The emergence of the left-right asymmetry in predicting brain activity from LLMs' representations specifically correlates with their formal linguistic competence

ICLR 2026 Conference Submission18896 Authors

Keywords: large language models, training, language processing, brain lateralization, fMRI, phase transition, syntax

TL;DR: As training progresses, the internal representations of an LLM become increasingly better at predicting brain activity, with a sudden increase in asymmetry between left and right hemispheres coinciding with the LLM's acquisition of formal competence.

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, 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 the origin of this left-right asymmetry. Using the OLMo-2 7B language model, whose authors provide checkpoints during training, and fMRI data from English participants, we track the evolution of the left-right asymmetry and compare it with the evolution of the performance on several linguistic and non-linguistic benchmarks. We observe that the asymmetry co-emerges with the linguistic abilities of the LLM, attested either by its capacity to make accurate acceptability judgments on English sentences or to produce well-formed text. In addition, this change in left-right asymmetry does not correlate at all with arithmetic or formal language (Dyck) tasks, nor with text-based tasks involving world knowledge and reasoning. We generalize these results to another family of LLMs (Pythia) and another language, namely French. Our observations indicate that the left-right asymmetry in brain predictivity matches the progress in formal competence (knowledge of linguistic patterns) rather than functional competence (understanding and using language in the world).

Primary Area: applications to neuroscience & cognitive science

Submission Number: 18896