# Seeing past words: Testing the cross-modal capabilities of pretrained V&L models on counting tasks

arXiv: https://arxiv.org/abs/2012.12352 · HTML: https://mlsanity.com/paper/2012.12352/

## Key facts

- Cited by 34 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 22 Dec 2020.

## Built on by (most cited first, 30 of 34)

- [When and why vision-language models behave like bags-of-words, and what to do about it?](https://mlsanity.com/paper/2210.01936/) (2022)
- [Winoground: Probing Vision and Language Models for Visio-Linguistic Compositionality](https://mlsanity.com/paper/2204.03162/) (2022)
- [What's "up" with vision-language models? Investigating their struggle with spatial reasoning](https://mlsanity.com/paper/2310.19785/) (2023)
- [VALSE: A Task-Independent Benchmark for Vision and Language Models Centered on Linguistic Phenomena](https://mlsanity.com/paper/2112.07566/) (2021)
- [Vision-Language Pre-training: Basics, Recent Advances, and Future Trends](https://mlsanity.com/paper/2210.09263/) (2022)
- [Vision-and-Language or Vision-for-Language? On Cross-Modal Influence in Multimodal Transformers](https://mlsanity.com/paper/2109.04448/) (2021)
- [Recent Advances of Continual Learning in Computer Vision: An Overview](https://mlsanity.com/paper/2109.11369/) (2021)
- [MM-SHAP: A Performance-agnostic Metric for Measuring Multimodal Contributions in Vision and Language Models & Tasks](https://mlsanity.com/paper/2212.08158/) (2022)
- [MultiViz: Towards Visualizing and Understanding Multimodal Models](https://mlsanity.com/paper/2207.00056/) (2022)
- [DIME: Fine-grained Interpretations of Multimodal Models via Disentangled Local Explanations](https://mlsanity.com/paper/2203.02013/) (2022)
- [Text encoders bottleneck compositionality in contrastive vision-language models](https://mlsanity.com/paper/2305.14897/) (2023)
- [What Do VLMs NOTICE? A Mechanistic Interpretability Pipeline for Gaussian-Noise-free Text-Image Corruption and Evaluation](https://mlsanity.com/paper/2406.16320/) (2024)
- [On Explaining Multimodal Hateful Meme Detection Models](https://mlsanity.com/paper/2204.01734/) (2022)
- [ViLMA: A Zero-Shot Benchmark for Linguistic and Temporal Grounding in Video-Language Models](https://mlsanity.com/paper/2311.07022/) (2023)
- [Can Linguistic Knowledge Improve Multimodal Alignment in Vision-Language Pretraining?](https://mlsanity.com/paper/2308.12898/) (2023)
- [Generative Models as a Complex Systems Science: How can we make sense of large language model behavior?](https://mlsanity.com/paper/2308.00189/) (2023)
- [FINEMATCH: Aspect-based Fine-grained Image and Text Mismatch Detection and Correction](https://mlsanity.com/paper/2404.14715/) (2024)
- [DEAL: Disentangle and Localize Concept-level Explanations for VLMs](https://mlsanity.com/paper/2407.14412/) (2024)
- [Vision-Language Integration in Multimodal Video Transformers (Partially) Aligns with the Brain](https://mlsanity.com/paper/2311.07766/) (2023)
- [What Vision-Language Models `See' when they See Scenes](https://mlsanity.com/paper/2109.07301/) (2021)
- [Do Vision-and-Language Transformers Learn Grounded Predicate-Noun Dependencies?](https://mlsanity.com/paper/2210.12079/) (2022)
- [Finding Structural Knowledge in Multimodal-BERT](https://mlsanity.com/paper/2203.09306/) (2022)
- [ColorFoil: Investigating Color Blindness in Large Vision and Language Models](https://mlsanity.com/paper/2405.11685/) (2024)
- [MARVEL-40M+: Multi-Level Visual Elaboration for High-Fidelity Text-to-3D Content Creation](https://mlsanity.com/paper/2411.17945/) (2024)
- [Do Medical Vision Language Models Actually See? A Counterfactual Grounding Framework and Hard-Negative Contrastive Training for Visually-Reliant Medical VLMs](https://mlsanity.com/paper/2607.03647/) (2026)
- [The Role of Linguistic Priors in Measuring Compositional Generalization of Vision-Language Models](https://mlsanity.com/paper/2310.02777/) (2023)
- [TraVLR: Now You See It, Now You Don't! A Bimodal Dataset for Evaluating Visio-Linguistic Reasoning](https://mlsanity.com/paper/2111.10756/) (2021)
- [VAQUUM: Are Vague Quantifiers Grounded in Visual Data?](https://mlsanity.com/paper/2502.11874/) (2025)
- [Probing Cross-modal Semantics Alignment Capability from the Textual Perspective](https://mlsanity.com/paper/2210.09550/) (2022)
- [The Scenario Refiner: Grounding subjects in images at the morphological level](https://mlsanity.com/paper/2309.11252/) (2023)

## Abstract (arXiv)

We investigate the reasoning ability of pretrained vision and language (V&L) models in two tasks that require multimodal integration: (1) discriminating a correct image-sentence pair from an incorrect one, and (2) counting entities in an image. We evaluate three pretrained V&L models on these tasks: ViLBERT, ViLBERT 12-in-1 and LXMERT, in zero-shot and finetuned settings. Our results show that models solve task (1) very well, as expected, since all models are pretrained on task (1). However, none of the pretrained V&L models is able to adequately solve task (2), our counting probe, and they cannot generalise to out-of-distribution quantities. We propose a number of explanations for these findings: LXMERT (and to some extent ViLBERT 12-in-1) show some evidence of catastrophic forgetting on task (1). Concerning our results on the counting probe, we find evidence that all models are impacted by dataset bias, and also fail to individuate entities in the visual input. While a selling point of pretrained V&L models is their ability to solve complex tasks, our findings suggest that understanding their reasoning and grounding capabilities requires more targeted investigations on specific phenomena.