Humans may tell apart a trillion smells, or far fewer

Humans may be able to tell apart more than a trillion different smells, according to a 2014 study in Science, but the figure is disputed: a 2015 reanalysis argued that the same data could support almost any number, and that the true count is still unknown. What is not in doubt is that the old textbook figure of 10,000 odors was never based on data.

That 10,000 estimate dates to the 1920s. To test it, Andreas Keller and Leslie Vosshall at Rockefeller University blended 128 odor molecules into random mixtures of 10, 20 or 30 components. Each of 26 volunteers made 264 comparisons, picking the odd one out from three vials.

From how often mixtures could be told apart, the team extrapolated to every possible blend and arrived at at least one trillion, against several million colors and almost half a million tones for sight and hearing.

The trillion is shaky, though. Neuroscientists Richard Gerkin and Jason Castro reran the calculation in eLife in 2015. Modest, reasonable changes to the statistical choices swung the answer by tens of orders of magnitude in either direction. They also pointed out that the formula gives an upper bound, not a lower one, so taken at face value the data say humans discriminate at most a trillion odor stimuli.

The broader belief that humans are poor smellers is shakier still. In a 2017 review in Science, Rutgers neuroscientist John McGann traced it not to experiments but to the 19th-century French anatomist Paul Broca, who linked a proportionally small olfactory bulb to human free will. In fact, McGann notes, the human bulb holds a similar number of neurons to other mammals’, and people are more sensitive than rodents and dogs to some odors.