# Lost in Perception: Isolating Perceptual and Reasoning Failures in Multimodal Physics and Geometry Reasoning

arXiv: https://arxiv.org/abs/2609.18991 · HTML: https://mlsanity.com/paper/2609.18991/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Placed in Multimodal reasoning, one of 17 directions in LLM reasoning.
- Builds on 23 of its 29 references that are in the map.
- First posted on arXiv 17 Jul 2026.

## Placed in

- LLM reasoning / [Multimodal reasoning](https://mlsanity.com/lr/multimodal-reasoning/)
- Multimodal LLMs / vision-language models / [Embodied and 3D vision-language models](https://mlsanity.com/mllm/embodied-and-3d-vision-language-models/)

## Builds on (most cited first, 23 of 23)

- [GPT-4 Technical Report](https://mlsanity.com/paper/2303.08774/) (2023)
- [The Llama 3 Herd of Models](https://mlsanity.com/paper/2407.21783/) (2024)
- [Chain-of-Thought Prompting Elicits Reasoning in Large Language Models](https://mlsanity.com/paper/2201.11903/) (2022)
- [Qwen2.5-VL Technical Report](https://mlsanity.com/paper/2502.13923/) (2025)
- [Gemini 2.5: Pushing the Frontier with Advanced Reasoning, Multimodality, Long Context, and Next Generation Agentic Capabilities](https://mlsanity.com/paper/2507.06261/) (2025)
- [Let's Verify Step by Step](https://mlsanity.com/paper/2305.20050/) (2023)
- [MMBench: Is Your Multi-modal Model an All-around Player?](https://mlsanity.com/paper/2307.06281/) (2023)
- [Gemma 3 Technical Report](https://mlsanity.com/paper/2503.19786/) (2025)
- [OlympiadBench: A Challenging Benchmark for Promoting AGI with Olympiad-Level Bilingual Multimodal Scientific Problems](https://mlsanity.com/paper/2402.14008/) (2024)
- [Language Models Don't Always Say What They Think: Unfaithful Explanations in Chain-of-Thought Prompting](https://mlsanity.com/paper/2305.04388/) (2023)
- [Measuring Faithfulness in Chain-of-Thought Reasoning](https://mlsanity.com/paper/2307.13702/) (2023)
- [Towards Reasoning in Large Language Models: A Survey](https://mlsanity.com/paper/2212.10403/) (2022)
- [MathVerse: Does Your Multi-modal LLM Truly See the Diagrams in Visual Math Problems?](https://mlsanity.com/paper/2403.14624/) (2024)
- [Inter-GPS: Interpretable Geometry Problem Solving with Formal Language and Symbolic Reasoning](https://mlsanity.com/paper/2105.04165/) (2021)
- [PhysReason: A Comprehensive Benchmark towards Physics-Based Reasoning](https://mlsanity.com/paper/2502.12054/) (2025)
- [A Survey of Mathematical Reasoning in the Era of Multimodal Large Language Model: Benchmark, Method & Challenges](https://mlsanity.com/paper/2412.11936/) (2024)
- [Intern-S1: A Scientific Multimodal Foundation Model](https://mlsanity.com/paper/2508.15763/) (2025)
- [SeePhys: Does Seeing Help Thinking? -- Benchmarking Vision-Based Physics Reasoning](https://mlsanity.com/paper/2505.19099/) (2025)
- [Improving Physics Reasoning in Large Language Models Using Mixture of Refinement Agents](https://mlsanity.com/paper/2412.00821/) (2024)
- [PhysUniBench: A Multi-Modal Physics Reasoning Benchmark at Undergraduate Level](https://mlsanity.com/paper/2506.17667/) (2025)
- [Seeing is Fixing: Cross-Modal Reasoning with Multimodal LLMs for Visual Software Issue Fixing](https://mlsanity.com/paper/2506.16136/) (2025)
- [Diagnosing Visual Reasoning: Challenges, Insights, and a Path Forward](https://mlsanity.com/paper/2510.20696/) (2025)
- [Beyond Correctness: Exposing LLM-generated Logical Flaws in Reasoning via Multi-step Automated Theorem Proving](https://mlsanity.com/paper/2512.23511/) (2025)

## Abstract (arXiv)

Multimodal LLMs report strong performance on scientific reasoning benchmarks, yet most treat perception and reasoning as a single measurable process. We introduce a five-task diagnostic experiment across physics and geometry benchmarks that isolates failures to perception, reasoning, or both. Incorrect diagram interpretation degrades performance even on problems models solve correctly from text alone, and accuracy generally rises from raw images to human-authored captions. Recovery under corrected captions is high for some models, separating perception-blocked failures from genuine reasoning bottlenecks. Which reasoning error follows a perception failure depends on domain: physics failures resolve into calculation errors, geometry into conceptual misapplication. As a discussion beyond our core experiments, InternS1-mini, despite heavy scientific pretraining and thinking capabilities, falls below the weakest model from experiments on every task, with reasoning traces frequently truncating before completion.