289 of 579 recorded neurons responded selectively to linguistic features, giving researchers a single-neuron map of language encoding across frontal, anterior temporal and posterior temporal cortex.
Left-hemisphere neurons showed significantly stronger modulation than right-hemisphere cells even though selective neurons were spread across both sides, pointing to distributed processing with clear hemispheric dominance.
Posterior temporal cortex showed the sharpest left-right gap, while prefrontal cortex produced the strongest overall modulation, suggesting regional specialization within that broader network.
Language-model-based analyses most accurately predicted activity in the left anterior temporal cortex, linking abstract model representations to neuronal responses during natural language processing.
The Nature study challenges narrow 'language center' models and could guide future work on language disorders, conversation and bilingual processing.
Have scientists found the biological source of human language, or just the brain's complex code for it?
If AI can decode words from single neurons, can it soon read our unspoken thoughts?
From Neuron to Network: How Single-Cell Language Mapping is Redefining Brain Science and AI
Overview
A groundbreaking study published in June 2026 used advanced neurotechnology to map language-encoding neurons in the human brain at single-cell resolution. By placing specialized microelectrode arrays in patients already undergoing neurosurgery for epilepsy, researchers were able to record precise neural activity directly from individual neurons in the frontotemporal cortex. This innovative approach provided a detailed look at how the brain processes language, offering new insights into the neural basis of speech. The findings mark a major step forward in understanding the complex mechanisms that enable human language.