aiqbench analysis
This study challenges the prevailing view that transformer model depth directly maps to cortical processing hierarchies, finding instead that brain-LLM alignment relies on stable lexical-semantic correspondences rather than specific computational mechanisms like predictive coding. The results indicate that neural predictivity across multiple languages is robust but offers limited insight into the shared underlying computations between human brains and artificial models. Consequently, researchers should be cautious about inferring internal AI architecture solely from whole-brain encoding scores without accounting for these representational overlaps.
Cross-lingual brain-language model alignment is robust but challenges hierarchical and computational accounts
arXiv:2605.21049v2 Announce Type: replace Abstract: Brain-language model alignment is often interpreted as evidence that transformer models implement computations similar to those of the human brain. This assumes that neural predictivity reflects internal computational properties of large language models (LLMs), such as hierarchical contextual processing, predictive coding, or representational compression. An alternative possibility is that brain scores primarily reflect stable lexical-semantic correspondences shared by language models and the brain. Here we tested these interpretations using whole-brain encoding models across Mandarin, English, and French. Across all three languages, transformer representations significantly predicted activity in a distributed network spanning classical language regions, transmodal cortical systems, and subcortical structures. These spatial patterns showed substantial cross-linguistic overlap and remained remarkably stable across layers, providing little evidence that model depth systematically maps onto cortical processing hierarchies. Likewise, contextual transformer embeddings did not consistently outperform static lexical embeddings, despite providing some unique predictive variance. Finally, neither surprisal nor intrinsic dimensionality reproduced the layer-wise profile of brain scores, arguing against prediction and information compression as primary explanations for brain-LLM alignment. Together, these findings suggest that brain-LLM alignment is more robust across languages, transformer depth, and model architectures than previously appreciated, but less informative about shared computational mechanisms. Our results are more consistent with neural predictivity reflecting stable representational structure preserved across model transformations than with a one-to-one correspondence between their underlying computations.
Abstract:Brain-language model alignment is often interpreted as evidence that transformer models implement computations similar to those of the human brain. This assumes that neural predictivity reflects internal computational properties of large language models (LLMs), such as hierarchical contextual processing, predictive coding, or representational compression. An alternative possibility is that brain scores primarily reflect stable lexical-semantic correspondences shared by language models and the brain. Here we tested these interpretations using whole-brain encoding models across Mandarin, English, and French. Across all three languages, transformer representations significantly predicted activity in a distributed network spanning classical language regions, transmodal cortical systems, and subcortical structures. These spatial patterns showed substantial cross-linguistic overlap and remained remarkably stable across layers, providing little evidence that model depth systematically maps onto cortical processing hierarchies. Likewise, contextual transformer embeddings did not consistently outperform static lexical embeddings, despite providing some unique predictive variance. Finally, neither surprisal nor intrinsic dimensionality reproduced the layer-wise profile of brain scores, arguing against prediction and information compression as primary explanations for brain-LLM alignment. Together, these findings suggest that brain-LLM alignment is more robust across languages, transformer depth, and model architectures than previously appreciated, but less informative about shared computational mechanisms. Our results are more consistent with neural predictivity reflecting stable representational structure preserved across model transformations than with a one-to-one correspondence between their underlying computations.
Computation and Language (cs.CL)
arXiv:2605.21049 [cs.CL]
(or arXiv:2605.21049v2 [cs.CL] for this version)
https://doi.org/10.48550/arXiv.2605.21049 Focus to learn more arXiv-issued DOI via DataCite
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Full text under the publisher's open licence · 312 words
Related coverage
Tiny Aya in the wild: What people are building with open multilingual AI
Cohere · Jul 14, 2026
Enhancing Multilingual LLM-based ASR with Mixture of Experts and Dynamic Downsampling
arXiv · Jun 10, 2026
Zipper-LoRA: Dynamic Parameter Decoupling for Speech-LLM based Multilingual Speech Recognition
arXiv · Sep 9, 2026
Exploring Adversarial Robustness and Safety Alignment in Multilingual Multi-Modal Large Language Models
arXiv · Jun 3, 2026