# V-SEAM: Visual Semantic Editing and Attention Modulating for Causal Interpretability of Vision-Language Models

arXiv: https://arxiv.org/abs/2509.14837 · HTML: https://mlsanity.com/paper/2509.14837/

## Key facts

- Cited by 8 papers in the map (data through 14 Oct 2026).
- Placed in VLM benchmarks and visual reasoning, one of 16 directions in Multimodal LLMs / vision-language models.
- Builds on 22 of its 41 references that are in the map.
- First posted on arXiv 18 Sep 2025.

## Placed in

- Multimodal LLMs / vision-language models / [VLM benchmarks and visual reasoning](https://mlsanity.com/mllm/vlm-benchmarks-and-visual-reasoning/)
- Mechanistic interpretability / [Interpreting vision and multimodal models](https://mlsanity.com/mechinterp/interpreting-vision-and-multimodal-models/)

## Builds on (most cited first, 22 of 22)

- [Improved Baselines with Visual Instruction Tuning](https://mlsanity.com/paper/2310.03744/) (2023)
- [InstructBLIP: Towards General-purpose Vision-Language Models with Instruction Tuning](https://mlsanity.com/paper/2305.06500/) (2023)
- [LLaVA-OneVision: Easy Visual Task Transfer](https://mlsanity.com/paper/2408.03326/) (2024)
- [Locating and Editing Factual Associations in GPT](https://mlsanity.com/paper/2202.05262/) (2022)
- [InternVL: Scaling up Vision Foundation Models and Aligning for Generic Visual-Linguistic Tasks](https://mlsanity.com/paper/2312.14238/) (2023)
- [Qwen-VL: A Versatile Vision-Language Model for Understanding, Localization, Text Reading, and Beyond](https://mlsanity.com/paper/2308.12966/) (2023)
- [Sparse Autoencoders Find Highly Interpretable Features in Language Models](https://mlsanity.com/paper/2309.08600/) (2023)
- [Understanding intermediate layers using linear classifier probes](https://mlsanity.com/paper/1610.01644/) (2016)
- [Vision-and-Language Navigation: Interpreting visually-grounded navigation instructions in real environments](https://mlsanity.com/paper/1711.07280/) (2017)
- [Probing Classifiers: Promises, Shortcomings, and Advances](https://mlsanity.com/paper/2102.12452/) (2021)
- [Dissecting Recall of Factual Associations in Auto-Regressive Language Models](https://mlsanity.com/paper/2304.14767/) (2023)
- [Towards Best Practices of Activation Patching in Language Models: Metrics and Methods](https://mlsanity.com/paper/2309.16042/) (2023)
- [Hierarchical Question-Image Co-Attention for Visual Question Answering](https://mlsanity.com/paper/1606.00061/) (2016)
- [Interpreting CLIP's Image Representation via Text-Based Decomposition](https://mlsanity.com/paper/2310.05916/) (2023)
- [Probing Image-Language Transformers for Verb Understanding](https://mlsanity.com/paper/2106.09141/) (2021)
- [Why Is Spatial Reasoning Hard for VLMs? An Attention Mechanism Perspective on Focus Areas](https://mlsanity.com/paper/2503.01773/) (2025)
- [Towards Vision-Language Mechanistic Interpretability: A Causal Tracing Tool for BLIP](https://mlsanity.com/paper/2308.14179/) (2023)
- [From Redundancy to Relevance: Information Flow in LVLMs Across Reasoning Tasks](https://mlsanity.com/paper/2406.06579/) (2024)
- [A Survey on Mechanistic Interpretability for Multi-Modal Foundation Models](https://mlsanity.com/paper/2502.17516/) (2025)
- [What Do VLMs NOTICE? A Mechanistic Interpretability Pipeline for Gaussian-Noise-free Text-Image Corruption and Evaluation](https://mlsanity.com/paper/2406.16320/) (2024)
- [LVLM-Interpret: An Interpretability Tool for Large Vision-Language Models](https://mlsanity.com/paper/2404.03118/) (2024)
- [Where do Large Vision-Language Models Look at when Answering Questions?](https://mlsanity.com/paper/2503.13891/) (2025)

## Built on by (most cited first, 8 of 8)

- [When Sinks Help or Hurt: Unified Framework for Attention Sink in Large Vision-Language Models](https://mlsanity.com/paper/2604.03316/) (2026)
- [Mechanistic Insights into Functional Sparsity in Multimodal LLMs via CoRe Heads](https://mlsanity.com/paper/2606.05843/) (2026)
- [When Structured Sparse Autoencoders Learn Consistent Concepts Across Modalities](https://mlsanity.com/paper/2607.08605/) (2026)
- [ProjLens: Unveiling the Role of Projectors in Multimodal Model Safety](https://mlsanity.com/paper/2604.19083/) (2026)
- [Multimodal Knowledge Edit-Scoped Generalization for Online Recursive MLLM Editing](https://mlsanity.com/paper/2607.01978/) (2026)
- [EDCT-Bench: Uncovering Faithfulness Gaps in VLMs via Explanation-Driven Counterfactual Testing](https://mlsanity.com/paper/2609.17953/) (2026)
- [Who Is Left of Whom? Tracing Spatial Evidence and Role Binding in Relative-Position Reasoning](https://mlsanity.com/paper/2609.35486/) (2026)
- [Cross-Modal Latent Influence: Patch-Level Training Sample Attribution for Vision-Language Models](https://mlsanity.com/paper/2610.14190/) (2026)

## Abstract (arXiv)

Recent advances in causal interpretability have extended from language models to vision-language models (VLMs), seeking to reveal their internal mechanisms through input interventions. While textual interventions often target semantics, visual interventions typically rely on coarse pixel-level perturbations, limiting semantic insights on multimodal integration. In this study, we introduce V-SEAM, a novel framework that combines Visual Semantic Editing and Attention Modulating for causal interpretation of VLMs. V-SEAM enables concept-level visual manipulations and identifies attention heads with positive or negative contributions to predictions across three semantic levels: objects, attributes, and relationships. We observe that positive heads are often shared within the same semantic level but vary across levels, while negative heads tend to generalize broadly. Finally, we introduce an automatic method to modulate key head embeddings, demonstrating enhanced performance for both LLaVA and InstructBLIP across three diverse VQA benchmarks. Our data and code are released at: https://github.com/petergit1/V-SEAM.