Beyond the Literal: Decomposing Pragmatic Intent in Multimodal Meme Understanding
About
When asked what a meme or sarcastic post means, Large Vision Language Models (LVLMs) tend to describe what the image shows rather than what the author is trying to communicate. Standard instruction tuning entangles a post's literal content with its pragmatic meaning, letting surface-level details contaminate the final response. We reframe meme understanding as a problem of literal-pragmatic decomposition and propose \textbf{Intent Projection}, a framework that separates the two signals at the representation, output, and objective levels within a single LVLM backbone. At the representation level, an orthogonal projection module removes dominant unimodal directions from the fused image-text representation, retaining only the pragmatic residual, while a surface-real affect classifier anchors the decoder with a discrete tag that names the polarity gap. At the output level, the model externalizes a structured reasoning chain, and at the objective level a contrastive reward explicitly penalizes answers that restate the literal description. Across six multimodal benchmarks, Intent Projection consistently outperforms open-source baselines and narrows the gap to proprietary models, with the largest gains on high-divergence posts where literal collapse is most damaging.
Related benchmarks
| Task | Dataset | Result | Rank | |
|---|---|---|---|---|
| Meme Generation | MemeReaCon PC-G (test) | BERTScore F159.8 | 21 | |
| Meme Generation | MemeReaCon PI-G (test) | BERTScore F141.3 | 21 | |
| Meme Generation | MemeCap (test) | BERTScore-F154.5 | 21 | |
| Meme Generation | MET-Meme (test) | BERTScore F151.7 | 21 | |
| Meme Generation | MUSE (test) | BERTScore F158 | 21 | |
| Meme Generation | GOAT-G (test) | BERTScore F153.1 | 21 | |
| Meme Classification | GOAT-C (test) | Accuracy79.4 | 12 | |
| Meme Classification | MemeReaCon CMI-C (test) | Accuracy79.6 | 12 | |
| Meme Classification | MemeReaCon CSAC-C (test) | Accuracy65 | 12 | |
| Meme Classification | MMSD 2.0 (test) | Accuracy84.8 | 12 |