Enregistré dans:
Détails bibliographiques
Auteurs principaux: Wen, Bo, Wang, Chen, Bilal, Erhan
Format: Preprint
Publié: 2025
Sujets:
Accès en ligne:https://arxiv.org/abs/2511.15717
Tags: Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
_version_ 1866912719498641408
author Wen, Bo
Wang, Chen
Bilal, Erhan
author_facet Wen, Bo
Wang, Chen
Bilal, Erhan
contents ARC-AGI and ARC-AGI-2 measure generalization-through-composition on small color-quantized grids, and their prize competitions make progress on these harder held-out tasks a meaningful proxy for systematic generalization. Recent instruction-first systems translate grids into concise natural-language or DSL rules executed in generate-execute-select loops, yet we lack a principled account of how encodings shape model perception and how to separate instruction errors from execution errors. We hypothesize that modality imposes perceptual bottlenecks -- text flattens 2D structure into 1D tokens while images preserve layout but can introduce patch-size aliasing -- thereby shaping which grid features are reliably perceived. To test this, we isolate perception from reasoning across nine text and image modalities using a weighted set-disagreement metric and a two-stage reasoning pipeline, finding that structured text yields precise coordinates on sparse features, images capture 2D shapes yet are resolution-sensitive, and combining them improves execution (about 8 perception points; about 0.20 median similarity). Overall, aligning representations with transformer inductive biases and enabling cross-validation between text and image yields more accurate instructions and more reliable execution without changing the underlying model.
format Preprint
id arxiv_https___arxiv_org_abs_2511_15717
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle How Modality Shapes Perception and Reasoning: A Study of Error Propagation in ARC-AGI
Wen, Bo
Wang, Chen
Bilal, Erhan
Artificial Intelligence
Computer Vision and Pattern Recognition
Multiagent Systems
ARC-AGI and ARC-AGI-2 measure generalization-through-composition on small color-quantized grids, and their prize competitions make progress on these harder held-out tasks a meaningful proxy for systematic generalization. Recent instruction-first systems translate grids into concise natural-language or DSL rules executed in generate-execute-select loops, yet we lack a principled account of how encodings shape model perception and how to separate instruction errors from execution errors. We hypothesize that modality imposes perceptual bottlenecks -- text flattens 2D structure into 1D tokens while images preserve layout but can introduce patch-size aliasing -- thereby shaping which grid features are reliably perceived. To test this, we isolate perception from reasoning across nine text and image modalities using a weighted set-disagreement metric and a two-stage reasoning pipeline, finding that structured text yields precise coordinates on sparse features, images capture 2D shapes yet are resolution-sensitive, and combining them improves execution (about 8 perception points; about 0.20 median similarity). Overall, aligning representations with transformer inductive biases and enabling cross-validation between text and image yields more accurate instructions and more reliable execution without changing the underlying model.
title How Modality Shapes Perception and Reasoning: A Study of Error Propagation in ARC-AGI
topic Artificial Intelligence
Computer Vision and Pattern Recognition
Multiagent Systems
url https://arxiv.org/abs/2511.15717