VeriCoT: Neuro-symbolic Chain-of-Thought Validation via Logical Consistency Checks
This addresses the issue of unreliable reasoning verification in LLMs for applications requiring trust, such as legal or biomedical domains, though it is an incremental improvement by combining existing neuro-symbolic and fine-tuning techniques.
The paper tackles the problem that LLMs cannot reliably verify their own Chain-of-Thought reasoning, which undermines trust in high-stakes scenarios, by introducing VeriCoT, a neuro-symbolic method that extracts and validates logical arguments from CoT reasoning, showing it effectively identifies flawed reasoning and improves reasoning validity and accuracy through experiments on datasets like ProofWriter, LegalBench, and BioASQ.
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.