arXiv：cs.AI· Hyewon Kang, Jungmin Lee, Ilgyu Lee, Seok-Jun Hong·· 6 小时前AI 评分31
ReGraph：视觉"what"与"where"双流中涌现泛化能力的计算解释
ReGraph: A Computational Account of Emergent Generalization in the "what" and "where" Dual Visual Streams
AI 导读
研究者提出 ReGraph——一种带生物归纳偏置的循环双流图模型，在 Something-Something V2 上训练后，情境无关编码与网格样空间基只在扩展背侧通路上共同涌现。单流、无调制变体及标准基线中均未出现这一现象，表明视网膜驱动的流特化编码、背侧到腹侧调制等归纳偏置是关系结构形成的前提。事后分析显示，这些网格样基通过侧向连接充当可复用的信息路由模板。
正文
Abstract:Where generalization capacity--the ability to extract context-invariant relational structures--first emerges remains a central question in AI and neuroscience. The foundation for this capacity lies upstream of the hippocampus, within the entorhinal cortex, where parallel pathways dissociate relational structure in the medial entorhinal cortex (MEC) from sensory content in the lateral entorhinal cortex. However, as Eichenbaum argued, such factorization likely originates earlier, driven by the segregation of the dorsal ('where') and ventral ('what') visual streams. Supporting this, grid-like firing patterns--a signature of MEC (context-invariant codes)--also appear in preceding neocortical regions along the dorsal pathway. Yet, how such representations are computationally formed along upstream pathways remains unknown. To investigate this in silico, we developed ReGraph, a recurrent dual-stream graph model with biological inductive biases, including retina-driven stream-specialized encoding, dorsal-to-ventral modulation, and dynamic lateral connectivity. Trained on the action benchmark Something-Something V2, ReGraph revealed a pathway-specific emergence of relational mapping: context-invariant codes and grid-like spatial bases uniquely co-emerged along the extended dorsal stream. In contrast, their absence in single-stream, unmodulated variants, and standard baselines implies that these inductive biases are prerequisites for relational structures. Crucially, our post-hoc analyses demonstrated that these grid-like bases serve as reusable routing templates for information processing via lateral connectivity. Together, our findings provide a computational account that generalization may not be a faculty that emerges abruptly within a dedicated region, but a property that already takes shape as sensory information is parsed into factorized streams of hierarchical visual processing.
| Comments: | 29 pages, 6 figures |
| Subjects: | Neural and Evolutionary Computing (cs.NE); Artificial Intelligence (cs.AI); Neurons and Cognition (q-bio.NC) |
| Cite as: | arXiv:2610.07962 [cs.NE] |
| (or arXiv:2610.07962v1 [cs.NE] for this version) | |
| https://doi.org/10.48550/arXiv.2610.07962 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Hyewon Kang [view email]
[v1]
Tue, 6 Oct 2026 08:30:45 UTC (15,805 KB)
来源：arXiv：cs.AI · arxiv.org