What the Nose Knows - Avery Gilbert [22]
So-called aroma models take this insight even further. To create an aroma model for french fries, for example, scientists run a batch through the GC/MS and generate a complete list of all the volatiles. Their goal is to create a fully realistic french-fry aroma using as few of the volatiles as possible. They begin by selecting odorants present at concentrations well above our sensory threshold. If a blend of those doesn’t match the original aroma, they extend the list to include odorants at or below the sensory threshold. Once a blend closely matches the full aroma, it is tested further. One by one, odorants are subtracted from the formula. If the resulting formula smells less realistic, the subtracted odorant is restored. If the subtraction makes no difference, that odorant is dropped. The final aroma model is one of irreducible simplicity—a stripped-down formula that smells complete to the nose. An authentic french-fry smell, for example, can be made from nineteen ingredients. This includes a trace of stinky methyl mercaptan—without it, the formula lacks the necessary boiled-potato character.
Aroma models have been developed for Swiss cheese, Camembert, basil, olive oil, and baguette crust, among other things. These whittled-down formulas all point to the same conclusion—most volatiles in a food add nothing to its smell. A high-fidelity odor replica can be created from one or two dozen ingredients. A classic example is the cup of coffee. Chemists have been analyzing coffee aroma for more than 100 years and have found more than 800 different molecules. Using aroma models, German scientists found a mere twenty-seven high-impact molecules in medium-roasted Arabica coffee; they made a high-fidelity model using only sixteen of them.
The sensory logic of aroma models can be extended to nonfood areas, and may even have applications for environmental issues. Livestock feeding operations, for example, generate a big, messy stink that can annoy nearby residents. A typical Iowa swine barn contains more than 300 different volatiles, which sounds like a lot of bad news for the downwind neighbors. Yet a recent study found that four molecules account for about 85 percent of the piggy odor. One of these—para cresol—has a smell that by itself closely resembles the overall barnyard odor. This discovery may turn an overwhelming odor problem into a manageable project. Instead of going after all 300 suspect chemicals in the swine barn, one might suppress a handful of character-defining molecules. Pinpoint sensory targeting could produce bigger benefits at less cost.
THE SUCCESS OF aroma models—those minimalist imposters—casts nature’s abundance in a new light. Lifelike smells can be made from a handful of molecules, and the same molecules turn up in smell after smell. Is nature’s chemical cornucopia really so impressive if only a tiny portion of it matters? And what does it say about our sensory abilities if there is so much more out there than meets the nose?
Terry Acree, the Cornell scientist who helped develop GC-olfactometry, has the numbers to back this up. He searched through hundreds of food-aroma studies and made a list of volatiles present at smellable concentrations. The first edition of the FlavorNet list was posted online in 1997. It contained three hundred chemicals. Today he has posted about eight hundred. Acree expects the list to top out at fewer than one thousand. In other words, all the smells in nature are built from fewer than one thousand smellable chemicals. What are those other thousands of volatiles doing? They may subtly round out a scent, give it shading and complexity. Acree speculates that many of them are intended for the noses of creatures