Interaction Locality in Hierarchical Recursive Reasoning
The paper discusses a framework called interaction locality for measuring information flow in spatial reasoning tasks. It highlights how this framework applies to hierarchical and recursive reasoning models, demonstrating patterns of local information processing. Additionally, the study explores the implications of interaction locality in embodied 3D models, suggesting a distinction in causal spatial structure at module boundaries.
- ▪Interaction locality is a framework designed to assess information flow in spatial reasoning tasks.
- ▪The framework was applied to hierarchical and recursive reasoning models, revealing patterns of local information processing.
- ▪The study also examined embodied 3D models, indicating that causal spatial structure may concentrate at module boundaries.
Opening excerpt (first ~120 words) tap to expand
Computer Science > Artificial Intelligence arXiv:2605.20784 (cs) [Submitted on 20 May 2026] Title:Interaction Locality in Hierarchical Recursive Reasoning Authors:Yosuke Miyanishi, Tetsuro Morimura View a PDF of the paper titled Interaction Locality in Hierarchical Recursive Reasoning, by Yosuke Miyanishi and 1 other authors View PDF HTML (experimental) Abstract:Spatial reasoning requires both location-bound computation and location-invariant structure: agents must make local moves while preserving route, object, or constraint-level plans. We propose interaction locality, a task-geometry-aware framework for measuring whether information flow stays within nearby cells or semantic segments, or crosses them.
…
Excerpt limited to ~120 words for fair-use compliance. The full article is at arXiv cs.AI.
Discussion
0 commentsMore from arXiv cs.AI
