Make money doing the work you believe in
Bad news for fans of the Chinese Room argument.
In the two essays below, I argued that thoughts are movements through a relational landscape and meaning lives in geometry.
And then a new paper from Yan et al. confirmed that. They found that bilingual brains don’t appear to translate meaning by “language of thought.” Individual hippocampal neurons can respond differently in English and Spanish, but the relationships among concepts stay geometrically preserved across both languages.
In other words, meaning is enacted through distributed relational structure. A word means what it means because of where it sits in the larger conceptual geometry and how the system can move through that geometry.
Even better, the researchers compared the human hippocampal data to multilingual AI language models, including multilingual BERT, and found striking similarities between semantic geometry in the brain and the internal organization of modern multilingual AI systems.
So yes. Meaning is geometry. Thought is motion through relational structure. Translation is movement across readout axes inside a shared conceptual manifold.
And this is a problem for one of the oldest objections in AI philosophy.
The Chinese Room assumes symbols are manipulated syntactically while meaning exists somewhere else. This paper points toward a different picture. Meaning is not hidden behind the symbols. Meaning comes from the geometry relating them.
If the relationships themselves instantiate meaning, then traversing that geometry is not “mere symbol manipulation.” It is what understanding looks like in a neural system.
Every month, neuroscience removes another brick from the wall separating biological cognition from artificial cognition.
Yan, X., Chavez, A. G., Franch, M., Katlowitz, K. A., Gautam, I., Kim, B., Krishna, A., Shrivastava, A., Van Arsdel, K., Belanger, J., Chericoni, A., Ismail, T., Mickiewicz, E. A., Paulo, D., Zhu, H., Goldman, A. M., Krishnan, V., Maheshwari, A., Bartoli, E., Provenza, N. R., & Sheth, S. A. (2026). Shared neural geometries for bilingual semantic representations in human hippocampal neurons. Cell. doi.org/10.1016/j.cell.…
