We have long known that the brain is a mapmaker. From the simple mapping of visceral states in the brainstem to the elaborate representations of the external world in the cerebral cortex, our consciousness relies on these internal topographies. But language presents a peculiar puzzle. When a person speaks two languages, how does the brain prevent them from collapsing into a chaotic, untranslatable noise? How does the mind know that 'dog' and 'perro' refer to the exact same creature of flesh, bone, and loyalty, without getting the two signals tangled?
For years, the standard explanation was that we must possess some sort of master translator neurons—cells that fire whenever the concept itself is accessed, regardless of the language used to express it. It was a neat, comforting theory. But it was wrong.
A remarkable study published in the journal Cell by researchers at the Baylor College of Medicine—including Xinyuan Yan, Sameer Sheth, and Benjamin Hayden—tears up that neat script. By peering directly into the human hippocampus at the resolution of single cells, they discovered a reality both more complex and far more beautiful. The brain does not translate in the way a computer dictionary does. Instead, it relies on a shared, underlying geometry.
The Rare Glimpse Into the Living Brain
To understand this discovery, we have to look closely at the methodology. Recording clean, real-time activity from individual neurons inside a living, thinking human brain is incredibly difficult. You cannot simply open up a healthy subject to satisfy scientific curiosity. The researchers had to wait for a rare clinical window: patients undergoing evaluation for severe, treatment-resistant epilepsy.
Four early-stage English-Spanish bilinguals, who had spoken both languages fluently since childhood, had high-precision electrodes temporarily implanted in their temporal lobes, specifically the hippocampus, to locate the origin of their seizures. This clinical setup allowed the team to record the electrical crackle of single neurons while the participants listened, read, and conversed naturally in both English and Spanish.
When the researchers looked at how these single cells behaved, they did not find a large population of universal "concept neurons." They expected cells that would fire equally whether the patient heard "dog" or "perro." Instead, they found almost nothing of the sort. Only a tiny fraction of cells showed any cross-linguistic correlation. Overwhelmingly, individual neurons were stubbornly language-specific. A neuron that fired enthusiastically for "dog" was silent when the word "perro" was spoken. The hardware, it turned out, is segregated. We do not have a bilingual dictionary written into individual cells.
The Geometry of Meaning
This brings us to the core mystery. If individual neurons are language-specific, how is meaning preserved across languages? If one set of cells handles Spanish and another handles English, what prevents the bilingual mind from splitting into two completely disconnected conceptual universes?
The answer lies in what the researchers call shared neural geometries.
Instead of relying on single "bridge" neurons, the hippocampus maps entire networks of words based on their semantic relationships. Think of it not as a dictionary, but as a three-dimensional landscape. In the English-language map, the word "dog" is located very close to "wolf" and "coyote," because they are semantically related. The word "fork" is located far away, across a valley of unrelated meanings.
This map of relationships is not arbitrary. When the researchers examined the Spanish-language map, they found a stunning structural parallel. [This structural parallel informs broader research on how belief-based mental models are updated, a topic explored in our work on belief stickiness.] The word "perro" was located relative to "lobo" (wolf) and "tenedor" (fork) in the exact same configuration as the English words. The spatial relationships—the geometry—remained identical.
This is the secret of translation. As senior author Sameer Sheth noted, it is like looking into the same room through two different windows. The furniture is in the exact same place, even if your angle of view is different. By measuring the coordinates of English words, the team could mathematically predict where the equivalent Spanish words would land in the neural space. The conceptual map is universal; only the labels are local.
The Collective Choir of the Hippocampus
This finding changes how we think about neural computation. The brain is not a collection of isolated switches. It is an ensemble.
First author Xinyuan Yan describes this mechanism as a form of distributed population coding. In this view, individual cells are like members of a choir. One singer might have a high register, another a deep bass. No single singer represents the harmony of a song on their own. But when they sing together, their combined voices create a complex chord. The hippocampus represents "dog" and "perro" using two different choirs—different groups of language-specific neurons—but both choirs sing the exact same semantic chord.
This is an incredibly elegant solution to a difficult computational problem. If the brain used the same neurons for both languages, the risk of interference would be high. You might start a sentence in Spanish and accidentally slip into English because the shared cells became confused. By utilizing distinct readout axes for the exact same underlying geometry, the brain can keep the two streams separate while maintaining a unified sense of meaning. It allows for swift, effortless code-switching without cognitive spillover.
Pre-Wired for the World: The AI Parallel
This brings us to a fascinating connection with artificial intelligence. The researchers decided to compare their biological data with mBERT, a large language model trained to understand over a hundred languages.
Unlike simple translation programs, mBERT does not translate word-for-word. Instead, it places words into a high-dimensional vector space. When the researchers analyzed the geometric relationships in mBERT's vector space, they found a structure that was strikingly similar to the human hippocampus. Both the natural brain and the artificial network converged on the same solution: representing meaning as relationships in space.
This suggests a profound truth about human nature. [This adaptability, much like the patterns found in AI-assisted brainwave analysis, highlights the importance of analyzing neural baseline patterns.] The brain is not a blank slate that must build a new conceptual framework for every language it learns. It is intrinsically pre-wired for multilingualism. Once a child builds a core map of how the world works—how objects, actions, and feelings relate to one another—that map becomes the foundation. Learning a second or third language is not a matter of building a new map from scratch. It is simply a matter of projecting a new set of labels onto the existing topography. The geography is already there. We only need to learn the new names for the mountains.
The Deep Roots of the Self
What does this tell us about the nature of mind and self?
As someone who has spent decades studying how the brain maps the body and emotions to create the conscious self, I find these results deeply satisfying. We often treat language as the ultimate expression of human intelligence, a clean, abstract realm separate from our animal origins. But these findings suggest otherwise.
The hippocampus is an ancient structure. Long before it was mapping Spanish and English nouns, it was mapping physical space—helping our ancestors locate food, remember pathways, and navigate physical landscapes. The brain has taken this raw, spatial mapping hardware and co-opted it to map the landscape of thought itself. Our conceptual structures are literally built out of spatial code.
More than that, this tells us that our shared human experience runs far deeper than the superficial barriers of language. Whether we speak English, Spanish, or any of the thousands of tongues spoken across this planet, our brains are navigating the same internal landscape of meaning. We are all walking the same paths, using the same universal cartography of the mind, even if we describe the journey in different words. The bilingual brain is not two minds crammed into one skull; it is a single, deeply unified mind viewing a single, beautiful world through more than one window.