IFDNS: An Iterative Feedback-Driven Neuro-Symbolic Method for Faithful Logical Reasoning

2026-01-12

AI Summary

This work addresses the limitations of large language models in logical reasoning, where inconsistencies between reasoning chains and conclusions often lead to failures, and existing neuro-symbolic approaches frequently lose critical details during causal information extraction. To mitigate these issues, the authors propose a neuro-symbolic method grounded in a multi-round feedback mechanism that iteratively refines causal relation extraction and propositional logic translation, thereby substantially enhancing both faithfulness and accuracy in reasoning. The approach seamlessly integrates with established prompting strategies—including Chain-of-Thought (CoT) and its self-consistent variants—and demonstrates significant performance gains across six logical reasoning benchmarks. Notably, it improves CoT accuracy by 9.40% on LogiQA and boosts CoT-SC performance by 11.70% on PrOntoQA.

Technology Category

• Machine Learning: Neuro-Symbolic Learning
• Reasoning under Uncertainty: Causality
• Knowledge Representation and Reasoning: Action, Change, and Causality

Application Category

• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs

Abstract

Large language models (LLMs) have demonstrated impressive capabilities across a wide range of reasoning tasks, including logical and mathematical problem-solving. While prompt-based methods like Chain-of-Thought (CoT) can enhance LLM reasoning abilities to some extent, they often suffer from a lack of faithfulness, where the derived conclusions may not align with the generated reasoning chain. To address this issue, researchers have explored neuro-symbolic approaches to bolster LLM logical reasoning capabilities. However, existing neuro-symbolic methods still face challenges with information loss during the process. To overcome these limitations, we introduce Iterative Feedback-Driven Neuro-Symbolic (IFDNS), a novel prompt-based method that employs a multi-round feedback mechanism to address LLM limitations in handling complex logical relationships. IFDNS utilizes iterative feedback during the logic extraction phase to accurately extract causal relationship statements and translate them into propositional and logical implication expressions, effectively mitigating information loss issues. Furthermore, IFDNS is orthogonal to existing prompt methods, allowing for seamless integration with various prompting approaches. Empirical evaluations across six datasets demonstrate the effectiveness of IFDNS in significantly improving the performance of CoT and Chain-of-Thought with Self-Consistency (CoT-SC). Specifically, IFDNS achieves a +9.40% accuracy boost for CoT on the LogiQA dataset and a +11.70% improvement for CoT-SC on the PrOntoQA dataset.

Problem

• faithful reasoning
• logical reasoning
• information loss
• neuro-symbolic methods
• large language models

Innovation

• Neuro-Symbolic Reasoning
• Iterative Feedback
• Faithful Reasoning
• Logical Implication Extraction
• Prompt Engineering