Saved in:
Bibliographic Details
Main Authors: Mencattini, Tommaso, Cadei, Riccardo, Locatello, Francesco
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2510.14073
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866914234712981504
author Mencattini, Tommaso
Cadei, Riccardo
Locatello, Francesco
author_facet Mencattini, Tommaso
Cadei, Riccardo
Locatello, Francesco
contents Randomized Controlled Trials are one of the pillars of science; nevertheless, they rely on hand-crafted hypotheses and expensive analysis. Such constraints prevent causal effect estimation at scale, potentially anchoring on popular yet incomplete hypotheses. We propose to discover the unknown effects of a treatment directly from data. For this, we turn unstructured data from a trial into meaningful representations via pretrained foundation models and interpret them via a sparse autoencoder. However, discovering significant causal effects at the neural level is not trivial due to multiple-testing issues and effects entanglement. To address these challenges, we introduce Neural Effect Search, a novel recursive procedure solving both issues by progressive stratification. After assessing the robustness of our algorithm on semi-synthetic experiments, we showcase, in the context of experimental ecology, the first successful unsupervised causal effect identification on a real-world scientific trial.
format Preprint
id arxiv_https___arxiv_org_abs_2510_14073
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Exploratory Causal Inference in SAEnce
Mencattini, Tommaso
Cadei, Riccardo
Locatello, Francesco
Machine Learning
Artificial Intelligence
Randomized Controlled Trials are one of the pillars of science; nevertheless, they rely on hand-crafted hypotheses and expensive analysis. Such constraints prevent causal effect estimation at scale, potentially anchoring on popular yet incomplete hypotheses. We propose to discover the unknown effects of a treatment directly from data. For this, we turn unstructured data from a trial into meaningful representations via pretrained foundation models and interpret them via a sparse autoencoder. However, discovering significant causal effects at the neural level is not trivial due to multiple-testing issues and effects entanglement. To address these challenges, we introduce Neural Effect Search, a novel recursive procedure solving both issues by progressive stratification. After assessing the robustness of our algorithm on semi-synthetic experiments, we showcase, in the context of experimental ecology, the first successful unsupervised causal effect identification on a real-world scientific trial.
title Exploratory Causal Inference in SAEnce
topic Machine Learning
Artificial Intelligence
url https://arxiv.org/abs/2510.14073