Position Paper: Logical Soundness Is Not a Reliable Criterion for Neurosymbolic Fact-Checking with LLMs

Forum topic · 小凯 · 2026-04-07

Summary

A position paper by Jason Chan, Robert Gaizauskas, and Zhixue Zhao argues that formal logic is not a reliable criterion for neurosymbolic fact-checking with large language models (LLMs). Many fact-checking pipelines translate natural language into logical formulae via LLMs and verify claims by checking whether they are logically sound—validly derivable from verified premises. The authors argue these approaches structurally fail to detect misleading claims because logically sound conclusions can systematically diverge from the inferences humans typically make and accept. Drawing on cognitive science and pragmatics, the paper presents a typology of cases where logically sound conclusions elicit human inferences unsupported by the underlying premises. The authors advocate a complementary approach: treating LLMs' human-like reasoning tendencies as a feature rather than a bug, and using these models to validate the outputs of formal components in neurosymbolic systems against potentially misleading conclusions.

Position: Logical Soundness is not a Reliable Criterion for Neurosymbolic Fact-Checking with LLMs

Field: NLP Authors: Jason Chan, Robert Gaizauskas, Zhixue Zhao

Overview

As large language models (LLMs) are increasingly integrated into fact-checking pipelines, formal logic is often proposed as a rigorous means to mitigate bias, errors, and hallucinations in these models' outputs. For example, some neurosymbolic systems verify claims by using LLMs to translate natural language into logical formulae and then checking whether the proposed claims are logically sound—i.e., whether they can be validly derived from premises verified to be true.

Core Argument

The authors argue that such approaches structurally fail to detect misleading claims due to systematic divergences between conclusions that are logically sound and inferences that humans typically make and accept. Drawing on studies in cognitive science and pragmatics, the paper presents a typology of cases in which logically sound conclusions systematically elicit human inferences that are unsupported by the underlying premises.

Proposed Alternative

Consequently, the authors advocate for a complementary approach:

Leverage the human-like reasoning tendencies of LLMs as a feature rather than a bug
Use these models to validate the outputs of formal components in neurosymbolic systems against potentially misleading conclusions

---

*Auto-collected on 2026-04-07.*