Bridging Internal Consistency and External Alignment: A Causal and Dynamic Interpretability Framework for LLM Generation

Shuyao Xiao, Shengling Wang, Ke Chao


Abstract
Large Language Models (LLMs) are widely used in high-stakes applications, making their interpretability increasingly important. Existing interpretability methods are typically categorized into internal and external perspectives, which are often studied in isolation and tend to overlook two key aspects: causality and temporal dynamics. Explanations are often limited to surface correlations or static dependencies, failing to capture how influences evolve during autoregressive generation. To address these limitations, we propose a causal and dynamic interpretability framework for LLM generation. We first characterize the backdoor-adjusted causal effects of both the generated prefix and the prompt on the current token using the Structural Causal Model. Next, we introduce two metrics to quantify contextual causal influence and question–answer causal influence. Overall, our work provides a unified causal view of internal consistency and external alignment in LLM generation dynamics.
Anthology ID:
2026.acl-long.933
Volume:
Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)
Month:
July
Year:
2026
Address:
San Diego, California, United States
Editors:
Maria Liakata, Viviane P. Moreira, Jiajun Zhang, David Jurgens
Venue:
ACL
SIG:
Publisher:
Association for Computational Linguistics
Note:
Pages:
20378–20392
Language:
URL:
https://aclanthology.org/2026.acl-long.933/
DOI:
10.18653/v1/2026.acl-long.933
Bibkey:
Cite (ACL):
Shuyao Xiao, Shengling Wang, and Ke Chao. 2026. Bridging Internal Consistency and External Alignment: A Causal and Dynamic Interpretability Framework for LLM Generation. In Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 20378–20392, San Diego, California, United States. Association for Computational Linguistics.
Cite (Informal):
Bridging Internal Consistency and External Alignment: A Causal and Dynamic Interpretability Framework for LLM Generation (Xiao et al., ACL 2026)
Copy Citation:
PDF:
https://aclanthology.org/2026.acl-long.933.pdf
Checklist:
 2026.acl-long.933.checklist.pdf