Thinking with Drafting: Optical Decompression via Logical Reconstruction

2026-02-12

AI Summary

This work addresses the imprecision in visual outputs of multimodal large language models during complex visual reasoning, often stemming from their neglect of underlying logical structure. The authors propose a “parsing-as-reasoning” paradigm that reframes visual reasoning as an optical decompression process, leveraging a minimal domain-specific language (DSL) as an intermediate representation. This DSL guides the model to generate executable code sketches, which in turn drive a self-verification mechanism for deterministic visual proofs. By centering the reasoning loop around executable DSL constructs, the framework ensures that visual generation serves logical validation rather than creative expression. Evaluated on the newly introduced VisAlg benchmark for visual algebra, the method demonstrates significant improvements in both reasoning accuracy and logical rigor.

Technology Category

• Computer Vision: Visual Reasoning & Symbolic Representations
• Machine Learning: Large Multimodal Models (LMMs)
• Knowledge Representation and Reasoning: Geometric, Spatial, and Temporal Reasoning

Application Category

• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG

Abstract

Existing multimodal large language models have achieved high-fidelity visual perception and exploratory visual generation. However, a precision paradox persists in complex reasoning tasks: optical perception systems transcribe symbols without capturing logical topology, while pixel-based generative models produce visual artifacts lacking mathematical exactness. To bridge this gap, we propose that reasoning over visual inputs be reconceptualized as optical decompression-the process of reconstructing latent logical structures from compressed visual tokens. Guided by the axiom that Parsing is Reasoning, we introduce Thinking with Drafting (TwD), which utilizes a minimalist Domain-Specific Language (DSL) as a grounding intermediate representation. Unlike standard approaches that hallucinate answers directly, TwD forces the model to draft its mental model into executable code, rendering deterministic visual proofs for self-verification. To validate this, we present VisAlg, a visual algebra benchmark. Experiments demonstrate that TwD serve as a superior cognitive scaffold. Our work establishes a closed-loop system where visual generation acts not as a creative output but as a logical verifier, offering a generalizable path for visual reasoning.

Problem

• visual reasoning
• logical topology
• precision paradox
• optical perception
• visual artifacts

Innovation

• optical decompression
• logical reconstruction
• Domain-Specific Language (DSL)
• visual reasoning
• deterministic visual proofs