arXiv cs.AI· Alexa R. Tartaglini, Zhengxuan Wu, Thadd\"aus Wiedemer, Martin Wattenberg, Been Kim, Yuxuan Li·· 2 小时前AI 评分52
arXiv 论文：VLM 中红心符号可隐式引导递归解法，但模型间视觉通信存在生成解释鸿沟
A Heart Means Recursion: Exploring Visual Meaning and Communication in VLMs
AI 导读
arXiv 论文（arXiv:2610.36449）探讨视觉语言模型（VLM）的视觉意义与通信能力。研究发现红心等简单视觉符号能隐式引导模型在简单编码问题上倾向递归解法，而模型的显式解释和思考轨迹与实际行为明显不一致，形成生成解释鸿沟：模型倾向发送自己觉得直观的形状基元，但这些视觉线索往往无法引导其他模型。在抽象概念与故事的通信游戏中，模型能创造性地用视觉传递意义，尽管部分通信对人类并不直观。
正文 · 原文
Abstract:Do vision-language models (VLMs) share our ability to communicate visually? We share preliminary findings, starting from a surprising observation: simple visual symbols like a red heart can strongly influence model behavior on seemingly unrelated tasks, such as steering models towards recursive solutions on simple coding problems. Notably, this steering effect is largely implicit, as models' direct interpretations and thinking traces about visual symbols diverge substantially from their observed behavior. In visual communication between models, this leads to a generation--interpretation gap: while models tend to send shape primitives that they (and humans) find intuitive, these visual cues often fail to steer other models. In further communication games involving abstract concepts and stories, models can creatively use visuals to transmit meaning to one another, although some communications remain non-intuitive to humans. These findings reveal that VLMs might be capable of rich, creative, and visually-grounded communication, but their interpretation and production of visual meaning diverges from human conventions in surprising ways. Together this work highlights both the promise and challenges of visual communication between humans and VLMs.
| Subjects: | Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Computer Vision and Pattern Recognition (cs.CV) |
| Cite as: | arXiv:2610.36449 [cs.AI] |
| (or arXiv:2610.36449v1 [cs.AI] for this version) | |
| https://doi.org/10.48550/arXiv.2610.36449 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Alexa Tartaglini [view email]
[v1]
Tue, 27 Oct 2026 00:51:17 UTC (14,761 KB)
来源：arXiv cs.AI · arxiv.org