# Is CLIP Cross-Eyed? Revealing and Mitigating Center Bias in the CLIP Family

arXiv: https://arxiv.org/abs/2604.05971 · HTML: https://mlsanity.com/paper/2604.05971/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Placed in Prompt learning and adaptation of CLIP, one of 16 directions in Multimodal LLMs / vision-language models.
- Builds on 23 of its 31 references that are in the map.
- First posted on arXiv 7 Apr 2026.

## Placed in

- Multimodal LLMs / vision-language models / [Prompt learning and adaptation of CLIP](https://mlsanity.com/mllm/prompt-learning-and-adaptation-of-clip/)

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

- [Learning Transferable Visual Models From Natural Language Supervision](https://mlsanity.com/paper/2103.00020/) (2021)
- [Fashion-MNIST: a Novel Image Dataset for Benchmarking Machine Learning Algorithms](https://mlsanity.com/paper/1708.07747/) (2017)
- [Grounding DINO: Marrying DINO with Grounded Pre-Training for Open-Set Object Detection](https://mlsanity.com/paper/2303.05499/) (2023)
- [Sigmoid Loss for Language Image Pre-Training](https://mlsanity.com/paper/2303.15343/) (2023)
- [Describing Textures in the Wild](https://mlsanity.com/paper/1311.3618/) (2013)
- [CoCa: Contrastive Captioners are Image-Text Foundation Models](https://mlsanity.com/paper/2205.01917/) (2022)
- [Reproducible scaling laws for contrastive language-image learning](https://mlsanity.com/paper/2212.07143/) (2022)
- [Eyes Wide Shut? Exploring the Visual Shortcomings of Multimodal LLMs](https://mlsanity.com/paper/2401.06209/) (2024)
- [When and why vision-language models behave like bags-of-words, and what to do about it?](https://mlsanity.com/paper/2210.01936/) (2022)
- [EVA-02: A Visual Representation for Neon Genesis](https://mlsanity.com/paper/2303.11331/) (2023)
- [What's "up" with vision-language models? Investigating their struggle with spatial reasoning](https://mlsanity.com/paper/2310.19785/) (2023)
- [SugarCrepe: Fixing Hackable Benchmarks for Vision-Language Compositionality](https://mlsanity.com/paper/2306.14610/) (2023)
- [What does CLIP know about a red circle? Visual prompt engineering for VLMs](https://mlsanity.com/paper/2304.06712/) (2023)
- [Vision language models are blind: Failing to translate detailed visual features into words](https://mlsanity.com/paper/2407.06581/) (2024)
- [Interpreting CLIP with Sparse Linear Concept Embeddings (SpLiCE)](https://mlsanity.com/paper/2402.10376/) (2024)
- [SUGARCREPE++ Dataset: Vision-Language Model Sensitivity to Semantic and Lexical Alterations](https://mlsanity.com/paper/2406.11171/) (2024)
- [ViTamin: Designing Scalable Vision Models in the Vision-Language Era](https://mlsanity.com/paper/2404.02132/) (2024)
- [A Sober Look at the Robustness of CLIPs to Spurious Features](https://mlsanity.com/paper/2403.11497/) (2024)
- [CLIP Behaves like a Bag-of-Words Model Cross-modally but not Uni-modally](https://mlsanity.com/paper/2502.03566/) (2025)
- [Zero-Shot Robustification of Zero-Shot Models](https://mlsanity.com/paper/2309.04344/) (2023)
- [DetailCLIP: Detail-Oriented CLIP for Fine-Grained Tasks](https://mlsanity.com/paper/2409.06809/) (2024)
- [Can CLIP Count Stars? An Empirical Study on Quantity Bias in CLIP](https://mlsanity.com/paper/2409.15035/) (2024)
- [SuperCLIP: CLIP with Simple Classification Supervision](https://mlsanity.com/paper/2512.14480/) (2025)

## Abstract (arXiv)

Recent research has shown that contrastive vision-language models such as CLIP often lack fine-grained understanding of visual content. While a growing body of work has sought to address this limitation, we identify a distinct failure mode in the CLIP family, which we term center bias, that persists even in recent model variants. Specifically, CLIP tends to disproportionately focus on the central region of an image, overlooking important objects located near the boundaries. This limitation is fundamental as failure to recognize relevant objects makes it difficult to perform any sophisticated tasks that depend on those objects. To understand the underlying causes of the limitation, we conduct analyses from both representation and attention perspectives. Using interpretability methods, i.e., embedding decomposition and attention map analysis, we find that relevant concepts especially those associated with off-center objects vanish from the model's embedding in the final representation due to information loss during the aggregation of visual embeddings, particularly the reliance on pooling mechanisms. Finally, we show that this bias can be alleviated with training-free strategies such as visual prompting and attention redistribution by redirecting models' attention to off-center regions.