Reproduce “Flexible inference for animal learning rules using neural networks” 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
Yuhan Helena Liu, Victor Geadah, Jonathan Pillow
The authors infer mouse trial-by-trial updates to an interpretable choice policy, using a feedforward network for current-trial information and a recurrent network for trial history. In a 12-mouse IBL task, the recurrent model achieved higher held-out log-likelihood than a feedforward model given the previous trial (−59,892 versus −60,168; p = 3.84 × 10⁻⁶). This supports prediction, not a unique mechanism.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.