VeriCoT: Neuro-symbolic Chain-of-Thought Validation via Logical Consistency Checks

TL;DR

VeriCoT introduces a neuro-symbolic approach that extracts formal logical arguments from chain-of-thought reasoning in large language models, enabling verification of logical validity and improving reasoning accuracy in high-stakes tasks.

Contribution

The paper presents VeriCoT, a novel method that formalizes chain-of-thought reasoning into first-order logic for automated verification and improved model performance.

Findings

VeriCoT effectively detects flawed reasoning in multiple datasets.

The verification signal enhances model accuracy through self-reflection and fine-tuning.

VeriCoT serves as a predictor of answer correctness.

Abstract

LLMs can perform multi-step reasoning through Chain-of-Thought (CoT), but they cannot reliably verify their own logic. Even when they reach correct answers, the underlying reasoning may be flawed, undermining trust in high-stakes scenarios. To mitigate this issue, we introduce VeriCoT, a neuro-symbolic method that extracts and verifies formal logical arguments from CoT reasoning. VeriCoT formalizes each CoT reasoning step into first-order logic and identifies premises that ground the argument in source context, commonsense knowledge, or prior reasoning steps. The symbolic representation enables automated solvers to verify logical validity while the NL premises allow humans and systems to identify ungrounded or fallacious reasoning steps. Experiments on the ProofWriter, LegalBench, and BioASQ datasets show VeriCoT effectively identifies flawed reasoning, and serves as a strong predictor of final answer correctness. We also leverage VeriCoT's verification signal for (1) inference-time self-reflection, (2) supervised fine-tuning (SFT) on VeriCoT-distilled datasets and (3) preference fine-tuning (PFT) with direct preference optimization (DPO) using verification-based pairwise rewards, further improving reasoning validity and accuracy.