Sensory Evaluation of Food Flavors - ACS Symposium Series (ACS

1 Sensory Evaluation of Food Flavors M . R. McDaniel

Downloaded by 117.255.238.10 on March 28, 2016 | http://pubs.acs.org Publication Date: November 8, 1985 | doi: 10.1021/bk-1985-0289.ch001

Department of Food Science and Technology, Oregon State University, Corvallis, OR 97331-6602

Many changes have occurred in the sensory analysis of flavor in the past half century, beginning with the phasing out of the inappropriate term, "Organoleptic," and the utilization of more appropriate terms to describe the field of study such as "Sensory Analysis" or "Sensory Science." In the food and flavor industries, bench scientists using unsophisticated methods have been replaced by highly educated specialists (M.S. and Ph.D.'s in Food Science and Psychophysics) directing large sensory groups. These specialists utilize highly trained judges and sophisticated test designs and analysis to solve sensory problems. Consumer testing in most industry applications now is done with consumers. Even with the continuing development of instrumentation to replace the human judge, sensory analysis continues to expand its contribution to flavor analysis.

Sensory evaluation (sensory science) i s a s c i e n t i f i c d i s c i p l i n e that concerns the presentation o f a stimulus ( i n t h i s case a flavor compound, a f l a v o r , or flavored product) to a subject and then evaluation o f the s u b j e c t s response. The response i s expressed as, or translated i n t o , a numerical form so that the data can be s t a t i s t i c a l l y analyzed. The sensory s c i e n t i s t then collaborates with the research or product development team to i n t e r p r e t the results and to reach decisions. Sensory s c i e n t i s t s stress that decisions, such as product formulation, are made by people, not by the r e s u l t s of a sensory t e s t , although such r e s u l t s may provide powerful guidance i n the decision-making process. Sensory science i s unique i n that i t requires human subjects. This i n i t s e l f creates challenges, some of which w i l l be discussed i n t h i s paper. The sensory s c i e n t i s t , often working as a part of a research team, also i s unique because t r a i n i n g i n a number o f f i e l d s i s necessary to the success o f the program. The t r a i n i n g of sensory s c i e n t i s t s has not proceeded as rapidly as has the appreciation of and need for sensory s c i e n t i s t s i n the flavor and 1

0097-6156/ 85/ 0289-0001 $06.00/ 0 © 1985 American Chemical Society

Bills and Mussinan; Characterization and Measurement of Flavor Compounds ACS Symposium Series; American Chemical Society: Washington, DC, 1985.

2

C H A R A C T E R I Z A T I O N A N D M E A S U R E M E N T OF FLAVOR C O M P O U N D S

food i n d u s t r i e s . In the past, v i r t u a l l y anyone, regardless of t r a i n i n g and background, might be handed a sensory methods manual, and t o l d that they were i n charge of the sensory program. Fortunately this rarely takes place today. Sensory s c i e n t i s t s are s t i l l i n r e l a t i v e l y short supply, but there are dozens of h i g h l y trained men and women who are currently d i r e c t i n g sensory programs i n both flavor and fragrance companies and t h e i r c l i e n t s ' companies. A sensory s c i e n t i s t working i n the area of flavor would be expected to have a background i n food science and at l e a s t basic knowledge i n the areas of physiology, psychology and s t a t i s t i c s .


History Forss (1) reviewed the r e l a t i o n s h i p between sensory analysis and flavor chemistry, and provided a discussion of sensory characteri z a t i o n of f l a v o r . Williams e t a l . (2) also addressed the problem of r e l a t i n g the many known f l a v o r compounds to what i s a c t u a l l y perceived by the i n d i v i d u a l on a physico-chemical basis. This chapter w i l l center more on reviewing advances i n sensory methodology as they have been adapted by the flavor and food industry. Moskowitz (3) i n h i s book "Product Testing and Sensory Evaluation of Foods" reviewed the h i s t o r y of sensory evaluation beginning with the study of psychophysics. Psychophysics i s the study of the r e l a t i o n s h i p between a p h y s i c a l stimulus and a subject's psychological response to that stimulus. Very e a r l y work was done i n Germany by E.M. Weber and G.T. Feckner over a hundred years ago. They were seeking quantitative laws o f human perception, s p e c i f i c a l l y how to measure our a b i l i t y to discriminate. The study of psychophysics advanced r a p i d l y over the following years and along with i t , so d i d testing methodology. As the psychologists developed and tested t h e i r methods, food s c i e n t i s t s and sensory analysts borrowed those methods and applied them to the study of food. Moskowitz {3) further observed that sensory analysis evolved greatly during World War I I . Much of t h i s work was done by the U.S. Army Quarter-Master Corps, l e d by Peryam and P i l g r i m , developers of the 9-point Hedonic scale. Also during t h i s time, Stevens (4), a psychologist at Harvard U n i v e r s i t y , was working on the psychophysics of hearing, and h i s work l e d to the development of a theory of scales of measurement, and ultimately to the development of the power law. The power law i s generally accepted to define the r e l a t i o n s h i p between psychol o g i c a l response and physical stimulus. The l o g of t h i s equation i s readily recognizable as the equation for a s t r a i g h t l i n e , therefore p l o t s of psychological response versus physical stimulus concentration on a log-log p l o t r e s u l t i n a s t r a i g h t l i n e . These p l o t s , depending upon which sensation i s being measured, be i t e l e c t r i c a l shock, brightness or sweetness, have d i f f e r e n t slopes, and the slope i s defined as the power of function. Food s c i e n t i s t s and psychophysicists have applied these laws to food evaluations and have obtained valuable information. Although knowledge of the power law has been a v a i l a b l e for many years, food s c i e n t i s t s have used t h i s information only r e cently to define how the various sensory q u a l i t i e s i n foods


1.

McDANIEL

Sensory Evaluation of Food Flavors

change, with changes i n concentration. I t i s s u r p r i s i n g that many food companies with only a few major products s t i l l do not know what the sensory properties of their products are or how they change with changes i n formulation.


Scaling Methodology A new method c a l l e d magnitude estimation, a form of r a t i o s c a l i n g , also emerged from Stevens (4J work. To estimate the magnitude of some stimulus, p a n e l i s t s simply assign a number to r e f l e c t the stimulus i n t e n s i t y . P r i o r to the use of magnitude estimation, food s c i e n t i s t s and other psychologists used rather a r b i t r a r y category scales i n order to quantify perceptions. The scales could vary i n length, from a three point scale to perhaps a nine point scale or even l a r g e r , usually with each scale point representing some i n t e n s i t y and designated as small, moderate, l a r g e , extreme or by some other d e s c r i p t i v e word. However, because of the nature of the scale i t s e l f and the way i n which i t was used by the judge, i f one p l o t t e d the r e l a t i o n s h i p between psychological response and stimulus i n t e n s i t y , one usually would not obtain a l i n e a r response. A log-log p l o t of stimulus i n t e n s i t y versus psychological response for magnitude estimation usually r e s u l t s i n a s t r a i g h t l i n e . Many studies have attempted to compare the r e s u l t s of magnitude estimation versus some of the more standard s c a l i n g techniques or semi-structure or unstructured l i n e techniques and d i f f e r e n t r e s u l t s have arisen from these various studies (5-8). Based on my experience, the more important consideration i s the degree of t r a i n i n g of the panel u t i l i z e d i n trained panel work rather than the type of scale that i s used. However, magnitude estimation results do tend to show more differences when the differences are very small than do other methods. Consumer Testing Perhaps the biggest advance i n consumer t e s t i n g i s the use of consumers instead of t r y i n g to obtain consumer data from people who do not use the product or who are too close to the product (a company's own workers). I t once was standard procedure to do in-house consumer t e s t i n g with company employees evaluating the products that produced t h e i r paychecks. This was not an unbiased sample and l e d to many expensive mistakes by industry. Most companies now use a marketing organization to do c e n t r a l l o c a t i o n testing or home placement t e s t i n g of their products. I h i s i s not to say that a l l in-house consumer t e s t i n g i s i n c o r r e c t . The concern i s that d e f i n i t e r i s k s are involved, and one must understand that misinformation may r e s u l t . Some very valuable d i r e c t i o n a l information can be gained i n the r i g h t circumstances by using in-house panels. V a l i d i t y of t h i s information can only be judged through experience with the i n d i v i d u a l i n d u s t r i e s and products involved. One of the most important things to learn i n sensory evaluat i o n i s that "experience" i s c r i t i c a l i n making methodology decisions. Years of sensory t e s t i n g on one product or one product


3


4


l i n e y i e l d s valuable information that should be used by the researcher and the product developer i n making decisions regarding their current l i n e . However, using the same methodology j u s t because i t ' s been used for years and years i s not necessarily a wise decision. Some methodologies are incorrect or i n s e n s i t i v e , or the words used i n scales have l i t t l e r e l a t i v e meaning. One should not be a f r a i d to t e s t new methodology versus o l d methodology to see i f valuable information can be gained by switching to another method. The food industry also has learned that they must test consumers i n large numbers. However, unfortunately, there i s no magic "large" number, and the number of judgments that i s decided upon must be based, again, on the r i s k they are w i l l i n g to take. Another valuable lesson learned over the years i n t e s t i n g products i s that consumers are a l l very d i f f e r e n t , and these very d i f f e r e n t segments of the population, where a product market e x i s t s , must be tested. I f a company i s considering changing the formulation of a product, they must t e s t current consumers of the product to see i f the change makes a difference to them. They aJlso may wish to test consumers of t h e i r product versus consumers of a competitor's product, but t h i s y i e l d s d i f f e r e n t information. I f the consumers of their products are c h i l d r e n , they must t e s t c h i l d r e n , and perhaps also the parent who purchases the product. I f the consumers of a product are on s p e c i a l d i e t s , people on s p e c i a l d i e t s should be the subjects of the t e s t . For example, a non-gluten bread may be highly appreciated by one on a gluten-free d i e t , but t o t a l l y unacceptable to those who have no need to r e s t r i c t gluten i n t h e i r d i e t s . Experience has shown that the order of sample presentation, when a number of samples are to be presented at one time, i s very important. The f i r s t product may strongly bias the evaluation of the product following i t , and one may find a s i g n i f i c a n t order e f f e c t with some products. The sample presentation order i n any test must be balanced or randomized. In analyzing data from such tests, one should consider the average score of one sample presented f i r s t versus i t s average score when presented second. Ihe f a c t that your competitor's product presented f i r s t somehow causes judges to score your product presented second lower, can be valuable information. What was i t about the competitor's product or about your product that caused t h i s difference? The amount of sample i s also c r i t i c a l . Very often i n a taste t e s t , people are given j u s t a "taste" and t h i s may not be enough for them to t r u l y evaluate the product. For example, when testing soup, where the flavor w i l l b u i l d up, mouthful a f t e r mouthful, i t may be necessary to have each p a n e l i s t consume an e n t i r e bowl of soup. The f i r s t impression gained from one or two b i t e s of a product may be t o t a l l y d i f f e r e n t than the f i n a l impression a f t e r consuming an e n t i r e serving. This can be true with many products, e s p e c i a l l y highly flavored products.


M c DAN I EL



Difference Testing Difference t e s t i n g has not changed greatly over the years. The t r i a n g l e t e s t and duo-trio t e s t remain popular and well-accepted, although much e f f o r t has been extended to prove one better than the other. The best advice i s to use the one that f i t s your test situation. When conducting one o f the pure difference t e s t s , the triangle or duo-trio t e s t , one must r e a l i z e that there are two d i s t i n c t types, one being a s i m i l a r i t y taste and the other being a difference test. In the case of the s i m i l a r i t y t e s t the experimental samples are a c t u a l l y d i f f e r e n t from the c o n t r o l . However, the purpose i s not to create a perceptible difference but to e f f e c t a cost reduction or to change suppliers of raw materials without changing the product i d e n t i t y . The actual goal of such a p r o j e c t i s to change the product without i n f l u encing consumer perception and acceptance of it» The s t a t i s t i c a l error of concern here i s the beta error which i s the p r o b a b i l i t y of concluding that the two samples are not d i f f e r e n t when they are d i f f e r e n t . In such t e s t i n g , i t i s desirable to keep the beta r i s k very low, but keeping both the alpha and beta r i s k s low i s d i f f i c u l t . When alpha i s low, beta tends to be high, and when beta i s low, alpha tends to be high. The only way to insure that both errors are small i s to t e s t a very large number of people. In a difference t e s t , rather than a s i m i l a r i t y t e s t , one would i n t e n t i o n a l l y make the samples d i f f e r e n t and then ascertain whether judges could detect the difference. Product D r i f t A major concern i n the food industry i s product d r i f t or subtle, step-wise changes that take place i n the product over time. Product d r i f t can occur when the o r i g i n a l ( f i r s t ) product i s tested and found not d i f f e r e n t from the new (second) product. The second product i s changed i n y e t another way, but i s found not to be d i f f e r e n t from the newest (third) product. However, the t h i r d product may be d i f f e r e n t from the f i r s t product. Hie best way to avoid product d r i f t i s to intimately know a product. This means knowing exactly what sensory c h a r a c t e r i s t i c s are present i n your product and a t what i n t e n s i t i e s . This type of sensory analysis requires the use of descriptive analysis techniques. Because d e s c r i p t i v e analysis involves using a trained panel, developing a set of descriptors, and r a t i n g their i n t e n s i t i e s , i t can be very expensive, but so can product drift. Descriptive Analysis One of the most e x c i t i n g developments i n sensory evaluation over the past decades has been the emergence and popularity of descriptive analysis or the use of highly trained panels to describe the sensory c h a r a c t e r i s t i c s of foods. This i s perhaps the most important development i n sensory evaluation methodology.



6


Only when products are described i n d e t a i l and the i n t e n s i t y of descriptors rated can true product differences or d r i f t be noted. The type of d e s c r i p t i v e analysis chosen should be based on the v a r i e t y of products produced by the company. I f a company produces only one type of product, a more l i m i t e d or s p e c i f i c t r a i n i n g such as quantitative d e s c r i p t i v e analysis described by Stone et a l . (9) may s u f f i c e . I f a company has a large variety of products, flavor p r o f i l e t r a i n i n g established by the Arthur D. L i t t l e Company may be more e f f i c i e n t . Many laboratories combine the best of both methods and develop t h e i r own method. The key to success with e i t h e r method i s panel t r a i n i n g and the e s t a b l i s h ment of appropriate terminology. To i l l u s t r a t e d e s c r i p t i v e a n a l y s i s , I w i l l draw from both the wine and beer industry. Oregon State University's Sensory Science Laboratory, located i n the Department of Food Science and Technology, i s heavily involved i n wine and beer research. The p r i n c i p l e problems and solutions i n the sensory analysis of wine and beer should be transferable to other products. Common wine d e s c r i p t o r s , such as s o f t , hard, f a t , are ambiguous. What do s o f t or hard mean when r e f e r r i n g to wine? The goal of d e s c r i p t i v e analysis i s to use precise terms, even r e f e r r i n g to s p e c i f i c chemical e n t i t i e s when possible. In the wine industry, objective sensory analysis must overcome the h i s t o r i c a l romance of wine. A c l a s s i c example i s the following description of a wine, " I t ' s a naive domestic burgundy without any breeding but I think y o u ' l l be amused by i t s presumption." Such a d e s c r i p t i o n obviously lacks meaningful sensory terms that convey an impression of the wine's aroma and taste. In work on Pinot Noir q u a l i t i e s i n our laboratory, a set of sensory descriptors were developed to aid i n evaluating the e f f e c t of several processing v a r i a b l e s , Henderson and McDaniel (10_) . A trained panel used the b a l l o t shown i n Figure 1 to describe wine produced by d i f f e r e n t malolactic cultures. There are several ways to display this type of i n f o r mation. The QDA method joins descriptor i n t e n s i t y points together to v i s u a l l y display difference. This works very n i c e l y for two to three comparisons. When one has more than two to three samples to compare, other types of s t a t i s t i c a l analyses and methods of displaying the r e s u l t s may be employed. This w i l l be covered i n a l a t e r section. S t a t i s t i c a l Analysis Sensory s c i e n t i s t s r e l y greatly on s t a t i s t i c a l analysis to aid i n i n t e r p r e t a t i o n of data. Ihey also continue to argue endlessly about what i s correct and i n c o r r e c t . Some of the questions that are posed include: which i s the r i g h t analysis; are the assumptions of the t e s t being v i o l a t e d ; i s the data good enough i n the f i r s t place to have s t a t i s t i c a l analyses applied to i t . Some of the most i n t e r e s t i n g advances i n sensory analysis i n the l a s t 20 years have been i n the area of s t a t i s t i c a l evaluation of the r e s u l t s . M u l t i v a r i a t e analysis i s an example of a new type of s t a t i s t i c a l analysis applied to food system. M u l t i v a r i a t e


1.

M c DAN I EL

1


NAME SAMPLE # DATE Using the 9-point intensity scale shown below, rate each sample for all attributes listed. FOR AROMA ONLY 1 - none 2 - threshold 3 - slight 4 - slight to moderate Overall Intensity

5 - moderate 6 - moderate to large


7 - large 8 - large to extreme 9 - extreme 1st tier Frui ty

2nd tier Citrus Berry

3rd tier qrapefruit blackberry strawberry raspberry

Tree Fruit Dried Fruit

cherry strawberry jam raisin fig prune

Spicy

Spicy

black pepper cloves

Vegetative

Canned/cooked

Earthy Caramelized

Caramelized

honey buttery butterscotch

Chemical

Pungent

Ethanol

Sul fur Microbiological

lactic

Figure 1. B a l l o t used by trained panel f o r evaluation of wine.



8


analysis i s used when observations are c o l l e c t e d on many d i f f e r e n t v a r i a b l e s , and i s p a r t i c u l a r l y h e l p f u l when one may be overwhelmed by the sheer bulk of the data that has been c o l l e c t e d . There are several types of multivariate a n a l y s i s , 1) p r i n c i p a l component analysis; 2) factor analysis; 3) c l u s t e r a n a l y s i s ; 4) discriminate a n a l y s i s . P r i n c i p l e component analysis and factor analysis are very s i m i l a r , a major difference being that factor analysis requires assumption of normality. Both methods take a l i s t of variables and reduce t h i s to a smaller number of factors (made up of these o r i g i n a l v a r i a b l e s ) . Such analyses have been applied in wine research (13, 14). An example would be i n the evaluation of wine where sensory and/or a n a l y t i c a l measurements have been taken on many samples. The goal would be to reduce the number of variables necessary to a c t u a l l y show meaningful differences between the wine samples. Factor analysis with vector loading i s h e l p f u l i n t h i s s i t u a t i o n . The loadings t e l l you how each variable f i t s under each f a c t o r . Cluster analysis allows one to see how close samples are together i n a multidimensional space. An analogy can be drawn to playing cards, which may c l u s t e r i n obvious ways by s u i t or card, or by other means inherent i n some game playing r u l e s . Cluster analysis i s generally a stepwise progression. An example, from hop variety research involving eight v a r i e t i e s and many samples of each, i s an appropriate subject for a p p l i c a t i o n of t h i s technique (15). The samples clustered near one side would be most s i m i l a r . Also, any two compounds whose concentration r a t i o s were r e l a t i v e l y constant i n a l l or most samples would have a high s i m i l a r i t y value and would c l u s t e r on the same side. With discriminate analysis one i s concerned about how observations d i f f e r and one sets the rules to d i s t i n g u i s h between populations. The r e s u l t i s a type of c l a s s i f i c a t i o n or sorting of observation into groups. For example, an unknown wine v a r i e t y may be c l a s s i f i e d among known v a r i e t i e s . Another example i s the a p p l i c a t i o n of response surface methodology (16_, 17) . The response i s some function of the design variables i n the t e s t and a l l of the variables are well c o n t r o l l e d and p r e c i s e l y measurable. In order to v i s u a l i z e response surface methodology, imagine that you are viewing a mountain on the horizon. This would be a single variable χ and i t s response, y showing the maximum. Imagine adding a second dimension, 2, coming s t r a i g h t out at you, to give the mountain dimension, and then s t a r t i n g at the very top of the mountain, taking s l i c e s of that mountain on a h o r i z o n t a l axis at equal response l i n e s . I f you then look down on that s l i c e of the mountain, you w i l l see c i r c l e s , the smallest inner c i r c l e equaling the l a r g e s t response. Response surface procedures are not used to understand the mechanism of the underlying system, but rather to determine what optimum operating conditions are or to determine a region of the t o t a l space of the factors i n which c e r t a i n operation s p e c i f i c a t i o n s are met.


1. M e DANIEL



Conclusion Almost everyone i s now u t i l i z i n g the computer for s t a t i s t i c a l analysis o f sensory data. Some laboratories also are using computers to gather the data as w e l l (18). A computerized sensory system would benefit most laboratories by freeing workers from laborious data entry and a n a l y s i s . A l s o , i t would allow for a more thorough analysis o f the data. I t should not replace inspection o f the raw data by the sensory s c i e n t i s t , but allow t h i s to occur more e a s i l y . The f i e l d of sensory evaluation has matured over the years. We have learned through expensive mistakes to rigorously control test s i t u a t i o n s to obtain v a l i d data and to analyze the data as thoroughly as possible to maximize understanding o f products. I believe the future o f sensory evaluation w i l l involve an expansion o f the use o f descriptive analysis i n many d i f f e r e n t s i t u a t i o n s , such as i n plant q u a l i t y c o n t r o l , as w e l l as product development and research applications. Because o f the increased competition i n the flavor industry, flavor companies are increasingly expanding their sensory work and sensory capa b i l i t i e s . This i s necessary, not only for the flavor company to understand the products they are producing but to be able to s a t i s f a c t o r i l y service t h e i r c l i e n t companies. Literature Cited 1. 2.

3. 4. 5. 6. 7. 8. 9. 10. 11.

Forss, D.A., in Flavor Research Recent Advances. (S.R. Tannenbaum and P. Watson, ed.). Marcel Dekker, Inc. 1981. p. 125 Williams, Α.Α.: Lea, A.G.H.; Timberlake, C . F . , in Flavor Quality Objective Methods. (R.A. Scanlan, ed.) ACS Symposium Series No. 51, Washington, D.C., American Chemical Society, 1977. p. 71. Moskowitz, H.R. Product Testing and Sensory Evaluation of Foods. Food and Nutrition Press, Inc. Westport, CT. 1983. Stevens, S.S. Science 1946, 103, 677-678. Moskowitz, H.R. and Sidel, J . L . J . Food Sci. 1971, 36, 677-680. McDaniel, M.R. and Sawyer, F.M. J . Food Sci. 1981, 46, 178-181. Giovanni, M.E. and Pangborn, R.M. J . Food Sci. 1983, 48, 1175-1182. Shand, P . J . ; Hawrysh, A . J . ; Hardin, R.T.; and Jeremiah, L . E . J . Food Sci. 1985, 50, 495-500. Stone, H.; Sidel, J.; Oliver, S.; Woolsey, H.; and Singleton, R.C. Food Technology 1974, 28(11), 24-34. Henderson, L.A. and McDaniel, M.R. Personal communication, 1985. Ennis, D.M.; Boelens, H.; Haring, H.; and Bowman, P. Food Technol. 1982, 36(11), 83-90.


9

10 12. 13. 14. 15. 16. 17.


18.


Johnson, R.A. and Wichern, D.W. Applied Multivariate Statistical Analysis. Prentice-Hall, Inc. Englewood Cliffs, New Jersey, 1982. Wu, L . S . ; Bargmann, R.E.; and Powers, J.J. J . Food Sci. 1977, 42, 944-952. Williams, A.A. J . Inst. Brew. 1982, 88, 43. Stenroos, L . E . and Siebert, K . J . ASBC Journal 1984, 34, 55. Henika, R.G. Food Technology 1982, 36(11), 96-101. Myers, R.H. Response Surface Methodology. Allyn and Bacon, Inc. Boston, 1984. Brady, P . L . ; Ketelsen, S.M.; and L . J . P . Ketelsen. Food Technology 1985, 39 (5), 82-88.

RECEIVED

July 29, 1985


Sensory Evaluation of Food Flavors - ACS Symposium Series (ACS

Recommend Documents