The best-known argument for reading fiction rests on a study that later failed to replicate. That does not make the argument wrong. It makes the popular version of it wrong, and the corrected version is more useful to anyone whose working life runs on incomplete information.
Harvard Business Review published the case in March 2020. The author was Christine Seifert, a professor of communication at Westminster College in Salt Lake City. Her claim was that fiction builds what employers say they cannot train: judgment, flexibility, reading a room. She linked three studies and named none of the researchers, and the story gets better than her article once you follow those links.
The study everyone quotes stopped working when other labs ran it
The first link goes to Science, October 2013. David Comer Kidd and Emanuele Castano ran five experiments using the Reading the Mind in the Eyes Test. It shows you a cropped photograph of someone's eyes and asks you to name the emotion. People who had just read a few pages of literary fiction scored better on it. Popular fiction did not produce the effect. Non-fiction did not. Reading nothing did not.
It travelled fast, because it offered something rare. A short, cheap, pleasant intervention with a measurable effect on a social skill.
In 2016, seven researchers working in three independent groups tried to reproduce it with 792 participants. Maria Eugenia Panero and her co-authors published the result in the Journal of Personality and Social Psychology. Literary fiction showed no significant advantage over popular fiction, over non-fiction, or over reading nothing at all.
That is not a scandal. It is the process working as designed. A single striking result is a hypothesis, and this hypothesis did not survive a larger sample.
Almost everyone who reported the failure stopped there. The interesting part is what the same paper found while it was failing.
The sentence nobody quoted is the one worth keeping
Panero's team measured one more thing, as a control rather than as the point. They asked participants how much fiction they had read across their lives.
That number predicted the eye test score. It predicted it consistently, across every condition, including among people who had been handed nothing to read, so the effect was real and simply had nothing to do with the passage they were handed.
This rearranges the question rather than settling it. A study that finds nothing after ten minutes of a short story has not shown that short stories do nothing. It has shown that ten minutes is the wrong unit, in the way that one gym session is the wrong unit for fitness. Nobody would design that experiment about exercise and then publish the null result as news.
The writing used in those literary conditions is built a particular way, and the structure is the point. Chekhov ends before the verdict. A man decides something, a marriage bends around it, and the story stops while you are still waiting to be told what it meant. Read three in a row from The Greatest Short Stories of Anton Chekhov and the discomfort is unmistakable. You keep turning the page expecting the paragraph that tells you who was right. It never arrives. Our Short Stories shelf is full of writers doing that on purpose, and ten minutes of it changes nothing, which is rather the point: what accumulates over years is the willingness to sit with an answer that has not arrived.
A habit only counts when the book actually takes you in
If the unit is a habit rather than an evening, the next question is what makes an evening count towards it. The second study Harvard Business Review links answers that, and the answer has a sting in it.
P. Matthijs Bal and Martijn Veltkamp published in PLOS ONE in 2013. Two experiments, with empathy measured three times: before reading, immediately after, and again a week later.
Immediately after, nothing moved. A week later, empathy had risen, but only among readers who reported being emotionally transported into the story. Transportation is the researchers' word for the state where you stop noticing the sentences and forget the room you are sitting in.
Now the part that should change how you pick a book. In both studies, the fiction readers who were not transported came out lower than they started.
A novel you grind through without ever falling into is not neutral. On this evidence it costs you something. Worth knowing, if you keep a shelf of books you feel you ought to finish. It argues for the move most reading advice forbids: abandon it at page forty and start something else.
Which means the right novel is the one that grips you
The condition is transportation, and transportation is not the same as prestige. Those two come apart constantly. Separate them and most reading-list advice starts to look backwards.
Take the honest case for the books that get patronised. Angels & Demons is engineered to close every question it opens. By the last page nothing is left unresolved, which is precisely the openness the literary category was supposed to supply. On the theory as originally stated, it should do nothing at all.
On the transportation evidence, for a reader it genuinely absorbs, it plausibly beats the prize-winning novel that same reader abandons in week two. A book that takes you in is doing the thing. A book that impresses your shelf is not doing anything.
This is not an argument that all books are equal. It puts grip first and ambition second, rather than the other way round. Our Literary Fiction shelf and the thriller shelf beside it are not two quality tiers, whatever anyone tells you at a party. They are two different bets about how a book holds a reader, and the one that holds you is the one that counts.
The strongest objection, and the one study that survives it
The finding that survived replication, lifetime fiction exposure, is self-reported and correlational. Nobody randomly assigned anyone a childhood of reading. The obvious alternative is selection: people who are naturally good at reading others find novels more rewarding, so they read more of them. On that account the novels are a symptom of the skill, not a cause. The whole argument becomes flattery for people who already liked books.
The objection is strong and it is not fully answerable. Anyone who tells you otherwise is selling something, which is the trap in reading a fourth self-help book instead of changing one thing.
One study does escape it, and it is not one of the famous ones. In 2013, Maja Djikic, Keith Oatley and Mihnea Moldoveanu published in the Creativity Research Journal. They took 100 people and randomly assigned them to read either short stories or non-fiction essays. Then they measured need for cognitive closure, the psychological term for how uncomfortable you are leaving a question open. The fiction readers came out significantly lower.
Random assignment, so the direction of causation is not in doubt. And notice what it measures. Not empathy. Not mind reading. Tolerance for not yet knowing.
Tolerance is a far smaller claim than the one on the poster, and it is the claim that pays rent in a non-fiction life. Most hard moments at work are not failures of empathy. They are moments when the information is incomplete and the answer will not arrive today. The instinct is to force a conclusion so that the discomfort stops.
Fiction rehearses exactly that discomfort. Never Let Me Go is the clearest case in modern fiction. Kathy narrates something monstrous in a calm, ordinary voice. Nobody escapes, and Ishiguro never once explains why nobody runs. Readers who need the mechanism explained find the book intolerable, and that reaction is the measurement.
Alice Munro does it in twenty pages instead of three hundred. In Runaway, a young woman leaves her husband and gets most of the way to a new life. Then she comes back. Munro plainly knows whether that was cowardice or clear sight, and she will not tell you, because the story is about the coming back rather than about how to score it.
Read for the thing you cannot summarise
Pick one novel this month by grip alone. Read the first ten pages standing up, in a shop or on your phone. If you are not somewhere else by the end of them, put it down without guilt and try the next one. That is the only selection rule this evidence actually supports.
Then, when you finish, try to say what it meant in one sentence. If you can, it was a good evening and not much else. If you cannot, and the ending is still sitting there refusing to resolve, that is the book doing its work. You have just spent several hours holding a question open without forcing it shut. That is the muscle you will want on Thursday, when the shipment is late and nobody can tell you why.
The Lost Daughter is a fair place to feel it quickly. Ferrante gives you a woman who does something small and inexplicable on holiday, then spends the book refusing to make it make sense. You will finish it in two evenings and argue with yourself for a week.
0 comments