Large language models (LLMs) suffer from serious unfaithful chain-of-thought
(CoT) issues. Previous work attempts to measure and explain it but lacks
in-depth analysis within CoTs and does not consider the interactions among all
reasoning components jointly. In this paper, we first study the CoT
faithfulness issue at the granularity of CoT steps, identify two reasoning
paradigms: centralized reasoning and distributed reasoning, and find their
relationship with faithfulness. Subsequently, we conduct a joint analysis of
the causal relevance among the context, CoT, and answer during reasoning. The
result proves that, when the LLM predicts answers, it can recall correct
information missing in the CoT from the context, leading to unfaithfulness
issues. Finally, we propose the inferential bridging method to mitigate this
issue, in which we use the attribution method to recall information as hints
for CoT generation and filter out noisy CoTs based on their semantic
consistency and attribution scores. Extensive experiments demonstrate that our
approach effectively alleviates the unfaithful CoT problem.

大型语言模型（LLMs）存在严重的不忠实的思维链 (CoT) 问题。本文首先从 CoT 步骤的粒度研究 CoT 忠实性问题，确定了集中式推理和分布式推理两种推理范式，并发现它们与忠实性的关系。随后，我们对环境、CoT 和答案之间的因果关系进行了联合分析。结果证明，当 LLM 预测答案时，它可以从上下文中回忆起 CoT 中缺失的正确信息，导致不忠实的问题。最后，我们提出了推理桥接方法来缓解这个问题，其中我们使用归因方法来作为 CoT 生成的提示并根据其语义一致性和归因分数过滤噪声 CoTs。大量实验证明我们的方法有效地减轻了不忠实的 CoT 问题。