Uncovering Grounding IDs: How External Cues Shape Multi-Modal Binding

Amirmohammad Izadi, Hosein Hasani, Fatemeh Askari, Mobin Bagherian, Sadegh Mohammadian, Mohammad Izadi, Mahdieh Soleymani Baghshah
Proceedings of the 43rd International Conference on Machine Learning, PMLR 306:50205-50229, 2026.

Abstract

Large vision–language models (LVLMs) perform well on multimodal tasks, but their ability to reason and precisely align visual and textual information still has room for improvement. In this study, we show that external visual cues, such as symbols or grid lines, help LVLMs form more accurate connections between visual components, such as objects, and their corresponding textual descriptions, improving their grounding and reasoning abilities. We introduce the concept of Grounding IDs, which are latent identifiers that arise within the model as a result of external cues structuring both visual and textual modalities. Our analysis reveals that partition-inducing external cues lead to Grounding IDs that make better alignment between corresponding visual and text representations, helping the model focus on relevant information. We find that Grounding IDs enhance attention between related components, improving cross-modal grounding and reducing hallucinations. Overall, our results show that Grounding IDs are a key mechanism that enables external cues to improve cross-modal alignment, reduce errors, and enhance the overall performance of LVLMs across a range of multimodal tasks.

Cite this Paper

BibTeX
@InProceedings{pmlr-v306-izadi26a, title = {Uncovering Grounding {ID}s: How External Cues Shape Multi-Modal Binding}, author = {Izadi, Amirmohammad and Hasani, Hosein and Askari, Fatemeh and Bagherian, Mobin and Mohammadian, Sadegh and Izadi, Mohammad and Baghshah, Mahdieh Soleymani}, booktitle = {Proceedings of the 43rd International Conference on Machine Learning}, pages = {50205--50229}, year = {2026}, editor = {Zhang, Tong and Dudik, Miroslav and Jaggi, Martin and Agarwal, Alekh and Li, Sharon and Schuurmans, Dale and Zhu, Jerry and Berkenkamp, Felix and Dong, Hanze and Bietti, Alberto}, volume = {306}, series = {Proceedings of Machine Learning Research}, month = {06--11 Jul}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v306/main/assets/izadi26a/izadi26a.pdf}, url = {https://proceedings.mlr.press/v306/izadi26a.html}, abstract = {Large vision–language models (LVLMs) perform well on multimodal tasks, but their ability to reason and precisely align visual and textual information still has room for improvement. In this study, we show that external visual cues, such as symbols or grid lines, help LVLMs form more accurate connections between visual components, such as objects, and their corresponding textual descriptions, improving their grounding and reasoning abilities. We introduce the concept of Grounding IDs, which are latent identifiers that arise within the model as a result of external cues structuring both visual and textual modalities. Our analysis reveals that partition-inducing external cues lead to Grounding IDs that make better alignment between corresponding visual and text representations, helping the model focus on relevant information. We find that Grounding IDs enhance attention between related components, improving cross-modal grounding and reducing hallucinations. Overall, our results show that Grounding IDs are a key mechanism that enables external cues to improve cross-modal alignment, reduce errors, and enhance the overall performance of LVLMs across a range of multimodal tasks.} }
Endnote
%0 Conference Paper %T Uncovering Grounding IDs: How External Cues Shape Multi-Modal Binding %A Amirmohammad Izadi %A Hosein Hasani %A Fatemeh Askari %A Mobin Bagherian %A Sadegh Mohammadian %A Mohammad Izadi %A Mahdieh Soleymani Baghshah %B Proceedings of the 43rd International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2026 %E Tong Zhang %E Miroslav Dudik %E Martin Jaggi %E Alekh Agarwal %E Sharon Li %E Dale Schuurmans %E Jerry Zhu %E Felix Berkenkamp %E Hanze Dong %E Alberto Bietti %F pmlr-v306-izadi26a %I PMLR %P 50205--50229 %U https://proceedings.mlr.press/v306/izadi26a.html %V 306 %X Large vision–language models (LVLMs) perform well on multimodal tasks, but their ability to reason and precisely align visual and textual information still has room for improvement. In this study, we show that external visual cues, such as symbols or grid lines, help LVLMs form more accurate connections between visual components, such as objects, and their corresponding textual descriptions, improving their grounding and reasoning abilities. We introduce the concept of Grounding IDs, which are latent identifiers that arise within the model as a result of external cues structuring both visual and textual modalities. Our analysis reveals that partition-inducing external cues lead to Grounding IDs that make better alignment between corresponding visual and text representations, helping the model focus on relevant information. We find that Grounding IDs enhance attention between related components, improving cross-modal grounding and reducing hallucinations. Overall, our results show that Grounding IDs are a key mechanism that enables external cues to improve cross-modal alignment, reduce errors, and enhance the overall performance of LVLMs across a range of multimodal tasks.
APA
Izadi, A., Hasani, H., Askari, F., Bagherian, M., Mohammadian, S., Izadi, M. & Baghshah, M.S.. (2026). Uncovering Grounding IDs: How External Cues Shape Multi-Modal Binding. Proceedings of the 43rd International Conference on Machine Learning, in Proceedings of Machine Learning Research 306:50205-50229 Available from https://proceedings.mlr.press/v306/izadi26a.html.

Related Material