ARIA: A Causal-Aware Framework for Rescuing LLM Reasoning in Trustworthy Materials Discovery

2026-06-21

AI Summary

This work addresses the susceptibility of large language models (LLMs) to “contextual tunneling” in materials discovery, wherein overreliance on locally retrieved information undermines global physical causal reasoning. To overcome this limitation, the authors propose ARIA, a framework that implements causally aware reasoning through a three-stage cascaded mechanism: direct causal inference when complete property–structure–performance (PSP) chains are available, physics-guided analogical transfer for sparse or novel materials, and explicit parameter fallback when evidence is insufficient. ARIA uniquely conditions knowledge invocation on mechanistic completeness, thereby constructing auditable causal reasoning pathways that integrate LLMs, domain-specific knowledge graphs, and real-time literature retrieval. Experiments demonstrate that ARIA significantly outperforms baseline methods in both forward prediction and inverse design tasks for two-dimensional materials, while producing interpretable causal trajectories.

Abstract

Generative models have revolutionized the process of materials discovery, yet they often fail to satisfy underlying physical causality. Through an analysis of Large Language Models (LLMs) augmented with knowledge graphs derived from current literature, we uncover a phenomenon termed contextual tunneling, where models "over-anchor" on narrow, retrieved evidence while suppressing global physical reasoning. To address this problem, we introduce ARIA, a causal-aware framework that conditions knowledge use on mechanistic completeness. ARIA routes each query through a three-tier cascade: (i) direct causal reasoning when complete evidence chains of Process-Structure-Property (PSP) are available, (ii) physics-informed analogical transfer for sparse or novel material systems, and (iii) explicit parametric fallback when external evidence is incomplete. As a proof of concept, we construct a Knowledge Graph (KG) containing 2,839 extracted PSP relations from peer-reviewed articles in the materials literature and evaluate ARIA on forward prediction and inverse design tasks for two-dimensional (2D) materials. ARIA mitigates contextual tunneling, improves over unaugmented and naive KG-augmented baselines, and provides further gains when an online literature search is used for evidence enrichment. Crucially, ARIA produces auditable causal traces, enabling physically grounded and trustworthy AI-assisted materials discovery.

Problem

• contextual tunneling
• causal reasoning
• materials discovery
• Large Language Models
• physical causality

Innovation

• causal reasoning
• contextual tunneling
• knowledge graph
• materials discovery
• physics-informed AI