards the idea that a healthy diet of fiction makes you better at interpreting other *actual* people came up in a thread below with Kenny Easwaran. In the case of scientific models, it's a non-trivial thing to be good at fitting the right model to the right real-world scenario. I like Dani Rodrik's book "Economics Rules" a lot, and he emphasizes how much art rather than science goes into decisions about which models to use to think about which real-world economic problems. But at least in econ, you get to fit models to data, and you get feedback about which models perform poorly in which settings. So there's some reason for optimism about economists collectively converging over time on better models. (This will come up in a future post, as it turns out.) If we think what fiction gives you is a bunch of models that you can use for interpreting people, there's still the highly non-trivial skill of being good at picking which model to use for interpreting which people, and I tend to most of us don't get enough feedback about how well we're doing at fitting interpretive models to people to get all that good at it.
Thanks for that link - looks interesting!
"If we think what fiction gives you is a bunch of models that you can use for interpreting people, there's still the highly non-trivial skill of being good at picking which model to use for interpreting which people."
This is exactly what I was getting at, in less precise language. The only connection the fiction model originally had to the real world in terms of data was the writer's own personal experiences and observations; meanwhile, the reader is getting some of that data secondhand and greatly transformed through the story, while simultaneously interpreting this firsthand through the very different data of their *own* prior experiences and observations. And then on top of that the reader must apply this to a specific situation back out in the world. So, many sources of distortion.
There is actually some empirical support for the premise that fiction reading increases social cognition and activates the default mode (aka mentalizing) network in the brain-- but you are totally right, this is a non-random sample.
https://academic.oup.com/scan/article/11/2/215/2375122