What the Study Found
- Readers who saw inside an at-fault repairman’s remorse rated forgiving him higher: 7.33 versus 6.28 out of 10 in the first experiment.
- When the repairman had an excuse, seeing his thoughts made no difference, so the later experiments used only the at-fault story.
- More forgiveness of the character statistically tracked with more forgiveness of a real past wrong, though participants self-reported both.
- Gaps were small, came from one online worker pool, and no one tested wellbeing or whether any shift lasted beyond the session.
MARY is 85, and she has finally risked another puppy, decades after watching her childhood dog get hit by a car. She has told the repairman, carefully, to keep the door shut. In three online experiments, readers who got inside a careless repairman’s head said they were a little more willing to forgive him than readers who only heard him apologize aloud, and only when he was plainly to blame. The experiments, run at Ohio State University and published in Frontiers in Psychology, also asked something harder: whether forgiveness borrowed from a fictional man could leak into real grudges.
People have credited literature with teaching empathy for more than 2,000 years. In 2013, a study reported that reading literary fiction sharpened readers’ grasp of what other people believe and want, but efforts to replicate it, by other labs and by the original researchers, haven’t panned out, and empathy has proved slippery to pin down elsewhere too: ScholarPeer has covered a brain-imaging study in which the response to lives at risk generally fell as the numbers grew.
So Angus Fletcher, a professor of English at Ohio State’s Project Narrative, and Mark Matthews, a recent Ohio State PhD graduate in social psychology, stopped asking whether literature works and isolated one technique instead, then tested it on its own. It is called internal focalization: narration that hands the reader a character’s innermost thoughts. They built a four-part story about Mary and her puppy, randomly assigned readers to versions, and changed only the final part, where the repairman apologizes. One version read, “The repairman thought to himself, I wish I didn’t leave the door open.” The other: “The repairman said to Mary, ‘I wish I didn’t leave the door open.’”
The words of regret are identical; what changes is whether they reach the reader as a private thought or as something said out loud to a woman who has just collapsed in horror. Readers then rated two kinds of forgiveness, the decision to forgive (scored out of 10) and the feeling that goes with it, a drop in resentment (scored out of 7).
Give the Repairman an Excuse and the Effect Fades
In the first experiment, 371 people from Amazon Mechanical Turk (an online crowd-work marketplace) also got one of two accounts of the repairman’s blame: either he had disregarded Mary’s warning, or he was distracted by trouble at home. Where he was at fault, readers who saw his thoughts scored 7.33 out of 10 on the decision to forgive and 3.49 out of 7 on feeling forgiving, against 6.28 and 3.08 for readers who only heard him speak. Where he had an excuse, seeing his thoughts changed nothing.
Blame has to be on the table first, Matthews says. “Once we know the transgression is someone’s fault, what we discover is that if we can create a condition where you can see the internal remorse, the internal world of the person, people can feel the sincerity of it.” Why would an excuse matter so much? Perhaps because it does the forgiving for you: readers averaged 8.23 out of 10 on the decision to forgive when the repairman was distracted, against 6.81 when he wasn’t, which may be why the inner voice had little left to add.
From a Fictional Grudge to Real Ones
The next two experiments dropped the excuse version and kept only the at-fault story, with 449 people in the second and 743 in the third, and randomly assigned readers to see the repairman’s thoughts or hear his spoken words. The gap held, if anything narrower: 6.97 versus 6.09 on the decision to forgive in the second experiment and 6.92 versus 6.30 in the third, with feelings at 3.29 versus 3.00 and 3.15 versus 2.91.
Then came the twist. After the story, readers described a time someone close had badly wronged them and rated how far they had forgiven that person. The more a reader had forgiven the repairman, the more they reported forgiving the real offender, and the inner-voice version raised forgiveness of the repairman, so a statistical pathway ran from the technique through the character to the real grudge (a mediation analysis, in the jargon, found in both the second and third experiments). A control group in the third experiment, 220 people who read a dull technical report about a fictional company instead, turned up no sign that reading any story at all would do the job.
The shifts are small, though. In the second and third experiments the gap on the ten-point scale was under a point, emotional forgiveness never topped 3.5 out of 7 even among readers who saw inside the repairman’s head, and the first experiment’s emotional gap only just cleared the usual bar for statistical significance. In the third experiment, the check on whether readers felt they had access to the repairman’s mind was merely marginal.
The real-life measure was thin, too: one question asking whether a reader had fully forgiven the person, plus the emotion scale, answered within the same online session as the story. In the second experiment the technique had no detectable direct effect on real-life forgiveness at all; it showed up only along the path through the repairman.
The university’s news release goes further, saying the work suggests stories may be a prescription for better emotional health (a pitch ScholarPeer has weighed before with salsa classes for low mood). The experiments measured forgiveness and nothing else, with no wellbeing outcome, no follow-up and no sign of how long any shift lasts, and the order of events the release describes rests on statistical modeling of a measured middle step rather than on manipulating it. “What’s really great about literature is you don’t have to be at a crisis point to use it,” Fletcher says, and that hope is lighter than the claim. It is also, so far, untested.
The authors’ own caution is plainer. Forgiving indiscriminately can trap people in cycles of harm, they note (a questionnaire study of women in domestic violence shelters found that forgiveness predicted their intention to return to a partner), and their samples came from online worker pools that skew Western, with no demographics collected, so cultures that treat forgiveness differently may respond differently. As for the other techniques in literature’s toolkit, they concede that not all of them will do the same thing, or anything at all.
Reference
Fletcher, A., & Matthews, M. (2026). Internal focalization influences real-life empathy: a literary contribution to wellbeing. Frontiers in Psychology, 17. https://doi.org/10.3389/fpsyg.2026.1968457
- Study type: Three preregistered, randomized online experiments with mediation analysis; peer-reviewed journal article, Frontiers in Psychology (Health Psychology section)
- Sample size: 1,563 participants after exclusions (371, 449 and 743 across the three experiments), all recruited through Amazon Mechanical Turk
- Intervention: A four-part story in which a repairman’s apology was written as the character’s private thought (internal focalization)
- Comparator: The same apology spoken aloud to Mary; the first experiment also varied blame (at fault versus distracted), and the third added 220 participants who read a technical report instead of a story
- Duration: One online session per participant; no follow-up
- Funding / conflicts of interest: Funded by a grant from the Ohio State Department of English, covering research salaries and participant recruitment; authors declared no conflicts of interest
- Data availability: Raw data to be made available by the authors on request
- Preregistration: All three experiments preregistered on the Open Science Framework before data collection
- Main limitation: Effects were small and self-reported, measured once in a single session, in a Western-leaning online sample with no demographics collected; wellbeing was not measured and the third experiment’s manipulation check was only marginal
FAQ
Does reading fiction make people more forgiving in real life?
Reading fiction did not, by itself, make people more forgiving in real life in these experiments. Readers who simply read a story showed no more forgiveness of a real grudge than readers of a technical report, and only the version that revealed an at-fault repairman’s private thoughts shifted forgiveness, through how much readers forgave him. Everyone was tested in a single online session.
Why did the repairman’s inner voice only work when he was at fault?
The repairman’s inner voice worked only when he was at fault because an excuse already made readers far more forgiving, leaving little for his thoughts to add. Mark Matthews, one of the Ohio State researchers, says that once someone is known to be at fault, seeing their remorse from the inside lets people feel its sincerity. The first experiment, with 371 people, found the gap only in the at-fault version.
Is forgiving always a healthy goal?
Forgiving is not always a healthy goal, and the researchers say so themselves. Fletcher and Matthews note that people who forgive indiscriminately can be harmed again and can become trapped in cycles of harm. The experiments measured forgiveness only, so they cannot say whether the technique helps anyone forgive safely.
Could stories help with anger, depression or anxiety?
These experiments cannot say whether stories help with anger, depression or anxiety, because they measured only how forgiving participants said they were. Angus Fletcher, a professor of English at Ohio State, hopes literature can help people without waiting for a crisis point, but that is his view rather than a tested result. There was no wellbeing outcome and no follow-up.
Would the results be the same in other cultures?
The results might differ in other cultures, the authors say. All three samples came from online worker pools that skew Western, and no demographic details were collected. Cultures that treat forgiveness differently may respond differently to the same story.
Cite This Page