Alignment between Brains and AI: Evidence for Convergent Evolution across Modalities, Scales and Training Trajectories

Guobin Shen, Dongcheng Zhao, Yiting Dong, Qian Zhang, Yi Zeng
Proceedings of the 43rd International Conference on Machine Learning, PMLR 306:111522-111541, 2026.

Abstract

Artificial and biological systems may converge on similar computational strategies despite different architectures and learning mechanisms—a form of convergent evolution. We test this at scale by comparing internal representations of 630 AI models (language and vision; 1.33M–72B parameters) against fMRI from the Natural Scenes Dataset, producing over 60 million alignment measurements. Within each modality, higher-performing models spontaneously develop stronger brain correspondence (language: r = 0.89; vision: r = 0.53); because the inputs are image-evoked, the language results reflect visual-semantic alignment rather than a direct cross-modal comparison. Longitudinal analysis combined with bidirectional Granger tests further shows that past alignment predicts future performance more reliably than the reverse, identifying brain-like representations as a robust early-emerging correlate of learning. Modality-specific organization also emerges: language models align with limbic and integrative regions, vision models with visual cortical hierarchies.

Cite this Paper

BibTeX
@InProceedings{pmlr-v306-shen26ae, title = {Alignment between Brains and {AI}: Evidence for Convergent Evolution across Modalities, Scales and Training Trajectories}, author = {Shen, Guobin and Zhao, Dongcheng and Dong, Yiting and Zhang, Qian and Zeng, Yi}, booktitle = {Proceedings of the 43rd International Conference on Machine Learning}, pages = {111522--111541}, 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/shen26ae/shen26ae.pdf}, url = {https://proceedings.mlr.press/v306/shen26ae.html}, abstract = {Artificial and biological systems may converge on similar computational strategies despite different architectures and learning mechanisms—a form of convergent evolution. We test this at scale by comparing internal representations of 630 AI models (language and vision; 1.33M–72B parameters) against fMRI from the Natural Scenes Dataset, producing over 60 million alignment measurements. Within each modality, higher-performing models spontaneously develop stronger brain correspondence (language: r = 0.89; vision: r = 0.53); because the inputs are image-evoked, the language results reflect visual-semantic alignment rather than a direct cross-modal comparison. Longitudinal analysis combined with bidirectional Granger tests further shows that past alignment predicts future performance more reliably than the reverse, identifying brain-like representations as a robust early-emerging correlate of learning. Modality-specific organization also emerges: language models align with limbic and integrative regions, vision models with visual cortical hierarchies.} }
Endnote
%0 Conference Paper %T Alignment between Brains and AI: Evidence for Convergent Evolution across Modalities, Scales and Training Trajectories %A Guobin Shen %A Dongcheng Zhao %A Yiting Dong %A Qian Zhang %A Yi Zeng %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-shen26ae %I PMLR %P 111522--111541 %U https://proceedings.mlr.press/v306/shen26ae.html %V 306 %X Artificial and biological systems may converge on similar computational strategies despite different architectures and learning mechanisms—a form of convergent evolution. We test this at scale by comparing internal representations of 630 AI models (language and vision; 1.33M–72B parameters) against fMRI from the Natural Scenes Dataset, producing over 60 million alignment measurements. Within each modality, higher-performing models spontaneously develop stronger brain correspondence (language: r = 0.89; vision: r = 0.53); because the inputs are image-evoked, the language results reflect visual-semantic alignment rather than a direct cross-modal comparison. Longitudinal analysis combined with bidirectional Granger tests further shows that past alignment predicts future performance more reliably than the reverse, identifying brain-like representations as a robust early-emerging correlate of learning. Modality-specific organization also emerges: language models align with limbic and integrative regions, vision models with visual cortical hierarchies.
APA
Shen, G., Zhao, D., Dong, Y., Zhang, Q. & Zeng, Y.. (2026). Alignment between Brains and AI: Evidence for Convergent Evolution across Modalities, Scales and Training Trajectories. Proceedings of the 43rd International Conference on Machine Learning, in Proceedings of Machine Learning Research 306:111522-111541 Available from https://proceedings.mlr.press/v306/shen26ae.html.

Related Material