Left-right asymmetry in predicting brain activity from LLMs' representations emerges with their formal linguistic competence
#Large Language Models #Brain Activity #Left-right Asymmetry #fMRI #Linguistic Competence #AI-Human Comparison #Neural Correlates
📌 Key Takeaways
- LLMs' internal activations correlate with human brain activity during text processing
- Left hemisphere predictions improve more than right hemisphere as LLM training progresses
- The asymmetry emerges specifically when models develop formal linguistic competence
- Research aims to understand factors contributing to this hemispheric difference
📖 Full Retelling
Researchers at an unspecified institution have published a groundbreaking study (arXiv:2602.12811v1) in February 2026 revealing that large language models (LLMs) exhibit left-right asymmetry in predicting human brain activity, with the left hemisphere showing stronger correlations as the models' training progresses, aiming to understand the fundamental differences between artificial and human language processing mechanisms. The study demonstrates that when humans and LLMs process identical text, the internal activations within the AI models correlate with brain activity measured through techniques like functional magnetic resonance imaging (fMRI). Notably, researchers discovered that as LLM training advances, the models' ability to predict brain activity from their internal representations improves significantly more in the left hemisphere compared to the right hemisphere, suggesting a specialized relationship between artificial neural networks and human brain lateralization. This asymmetry emerges specifically when the models develop formal linguistic competence, indicating that the way AI learns language may mirror, in some respects, the specialized processing that occurs in the human brain, particularly in the left hemisphere which is traditionally associated with language processing in humans.
🏷️ Themes
Artificial Intelligence, Neuroscience, Language Processing
📚 Related People & Topics
Large language model
Type of machine learning model
A large language model (LLM) is a language model trained with self-supervised machine learning on a vast amount of text, designed for natural language processing tasks, especially language generation. The largest and most capable LLMs are generative pre-trained transformers (GPTs) that provide the c...
Entity Intersection Graph
Connections for Large language model:
🌐
Educational technology
4 shared
🌐
Reinforcement learning
3 shared
🌐
Machine learning
2 shared
🌐
Artificial intelligence
2 shared
🌐
Benchmark
2 shared
Original Source
arXiv:2602.12811v1 Announce Type: cross
Abstract: When humans and large language models (LLMs) process the same text, activations in the LLMs correlate with brain activity measured, e.g., with functional magnetic resonance imaging (fMRI). Moreover, it has been shown that, as the training of an LLM progresses, the performance in predicting brain activity from its internal activations improves more in the left hemisphere than in the right one. The aim of the present work is to understand which ki
Read full article at source