December 22, 2025 – The human brain processes spoken language in a sequence that closely mirrors the structure of advanced artificial intelligence (AI) language models, according to a new study that challenges rule-based theories of language comprehension and introduces a publicly available neural dataset for studying how the brain constructs meaning. 

In a study published in Nature Communications, researchers led by Dr. Ariel Goldstein of the Hebrew University of Jerusalem Department of Cognitive and Brain Sciences, uncovered a surprising connection between the way our brains make sense of spoken language and the way advanced AI models analyze text. 

What the Study Found 

When we listen to someone speak, our brain transforms each incoming word through a cascade of neural computations. Goldstein’s team discovered that these transformations unfold over time in a pattern that parallels the tiered layers of AI language models. Early AI layers track simple features of words, while deeper layers integrate context, tone, and meaning. The study found that human brain activity follows a similar progression: early neural responses aligned with early model layers, and later neural responses aligned with deeper layers. 

This alignment was especially clear in high-level language regions such as Broca’s area, where the peak brain response occurred later in time for deeper AI layers. According to Dr. Goldstein, “What surprised us most was how closely the brain’s temporal unfolding of meaning matches the sequence of transformations inside large language models. Even though these systems are built very differently, both seem to converge on a similar step-by-step buildup toward understanding.”  

The findings suggest that AI is not just a tool for generating text. It may also offer a new window to understand how the human brain processes meaning. For decades, scientists believed that language comprehension relied on symbolic rules and rigid linguistic hierarchies. However, this study challenges that view. Instead, it supports a more dynamic and statistical approach to language, in which meaning emerges gradually through layers of contextual processing. 

The researchers also found that classical linguistic features such as phonemes and morphemes did not predict the brain’s real-time activity as well as AI-derived contextual embeddings. This strengthens the idea that the brain integrates meaning in a more fluid and context-driven way than previously believed. 

A New Benchmark for Neuroscience 

To advance the field, the team publicly released the full dataset of neural recordings paired with linguistic features. This new resource enables scientists worldwide to test competing theories of how the brain understands natural language, paving the way for computational models that more closely resemble human cognition. 

The research paper titled “Temporal structure of natural language processing in the human brain corresponds to layered hierarchy of large language models” is now available in Nature Communications and can be accessed here.

Researchers:

Ariel Goldstein ^1,2,3, Eric Ham^4,5, Mariano Schain³, Samuel A. Nastase ⁴, Bobbi Aubrey^4,6, Zaid Zada ⁴, Avigail Grinstein-Dabush³, Harshvardhan Gazula⁴, Amir Feder³, Werner Doyle⁶, Sasha Devore⁶, Patricia Dugan⁶, Daniel Friedman⁶, Michael Brenner ^3,7, Avinatan Hassidim³, Yossi Matias ³, Orrin Devinsky ⁶, Noam Siegelman ^1,8, Adeen Flinker^6,9, Omer Levy¹⁰, Roi Reichart¹¹, Uri Hasson ^3,4

Institutions:

1. Department of Cognitive and Brain Sciences, Hebrew University, Jerusalem, Israel
2. Business School, Hebrew University, Jerusalem, Israel
3. Google Research, Tel-Aviv, Israel
4. Department of Psychology and the Neuroscience Institute, Princeton University, Princeton, NJ, USA.
5. Bioinformatics Interdepartmental Program, University of California, Los Angeles, Los Angeles, CA, USA
6. New York University Grossman School of Medicine, New York, NY, USA
7. School of Engineering and Applied Science, Harvard University, Cambridge, MA, USA
8. Department of Psychology, Hebrew University, Jerusalem, Israel
9. New York University Tandon School of Engineering, Brooklyn, NY, USA. 10Blavatnik School of Computer Science, Tel-Aviv University, Tel-Aviv, Israel
10. Technion—Israel Institute of Technology, Haifa, Israel