Reproduce “Stand-In Effects and Pooling Amplification: Design-Based Tests of Whether Biological World Models Use Actions” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Bowen Xu, Boyu Chen
The authors develop a randomization-based test of whether world models capture causal action effects. The test compares a fixed model score for observed action blocks with scores under blocks resampled from the experiment’s design law. Pooling site scores can suppress evidence; registered tempering enabled rejection on one ERK field at ten-minute horizon, conditional on the reconstructed design law.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.