Perception verbs alone identify fiction with 90% accuracy, decision tree shows
_akpiper · x · 2026-11-20
The "Perception is all you need" edition of the Science of Storytelling newsletter shows a decision tree distinguishing fiction from nonfiction.
- Asked to find the most predictive linguistic feature, the model picked verbs of perception (seeing, hearing, touching, feeling).
- Books averaging ≥3.4 perception verbs per page are fiction 90% of the time; below the threshold, they're nonfiction 91% of the time.
- Adding two more features — action nouns and time nouns — raises accuracy to 93%.
The takeaway: fiction doesn't just convey information about the world; it invites readers to experience it through others' perceptions, with language organized around representing experience.
More from Research
- The core dilemma of agentic benchmarks: web access gives realism but kills replicability — davidmanheim · 2026-11-20
- New 79-minute review maps deep learning for gene regulation across cellular contexts, from maxATAC to Bolero — anshulkundaje · 2026-11-20
- Same harness trains optical encoders; paradigm generalizes to any controllable physical system — GordonWetzstein · 2026-11-20
- Agent infers optics geometry from 48 probes, then tunes a few parameters instead of 100k+ pixels — GordonWetzstein · 2026-11-20
- Gordon Wetzstein: coding agent in the loop boosts optical NN accuracy 35% to 96.8% — GordonWetzstein · 2026-11-20
- AI2 launches Theorizer in Asta, an AI that proposes testable scientific theories from literature — allen_ai · 2026-11-20