Skip to main content
Cornell University
We gratefully acknowledge support from the Simons Foundation, member institutions, and all contributors. Donate
arxiv logo > cs > arXiv:2410.23884

Help | Advanced Search

arXiv logo
Cornell University Logo

quick links

  • Login
  • Help Pages
  • About

Computer Science > Machine Learning

arXiv:2410.23884 (cs)
[Submitted on 31 Oct 2024 (v1), last revised 15 Jun 2025 (this version, v5)]

Title:Failure Modes of LLMs for Causal Reasoning on Narratives

Authors:Khurram Yamin, Shantanu Gupta, Gaurav R. Ghosal, Zachary C. Lipton, Bryan Wilder
View a PDF of the paper titled Failure Modes of LLMs for Causal Reasoning on Narratives, by Khurram Yamin and 4 other authors
View PDF HTML (experimental)
Abstract:The ability to robustly identify causal relationships is essential for autonomous decision-making and adaptation to novel scenarios. However, accurately inferring causal structure requires integrating both world knowledge and abstract logical reasoning. In this work, we investigate the interaction between these two capabilities through the representative task of causal reasoning over narratives. Through controlled synthetic, semi-synthetic, and real-world experiments, we find that state-of-the-art large language models (LLMs) often rely on superficial heuristics -- for example, inferring causality from event order or recalling memorized world knowledge without attending to context. Furthermore, we show that simple reformulations of the task can elicit more robust reasoning behavior. Our evaluation spans a range of causal structures, from linear chains to complex graphs involving colliders and forks. These findings uncover systematic patterns in how LLMs perform causal reasoning and lay the groundwork for developing methods that better align LLM behavior with principled causal inference.
Comments: ICML 2025 Workshop on Scaling up Intervention Models
Subjects: Machine Learning (cs.LG); Computation and Language (cs.CL)
Cite as: arXiv:2410.23884 [cs.LG]
  (or arXiv:2410.23884v5 [cs.LG] for this version)
  https://doi.org/10.48550/arXiv.2410.23884
arXiv-issued DOI via DataCite

Submission history

From: Khurram Yamin [view email]
[v1] Thu, 31 Oct 2024 12:48:58 UTC (3,343 KB)
[v2] Tue, 24 Dec 2024 23:07:13 UTC (3,448 KB)
[v3] Wed, 4 Jun 2025 18:26:30 UTC (253 KB)
[v4] Wed, 11 Jun 2025 22:07:36 UTC (247 KB)
[v5] Sun, 15 Jun 2025 00:36:18 UTC (247 KB)
Full-text links:

Access Paper:

    View a PDF of the paper titled Failure Modes of LLMs for Causal Reasoning on Narratives, by Khurram Yamin and 4 other authors
  • View PDF
  • HTML (experimental)
  • TeX Source
  • Other Formats
license icon view license
Current browse context:
cs.LG
< prev   |   next >
new | recent | 2024-10
Change to browse by:
cs
cs.CL

References & Citations

  • NASA ADS
  • Google Scholar
  • Semantic Scholar
a export BibTeX citation Loading...

BibTeX formatted citation

×
Data provided by:

Bookmark

BibSonomy logo Reddit logo

Bibliographic and Citation Tools

Bibliographic Explorer (What is the Explorer?)
Connected Papers (What is Connected Papers?)
Litmaps (What is Litmaps?)
scite Smart Citations (What are Smart Citations?)

Code, Data and Media Associated with this Article

alphaXiv (What is alphaXiv?)
CatalyzeX Code Finder for Papers (What is CatalyzeX?)
DagsHub (What is DagsHub?)
Gotit.pub (What is GotitPub?)
Hugging Face (What is Huggingface?)
Papers with Code (What is Papers with Code?)
ScienceCast (What is ScienceCast?)

Demos

Replicate (What is Replicate?)
Hugging Face Spaces (What is Spaces?)
TXYZ.AI (What is TXYZ.AI?)

Recommenders and Search Tools

Influence Flower (What are Influence Flowers?)
CORE Recommender (What is CORE?)
IArxiv Recommender (What is IArxiv?)
  • Author
  • Venue
  • Institution
  • Topic

arXivLabs: experimental projects with community collaborators

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.

Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.

Which authors of this paper are endorsers? | Disable MathJax (What is MathJax?)
  • About
  • Help
  • contact arXivClick here to contact arXiv Contact
  • subscribe to arXiv mailingsClick here to subscribe Subscribe
  • Copyright
  • Privacy Policy
  • Web Accessibility Assistance
  • arXiv Operational Status
    Get status notifications via email or slack