MentalThink: Shaping Thoughts in Mental SVG World

• URL: http://arxiv.org/abs/2607.03530v1
• Date: Fri, 03 Jul 2026 17:59:58 GMT
• Title: MentalThink: Shaping Thoughts in Mental SVG World
• Abstract summary: MentalThink is a visual-symbolic reasoning paradigm that equips Multimodal LLMs with an executable mechanism for "mental" visualization.<n>We instantiate this paradigm through a two-stage training framework, combining Supervised FineTuning (SFT) for syntactic alignment with multiturn SVG-Reinforcement Learning (RL) to encourage iterative inspection, revision, and refinement of intermediate visual hypotheses.
• Score: 65.29580685085101
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: We introduce MentalThink, a visual-symbolic reasoning paradigm that equips Multimodal LLMs (MLLMs) with an executable mechanism for "mental" visualization. The core of MentalThink is a think-with-SVG pipeline, where the model learns to generate, render, and interpret scalable vector graphics (SVG) code as an intermediate visual representation for multi-turn reasoning. By creating structured vector sketches, the model can externalize spatial hypotheses, inspect them through deterministic rendering, and reason within a constrained geometric space, effectively mimicking the human process of mental imagery. We instantiate this paradigm through a two-stage training framework, combining Supervised Fine-Tuning (SFT) for SVG syntactic alignment with multi-turn Reinforcement Learning (RL) to encourage iterative inspection, revision, and refinement of intermediate visual hypotheses. Extensive evaluations demonstrate that MentalThink achieves superior performance on spatial understanding and reasoning benchmarks (e.g., 55.1% on VSIBench, 76.0% on MindCube), showing that executable vector graphics provide a verifiable visual workspace for dynamic perspective taking, visual reflection, and compositional scene construction.

Related papers

• CURV: Enhancing Chart Understanding Through Curriculum Visual Grounded Reasoning [58.582602613053204]
  Chart question answering (CQA) requires multimodal large language models (MLLMs) to integrate visual comprehension with logical reasoning.<n>Current MLLMs lack intrinsic visual grounded reasoning capabilities, leading to inaccurate perception and reasoning disconnected from visual evidence.<n>We propose CURV, a curriculum learning framework that develops intrinsic visual reasoning capabilities by reformulating CQA as multi-step visual grounded reasoning.
  arXiv  Detail & Related papers  (2026-08-03T19:48:00Z)
• LanteRn: Latent Visual Structured Reasoning [7.141402207573525]
  We introduce LanteRn, a framework that enables visual reasoning to occur directly in latent space.<n>LanteRn augments a vision-language transformer with the ability to generate and attend to continuous visual thought embeddings during inference.<n>We evaluate LanteRn on three perception-centric benchmarks (VisCoT, V*, and Blink)
  arXiv  Detail & Related papers  (2026-03-26T16:41:59Z)
• Visual-ERM: Reward Modeling for Visual Equivalence [59.317480168347664]
  Visual Equivalence Reward Model (Visual-ERM) is a multimodal generative reward model that provides fine-grained, interpretable, and task-agnostic feedback.<n>Visual-ERM improves Qwen3-VL-8B-Instruct by +8.4 on chart-to-code and yields consistent gains on table and SVG parsing.<n>VisualCritic-RewardBench (VC-RewardBench) is a benchmark for judging fine-grained image-to-image discrepancies on structured visual data.
  arXiv  Detail & Related papers  (2026-03-13T17:58:14Z)
• Thinking with Drafting: Optical Decompression via Logical Reconstruction [19.018138128118775]
  We present Thinking with Drafting (TwD) as a process of reconstructing latent logical structures from compressed visual tokens.<n>TwD forces the model to draft its mental model into executable code, rendering deterministic visual proofs for self-verification.<n>Our work establishes a closed-loop system where visual generation acts not as a creative output but as a logical verifier.
  arXiv  Detail & Related papers  (2026-02-12T08:54:02Z)
• Seeing through Imagination: Learning Scene Geometry via Implicit Spatial World Modeling [68.14113731953971]
  This paper introduces MILO, an Implicit spatIaL wOrld modeling paradigm that simulates human-like imagination.<n>We show that our approach significantly enhances spatial reasoning capabilities across multiple baselines and benchmarks.
  arXiv  Detail & Related papers  (2025-12-01T16:01:41Z)
• Hyperphantasia: A Benchmark for Evaluating the Mental Visualization Capabilities of Multimodal LLMs [22.46006112029019]
  Mental visualization is a critical cognitive skill in humans, supporting abilities such as spatial navigation, predicting physical trajectories, and solving complex visual problems through imaginative simulation.<n>We introduce Hyperphantasia, a synthetic benchmark designed to evaluate the mental visualization abilities of Multimodal Large Language Models (MLLMs) through four carefully constructed puzzles.<n>Our comprehensive evaluation of state-of-the-art models reveals a substantial gap between the performance of humans and MLLMs.
  arXiv  Detail & Related papers  (2025-07-16T05:54:37Z)
• Spatial Mental Modeling from Limited Views [71.57140964322559]
  Our new MindCube benchmark with 21,154 questions across 3,268 images exposes this critical gap.<n>Using MindCube, we evaluate how well Vision Language Models (VLMs) build robust spatial mental models.<n>We then explore three approaches to help VLMs approximate spatial mental models, including unseen intermediate views, natural language reasoning chains, and cognitive maps.
  arXiv  Detail & Related papers  (2025-06-26T16:38:19Z)
• Thinking with Generated Images [30.28526622443551]
  We present Thinking with Generated Images, a novel paradigm that transforms how large multimodal models (LMMs) engage with visual reasoning.<n>Our approach enables AI models to engage in the kind of visual imagination and iterative refinement that characterizes human creative, analytical, and strategic thinking.
  arXiv  Detail & Related papers  (2025-05-28T16:12:45Z)
• Visually Descriptive Language Model for Vector Graphics Reasoning [76.42082386029206]
  We propose the Visually Descriptive Language Model (VDLM) to bridge the gap between low-level visual perception and high-level language reasoning.<n>We show that VDLM significantly improves state-of-the-art LMMs like GPT-4o on various multimodal perception and reasoning tasks.
  arXiv  Detail & Related papers  (2024-04-09T17:30:18Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.