Skip to content

James Belanger

Research Technician · Hayden Lab, Baylor College of Medicine

Computational neuroscientist studying the geometry of language in single-neuron and LFP recordings of the human brain.

James Belanger
Download CV Google Scholar ORCID GitHub Email

I study how the human brain represents language. My work measures the geometry of syntax and semantics in single-neuron population codes and asks a question that bridges neuroscience and AI: can the brain’s compressed, low-dimensional language code serve as a design principle for more efficient, compositional machine-learning models?

My path was deliberate: a B.A. in Cognitive Sciences (Psychology and Linguistics) with a Data Science minor at Rice gave me the formal structure of language; a research position in the Hayden Lab at Baylor College of Medicine turned that into quantitative, code-driven neuroscience — building real Poisson encoding and LLM-alignment pipelines on human intracranial data.

syntax semantics
A rotating three-dimensional point cloud: two semi-orthogonal subspaces, one for syntax and one for semantics, sharing a single dominant low-dimensional population axis.

Syntax and semantics occupy semi-orthogonal subspaces along a single low-dimensional hippocampal population axis — the geometry at the heart of my research.

Research

All research →

The Language Manifold: how the hippocampus encodes syntax and semantics

In recordings from human hippocampus, syntactic and semantic information are not spread across separate, distributed populations — they are written into semi-orthogonal subspaces along a single, strikingly low-dimensional population axis. The brain’s solution is more compressed than the internal geometry of any state-of-the-art language model I compared it against. This is the core of my research: characterizing that geometry and asking what it teaches us about efficient computation.

Aligning language models to the brain (QA-Emb)

I built a question-answer embedding framework (QA-Emb) that interrogates the hidden states of large language models and aligns them against hippocampal population geometry. Across ten models, syntax is read out from shallower layers than semantics — a depth gap that survives length-controlled, grain-matched comparison in 9 of 10 models.

Poisson encoding across domains: music and grammar

The same population-geometry toolkit generalizes beyond English narrative. In a piano-listening task I characterized 704 hippocampal neurons and found that the Krumhansl–Kessler tonal hierarchy — not the Circle of Fifths — best predicts their geometry, while dissociating absolute from relative pitch. In bilingual listeners, SVM decoders read grammatical gender and conjugation from Spanish-evoked activity.

Selected publications

All publications →