Reasoning is a Modality

TL;DR

This paper proposes that reasoning functions as a distinct modality separate from the workspace, introducing a role-separated transformer that improves visual reasoning performance on ARC tasks beyond human accuracy.

Contribution

The paper introduces a novel role-separated transformer architecture that explicitly models reasoning as a separate modality, achieving state-of-the-art results on the ARC benchmark.

Findings

Achieved 62.6% accuracy on ARC-1, surpassing human performance.

Model exhibits more coherent rule-application structure than baseline.

Supports hypothesis that reasoning is a distinct modality.

Abstract

The Abstraction and Reasoning Corpus (ARC) provides a compact laboratory for studying abstract reasoning, an ability central to human intelligence. Modern AI systems, including LLMs and ViTs, largely operate as sequence-of-behavior prediction machines: they match observable behaviors by modeling token statistics without a persistent, readable mental state. This creates a gap with human-like behavior: humans can explain an action by decoding internal state, while AI systems can produce fluent post-hoc rationalizations that are not grounded in such a state. We hypothesize that reasoning is a modality: reasoning should exist as a distinct channel separate from the low-level workspace on which rules are applied. To test this hypothesis, on solving ARC tasks as a visual reasoning problem, we designed a novel role-separated transformer block that splits global controller tokens from grid workspace tokens, enabling iterative rule execution. Trained and evaluated within the VARC vision-centric protocol, our method achieved 62.6% accuracy on ARC-1, surpassing average human performance (60.2%) and outperforming prior methods significantly. Qualitatively, our models exhibit more coherent rule-application structure than the dense ViT baseline, consistent with a shift away from plausible probability blobs toward controller-driven reasoning.