MoRI: Learning Motivation-Grounded Reasoning for Scientific Ideation in Large Language Models

TL;DR

MoRI is a novel framework that enhances scientific reasoning in large language models by explicitly grounding motivation and methodology, leading to more rigorous and innovative research proposals.

Contribution

MoRI introduces a motivation-grounded reasoning framework with reinforcement learning to improve scientific ideation in large language models.

Findings

MoRI outperforms existing models in novelty, rigor, and feasibility.

The framework effectively grounds reasoning in scientific motivations and methodologies.

Code is publicly available at the provided GitHub link.

Abstract

Scientific ideation aims to propose novel solutions within a given scientific context. Existing LLM-based agentic approaches emulate human research workflows, yet inadequately model scientific reasoning, resulting in surface-level conceptual recombinations that lack technical depth and scientific grounding. To address this issue, we propose \textbf{MoRI} (\textbf{Mo}tivation-grounded \textbf{R}easoning for Scientific \textbf{I}deation), a framework that enables LLMs to explicitly learn the reasoning process from research motivations to methodologies. The base LLM is initialized via supervised fine-tuning to generate a research motivation from a given context, and is subsequently trained under a composite reinforcement learning reward that approximates scientific rigor: (1) entropy-aware information gain encourages the model to uncover and elaborate high-complexity technical details grounded in ground-truth methodologies, and (2) contrastive semantic gain constrains the reasoning trajectory to remain conceptually aligned with scientifically valid solutions. Empirical results show that MoRI consistently outperforms strong commercial LLMs and complex agentic baselines across multiple dimensions, including novelty, technical rigor, and feasibility. The code is available on \href{https://github.com/ECNU-Text-Computing/IdeaGeneration}{GitHub}.