From Language to Cognition: How LLMs Outgrow the Human Language Network

2025-03-03

AI Summary

This study investigates how neural representations in large language models (LLMs) evolve during training toward alignment with the human language brain network, and how such alignment relates to formal linguistic competence (syntactic rule knowledge) versus functional linguistic competence (world knowledge and reasoning). Method: Leveraging 34 training checkpoints spanning 300B tokens and eight model scales, we integrate fMRI/MEG brain activity alignment analysis, multi-scale neural language evaluation, and decoupled assessment of formal and functional capabilities. Contribution/Results: We first demonstrate that brain–model alignment tracks formal linguistic competence more closely than functional competence; model scale is not the primary driver of alignment; and the alignment–performance correlation sharply diminishes beyond human-level linguistic capability. These findings establish that the human language brain network prioritizes formal structural encoding. We further propose the first rigorously parameter-controlled framework for analyzing brain–model alignment–capability relationships, advancing cognitive modeling of language models.

Technology Category

• Natural Language Processing: (Large) Language Models
• Machine Learning: Large Multimodal Models (LMMs)
• Humans and AI: Brain-Sensing and Analysis

Application Category

• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• User Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs

Abstract

Large language models (LLMs) exhibit remarkable similarity to neural activity in the human language network. However, the key properties of language shaping brain-like representations, and their evolution during training as a function of different tasks remain unclear. We here benchmark 34 training checkpoints spanning 300B tokens across 8 different model sizes to analyze how brain alignment relates to linguistic competence. Specifically, we find that brain alignment tracks the development of formal linguistic competence -- i.e., knowledge of linguistic rules -- more closely than functional linguistic competence. While functional competence, which involves world knowledge and reasoning, continues to develop throughout training, its relationship with brain alignment is weaker, suggesting that the human language network primarily encodes formal linguistic structure rather than broader cognitive functions. We further show that model size is not a reliable predictor of brain alignment when controlling for feature size and find that the correlation between next-word prediction, behavioral alignment and brain alignment fades once models surpass human language proficiency. Finally, using the largest set of rigorous neural language benchmarks to date, we show that language brain alignment benchmarks remain unsaturated, highlighting opportunities for improving future models. Taken together, our findings suggest that the human language network is best modeled by formal, rather than functional, aspects of language.

Problem

• LLMs' brain alignment with human language
• Formal vs. functional linguistic competence
• Model size and brain alignment correlation

Innovation

• Benchmarking 34 training checkpoints
• Analyzing brain alignment evolution
• Using neural language benchmarks