I like fucked-up fiction. So this post is kind of a betrayal of my side in the What Things Should Authors Write About wars, which tends to produce posts like this:
“Fiction affects reality.”
Yes. But fiction is not reality and it doesn’t make us do anything.
Fiction affects our feelings. Regarding our emotions, our brains activate the same regions that respond to real events when we read fiction. That’s why we feel for the characters in stories.
That does not mean that fiction makes us do anything, and also, it doesn’t mean that the creators of fiction are somehow responsible for the mental health or morals of the audience. It’s always an individual’s responsibility to choose what fiction to consume, how to react to it (as in, what actions to take), and to make sure they have a healthy concept of reality.Fiction and its creators have in no way a responsibility to be some edifying moral influence in the audience’s life. Audiences have 100% of the responsibility to curate their fictional consumption wisely.
I’m not here to tell people they should or shouldn’t write anything. In particular:
Fiction having an effect on reality requires it having an audience: if only twenty of the author’s friends read a story, the effect it has on the world is negligible.
I don’t think there’s any evidence that horrific, disturbing, “triggering,” or otherwise fucked up fiction is more likely to have negative effects on the world than any other kind of fiction.
My guess is that the effects of a story are attenuated to the point of nonexistence if the story is obviously unrelated to reality in any way, such as some experimental fiction and outre pornography.
But I do think there’s pretty robust evidence that fiction has effects on its audience. This evidence is rarely talked about in a careful and nuanced way. Instead, it’s used for writerly self-congratulation or as a cudgel to condemn whatever fiction the author finds disgusting. So I’m going to talk about various effects of fiction, and then conclude with some thoughts about what writers should do.
Empathy and Fiction
People often claim that fiction creates empathy: that is, that fiction makes it easier for us to put ourselves in the shoes of people different from us. The studies of this are mostly bad. For example, there was a priming study that found that reading a brief passage of literary fiction improved people’s ability to tell emotions from faces. As you might expect, it failed to replicate.
One study found that people who were emotionally engaged in a fiction story had higher levels of empathetic concern: that is, warmth and compassion, measured by questions like “I often have tender, concerned feelings for people less fortunate than me” and “I care for my friends a great deal.” Conversely, emotionally disengaged readers had lower levels of empathetic concern. But the study was on a few dozen undergraduates and they were only followed up for a week. The most robust effect came from reading a Sherlock Holmes story—not what one would think of as an especially empathy-producing story. I smell p-hacking.
A meta-analysis reviewed the evidence on whether people who read fiction regularly have more empathy: that is, both empathetic concern and ability to take other people’s perspectives. Because these studies are correlational, we can infer very little from them: it’s possible (indeed likely) that people like reading more if they care more about others and have an easier time putting themselves into others’ shoes. The correlation between fiction reading and self-reported empathy was .07—that is, smaller than the correlation between extraversion and how much you spend on holiday gifts.
A larger correlation was found between fiction reading and the ability to tell emotions from faces (.21, about the same as the correlation between parents’ social class and children’s grades). But that shows that this entire line of research is misguided. Fiction might make you better able to understand why people behave the way they do, because you’re seeing inside the minds of strangers. But fiction books don’t have any pictures of faces to practice on. If anything, reading more fiction should make you less able to read people’s faces, because of all the conversations you’re grunting through while focusing on your book. Some confounder is clearly in play here.
But maybe we don’t expect fiction reading to increase empathy as a general trait. Maybe we expect it to increase empathy with the sort of people you’re reading about. There’s no reason to expect that reading about handsome and eligible Regency dukes with a dark and troubled past would increase your empathy with Indian Muslims from the slums of Mumbai. For the latter you must watch Slumdog Millionaire.
I haven’t been able to find any studies of this obvious hypothesis. It does seem pretty likely to me based on my experience reading fiction. But in his excellent Against Empathy, Paul Bloom quotes an unpublished article that eloquently warns of the risks of this sort of empathy:
For every Uncle Tom’s Cabin there is a Birth of a Nation. For every Bleak House there is an Atlas Shrugged. For every Color Purple there is a Turner Diaries, that white supremacist novel Timothy McVeigh left in his truck on the way to bombing the Oklahoma building. Every single one of these fictions plays on its readers’ empathy: not just high-minded writers like Dickens, who invite us to sympathize with Little Dorrit, but also writers of Westerns, who present poor helpless colonizers attacked by awful violent Native Americans; Ayn Rand, whose resplendent “job-creators” are constantly being bothered by the pesky spongers who merely do the real work; and so on and so on.1
Some writers do encourage the reader to empathize with everyone in their books.2 But most authors don’t. They create contemptible villains whose demises we root for, without making us reflect on whether the villain has a point. They create annoying and cringey side characters whose humiliation we laugh at, without prompting us to ask whether in real life this would be bullying. They create interchangeable hot girls for the hero to seduce to show off his manliness, without putting much thought into what the girls think about all this.3 And even the authors with the most all-encompassing empathy rarely put us in the head of Spearcarrier #14.
All this is natural. If nothing else, space limitations command that the reader empathize with some characters and not others. There’s a reasonable argument that we shouldn’t empathize with everyone: too much concern for the wellbeing of (say) Harvey Weinstein easily leads to a lack of concern for the wellbeing of his victims. But all this shows that fiction’s ability to provoke empathy—if it exists—isn’t a straightforward good. With whom are we empathizing? Whom is our empathy leaving out?
Social Norms
There is considerable evidence that fiction affects people’s behavior through affecting their understanding of social norms.
The most interesting studies are of Brazilian telenovelas. In the 1970s and 1980s, the Brazilian government expanded the TV network Rede Globo to cities based on politics, giveaways to powerful people’s loyal followers, and so on. We don’t expect this to be correlated with individual Brazilians’ romantic and family-formation behavior, so Rede Globo expansion is a natural experiment.
Telenovela exposure has a major effect on fertility. Brazilian telenovelas typically depict families with very few children, in order to keep the number of characters manageable: “in 62.2 percent of novelas the main female character does not have any children, in 20.7 percent she has one child, in 9 percent two children, 4.5 percent three children, and in the remaining cases has four or more children.” Women in areas with Rede Globo access were 0.6 percentage points less likely to give birth in a given year, equivalent to a woman getting two extra years of education.
Telenovela exposure also seems to have small but significant effects on divorce rates. Telenovelas overrepresented divorced characters: for example, about a fifth of female protagonists were divorced in the 1975-1984 period, although in 1984 only 3 in 100 Brazilian marriages ended in divorce. Being in an area with Rede Globo access raised divorce rates by about a tenth of a standard deviation—a small but noticeable effect.
There have also been a few randomized controlled trials of “edutainment.” In Uganda, economists put on free film festivals. Some festivals were randomized to include a short fictional film conveying a message about a social norm: parents should take action when teachers don’t show up at school; domestic violence should be reported to the authorities; and you should help people who have complications from an illegal abortion even if you’re opposed to abortion. The effect sizes were medium-sized:
In the teacher absenteeism condition, self-reported willingness to do something if a teacher regularly didn’t show up to school increased by half a standard deviation (a little more than the amount ibuprofen helps arthritis pain).
There was also a statistically significant increase in the number of people who thought education was among the most important goals for their village.
In the violence against women condition, self-reported willingness to help abuse victims and report abuse to the authorities increased by half a standard deviation.
In the abortion condition, self-reported willingness to help people who experienced complications from an abortion increased by a third of a standard deviation (a little less than the amount SSRIs help depression).
The spillover effects to people who didn’t go to the film festivals were very small, suggesting the effect comes from consuming the media.
In Rwanda, researchers randomized people to listen to a yearlong radio soap opera:
The soap opera launched in Rwanda one decade after the genocide featured a fictional story of two Rwandan communities. Due to government restrictions against public discussion of ethnicity, the story of the communities serves as a transparent allegory for the history of cooperation and conflict between Rwandan Tutsis and Hutus. Inter-community tensions created by a land shortage are set aflame by demagogic authorities who seek to accumulate power, and relations between the fictional communities disintegrate as the less prosperous community attacks its rival. Against this backdrop of communal violence, a romantic storyline unfolds between a young man and woman, each from a different community. Instead of falling victim to the violence and prejudices between their two communities, the Rwandan Romeo and Juliet form a coalition for peace with citizens from both communities. Their coalition defies the power-hungry authorities and seeks to mediate the conflict and help the victims.
Listeners to Rwandan Romeo and Juliet were consistently more likely to question authority across several different metrics:
Rwandan communities have a low level of social trust, but it’s taboo to admit that in public. Participants randomized to Rwandan Romeo and Juliet (as opposed to a comparable radio soap opera about health behaviors) were three times as likely to admit in a focus group that there was a lot of mistrust in their community.
When asked to roleplay how they thought a community should respond to a refugee crisis, all Rwandan Romeo and Juliet groups tried to solve the problem as a community, while health-behavior-soap-opera groups usually asked the government or an NGO to help out.
When given a radio as a reward for participation,4 all health-behavior-soap-opera groups quickly agreed to give the radio to the village’s local authority. Conversely, Rwandan Romeo and Juliet groups challenged the suggestion to give the radio to the local authority, instead suggesting that the group could share responsibility or vote for a member to manage it.
However, there was no difference in participants’ interest in “greeting, sharing beer with, working with, and intermarrying a child with a member of a different group”—the antiracist effect that the creators of the soap opera were hoping for. This suggests to me that the anti-authoritarian effects of the soap opera were real and not social-desirability effects.5
A Sidebar About Violence
You might think, based on the evidence about social norms, that we should be opposed to violent media. In fact, the evidence tentatively suggests that violent media reduces violence.
One study suggested that, between 1995 and 2004, violent movies in theaters prevented a thousand assaults per weekend. The study compared the number of assaults on weekends with violent blockbuster movies in theaters to the number of assaults on weekends with nonviolent blockbuster movies in theaters. They found a substantial decrease in violence on weekends when a popular violent movie was in theaters.
The reason is that violent people like violent media. If there’s a violent movie in theaters, they’re likely to go watch it. Conversely, if the popular movies are nonviolent, they’re more likely to decide the movie sounds boring, go out drinking and then assault someone. In my opinion, the effect size of violent media distracting violent people from committing violence far outweighs any reasonable estimate of how much violent media changes social norms, at least for adults. These findings could possibly be generalized to other kinds of behavior we’d rather people not do: for example, if teens can access high-quality pornography, they’ll probably lose their virginity at older ages.
In short, the most virtuous kind of writing to work on is extremely addictive first-person shooters.
Information Effects
People learn about the world from media they consume. It makes sense that they’d learn things from fiction and assume those things are true. However, the evidence that this happens is surprisingly equivocal.
The “Jaws effect” is the idea that people became more scared of sharks as a consequence of the popular book and film Jaws. However, Jaws was a product of a decades-long change in attitudes about sharks towards seeing them as deadly killers bent on hunting down humans. While Jaws-style shark hunts were more common after Jaws (a social norm shift), it’s unclear whether Jaws had any effect on people’s beliefs about sharks at all.
Similarly, many people have proposed a “CSI effect”: police procedurals either harm the prosecutor’s case by making people think that every case needs to have a lot of forensic evidence, or harm the defense’s case by making people think that forensic science is more reliable than it really is. However, a review of the CSI Effect in the Routledge International Handbook of Forensic Intelligence and Criminology found no particular evidence for this claim. Studies generally find no difference between viewers and nonviewers of forensic science TV shows with regards to their views of the accuracy or necessity of forensic evidence or their likelihood of acquitting a defendant. Even when effects existed, they were very small, and often pointed different ways in different studies. Similarly, while TV news tends to scare people about crime, there’s little evidence that fictional crime shows have the same effect.
That said, some studies do show an effect. For example, heavy science fiction consumers are more concerned about lethal autonomous weapons than people who don’t consume much science fiction, with consumers of scarier franchises being more concerned than consumers of less scary franchises.
The most interesting study is on the reality show 16 and Pregnant, which estimated the effect of 16 And Pregnant by comparing areas that watched more MTV before the show aired with areas that watched less MTV. The authors concluded that the show led to a 4.3% decline in teen births during its first year of airing, which means it accounted for about a quarter of the decline in teen births in that period.6
However, 16 and Pregnant is a reality show. Viewers might believe that 16 and Pregnant accurately depicts what it’s like to be a teen mom because it’s documenting the lives of real people. This would be similar to how people who watch TV news are more scared of crime, but people who watch fictional crime shows aren’t.
I suspect that people know that fiction is, well, not reality, so they’re more likely to discount facts they learn from fiction, compared to facts they learn from nonfictional sources such as reality shows and TV news. The most robust effect I found was from a 2018 study of lethal autonomous weapons—a subject that there just wasn’t much non-fictional information on at the time.
I did find a small study suggesting that people who watch more fictional TV are more likely to believe in the just-world fallacy (that is, that good things happen to good people and bad things to bad people). While this study is correlational, I can’t think of a reason why people who believe in the just world hypothesis would be more likely to watch TV, so I suspect that if this effect is real it’s causal. If fiction does cause people to be more likely to believe in the just-world fallacy, I think this might be because it’s more of a narrative and less of a fact—there’s not specific information to be skeptical of, the way you might be skeptical of information from CSI.
What Should Writers Do?
The most robust effects of fiction seem to be on social norms: people’s perception of what behavior is normal, acceptable, cool, or badass. There’s not much evidence of informational effects or empathy effects, although both seem prima facie plausible at least in some situations.
Some writers are just interested in telling a good story, but don’t want to cause harm. If you have a large audience,7 I think it makes sense to spend a little time thinking about the norms expressed by your story.
Many stories will have no relevant norms. Haunted houses don’t exist, so it doesn’t matter if your story glamorizes ineffective ways of responding to haunted houses. Most modern monarchies are powerless and most modern dictators are self-made men, so it seems harmless to write a story that expresses the norm that countries should be ruled by the rightful king who has royal blood flowing in his veins. Similarly, many stories are understood by their audience to not accurately reflect good behavior. For example, if a fanfiction is tagged “Dubious Consent” or “Idiots in Love,” you know you’re not supposed to do anything these people get up to.
Still, some stories express norms that are, in my opinion, very harmful: if a man ignores a woman’s preference not to speak to him, it means that he loves her; society is kept safe by strong men who make hard decisions like being willing to torture people; innocent people don’t care about having their civil liberties violated by police. I think it’s worth reworking your story in those cases to see if you can tell a story that’s equally good without reinforcing those norms.
I would also suggest that authors, when possible, try to include accurate details about the world; to include characters the reader empathizes with from marginalized groups (e.g. queer people, people of color, disabled people, homeless people, people from the developing world); and to avoid writing books where no empathy is extended to any female characters.8
Conversely, some writers want to make the world a better place through their writing. For these writers, I suggest the following guidelines:
Do lots of research and make sure all the facts in your book are accurate.9
Consider carefully who you want your reader to empathize with and who you don’t. Some writers might want to get readers to empathize with everyone on every side of a conflict.10 Other writers might want their readers not to empathize with specific people that they think are causing harm. Make these decisions intentionally.
Choose themes for your book that advance ideas that you agree with. Robert McKee’s Story—long considered the Bible of Hollywood—extensively discusses how to have a strong theme in your story without coming off as preachy.
Subtly convey social norms through your characters’ behavior. Remember, the telenovelas never explicitly said “you should have fewer children”: they simply depicted rich, successful, glamorous women who happened not to have kids. You can simply depict the coolest characters in your story doing behavior you approve of, without explicit editorialization or commentary.
[Ozy’s footnote:] if you’re one of my Ayn-Rand-sympathetic readers, you can replace this example with Upton Sinclair’s The Jungle.
I vividly remember the vertigo-inducing experience of reading Guy Gavriel Kay’s Tigana in middle school. I ended each section convinced that the point-of-view character was in the right and only a moral imbecile could doubt the righteousness of their cause, only to switch positions three pages into the next section.
Friends, I read speculative fiction from the 1970s and 1980s, this is an epidemic.
The participants didn’t know that the researchers were taking notes on this interaction.
Note that it’s possible that the health-behaviors soap opera made people more authoritarian—there wasn’t a neutral control group. Still, I think we can conclude some radio show was affecting people somehow.
This is prima facie plausible—71% of teenagers had seen an episode of 16 And Pregnant and it was consistently the #1 slot in its time period among women aged 12 to 34.
As I said above, if you’re writing for an audience of twenty of your friends, don’t worry about it.
There are many settings that legitimately don’t have very many people of color or people from the developing world, and queer and disabled people are minorities. But women are half the population and nearly all settings (outside of boarding schools, prisons, and some militaries) include them.
Or I guess you can lie if for some reason you want to advance beliefs that aren’t true.
The book series A Song of Ice and Fire is a masterclass in this. I also suggest studying Terry Pratchett to understand how to build reader empathy with minor characters without slowing down the plot of your book.
> I can’t think of a reason why people who believe in the just world hypothesis would be more likely to watch TV
As someone who does not believe in the just world hypothesis, it can sometimes get tedious to constantly consume media that does :P
One thing I have noticed about debates about how fiction affects people is that people who tend to argue it is positive tend to focus on the themes and messages in the story. By contrast, people who argue it is negative tend to claim that behavior that is portrayed in fiction will influence behavior in reality, regardless of whether it is portrayed ositive or negative. So, for example, in the famous comic book moral panic of the Fifties, there were a lot of horror and crime anyhology comics that portrayed villain protagonists getting a grisly comeuppance after doing something terrible. The moral panickers tended to assume people would do crime because crime is portrayed, even though it is condemned and punished in the narrative. The defenders of the comics tended to deny that this would happen, but did believe that explicit positive messages might have some impact, for example, one publisher was proud of publishing a science fiction story with an anti-racisr moral.
I suspect what's actually going on is that some people find the portrayal of certain subject matter to be viscerally upsetting and seek rationalizations for censoring it. The people in the Fifties who hated horror and crime comics asserted that they promoted violence, even though the narrative condemned violence. You can see the same dynamic today when someone condemns a work of fiction for being racist or homophobic when it portrays an evil villain doing racist and homophobic things.
Based on the studies Ozy has assembled, the anti-censorship side is largely correct. The Rwandan and Ugandan studies seem to indicate that people responded mildly to some of the explicit themes and messages in fiction. The Brazilian study does show that fiction can have unintentional effects, but they tended to promote stuff that was portrayed as "normal" and not condemned or promoted, rather than negative behavior that was explicitly portrayed as villainous.