For decades, textbooks repeated a tidy claim: the human nose could distinguish roughly 10,000 different smells. That number, never rigorously tested, shaped how scientists and physicians thought about olfaction. A peer-reviewed paper published in Science dismantled that assumption with experimental data showing humans can discriminate more than 1 trillion olfactory stimuli. The finding did not just revise an old estimate upward. It multiplied it by a factor of roughly 100 million, raising urgent questions about how smell testing could be used to detect neurological disease far earlier than current screening tools allow.
Why a trillion-smell threshold changes neurological screening
Smell loss ranks among the earliest symptoms of Parkinson’s disease, Alzheimer’s disease, and post-COVID neurological complications. Clinicians currently rely on identification-based tests, where patients sniff standardized scents and try to name them. These tests measure recognition, not discrimination, and they capture only a fraction of the nose’s actual capacity. If the olfactory system can parse a trillion distinct stimuli, then a test asking patients to identify 40 labeled odors is sampling a vanishingly small slice of that range.
The gap between what the nose can do and what clinical tools actually measure creates a blind spot. A patient might correctly name common scents like coffee or peppermint while already losing the ability to tell apart closely overlapping odor mixtures. That subtler deficit, detectable only through discrimination tasks, could signal cognitive decline months or years before standard identification tests flag a problem. The hypothesis that individual differences in fine-grained odor discrimination will predict early cognitive decline more accurately than current smell-identification batteries has not been tested at population scale, but the experimental framework now exists to do so.
How 128 molecules produced a trillion-smell estimate
The Science paper, cataloged under DOI 10.1126/science.1249168, used a psychophysics approach built on a library of 128 odorant molecules. Researchers created complex mixtures containing 10, 20, or 30 of those components, then varied the degree of overlap between pairs of mixtures. Volunteers performed an odd-one-out task: they sniffed three vials, two containing the same mixture and one containing a different blend, and tried to pick the outlier.
Even when two mixtures shared most of their components, subjects could reliably tell them apart as long as the overlap stayed below a certain threshold. The team then extrapolated from the discrimination rates across all tested overlap levels to estimate the total number of distinguishable stimuli. The resulting lower bound exceeded 1 trillion olfactory stimuli, a figure that dwarfed the old 10,000-odor conventional wisdom. That older number, widely cited in neuroscience courses and popular science writing, had never been derived from controlled human experiments. It traced back to rough theoretical estimates from the 1920s that went largely unchallenged for the better part of a century.
The National Institutes of Health summarized the finding for a general audience, framing it as relevant to health and disease research. The institutional endorsement signaled that the result was not a statistical curiosity but a data point with direct implications for how federal agencies think about sensory health.
Unresolved gaps in the trillion-odor claim
The estimate rests on extrapolation from a specific experimental design, and several questions remain open. The study tested mixtures drawn from 128 molecules, but the real world contains thousands of volatile compounds in varying concentrations, temperatures, and humidity levels. Whether the trillion figure holds outside a controlled laboratory setting has not been confirmed by independent replication at population scale.
Raw trial-by-trial response data from the original odd-one-out sessions have not been publicly released in the archived records. Without that granular data, outside researchers cannot fully audit the statistical model that produced the trillion-smell estimate. A separate analysis published in eLife examined the dimensionality of odor space and raised methodological questions about how discrimination rates translate into total counts, though it did not dispute that the old 10,000 figure was far too low.
Individual variation adds another layer of complexity. The original experiments involved a small number of participants, and the published record does not break out performance differences by age, sex, health status, or prior olfactory training. If the trillion-smell threshold varies widely across people, that variation itself could become a diagnostic signal, but only if future studies measure it systematically.
For readers tracking the intersection of sensory science and brain health, the practical next step is straightforward. Current smell-identification tests used in neurological clinics were designed around a model of olfaction that the Science paper showed to be orders of magnitude too conservative. Any screening tool built on the assumption that humans distinguish only 10,000 odors is measuring a tiny window of a far larger capacity. The next generation of clinical smell tests will need to incorporate discrimination tasks, not just identification, to catch the earliest signs of decline. Until those tools reach routine clinical use, patients concerned about smell loss should report even subtle changes to their physicians, because the nose is telling a more detailed story than most diagnostic instruments are equipped to hear.
More from Morning Overview
*This article was researched with the help of AI, with human editors creating the final content.
