While LLMs demonstrate impressive reasoning capabilities, they remain fragile in multi-step logical deduction, where a single transition error can propagate through the entire reasoning chain, leading to unstable performance. In this work, we identify logical connectives as primary points of this structural fragility. Through empirical analysis, we show that connective tokens function as high entropy forking points, at which models frequently struggle to determine the correct logical direction. Motivated by this observation, we hypothesize that intervening in logical connective selection can guide LLMs toward more correct logical direction, thereby improving the overall reasoning chain. To validate this hypothesis, we propose a multi-layered framework that intervenes specifically at these logic-critical junctions in the reasoning process. Our framework includes (1) Gradient-based Logical Steering to guide LLMs internal representations towards valid reasoning subspaces, (2) Localized Branching to resolve ambiguity via targeted look-ahead search, and (3) Targeted Transition Preference Optimization, a surgical reinforcement learning objective that selectively optimizes single-token preferences at logical pivots. Crucially, by concentrating intervention solely on logic-critical transitions, our framework achieves a favorable accuracy--efficiency trade-off compared to global inference time scaling methods like beam search and self-consistency.

翻译：虽然大语言模型展现出令人印象深刻的推理能力，但在多步逻辑推导中仍存在脆弱性，单个转换错误可能通过整个推理链传播，导致性能不稳定。本研究将逻辑连接词识别为此类结构脆弱性的主要根源。通过实证分析，我们证明连接词标记作为高熵分支点，模型在此类节点上常难以确定正确的逻辑方向。基于这一观察，我们假设干预逻辑连接词的选择可以引导大语言模型走向更正确的逻辑方向，从而改进整体推理链。为验证该假设，我们提出一个多层框架，专门在推理过程中这些逻辑关键节点进行干预。该框架包括：（1）基于梯度的逻辑引导，用于将大语言模型的内部表征导向有效的推理子空间；（2）局部分支策略，通过定向前瞻搜索解决歧义；（3）目标转换偏好优化，一种选择性优化逻辑枢轴处单标记偏好的精细强化学习目标。关键在于，通过将干预集中在逻辑关键转换上，我们的框架相比波束搜索和自一致性等全局推理时间扩展方法，实现了更优的准确率-效率权衡。