# Learning to Think Like a Cartoon Captionist: Incongruity-Resolution Supervision for Multimodal Humor Understanding

arXiv: https://arxiv.org/abs/2604.15210 · HTML: https://mlsanity.com/paper/2604.15210/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 25 of its 48 references that are in the map.
- First posted on arXiv 16 Apr 2026.

## Builds on (most cited first, 25 of 25)

- [GPT-4o System Card](https://mlsanity.com/paper/2410.21276/) (2024)
- [LlamaFactory: Unified Efficient Fine-Tuning of 100+ Language Models](https://mlsanity.com/paper/2403.13372/) (2024)
- [The FineWeb Datasets: Decanting the Web for the Finest Text Data at Scale](https://mlsanity.com/paper/2406.17557/) (2024)
- [Visual-RFT: Visual Reinforcement Fine-Tuning](https://mlsanity.com/paper/2503.01785/) (2025)
- [GLM-4.5V and GLM-4.1V-Thinking: Towards Versatile Multimodal Reasoning with Scalable Reinforcement Learning](https://mlsanity.com/paper/2507.01006/) (2025)
- [LLaVA-CoT: Let Vision Language Models Reason Step-by-Step](https://mlsanity.com/paper/2411.10440/) (2024)
- [LlamaV-o1: Rethinking Step-by-step Visual Reasoning in LLMs](https://mlsanity.com/paper/2501.06186/) (2025)
- [NVLM: Open Frontier-Class Multimodal LLMs](https://mlsanity.com/paper/2409.11402/) (2024)
- [BIG-Bench Extra Hard](https://mlsanity.com/paper/2502.19187/) (2025)
- [Do Androids Laugh at Electric Sheep? Humor "Understanding" Benchmarks from The New Yorker Caption Contest](https://mlsanity.com/paper/2209.06293/) (2022)
- [Vision-R1: Evolving Human-Free Alignment in Large Vision-Language Models via Vision-Guided Reinforcement Learning](https://mlsanity.com/paper/2503.18013/) (2025)
- [Let's Think Outside the Box: Exploring Leap-of-Thought in Large Language Models with Creative Humor Generation](https://mlsanity.com/paper/2312.02439/) (2023)
- [Perception-R1: Advancing Multimodal Reasoning Capabilities of MLLMs via Visual Perception Reward](https://mlsanity.com/paper/2506.07218/) (2025)
- [X-Reasoner: Towards Generalizable Reasoning Across Modalities and Domains](https://mlsanity.com/paper/2505.03981/) (2025)
- ["President Vows to Cut <Taxes> Hair": Dataset and Analysis of Creative Text Editing for Humorous Headlines](https://mlsanity.com/paper/1906.00274/) (2019)
- [Humor in AI: Massive Scale Crowd-Sourced Preferences and Benchmarks for Cartoon Captioning](https://mlsanity.com/paper/2406.10522/) (2024)
- [Can Large Multimodal Models Uncover Deep Semantics Behind Images?](https://mlsanity.com/paper/2402.11281/) (2024)
- [LongPerceptualThoughts: Distilling System-2 Reasoning for System-1 Perception](https://mlsanity.com/paper/2504.15362/) (2025)
- [Reverse-Engineering Satire, or "Paper on Computational Humor Accepted Despite Making Serious Advances"](https://mlsanity.com/paper/1901.03253/) (2019)
- [Bridging the Creativity Understanding Gap: Small-Scale Human Alignment Enables Expert-Level Humor Ranking in LLMs](https://mlsanity.com/paper/2502.20356/) (2025)
- [Cracking the Code of Juxtaposition: Can AI Models Understand the Humorous Contradictions](https://mlsanity.com/paper/2405.19088/) (2024)
- [Stimulating Creativity with FunLines: A Case Study of Humor Generation in Headlines](https://mlsanity.com/paper/2002.02031/) (2020)
- [Getting Serious about Humor: Crafting Humor Datasets with Unfunny Large Language Models](https://mlsanity.com/paper/2403.00794/) (2024)
- [HumorDB: Can AI understand graphical humor?](https://mlsanity.com/paper/2406.13564/) (2024)
- [On the Wings of Imagination: Conflicting Script-based Multi-role Framework for Humor Caption Generation](https://mlsanity.com/paper/2602.06423/) (2026)

## Abstract (arXiv)

Humor is one of the few cognitive tasks where getting the reasoning right matters as much as getting the answer right. While recent work evaluates humor understanding on benchmarks such as the New Yorker Cartoon Caption Contest (NYCC), it largely treats it as black-box prediction, overlooking the structured reasoning processes underlying humor comprehension. We introduce IRS (Incongruity-Resolution Supervision), a framework that decomposes humor understanding into three components: Incongruity Modeling, which identifies mismatches in the visual scene; Resolution Modeling, which constructs coherent reinterpretations of these mismatches; and Preference Alignment, which evaluates candidate interpretations under human judgments. Grounded in incongruity-resolution theory and expert captionist practice, IRS supervises intermediate reasoning process through structured traces that make the path from visual perception to humorous interpretation explicit and learnable. Across 7B, 32B, and 72B models on NYCC, IRS improves performance across caption matching and ranking, with IRS-72B achieving the strongest model performance on ranking (76.10%), surpassing both non-expert human performance and all evaluated open- and closed-source multimodal baselines. Zero-shot transfer further shows that IRS learns generalizable reasoning patterns.