Template-Type: ReDIF-Paper 1.0
Author-Name: Robin Musolff
Author-Name-First: Robin
Author-Name-Last: Musolff
Author-Email: robin.musolff@gmail.com
Author-Workplace-Name: University of Cologne
Author-Name: Christopher Roth
Author-Name-First: Christopher
Author-Name-Last: Roth
Author-Email: roth@wiso.uni-koeln.de
Author-Workplace-Name: University of Cologne, CESifo, NHH Norwegian School of Economics, CEPR & MPI for Behavioral Economics
Author-Name: Florian Zimmermann
Author-Name-First: Florian
Author-Name-Last: Zimmermann
Author-Email: florian.zimmermann@uni-bonn.de
Author-Workplace-Name: University of Bonn, CESifo & MPI for Behavioral Economics

Title: Learning From False Stories
Abstract: False information pervades societies across the globe. Using incentivized online experiments, we document a qualitative residue in learning from false information: transparently debunked quantitative signals barely move beliefs, whereas debunked stories leave a substantial residual imprint, robustly across several experiments. Making the prior salient mitigates the effect, and neutral rather than evaluative stories leave no detectable residue. We show directly that false stories heighten mental simulation, including via speech-extracted valence. The residue also appears among respondents who report being uninfluenced, consistent with stories shaping beliefs below conscious awareness. Finally, a transparently disclosed deepfake significantly reduces incentivized investment.
Length: 73 pages
Classification-JEL: D83, D91, C90
Keywords: Stories, Learning, Mental Simulation, Fake News

Creation-Date:  2026-05
File-URL: https://www.econtribute.de/RePEc/ajk/ajkdps/ECONtribute_412_2026.pdf
File-Format: Application/pdf
File-Function: Second version, 2026
Number: 412
Handle: RePEc:ajk:ajkdps:412