Shots - Health News
2:28 pm
Thu March 20, 2014

Never Mind Eyesight, Your Nose Knows Much More

Originally published on Mon March 24, 2014 6:14 am

The human eye can distinguish more than 2 million distinct colors. But scientists studying smell now say they have their vision colleagues beat: The human nose, they say, can distinguish more than a trillion different smells.

Yes, trillion with a T.

That new figure displaces a much more modest estimate. Until now, smell researchers have been saying the human nose can distinguish about 10,000 smells.

But Andreas Keller at The Rockefeller University says that number comes from a discredited idea from the early 20th century, which asserted that there are four primary smells, the way there are primary colors. That's just plain wrong.

"And I found that so interesting and ridiculous that I thought it would be time to do an experiment and test how many odors there really are," Keller says.

So he and his colleagues at The Rockefeller University and the Howard Hughes Medical Institute put together an experiment. They started with 128 chemicals with distinct smells, and started mixing them together —as many as 30 chemicals per bottle.

"Subjects come in and they're given three little bottles with odors in them," Keller explains. Two of the bottles are identical. The third is slightly different. "And the task is simply to tell which one is different."

Through this process, the scientists found that the human nose could do a remarkable job of distinguishing very small differences between smells. Once they figured out the percentage of these mixtures a person could distinguish, they called in a mathematician, who figured out how many possible unique odors you can make from elaborate mixtures of these 128 chemicals.

"And that's how the number of 1 trillion came about," Keller says.

It's true that a human being would never encounter a trillion different smells in a lifetime. "And most of the smells we tested in our test were probably never smelled by a human being before," Keller says.

But the human nose has presumably evolved to be able to detect tiny differences among smells — say to tell the difference between fresh food and something that's just beginning to spoil. That could be a matter of life and death.

So why do this experiment?

"This is just fun," Keller says. "It's just bothersome to see a number in the literature that's so clearly wrong."

The study was published Thursday in the journal Science.

The question now is whether the new number will be believable to other scientists in the field.

"I think it's actually a huge underestimate," says Joel Mainland, who works at the Monell Chemical Senses Center in Philadelphia. That's more or less the world capital of research on smell and taste.

The new study "at least puts a lower limit on it, and says it's much, much higher than people were thinking," he says.

Mainland, who wasn't involved in this project, says what intrigues him about the findings is not so much the huge number but that the scientists sometimes found mixtures of very different chemicals that ended up smelling the same.

This hints that there could be a basic logic of smell, he says. If scientists could decipher that, they could, in theory, reduce smells to a basic set of instructions — a universal code for smells.

Mainland imagines that someone someday will come to him and say, "Look, I have this beautiful strawberry smell, and I want you make this strawberry smell. But you can't use any of the molecules I have on this list. I want you to use this other set of molecules to make a strawberry."

He says, "If you have a universal code, then you can use a set of ingredients that were never in the strawberry to make the same strawberry smell."

And if you could do that, you could send smells on the Internet. That is, you could transmit instructions to a machine that has a small number of chemicals that can be combined to make a huge variety of smells. We could have Smell-O-Rama.

But first you have to figure out the code. Mainland says that's a daunting challenge, but the new research just might help frame that search.

Copyright 2014 NPR. To see more, visit http://www.npr.org/.

Transcript

AUDIE CORNISH, HOST:

From NPR News, this is ALL THINGS CONSIDERED. I'm Audie Cornish.

ROBERT SIEGEL, HOST:

And I'm Robert Siegel.

The human nose can distinguish 10,000 distinct smells; at least that's been the working rule among scientists. Ten thousand smells sounds like a lot. But a new study argues that it is a vast underestimate. A new report in Science magazine concludes that the human nose can distinguish more than a trillion unique smells.

NPR's Richard Harris reports.

RICHARD HARRIS, BYLINE: When scientists say people can smell 10,000 odors, that comes from a century-old idea that's just plain wrong. Smell researcher Andreas Keller says it comes from a now laughable notion that smell has four primary scents, the way light has primary colors.

ANDREAS KELLER: And I found that so interesting and ridiculous that I thought it would be time to do an empirical experiment and test how many odors there really are.

HARRIS: So, Keller and his colleagues at The Rockefeller University and the Howard Hughes Medical Institute put together an experiment. They started with 128 chemicals with distinct smells and started mixing them together in up to 30 chemicals per bottle.

KELLER: The experiment works in the way that the subjects come in and they're given three little bottles with odors in it. Two of the bottles have the same odor in it and one has a different one. And the task is simply to tell which one is different.

HARRIS: Through this process, the scientists found that the human nose could do a remarkable job of distinguishing very small differences between smells. Once they figured out what percentage of these mixtures a person could distinguish, they called in a mathematician who figured out how many possible unique odors you can make from elaborate mixtures of these 128 chemicals.

KELLER: And that's how the number of one trillion came about.

HARRIS: Granted, 10,000 seems low. But a trillion seems so high that, I mean within a human's lifetime, clearly you couldn't encounter a trillion smells. Could you?

KELLER: That's very true. You would never encounter a trillion smells. And most of the smells we tested in our test were probably never smelled by a human being ever before.

HARRIS: So is this just fun or is there some deeper point here?

KELLER: This is just fun.

HARRIS: Really?

KELLER: It's just bothersome to see in the literature a number that's so clearly wrong.

HARRIS: The question now is whether the new number will be believable to other scientists in the field.

JOEL MAINLAND: I think it's actually a huge underestimate.

HARRIS: Joel Mainland works at the Monell Chemical Senses Center in Philadelphia, which is more or less the world capital of research on smell and taste.

MAINLAND: But this at least puts a lower limit on it and says it's much, much higher than people were thinking.

HARRIS: And that's what the research paper says: A trillion odors is a lower limit, a minimum.

Mainland wasn't involved in this project. But what intrigues him about it is not so much the huge number but the fact the scientists sometimes found mixtures of very different chemicals that ended up smelling the same. This hints that there could be a basic logic of smell. If scientists could decipher that, they could in theory reduce smells to a basic set of instructions, a universal code for smells.

MAINLAND: Somebody comes to me and they say: Look, I have this beautiful strawberry smell and I want you make this strawberry smell but you can't use any of the molecules that we have on this list. I want you to use this other set of molecules to make a strawberry. If you have a universal code, then you can use a set of ingredients that were never in the strawberry to make the same strawberry smell.

HARRIS: And if you could do that, you could send smells on the Internet. That is, you could transmit instructions to a machine that has a small number of chemicals that can be combined to make a huge variety of smells. We could have smell-o-rama. But first, you have to figure out the code. Mainland says that's a daunting challenge but the new research just might help frame that search.

Richard Harris, NPR News. Transcript provided by NPR, Copyright NPR.

Related Program