Ah! A bold espresso that boasts intense flowery, winey, citrus, acid — and yes, even butter toffee notes. So says an electronic nose, anyway.
Behold, the coffee snob of the future.
Perhaps the machine assembled by scientists at the Nestlé Research Center in Switzerland isn’t quite ready to be called into daily demitasse-sipping service. But in an analytical test of its abilities, it predicted the range of aromas and intensities noted by a panel of experts for 11 different espressos, with few mismatches. And in a subsequent validation, the electronic nose nearly duplicated the panel’s opinion in characterizing an additional eight espressos.
In the quest for consistently high-quality java, the coffee industry stands to benefit enormously from any nose that really knows its stuff, whether attached to a person or a machine.
“We do not attempt to replace human tasters by instruments but to assist human tasters,” said Nestlé researcher and lead author Christian Lindinger, whose report was published last month in the journal Analytical Chemistry. “But in some cases we can use the approach as a pre-screening tool to eliminate those samples which would anyhow fail a sensory evaluation because of insufficient quality.”
The perfect espresso
Researchers already have developed prototype electronic noses to analyze perfumes and wines as well as to monitor landfill odors, distinguish between smoke from fires or cigarettes, warn of hazardous gases or narcotics and even detect the warning signs of pneumonia and asthma in a person’s breath.
With more than 1,000 organic compounds contributing to a roasted coffee’s aroma, sniffing out the perfect espresso has lagged behind a bit. Recent research, however, has suggested that only 50 of those chemicals might be necessary for mapping out the sensory profile of that aroma — and perhaps even less for accurately reproducing it.
For their study, the researchers used a standard espresso machine (made in Switzerland, of course) to prepare the different samples. The team poured each one into a setup that allowed the espresso’s heated gases to waft up from an oven to a chemical detector known as a proton-transfer reaction mass spectrometer. The mass spectrometer recorded the prevailing gas mixture and a mathematical model then translated the most useful chemical information into a more understandable description.
Lindinger said the model can predict the sensory profile of an unknown espresso based solely on the chemical measurements. It’s also possible, he said, to discern the mix that correlates with a specific attribute like, say, flowery notes, provided that the right balance of compounds is carefully recorded.
With their human noses to guide them, an in-house panel of 10 experts from across Europe also evaluated each espresso on its range of attributes, scoring each as weak, medium or high-intensity. One espresso was noted for its strong “butter toffee” quality, while another scored best on its “woody” attribute and a third was, alas, deemed particularly “bitter.”
'Nose' to nose with the pros
Lindinger and his co-authors cautioned that an intense coffee can be a headache when trying to dispassionately summarize not only its attributes but also the relative strengths of each. One taster might judge an overly bitter coffee as rather intense, while another expert could just as easily be influenced by a coffee’s evident acidity. Such quality versus quantity distinctions are important because the study included espressos made with varying amounts of water that could dilute or concentrate some smells.
Even so, the panel and machine matched up remarkably well on nearly every espresso’s qualities, although they did, on occasion, beg to differ about the intensities. Did that one really have a high-intensity roasted and bitter aroma, as the experts thought, or only a medium level, as the electronic nose judged?
Only one espresso seemed to defy a uniform judgment, as the panel pointed to its intense butter toffee and cereal qualities (the machine did not), while mostly thumbing their noses at what the machine judged to be its moderately bitter, cocoa and coffee elements.
After achieving a near-perfect match in describing a second set of eight espressos, the researchers concluded that their model was robust enough to withstand differences in how the espresso was prepared, when the coffee beans were harvested, and whether or not their mass spectrometer had been recently updated.
Going nose to nose with the pros in describing an espresso is one thing. But how could an analytical measurement be translated into something akin to the description accompanying Colombia’s Finca El Palacio coffee, acclaimed by one taster for its floral tones that are “rich and rose-like in both aroma and cup, with caramelly, almost custard-like support and hints of cedar and cocoa”?
Lindinger said the sensory attributes have to be adapted to the range of espressos being analyzed.
“The main limitation is the set of espressos used to train the model,” he conceded. “This is the range within which the model can work properly.” The researchers sought out a wide variety of aromas and tastes for their samples, he added, but ultimately, he can affirm only that their machine-based model works just as well for a specific set as a panel of human experts.
The future smells sweet
Despite the limitations, other coffee experts praised the line of research.
“I think this technology, the scientific way to analyze this — it’s just leaps and bounds ahead of what we’ve been doing,” said Susie Spindler, executive director of the Missoula, Mont.-based Alliance for Coffee Excellence. Spindler, whose nonprofit organization oversees coffee competitions in eight countries, said an electronic nose like the one developed by Nestlé could be enormously useful for sniffing out far less desirable attributes as well.
Consider an espresso boasting dirty, moldy, oniony, old — or perhaps worst of all — phenolic notes. Spindler compared the last odor to something like eau de iodine. Experts are still at a loss as to how it can invade certain batches, she said, but definitely know the coffee-ruining aroma when they smell it.
“There’s nothing good about it,” Spindler said.
She cautioned that flavor doesn’t always match a coffee’s aroma, but welcomed the potential of an electronic sniffer to effectively wrinkle up its nose when detecting a stinker. Likewise, she expressed hope that the technology could eventually trace particularly pleasant aromas all the way back to specific coffee plantations or growing conditions. “I would love to see more research done, to find out if there is a way to isolate some of the physical properties that people end up loving,” Spindler said. “What is it that creates this phenomenal flavor and how can we help farmers deliberately produce more of it rather than accidentally produce it?”
As for improving the general sniff-ability of electronic noses, collaborators at the University of Warwick and Leicester University in the United Kingdom have come up with another solution that’s nothing to sneeze at: artificial snot. Nasal mucous, after all, is known to dissolve scents and separate their components in a way that allows humans to pick out, say, a more jasmine flower-like espresso than one featuring hints of rose. The faux schnozz still has a bit of catching up to do, however; the ability of its snot-lined sensors to distinguish banana from milk was hailed as something of a breakthrough.
It may be difficult to imagine how such, well, unrefined tastes could ever compare to the Seattle roaster who compared a Rwandan coffee to “flapjacks covered in amber maple syrup,” nutmeg and water hyacinths. But for coffee lovers, the continued evolution of an electronic nose that rejects lousy espresso could bring a future that smells just as sweet.