Open-ended Scientific Discovery with Possibilistic Reasoning
Abstract
Autonomous scientific discovery with LLMs requires generating and testing hypotheses adaptively as evidence accumulates while maintaining statistical validity. Existing anytime-valid methods can handle data-dependent hypotheses, but open-ended discovery poses a deeper challenge: the best discovered hypothesis may still be the best of a bad lot, with better explanations yet undiscovered, while even background knowledge such as physical laws may require revision in light of new findings. In response, we formalize the problem as Abductive Autonomous Scientific Discovery (AASD) using possibility theory. We introduce abductive utility, a computable measure of discovery progress, and possibility frontier search, the first algorithm for AASD, which maintains anytime validity and achieves ε-optimal abductive utility asymptotically under suitable conditions. Experiments on synthetic and real-world scientific-discovery tasks show strong performance.
Related
- ExpEvolve: Structuring and Reusing Experience for Open-Ended Discovery
- Optimizing Open-Ended Interview Assessment Using Response-Based Semantic Indices
- MatrixReward: Reward from Rubric Matrix for Open-Ended Generation
- APOD: reasoning-guided agentic population ordinary differential equation discovery for pharmacological digital twins
- Beyond χ2: Ensemble null tests for reliable scientific discovery
BibTeX
@misc{yang2026openended,
title = {{Open-ended Scientific Discovery with Possibilistic Reasoning}},
author = {Anita Yang and Siu Lun Chau and Tomoya Wakayama and Krikamol Muandet and Masaki Adachi},
howpublished = {arXiv},
year = {2026},
month = oct,
doi = {10.48550/arxiv.2610.11289},
eprint = {2610.11289},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2610.11289},
note = {CC BY-NC-SA 4.0},
}
RIS
TY - UNPB
TI - Open-ended Scientific Discovery with Possibilistic Reasoning
AU - Anita Yang
AU - Siu Lun Chau
AU - Tomoya Wakayama
AU - Krikamol Muandet
AU - Masaki Adachi
PY - 2026
DA - 2026/10/08
PB - arXiv
DO - 10.48550/arxiv.2610.11289
UR - https://arxiv.org/abs/2610.11289
AB - Autonomous scientific discovery with LLMs requires generating and testing hypotheses adaptively as evidence accumulates while maintaining statistical validity. Existing anytime-valid methods can handle data-dependent hypotheses, but open-ended discovery poses a deeper challenge: the best discovered hypothesis may still be the best of a bad lot, with better explanations yet undiscovered, while even background knowledge such as physical laws may require revision in light of new findings. In response, we formalize the problem as Abductive Autonomous Scientific Discovery (AASD) using possibility theory. We introduce abductive utility, a computable measure of discovery progress, and possibility frontier search, the first algorithm for AASD, which maintains anytime validity and achieves ε-optimal abductive utility asymptotically under suitable conditions. Experiments on synthetic and real-world scientific-discovery tasks show strong performance.
ER -