As Large Language Models (LLMs) are increasingly deployed in real-world settings, correctness alone is insufficient. Reliable deployment requires maintaining truthful beliefs under contextual perturbations. Existing evaluations largely rely on point-wise confidence like Self-Consistency, which can mask brittle belief. We show that even facts answered with perfect self-consistency can rapidly collapse under mild contextual interference. To address this gap, we propose Neighbor-Consistency Belief (NCB), a structural measure of belief robustness that evaluates response coherence across a conceptual neighborhood. To validate the efficiency of NCB, we introduce a new cognitive stress-testing protocol that probes outputs stability under contextual interference. Experiments across multiple LLMs show that the performance of high-NCB data is relatively more resistant to interference. Finally, we present Structure-Aware Training (SAT), which optimizes context-invariant belief structure and reduces long-tail knowledge brittleness by approximately 30%. Code is available at https://github.com/zjunlp/belief.

翻译：随着大语言模型（LLMs）越来越多地部署于现实场景，仅凭正确性已不足以满足需求。可靠部署要求模型在上下文扰动下保持忠实信念。现有评估方法主要依赖基于自我一致性的逐点置信度，这往往会掩盖脆弱的信念。我们发现，即使是能以完美自我一致性回答的事实，在温和的上下文干扰下也会迅速崩塌。为弥补这一不足，我们提出邻域一致性信念（NCB），这是一种从结构上衡量信念鲁棒性的指标，可评估概念邻域内响应的连贯性。为验证NCB的有效性，我们引入一种新的认知压力测试协议，用于探测模型在上下文干扰下的输出稳定性。跨多个LLMs的实验表明，高NCB数据的性能相对更能抵抗干扰。最后，我们提出结构感知训练（SAT），该训练优化了上下文不变的信念结构，并将长尾知识脆弱性降低了约30%。代码已开源：https://github.com/zjunlp/belief。