From molecules to mind.
Model Torii is the world model Eybna is building for smell and taste, predicting what a molecular environment does to a person, and designing what should come next.
One relationship,
read in two directions.
The Core Prediction Problem
Rich in observation.
Ready for prediction.
Science already measures many pieces of the system. EEG captures neural dynamics. GC-MS maps chemistry. Psychophysics records preference. Physiology tracks autonomic response.
The problem is fragmentation. Each discipline publishes in isolation. None of them form one integrated predictive model.
What We’re Building
One representation.
Many questions.
Most models are narrow. Each one answers one question: one molecule, one marker, one context. The result is a library of unconnected predictions.
Model Torii learns a shared representation, a biological map of the chemical senses, from which many different questions can be asked of the same underlying knowledge.
The goal
Computational olfactory biology.
This is the field Model Torii belongs to. Molecular science, olfaction, neuroscience, physiology, human experimentation, and machine learning converge into one integrated system.
The Engine
Every experiment
sharpens the next.
Model Torii is not trained once and deployed. Every formulation Eybna runs through human studies becomes a probe. The model designs the experiment. The experiment improves the model. Each cycle, uncertainty decreases. The map expands.
Plant compounds mapped
Human receptor systems
Measurement layers
Production sites: IL · US · DE
From molecules to mind.
From measurement to prediction.
From prediction to design.
Model Torii is the world model Eybna is building for the chemical senses. Its goal is to understand the relationship between molecular inputs and human state well enough to predict, and eventually to design.
We are early. The loop is running. Every experiment sharpens the next.