Educated tastes. Move over, Goldilocks: Taste-testing is a science that demands a precision palate
Mount Vernon, N.Y.
THE long room is divided into private booths, each dimly bathed in red light. Winifred Conkling steps into chamber No. 7. Sitting down in a swivel chair, she stares at a small door in the wall and waits. Others are in booths to the left and right of her, but she can't see what they are doing. In the hush, all she can hear is the sound of chewing. Suddenly the little door pops open and a hand thrusts a steaming plate onto the counter before her - Sample 238. The red light distorts the color of the food, but her nose knows it's a pancake. She cuts a piece from the center, puts it in her mouth, and chews thoughtfully.
The taste and smell of wheat hit her senses; she finds ``wheat'' listed on the ballot served with the food, and she pencils in a 4 (out of a possible high of 15). The wheat has a slightly raw taste; she gives it a 1 for ``rawness.'' Sweetness is noticeable, but the taste of leavening is not; she writes in 3 for ``sweetness,'' 0 for ``leavening.'' There's a hint of dairy and egg tastes, so she scores 1 for each. As she continues chewing, the pancake balls up and has a slimy, greasy mouthfeel; she scores a 6 for ``cohesiveness,'' a 4 for ``greasiness.'' Then she unceremoniously spits out the wad of pancake and presses a button to signal that she's ready for the next sample.
Twice a day for half an hour, Ms. Conkling and 20 other adventurous recruits take time out from their jobs as engineers, technicians, writers, secretaries, and statisticians here at Consumers Union to explore the world of food, both flavor and texture. They sip beverages, savor spaghetti sauce, nibble turkey, and chomp chocolate bars, carefully analyzing and recording their every sensation.
The data that emerge are the basis for the product ratings of packaged, canned, and frozen foods in Consumer Reports, a monthly magazine whose judgments are closely watched by both subscribers and food manufacturers. In its testing labs here in suburban New York, the 51-year-old Consumers Union, which publishes the reports, rates 12 food products a year - everything from soup to peanut butter. (In-house taste testing, like other volunteer activities at CU, has been temporarily suspended while union contract negotiations are going on.)
Taste-testing may sound like fun, but when the stakes - a company's reputation - are high, the work is intense and thorough. Testers may be called upon to taste, for months on end, such things as breakfast cereal, syrup without pancakes, and pancakes without syrup.
``Sensory evaluation is a scientific discipline, so we have terminology; you can't say, `Mmmmm, this is good,' or `Yuck,''' says Louise Miller Mann, head of the Sensory Evaluation lab here.
``Learning to taste is like learning to appreciate music. At first [the listener] can't distinguish one symphony from another, but gradually he can pick out the sound of the oboe, the clarinet. In the same way [the taster] learns he can pick out the oats or wheat - raw or toasted - from the total product.''
For the testers, getting past the rigorous screening is a formidable hurdle.
When Mrs. Mann called for volunteers recently to serve on taste panels, 200 eager CU staff members responded. She gave each one a probing interview. ``I wanted to see if they were interested in what I was doing,'' she says, remarking that motivation is pretty important to get through six weeks of testing a food product. ``Did they take it seriously, or did they think my questions were silly?''
Each would-be taster had to identify the four basic tastes - sweet, sour, salty, and bitter. Some people get those confused, says Mann, who has a master's degree in food science. There were also tests for differing intensities of sweetness and firmness, and a test for common odors to see how well the volunteers identified smells.
After the month-long screening process, only 45 people were left, and half of those opted out because they couldn't spare the extra time the twice-daily tastings require.
Food snobs were not among the chosen. Media liaison Marnie Goodman, who got screened out, says ruefully, ``At first I thought it was because I had a tin palate, and I was heartbroken. But Louise told me it was because I had an aversion to processed food, which is mostly what they test.''
The 21 who did qualify spent the next three months in intensive training. For an hour each day, they sampled foods of all types, from lemon frosting to ketchup to corn bread. They even tasted (or at least smelled) rancid oil and other off flavors, and discussed their sensations in minute detail.
To calibrate their taste buds, they were introduced to the taste and texture scales CU uses as references. With the help of sweet, sour, salty, and bitter solutions mixed in various intensities according to fixed recipes, they learned to break any food taste into its component parts and position it accurately on these scales. Tasters must be able to identify a No. 9 sweet, for instance, whether it appears in a muffin, orange juice, or strawberry jam.
The crew emerged from sensory boot camp able to make such fine distinctions in taste as between chocolate and cocoa, vanilla and vanillin, and, in texture, between adhesiveness and cohesiveness.
``We learned how to spit, too,'' Conkling, a financial writer, says wryly. ``It's actually one of the most important things. ... I frankly was surprised how often in the testing you find food that isn't fresh.'' Tasters must also spit to avoid feeling full and to keep the taste of a sticky or piquant food from lingering.
The training has some side effects. ``It makes you fussier about bad foods,'' observes magazine editor Mike Echols, who finds he can no longer eat some of his old favorites. ``I couldn't feed myself hamburger steak TV dinners anymore. That stuff is universally ghastly. The meat isn't good, and it's not well prepared.''
``I probably make more comments at the dinner table than I used to, like, `H'mmm, this seems a little high on oregano,''' says Ned Groth, CU's associate technical director, with a laugh.
Before testing a particular type of food, a panel (usually seven tasters) spends five or six hours sampling a group of products in that category. They never see any labels - it's all done blind. As a team, they come up with a brainstorming list of words to describe each sensation they experience.
For peanut butter, the panel first tasted fresh peanuts and listed several kinds of peanut flavor - raw, roasted, overroasted, and peanut skins. Along with these, they found the peanut butters contained other flavors, like sweetness, saltiness, sweet aromatics (such as molasses or honey), rancidity, and bitterness. Textures, like adhesiveness, that stick-to-the-roof-of-your-mouth quality, were also noted.
The panelists often get into a lively dispute over their perceptions, but it is usually a problem of vocabulary, says Mann. Once they've clarified the definitions, the terms they've chosen become the ballot they will use during the actual taste test.
When they are at last ready to rate the products, they enter the soundproof chamber. ``We use red lights because they eliminate color,'' says Mann. ``We don't want panelists tasting with their eyes.''
Visual cues can mislead. In a test CU conducted back in the 1940s on dessert gelatins, blind-folded panelists were unable to distinguish between ``cherry,'' ``raspberry,'' and ``strawberry'' flavors. Without the help of color, they all tasted alike.
In individual booths, the tasters are served only four samples per sitting, twice a day. To control all the variables, they taste each sample according to a strict procedure they've agreed on ahead of time. To gauge spaghetti texture, for example, they bit with molars into two noodles folded twice.
Though CU's cold breakfast cereal ratings were based on nutrition rather than taste, the sensory testing was nonetheless painstaking - a little too painstaking for some appetites. After a month and a half of eating the 58 cereals dry, the panelists decided they should also try them with milk. They repeated the tests, adding a fixed amount of milk to a fixed amount of cereal, then waiting exactly 30 seconds before tasting. Three months and 464 bowls of cereal later, the panel had all the enthusiasm of a soggy cornflake. Even now, the mere mention of cereal brings groans. ``But you sort of forget after a while,'' one taster sighs hopefully.
The sensory panelists must check their personal likes and dislikes at the door.
``Think of tasters as measuring instruments,'' says Mr. Groth. Usually products are tasted four times each by seven people. That's a total of 28 tastings for each brand. A statistician then averages all those scores and comes up with a sensory profile of that brand.
Independently of the panelists, Mann or an expert taster in the food industry decides what an ideal example of the food should taste like. Here, for instance, is Consumer Reports' tantalizing profile of the perfect chocolate chip cookie:
``[It] provides an intense jolt of chocolate aroma and flavor, with just a hint of vanilla and the tastes and smells of milk and butter. The chips have the moderate `cocoa bitterness' typical of true chocolate, combined with come-hither tenderness. They melt with silky smoothness. The cookie smacks moderately of sweetness and caramel, with traces of salt, vanilla, and dairy flavors. It provides a double texture - a crisp edge, with a moderately firm, chewy center - and leaves a faint residual mouthcoating after each bite.''
The ideal sensory profile is not shared with the taste panelists. That might bias their observations. It is used for statistical comparison only. Each characteristic of the ideal is translated into a range of scores, so all the product samples can be rated against it mathematically.
``It can be appalling to see the range of quality in a type of product - say, among 50 types of orange juice,'' says Groth. ``As a taster, you get a much bigger slice of life than you would experience as a consumer.''
This test was a real turkey
Getting hundreds of food samples to the taste testers' booths fresh and piping hot is a complicated task no matter what the test subject. But frozen turkey posed problems that would have taxed the logistical skills of Napoleon.
``Turkeys were an absolute pain,'' moans Hilary Brown, senior project leader of the foods division at Consumers Union's testing labs.
First, CU's shoppers in locations all over the country were instructed to buy 36 brands of frozen turkeys - in retail stores, just as consumers buy them. Then the turkeys had to be packed in dry ice and shipped to CU headquarters in Mount Vernon, N.Y.
As birds of various sizes poured in - 1,200 in all - the test kitchen took on the look of a petrified-poultry farm. Turkeys were everywhere.
``Our freezer was jammed. We had to get extra chest freezers as well as our own walk-in freezer,'' Ms. Brown recalls with the relish of a war veteran recounting her toughest mission.
Because food samples must be served to the taste panel in the random order prescribed by a statistical design, each step of the preparation had to be carefully orchestrated. The thawing timetable alone required a three-inch-thick computer printout.
``We had little lists all over the freezer saying what went where, when, and for how long,'' says Brown with British good humor. ``It was like World War II, moving the Army around the country. I was in total sympathy with the field marshal.''
Before the turkeys could be roasted, both ovens and poultry thermometers had to be tested and calibrated. That took three weeks and three kinds of thermometers. Then Brown made pilot tests to find out how long to bake turkeys of each size - from three-pound breasts to large full birds - and how much time they had to sit before carving.
``All the samples have to reach the sensory panel in as close to the same condition as possible. You can't have one standing 30 minutes and one standing five minutes,'' Brown explains. ``Every source of bias has to be eliminated.''
Statistical design ensures that a particular brand isn't always served on the same day of the week, or in conjunction with the same three other brands.
Finally the roasting began, four turkeys at a time, according to another voluminous computer schedule. But the birds didn't bother to read the schedule. ``Every day there were changes in the statistical protocol,'' Brown says, ``because the turkeys weren't always ready in the planned-on order.''
When the recalcitrant birds were finally done to a turn and rested precisely, Brown and staff would swing into action, quickly carving the same right breast and thigh on each one. The light and dark meats were weighed and served to the waiting tasters. The turkeys were then carved down to the bone and each element weighed, sealed in coded freezer bags, and sent out for nutritional analysis.
In the afternoon, the staff would repeat the entire process.
After cooking eight turkeys a day for three months, a battle-worn Brown said, ``I haven't tasted turkey since then. We had duck for Christmas and goose for Thanksgiving.''