From Predictive Algorithms to the Automatic Generation of Anomalies
Econometrica · 2026 · 94 (6) · 2025-2047
DOI: 10.3982/ecta22958 ↗At a glance
Journal article in Econometrica, 2026. Written by Sendhil Mullainathan and 1 co-authors. Cited 1 times so far (Crossref); 43 references. The full text is freely available (open access, CC BY 4.0).
Abstract (original language)
How can we extract theoretical insights from machine learning algorithms? We take a familiar lesson: researchers often turn intuition into theoretical insight by constructing “anomalies”—specific examples highlighting hypothesized flaws in a theory, such as the Allais paradox. We develop procedures that replace researchers' intuitions with predictive algorithms: given a predictive algorithm and a theory, our procedures automatically generate anomalies for that theory. We illustrate our procedures by generating anomalies for expected utility theory. Based on a neural network that accurately predicts lottery choices, we recover known anomalies for expected utility theory and discover novel ones. In incentivized experiments, subjects violate expected utility theory on these algorithmically generated anomalies at rates similar to the Allais paradox and common ratio effect.
Cite
APA
Mullainathan, S., & Rambachan, A. (2026). From Predictive Algorithms to the Automatic Generation of Anomalies. Econometrica, 94(6), 2025-2047. https://doi.org/10.3982/ecta22958
Vancouver
Mullainathan S, Rambachan A. From Predictive Algorithms to the Automatic Generation of Anomalies. Econometrica. 2026;94(6):2025-2047. doi:10.3982/ecta22958
BibTeX
@article{Mullainathan2026,
author = {Mullainathan, Sendhil and Rambachan, Ashesh},
title = {From Predictive Algorithms to the Automatic Generation of Anomalies},
journal = {Econometrica},
year = {2026},
volume = {94},
number = {6},
pages = {2025--2047},
doi = {10.3982/ecta22958}
}
EUVIMED