COLUMBUS, Ohio -- Researchers at Ohio State University have shown that an electronic nose -- a computerized device that samples scents -- recognizes the unique flavors of different kinds of cheese.
Aside from determining the value of this technology for the food industry, this research may one day help scientists better understand the human sense of smell. W. James Harper, professor of food science and technology at Ohio State, explained that most companies employ human sensory panels -- professional taste and smell-testers -- to inspect products such as foods and beverages for which scent and flavor indicate quality. After testing a few samples, however, humans lose the ability to differentiate between similar scents and flavors and must take a rest.
Not so with electronic noses, which can test thousands of samples per day. Still, the machines can't independently gauge the quality of a scent or flavor. They can only compare it to what human noses and taste buds have previously deemed acceptable.
"An electronic nose is not a replacement for people -- it is a supplement," said Harper. "It allows companies to run a large number of samples without worrying about fatigue."
When electronic noses debuted commercially in 1995, the devices promised to test the quality of products ranging from foods to petrochemicals. Since then, few companies have embraced the expensive technology -- costs run as high as $100,000 per machine. That's why Harper and Kuen-Da Jou, a graduate student, decided to find out whether the devices really work.
Compared to the senses of sight, hearing, and touch, scientists know relatively little about how humans smell and taste. Designers of electronic noses have tried to mimic human noses by linking together sensors that detect a variety of volatile compounds.
"We wanted to prove whether the technology was real -- whether we could differentiate between aromas and relate those differences back to the way humans discern smell. That was essential as a first step to determining whether these devices would be useful in industry and research," said Harper.
The results appeared in a recent issue of the journal Milk Science International. Harper discussed the work last month at the 216th national meeting of the American Chemical Society in Boston. Harper and Jou compared the opinions of a human sensory panel to the scent analysis of a commercially-available electronic nose. The panel rated five different types of Swiss cheeses according to scent and flavor.
After the panel finished its work, Harper and Jou studied the cheese samples with gas chromatography, a technique that separates gases into constituent chemicals. The analysis revealed that each cheese contained different amounts of four fatty acids. The researchers then fitted a commercially-available electronic nose with an artificial neural network that trained the device to scan the cheeses for these key fatty acids.
The electronic nose correctly discriminated among the cheeses. It noted that the cheese which panelists described as especially sharp smelled very different from the one they described as especially bland. The device also agreed with the human panelists that the other three cheeses -- one fat-free, one reduced-fat, and one sharp Swiss variety -- all contained similar amounts of the volatile compounds and would taste virtually the same.
Based on these preliminary results, Harper and Jou decided that the electronic nose differentiated between scents well enough to act as a research or quality-assurance tool. "As long as human panels set the standards, then the electronic nose provides a good basis for acceptance or rejection," said Harper.
The researchers will continue this work, but so far they have observed in the machine a situation analogous to the way in which humans experience a scent -- in stages.
"A good human sensory panel will pick up what they call top notes, middle notes, and bottom notes. When people smell an aroma, they experience one sensation immediately, then another sensation later, and then another at the end," explained Harper. "We suspect that compounds that are present in a sample in very low quantities but possess a very strong affinity bind to the sensors first. Those are the top notes. Then, compounds with lower affinity but much higher concentration -- the middle notes -- reach the sensors and replace the top notes. We don't know exactly what's happening with the bottom notes yet."
That explains why some scents scramble the machine's sensors. "Electronic noses have been used to pick up off-odors in beer, but they don't work on whiskey. When a sample contains one compound such as alcohol in high concentrations, then the machine's sensitivity for other compounds goes down," said Harper.
An endowment from the dairy industry of Ohio funded this research, and Brewster Cheese Inc. of Brewster, Ohio, provided the cheeses. The commercially-available electronic nose Harper and Jou used was the Fox 2000 from Alpha M.O.S., a French instruments company.
Contact: W. James Harper, (614) 292-6281;
Harper.9@osu.eduM
Written by Pam Frost, (614) 292-9475; Frost.18@osu.edu