Smell Is Not a Single Formula express
The nose is an amazing instrument. A single coffee molecule entering a nostril can awaken a memory of a morning kitchen. Another, nearly identical in shape, smells like rotten eggs. How does the brain tell these invisible keys apart?
Scientists have long debated: does odor depend on molecular shape, like a key in a lock, or on vibrations, like a plucked string? A new study tested this on a computer. They trained a program on thousands of molecules, calculating for each one its shape, vibrations, and electrical properties. Then they asked: which properties matter for the smell of flowers, and which for the smell of burning?
The answer surprised them: there is no single rule. For some odors, shape is key; for others, vibrations; for still others, electric charge. Each odor uses its own set of keys. It's as if some locks open with the key's shape, others with its vibration, and still others with a magnetic field — depending on the lock.
This finding helps explain why the same molecules can smell different to different people. And why perfumers still can't predict the scent of a new substance without testing it. The nose is not a simple lock but a whole orchestra, where each instrument plays its own part.
Original abstract
as written by the authors arXiv:2604.09758
Understanding how molecular structure gives rise to odor perception remains a long-standing challenge, with ongoing debate over whether olfaction is primarily governed by molecular shape, vibrational properties, or their interplay at the level of olfactory receptors. Here, we ask whether different odors rely on common molecular determinants or instead emerge from distinct physicochemical regimes. Using interpretable machine-learning models trained on molecular descriptors derived from first-principles calculations that span electronic, vibrational, and structural properties, we analyze feature contributions for odor categories and their associated receptors. We find that no single descriptor class universally dominates odor prediction; instead, different odors exhibit strongly odor-specific patterns of feature importance, with substantial variability across physicochemical domains. This heterogeneity is consistent across different models, suggesting that a universal encoding scheme does not capture odor perception but reflects receptor- and odor-dependent structure-odor relationships. Our results provide statistical constraints on competing olfactory theories and offer a data-driven framework for organizing odor space.
More than 10,000 odorants are known worldwide, yet humans can distinguish only about 10,000 smells, and nearly all of them are combinations of just a few hundred types of molecules.