mlsanityData to 14 Oct 2026

Beyond Static Visual Tokens: Structured Sequential Visual Chain-of-Thought Reasoning

arXiv 2603.26737 ↗First posted 21 Mar 2026cs.CV cs.AIcited by 1

On the map: LLM reasoning → Multimodal reasoning

also in 1 more
Abstract (arXiv)

Current multimodal LLMs encode images as static visual prefixes and rely on text-based reasoning, lacking goal-driven and adaptive visual access. Inspired by human visual perception-where attention is selectively and sequentially shifted from the most informative regions to secondary cues-we propose Structural Sequential Visual CoT SSV-CoT. First, a question-relevant saliency map identifies and organizes key visual regions, explicitly modeling the spatial distribution of visual importance. Second, reasoning is performed following this discriminative order, inducing a curriculum-like semantic progression from primary to secondary cues. This method is trained end-to-end, using text cot and answer supervision, without relying on region-level annotations or specialized external tools. Experiments on diverse visual reasoning benchmarks show gains, validating structured and sequential visual cognition.

Key facts. Cited by 1 paper in the map (data through 14 Oct 2026). Placed in Multimodal reasoning, one of 17 directions in LLM reasoning. Builds on 12 of its 52 references that are in the map. First posted on arXiv 21 Mar 2026.

Builds on

12 of 52 references.

Closest ancestors · 5

Same direction, newest first

  1. 2025VIPER: Visual Perception and Explainable Reasoning for Sequential Decision-Making19 Mar
  2. 2024Expanding Performance Boundaries of Open-Source Multimodal Models with Model, Data, and Test-Time Scaling6 Dec
  3. 2024CoCoT: Contrastive Chain-of-Thought Prompting for Large Multimodal Models with Multiple Image Inputs5 Jan
  4. 2023MM-REACT: Prompting ChatGPT for Multimodal Reasoning and Action20 Mar
  5. 2023Visual ChatGPT: Talking, Drawing and Editing with Visual Foundation Models8 Mar

Other references · 7

  1. 2024Qwen2-VL: Enhancing Vision-Language Model's Perception of the World at Any Resolutioncited by 4,234
  2. 2024MiniCPM-V: A GPT-4V Level MLLM on Your Phonecited by 1,079
  3. 2023Shikra: Unleashing Multimodal LLM's Referential Dialogue Magiccited by 906
  4. 2024Ovis: Structural Embedding Alignment for Multimodal Large Language Modelcited by 169
  5. 2024POINTS1.5: Building a Vision-Language Model towards Real World Applicationscited by 12
  6. 2025Structured Outputs Enable General-Purpose LLMs to be Medical Expertscited by 7
  7. 2014Multiple Object Recognition with Visual Attentioncited by 4

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2026: 1.

1 paper.

  1. 2026Position Rebinding Cache Reuse: Replay-Free Visual Revisiting for Interleaved Multimodal Reasoningcited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2026SLVR: Structured Latent Visual Reasoning via Human-like Reasoning Flows
  2. 2026TVI-CoT: Text-Visual Interleaved Chain-of-Thought Reasoning for Multimodal Understanding
  3. 2026Distilling Visual Reasoning into Text Space
  4. 2025Understand, Think, and Answer: Advancing Visual Reasoning with Large Multimodal Models
  5. 2025VGR: Visual Grounded Reasoning
  6. 2025MM-CoT:A Benchmark for Probing Visual Chain-of-Thought Reasoning in Multimodal Models
  7. 2025Latent Visual Reasoning
  8. 2026UniVLR: Unifying Text and Vision in Visual Latent Reasoning for Multimodal LLMs