Experimental reasoning in large language models
- 1Duke University
Abstract
Scientific reasoning requires interaction with the physical world: designing experiments, interpreting outcomes, and deciding what experiment should come next. Here we show that general-purpose large language models (LLMs) already possess substantial capabilities for experimental reasoning. We connected LLMs to a liquid-handling robot and camera through PEARL, which provides perception, action, and persistent state. Given objectives rather than prescribed workflows, the system designed and successfully executed experiments, interpreted observations, and revised its approach as evidence accumulated. When information was insufficient, PEARL autonomously designed and performed intermediate experiments to acquire it. In quantitative enzyme characterization, it judged whether measurements were sufficient, new data were required, or conditions needed to change. Across eight scenarios, these behaviors were reliable with limited agent scaffolding and no supplied scenario-specific recovery recipes. More broadly, general-purpose LLMs may provide a reasoning layer between scientific objectives and physical experimentation, shifting laboratory automation from executing predefined procedures toward AI systems that participate directly in the experimental process.
Related
- Evidence Scaling for Zero-Shot Protein Reasoning with Large Language Models
- OmicsLM: A Multimodal Large Language Model for Multi-Sample Omics Reasoning
- The economics of accuracy for medical reasoning with large language models
- Evidence from formal logical reasoning reveals that the language of thought is not natural language
- Brain alignment of reasoning and action representations from vision-language and action models during naturalistic gameplay
BibTeX
@misc{kim2026experimental,
title = {{Experimental reasoning in large language models}},
author = {J. Kim and P. A. Romero},
howpublished = {bioRxiv},
year = {2026},
month = oct,
doi = {10.64898/2026.10.20.760214},
url = {https://www.biorxiv.org/content/10.64898/2026.10.20.760214v1},
note = {CC BY},
}
RIS
TY - UNPB
TI - Experimental reasoning in large language models
AU - J. Kim
AU - P. A. Romero
PY - 2026
DA - 2026/10/20
PB - bioRxiv
DO - 10.64898/2026.10.20.760214
UR - https://www.biorxiv.org/content/10.64898/2026.10.20.760214v1
AB - Scientific reasoning requires interaction with the physical world: designing experiments, interpreting outcomes, and deciding what experiment should come next. Here we show that general-purpose large language models (LLMs) already possess substantial capabilities for experimental reasoning. We connected LLMs to a liquid-handling robot and camera through PEARL, which provides perception, action, and persistent state. Given objectives rather than prescribed workflows, the system designed and successfully executed experiments, interpreted observations, and revised its approach as evidence accumulated. When information was insufficient, PEARL autonomously designed and performed intermediate experiments to acquire it. In quantitative enzyme characterization, it judged whether measurements were sufficient, new data were required, or conditions needed to change. Across eight scenarios, these behaviors were reliable with limited agent scaffolding and no supplied scenario-specific recovery recipes. More broadly, general-purpose LLMs may provide a reasoning layer between scientific objectives and physical experimentation, shifting laboratory automation from executing predefined procedures toward AI systems that participate directly in the experimental process.
ER -