Vision-language models (VLMs) aim to reason by jointly leveraging visual and textual modalities. While allocating additional inference-time computation has proven effective for large language models (LLMs), achieving similar scaling in VLMs remains challenging. A key obstacle is that visual inputs are typically provided only once at the start of generation, while textual reasoning (e.g., early visual summaries) is generated autoregressively, causing reasoning to become increasingly text-dominated and allowing early visual grounding errors to accumulate. Moreover, vanilla guidance for visual grounding during inference is often coarse and noisy, making it difficult to steer reasoning over long texts. To address these challenges, we propose \emph{Saliency-Aware Principle} (SAP) selection. SAP operates on high-level reasoning principles rather than token-level trajectories, which enable stable control over discrete generation under noisy feedback while allowing later reasoning steps to re-consult visual evidence when renewed grounding is required. In addition, SAP supports multi-route inference, enabling parallel exploration of diverse reasoning behaviors. SAP is model-agnostic and data-free, requiring no additional training. Empirical results show that SAP achieves competitive performance, especially in reducing object hallucination, under comparable token-generation budgets while yielding more stable reasoning and lower response latency than CoT-style long sequential reasoning.

翻译：视觉-语言模型（VLMs）旨在通过联合利用视觉与文本模态进行推理。尽管为大型语言模型（LLMs）分配额外的推理时间计算已被证明是有效的，但在VLMs中实现类似的扩展仍然具有挑战性。一个关键障碍在于：视觉输入通常仅在生成开始时提供一次，而文本推理（例如早期的视觉摘要）是以自回归方式生成的，这导致推理过程逐渐被文本主导，并使得早期的视觉基础错误不断累积。此外，推理过程中用于视觉基础的常规引导通常是粗糙且带有噪声的，难以在长文本推理中进行有效引导。为解决这些挑战，我们提出了**显著性感知原则**（SAP）选择方法。SAP作用于高层推理原则而非词元级轨迹，这使得在噪声反馈下能对离散生成进行稳定控制，同时允许后续推理步骤在需要重新建立基础时再次参考视觉证据。此外，SAP支持多路径推理，能够并行探索多样化的推理行为。SAP与模型无关且无需数据，不需要额外的训练。实证结果表明，在可比的词元生成预算下，SAP实现了有竞争力的性能（尤其在减少物体幻觉方面），同时比思维链（CoT）式的长序列推理产生了更稳定的推理过程和更低的响应延迟。