A theory of narrative empathy

What the evidence shows

This essay lays out Suzanne Keen’s foundational theory of “narrative empathy” — the vicarious sharing of feeling that fiction can evoke in readers. Keen surveys how psychologists and neuroscientists define and measure empathy (including then-emerging fMRI and mirror-neuron research), distinguishes empathy from related concepts like sympathy, personal distress, and emotional contagion, and asks a central question: does reading fiction that generates empathy actually lead to real-world prosocial or altruistic behavior (the “empathy-altruism hypothesis”)? Her central, somewhat provocative claim is that this link has never actually been proven, despite widespread assumptions among educators, publishers, and humanities advocates that it exists.

The essay then reviews specific narrative techniques often assumed to generate empathy — first-person narration, free indirect discourse/“narrated monologue,” character identification, “showing vs. telling” — and finds that the experimental evidence for most of these techniques is thin, inconsistent, or untested. Keen introduces her own conceptual framework of three “strategic empathy” types (bounded, ambassadorial, and broadcast), describing different ways authors direct empathetic address toward in-groups, outsiders, or universal audiences. She closes by cataloguing open research questions and cautioning against uncritical claims that any single literary technique reliably produces empathy or ethical/behavioral change in readers.

This is a literary-theory and psychology synthesis piece, not a climate- or advocacy-specific study — but it is directly relevant to Understories because it interrogates the core assumption underlying much “storytelling for social change” work: that evoking empathy through narrative reliably produces real-world prosocial action.

Key Findings

• The link between novel-reading, narrative empathy, and altruistic/prosocial behavior has not been empirically verified, despite being widely assumed by educators, philosophers, and literary advocates.

• In a study of 50 fiction writers, 92% of the authors reported some experience of the illusion of independent agency (IIA), and the more successful fiction writers who had published had more frequent and more intense experiences of it. Fiction writers as a group scored higher than general population norms on empathy scales, particularly on fantasy and perspective-taking subscales.

• Experimental research on first-person narration’s effect on reader empathy is weak: in studies of Dutch teenagers testing whether first-person narration created a greater sense of closeness and fusion with characters, researchers concluded it remained unclear why point of view had no more powerful an effect on readers, despite the effort authors put into using such devices — and found that any enhancement of sympathy through positive internal focalization actually weakened as the teenage subjects matured.

• fMRI research (Singer et al., Science, 2004) found that watching a loved one experience pain activated the affective (but not sensory) pain-processing regions of the brain, and that this activation could be triggered even by an abstract cue with no visible emotional expression — suggesting empathy operates through a distinct affective pathway rather than literal shared sensation.

• Most narrative techniques nominated by literary theorists as “empathy-inducing” (character naming, showing vs. telling, narrative pace, “roundness” of characters) have not actually been tested in controlled experiments, and where testing has occurred, results often fail to confirm the theoretical assumptions.

Stand out data points

• Qualitative: first-person narration’s assumed empathetic advantage over third-person narration was not supported in controlled reader studies — the effect was weak, inconsistent, and diminished with reader age.

Examples of how Understories uses this research

• Actionability Central relevance — this is the paper’s core caution: empathetic engagement with a story does not reliably translate into real-world action. Messages that rely purely on “if they just feel with the character/victim, they’ll act” should be scored down unless paired with a concrete action pathway.

• Emotional resonance Provides the theoretical vocabulary (empathy vs. sympathy vs. personal distress vs. emotional contagion) for evaluating whether a message’s emotional appeal is likely to land as intended.

• Personal relevance Keen’s concept of “situational empathy” (empathy triggered by matching a reader’s own prior experience) supports scoring messages higher when they connect to an audience’s lived context, and flags that “categorical empathy” (identifying with someone simply because they’re in one’s in-group) is weaker and more bias-prone.

• Beneficence This research is a useful check on overconfidence — it directly undermines the common assumption that a moving, empathy-triggering story is sufficient to drive real advocacy outcomes. Understories can use this to caution against overselling “emotionally powerful” messaging as inherently effective.

• Non-maleficence Keen’s concept of “empathic inaccuracy” (where a reader’s empathetic response diverges from or contradicts what the story or its subjects would actually want) is a useful flag: advocacy messaging that projects assumed feelings onto affected communities risks this kind of misrepresentation. Her discussion of critiques that empathy can become “a pornographic indulgence of sensation acquired at the expense of suffering others” is relevant for flagging exploitative or voyeuristic uses of others’ suffering in messaging.

Key quote

“No specific set of narrative techniques has yet been verified to over-ride the resistance to empathizing often displayed by members of an in-group regarding the emotional states of others marked out as different.”

© 2006 The Ohio State University — All rights reserved.

Library entry updated: 30 September 2026.