Caballero B. (ed.) Encyclopaedia of Food Science, Food Technology and Nutrition. Ten-Volume Set

Подождите немного. Документ загружается.

0011 By the 1990s, electronic mail encouraged speedy

interaction among sensory scientists worldwide.

Other tools continued to evolve. The beginning

years of the twenty-first century are bringing intri-

guing tools such as the electronic nose to the chal-

lenge of sensory research.

Disciplines

0012 Figure 2 describes the root disciplines involved in

sensory evaluation together with those that have

been most active in its applications. This picture

confirms the interdisciplinary nature of the field. As

each of these grows in sophistication, so does sensory

science.

0013 Some examples of the contributions of each of the

basic areas are:

0014 Psychology/Psychophysics: the study of the laws of

sensory perception, basic information on testing

methodologies and the requirements for effective

testing (from ballot design to sample presentation

to testing protocols), sources of psychological error

in sensory testing, the study of cognitive functions

in the development, and use of terminology in sens-

ory testing.

0015 Physiology: the functioning of the sensory systems,

sources of physiological error in sensory testing.

0016 Statistics: the data analysis methods used to date

and the development of new methods specific to

sensory analysis.

Further examples from the applied areas include:

0017Home economics/food science: applications of

sensory analysis to the production of foods as

basic commodities and in processed forms includ-

ing product stability, product quality, consumer

preference.

0018Chemistry: the instrumental techniques for the an-

alysis of foods and other products tested in sensory

science.

0019Management/marketing: methods for the assess-

ment of consumer behavior.

0020Engineering: the development of sensor-based tech-

nology in the forms of the electronic nose and the

electronic tongue, and of neural networks.

Classifications of Areas of Sensory

Testing

0021The most fundamental division of sensory measure-

ment is determined by whether the subject is asked

how they ‘fit’ relative to the stimulus or whether the

subject is required to make an analytical judgement of

the stimulus. This is given in sensory analysis by the

terms subjective and objective testing (see Table 1).

Answering a question using sensory analysis requires

the development of a strategy for the approach, in-

cluding the identification of the question that is really

being asked and formulating an experimental plan to

obtain appropriate data.

AppliedBasic

Sensory

Science

Marketing

Psychology

Chemistry

and Physics

Engineering

Physiology

Agriculture

Statistics and

Computer Science

Home

Economics

fig0002 Figure 2 Disciplines contributing to sensory science.

SENSORY EVALUATION/Sensory Characteristics of Human Foods 5127

Subjectivity and Objectivity

0022 Subjective tests are those that evaluate the behavior of

the assessor relative to the sample, i.e., they measure

the opinion of a person who is using a product and

are more commonly referred to as ‘consumer tests.’

They imply a response that is affected by personal

bias, emotional background, etc., and describe the

individual. These tests are not necessarily validated

through demonstration of sensory ability. The term

subjective is often incorrectly used to describe all

sensory tests.

0023 Objective tests are those that measure the presence

and/or intensity of specific characteristics within a

sample. These tests imply a response that is based

on external phenomena or events, and the approach

of the panelist is expected to be detached, impersonal,

and unprejudiced. This term is used in sensory

evaluation to define those tests that are based on

demonstrated results with selected, trained judges

appropriate to the test procedure.

0024 Sensory test methodologies have been classified in

different ways – by the overall purpose of sensory

testing, by format of test methodologies, and by the

type of primary ‘tool’ used, i.e., the type of sensory

assessor.

0025 For example, Michael O’Mahony in 1988 classifies

the world of sensory evaluation into four types of

testing, based on test function and experimental goals:

0026 Sensory Evaluation I: sensory evaluation used as a

basic tool to study chemical and physical charac-

teristics of food systems, e.g., a change in a

flavoring would result in a change in a volatile

chemical and thus a perceived change in odor.

Judges are selected for sensory ability.

0027 Sensory Evaluation II: testing with consumers to

measure the degree to which product changes can

be detected by untrained assessors. Assessors can

be trained in sensory methodology, but not in the

product characteristic.

0028Consumer testing: preference testing, etc.

0029Sensory psychophysics: studies of testing method-

ologies and assessor sensitivities.

Several workers have used a classification system by

the type of assessor chosen. In this classification, there

are four basic types of assessors:

0030Naı

ve untrained assessors: consumer testing; used

to measure subjective opinions about a sample.

0031Selected assessors trained for specific short-term

tasks: laboratory tests for specific attributes; ob-

jective tests including the difference tests and the

rating scales that measure specific product charac-

teristics.

0032Descriptive analysis panels: groups selected and

highly trained assessors that work as an expert

group; objective procedure to evaluate specific

products or product groups for complete ranges

of product characteristics.

0033Expert assessors: selected, highly trained, and

employed to work as individuals; objective testing

of specific product or product group by an indi-

vidual who has responsibility for the samples and

the data about them.

Sensory Science and Psychophysics

0034Psychophysics defines those aspects of sensory science

that include the way in which stimuli are perceived

and the way in which they are measured. Some of the

pioneering work in psychophysics was done by the

psychologist L. L. Thurstone, mainly in the areas of

psychology, but also in applications to food testing.

Psychophysics itself has been defined as ‘concerning

the functional relationship between stimulus and re-

sponse.’ Work in this area includes not only the rela-

tionship between the perceived intensity and physical

intensity of the stimulus, but also the ways in which

judge sensitivity can change in test situations, the

nature of language and terminology development

tbl0001 Table 1 General characteristics of subjective vs. objective tests

Subjective test Objective test

Measures properties of: Person (opinion) Products (level or intensity of attribute)

Approach to test: Emotional Analytical

Assessor selected for: Represents a population segment or a group

with particular product experience

Demonstrated ability to perform sensory

measurement tasks

Training level: Untrained (naı

ve response) Trained (educated response)

Number of assessors: Many (30–50 for preliminary test, up to

several hundred for final test)

Few (usually four to 10)

Performance evaluation of assessors: No Yes

Terms used on ballot: Value judgements (like/dislike, good/bad,

accept/reject)

Descriptive of phenomena (sweet, salty,

red, tough, etc.)

Method described in early sensory

evaluation literature as:

Organoleptic testing Sensory analysis

5128 SENSORY EVALUATION/Sensory Characteristics of Human Foods

and its relationship to everyday language, and the

development and processing of sensory concepts.

0035 An example of the understanding this discipline

provides is from the work by Stevens, which categor-

ized sensory measurement as involving three param-

eters – the task undertaken by the assessor, the

manner of the stimulus, and the statistical measure

used to describe the data.

0036 The task undertaken by the assessor was further

defined by four groupings, which are generally

accepted as defining the data collected in sensory

testing:

0037 Classification or nominal data: where the observer

judges whether the stimulus is present or not. The

data consist of meaningful categories that can

be clearly differentiated. No numerical values are

associated with the categories within this group –

the data analysis includes frequency data only.

One example of these data is given by the

numbers on football jerseys – numerical handling

of these data, e.g., a mean of the numbers, is

meaningless.

0038 Order or ordinal data: consisting of meaningful

categories that can be clearly differentiated and

placed in order of magnitude, e.g., greater or lesser

of some defined characteristic, ranked data.

0039 Interval data: involving judging the apparent dif-

ference between two or more perceptions and

assigning values to the intensity of the perceptions

using a continuum that is divided into apparently

equal intervals.

0040 Ratio or magnitude data: involving reporting the

ratio of magnitude between two or more percep-

tions or the apparent magnitude of a perception.

Another example of the contribution of psychophys-

ics is the development of a numerical expression of

the relationship between the physical intensity of a

stimulus to its perceived sensory intensity. This began

with the Weber Fraction and Fechner’s semilog rela-

tionship between these phenomena. This evolved into

the development of Stevens’ Power law as the basis

for sensory work with the definition:

I ¼ kS

where I ¼perceived intensity; S ¼stimulus intensity;

n ¼the exponent of the power function; k ¼acon-

stant. Once this function was established, Stevens’

Law offered useful predictive power for the study of

sensory perception and the relation to other product

attributes.

0041 Work on the application of Thurstonian methods to

sensory data analysis is currently underway through

the activities of one of the ASTM E18 Subcommittees

and will result in a standard for this application.

Sensory Science and Advancements in

Methodology Development

0042Developments in data collection methodologies in the

evolution of sensory science have included landmarks

such as the nine-point hedonic scale, category scaling

systems for descriptive analysis, the line scale method

of Quantitative Descriptive Analysis, ratio scaling

methods for both perception and preference, and the

labeled magnitude scale, first developed for the study

of pain perception and now applied to odor and taste

measurements as well as to preference measurements.

0043Advances in sensory methodology have been en-

hanced by the recent advances in computer technol-

ogy and the easy availability of both hardware and

software for the complex analyses that are often

required. This has provided sensory testing with a

greater flexibility in data collection and analysis.

The sensory laboratory is no longer limited to paper

ballots and the more cumbersome manual handling of

data. Laboratories can now be equipped with data

acquisition hardware and software in the form of

computer terminals that allow the ballots to be pre-

sented and recorded as data without requiring

manual handling of the data. Other systems allow

the collection of consumer information using tele-

phone and internet formats.

0044Consumer testing has evolved greatly since its first

uses in sensory science with the work of the US

Quartermaster Corps and the evaluation of military

rations. Part of this was the development of the

nine-point hedonic scale and the application of

this to preference testing. While this method is still

used, the current work in this field now includes

greater breadth in the considerations given to the

information which is needed to make product deci-

sions from consumer responses. This includes the

interactions of preference with underlying sensory

mechanisms and with various cultural experiences

that cause preference to vary between different

groups. These groups can be characterized by factors

such as sensory acuity to particular stimuli and par-

ticular demographics such as age, cultural back-

ground, etc. This has led to the developments in

methodology such as sensory segmentation, prefer-

ence mapping, applications of advanced design

methods such as conjoint analysis, and structural

equation modeling.

0045The use of expert assessors has developed as well.

Expert sensory panels for complex descriptive analy-

sis have evolved in both training and application

systems. The use of expert assessors in regulatory

and industrial applications is well established and,

in current practice, relies on the basic knowledge of

sensory science methodologies.

SENSORY EVALUATION/Sensory Characteristics of Human Foods 5129

Resources in Sensory Science

0046 Sensory evaluation is currently recognized as an es-

tablished discipline with courses and/or programs in

many universities all over the world. Major inter-

national food corporations have units specializing in

sensory analysis. Additionally, there are independent

companies with facilities specializing in sensory ap-

plications and advice in different areas of application.

0047 Two contemporary international events especially

for sensory scientists are the Pangborn Sensory Sci-

ence Symposium and meetings of the Sensometric

Society, which are now scheduled for alternate

years. Historical attendance at these meetings has

been impressive, with 600 registrants for the 2001

Pangborn Symposium in France.

0048 Publications are keys to communication in any

discipline. Chemoreception References circulated by

the European Chemoreception Research Organiza-

tion since 1984 has proved to be a particularly valu-

able resource for sensory scientists. In 2002, the

Sensometric Society chose Food Quality and Prefer-

ence as its official journal. The primary criterion for

acceptance for this journal is that the work include

some form of human measurement. Standards for,

and information on, testing methodology have been

developed by the ISO and ASTM and can be ordered

through their respective websites.

Conclusion

0049 Sensory evaluation is a good example of an interdis-

ciplinary success. Its maturation has been relatively

rapid in the history of science, due to the strong

leadership provided by imaginative researchers span-

ning a range of traditional and applied disciplines. Its

evolution continues in all areas in which human

perceptions are used to evaluate food and other

products.

See also: Sensory Evaluation: Food Acceptability and

Sensory Evaluation; Descriptive Analysis; Appearance;

Texture; Aroma; Taste; Spices and Flavoring

(Flavouring) Crops: Use of Spices in the Food Industry

Further Reading

Amerine MA, Pangborn RM and Roesler EB (1965)

Principles of Sensory Evaluation of Food. New York:

Academic Press.

Boring EG (1942) Sensation and Perception in the History

of Experimental Psychology. New York: D. Appleton

Century.

Chambers IVE and Wolf MB (1996) Sensory Testing

Methods, ASTM MNL 26. West Conshohocken, PA:

ASTM.

IFT Sensory Evaluation Division (1981) Sensory evaluation

guide for testing food and beverage products. Food

Technology 35: 50–59.

Lawless HT and Heymann H (1999) Sensory Evaluation of

Food – Principles and Practices. Gaithersburg, MD:

Aspen.

Lawless HT and Klein BP (1991) Sensory Science Theory

and Application in Foods. New York: Marcel Dekker.

McBride RL and MacFie HJH (1990) Psychological Basis

of Sensory Evaluation. London: Elsevier Applied

Science.

Meilgaard MC, Civille GV and Carr BT (1999) Sensory

Evaluation Techniques, 3rd edn. Boca Raton, FL: CRC

Press.

Moskowitz HR (1988) Applied Sensory Analysis of Foods,

vol. 1. Boca Raton, FL: CRC Press.

Munoz AM, Civille GV and Carr BT (1992) Sensory Evalu-

ation in Quality Control. New York: Van Nostrand

Reinhold.

O’Mahony M (1986) Sensory Evaluation of Food. Statis-

tical Methods and Procedures. New York: Marcel

Dekker.

O’Mahony M (1995) Sensory measurement in Food

Science: Fitting methods to goals. Food Technology 49:

72–82.

Pangborn RM (1989) The evolution of sensory science and

its interaction with IFT. Food Technology 43: 248–256,

307.

Peryam DR (1990) Sensory evaluation – The early days.

Food Technology 44: 86–91.

Schutz HG (1998) Evolution of the sensory science discip-

line. Food Technology 52: 42–46.

Stone H and Sidel JL (1993) Sensory Evaluation Practices,

2nd edn. New York: Academic Press.

Szczesniak AS (1998) Sensory texture profiling – Historical

and scientific perspectives. Food Technology 52: 54–57.

Yantis JE (1992) The Role of Sensory Analysis in Quality

Control. Philadelphia, PA: American Society for Testing

and Materials.

Food Acceptability and Sensory

Evaluation

J M Murray, Quest International Australia, Smithfield,

Australia

I A Baxter, Food Science Australia, North Ryde,

Australia

Background

0001Food acceptability is affected by many factors, which

may be related to the individual, the food, or the

environment in which the food is consumed. Accept-

ability is a subjective measure based on hedonics

5130 SENSORY EVALUATION/Food Acceptability and Sensory Evaluation

(pleasure), which in turn is influenced by the sensory

properties of the food, previous exposure to it and

subsequent expectations, contextual factors, an

individual’s culture, physiological status (i.e., hunger,

thirst, and presence/absence of illness), and many

other variables. The measurement of food acceptance

is highly complex and relies on psychometrics (scales)

and/or behavioral models (food-choice models). This

article aims to discuss some innate factors that

impact on foods’ sensory acceptability and how

these can be measured and understood using sensory

evaluation.

Food Properties and Sensory

Acceptability

0002 The acceptability of a food in sensory terms is influ-

enced by the intrinsic properties it possesses, that is,

the appearance, aroma, flavor, texture, aftertaste, and

auditory attributes of the food. The interaction be-

tween these variables is also an important consider-

ation, and the manifestation of sensory attributes can

be affected by factors such as temperature, portion

size, and so on.

0003 Appearance of a food provides cues regarding its

acceptability, e.g., indicating freshness or lack of (the

turgid appearance of lettuce, the wrinkled skin of an

apple that is past its best, the appearance of green

mold on cheese), ripeness (greening in tomatoes), or

skill in preparation (e.g., absence of burning, absence

of lumps in mashed potato). Color affects accept-

ability, influencing taste thresholds, taste perception,

expected food pleasantness, and acceptability. For

example, certain colors are associated with particular

flavors so that a yellow sweet may taste ‘sour’ as it is

associated with lemons, and foods that are intensely

colored may be perceived as more intense in flavor.

0004 Aroma suggests the flavor to be expected when

eating foods and thereby influences acceptance. Vola-

tile compounds are generally responsible for aroma

and vary from one food to another; for example,

brussel sprouts are rich in volatile sulfurous com-

pounds that contribute to their characteristic odor

and flavor, thus influencing acceptance among many

consumers. Aroma is also a good indicator of the

freshness of a product; think of yourself sniffing a

carton of milk to check the freshness before using it.

In this respect, aroma can also be regarded as a

‘defense mechanism,’ by providing the consumer

with information about the safety of a food prior to

consumption.

0005 Flavor is frequently regarded as the most influential

attribute for sensory acceptance, and perception of

flavor is a complex process involving the sense of

smell, taste, and chemesthesis. A distinction is often

made between the terms ‘taste’ and ‘flavor.’ The

former refers to the four basic tastes, sweet, sour,

salty, and bitter (a fifth, umami has also been pro-

posed), and the latter refers to a combination of

‘taste’ and ‘aroma,’ with the volatile components

of the food mainly contributing to the overall per-

ception. Other sensations such as ‘chilli burn,’

perceived by the trigeminal nerve, and the ‘cooling’

sensation of foods such as mints contribute to the

overall flavor.

0006The acceptability of a ‘flavor sensation’ changes

throughout life, and genetic components have been

proposed for sensitivity to certain bitter compounds,

e.g., phenylthiocarbamide (PTC) and 6-n-propylthi-

ouracil (PROP), with consumption of bitter-tasting

foods being inversely related to sensitivity. Humans

are born with particular taste preferences, indicated

by newborns whose facial expressions show a prefer-

ence for sweet, indifference to salt, and negative reac-

tions to bitter flavors. Evidence suggests that these are

innate responses, whilst throughout the development

of a child to adulthood, a combination of innate and

learned responses determine food acceptability. Aging

has a diminishing effect on our taste and aroma

thresholds and the ability to differentiate particular

flavors, and thus ultimately influences acceptability.

Cultural differences in the perception of flavor have

also been suggested but have yet to be proved. Flavor

perception and release has been the subject of a vast

amount of research in the last 10 years and is influ-

enced very much by the food matrix and individual

characteristics of the consumer (e.g., rate of saliva

production in the mouth). The interaction between

flavor and texture is vitally important, with the per-

ception of a flavor changing, depending on the tex-

ture of the food. For example, the same amount of

cheese flavor on a hard and crunchy corn chip will be

released and perceived differently to that on a puffed

(extruded) chip.

0007Texture is perceived by the senses of sight (visual

texture), touch (tactile texture), and sound (auditory

texture), and its contribution to overall food quality

can be relatively simple (a cup of coffee, pure

ed soup)

or complex (steak, potato chip). Consumers use tex-

ture information for classifying foods, but they do not

normally volunteer comments on texture, unless it is

definitely different from the consumers’ expectation,

it has been suggested that texture awareness is usually

on the subconscious level, with awareness heightened

only when expectations are violated, or nonfood as-

sociations are triggered. Auditory texture in some

foods is critical for overall acceptability: dried break-

fast cereals that do not snap, crackle, and pop, cola

that does not fizz, or snacks that do not crunch would

lose their inherent identity and acceptability. The

SENSORY EVALUATION/Food Acceptability and Sensory Evaluation 5131

acoustic properties of foods are also reliable indica-

tors of freshness; imagine an apple that did not crunch

when first bitten.

0008 Visual texture contributes significantly to overall

acceptability and suggests the texture to be perceived

upon consumption. Visual texture can suggest the

freshness of the food; for example, a cut slice of

cake may look moist or dry and thus may be per-

ceived as stale or fresh. Tactile texture perceived

upon consumption is related to the food matrix and

size and shape of the food being consumed. Mouth-

feel characteristics (which can be, for example, vis-

cosity-, carbonation-, or body-related) impact signifi-

cantly on acceptance depending on the food, and

factors such as melting are vital attributes for categor-

ies of food such as icecream and chocolate. The per-

ception of texture is dependent mainly on the teeth

and masticatory musculature. Infants are able to

consume only liquids until the musculature has de-

veloped sufficiently to allow them to ingest semi-

liquid (pure

ed) foods. The development of teeth

affords children the ability to deal with foods that

need biting, tearing, and chewing. In adulthood,

most texture challenges can be dealt with, but ageing

and the subsequent decline in the quality of dentition

and musclature impact on the acceptability of certain

textures.

0009 The manifestation of sensory properties can also be

affected by factors such as the temperature and por-

tion size, and thus influence perception and accept-

ability. Changes in temperature can alter the viscosity

of a sauce (and thus the texture sensation). The per-

ception of the basic tastes can also be altered: a cold,

sweetened beverage is perceived to be less sweet as the

tongue is cooled, but as the temperature of the drink

rises, the tongue is not cooled, and the drink’s sweet-

ness seems to increase. The amount of food placed in

the mouth also affects perception: flavor may be per-

ceived as more intense when a larger piece of food is

eaten, and less intense with a smaller portion.

Determining Acceptability by Sensory

Evaluation

0010 Many complex variables of the food interact with the

consumer to influence sensory acceptability, and as

such, objective measurement of sensory acceptability

proves difficult. However, there is an increasing re-

quirement to understand and measure sensory prefer-

ences reliably within the food industry. As the focus of

most food manufacturers’ product development,

quality assurance, and formulation efforts, it is vital

to ensure that the results of these sensory tests are

valid. Attention must therefore be paid to the type of

outcomes required from the test, the design of the

experiment, sources of bias in sensory evaluation,

and the use of appropriate participants. Today, there

are a number of approaches that can be used for

eliciting both qualitative and quantitative informa-

tion relating to foods sensory properties and their

subsequent acceptability. These range from simple,

‘I like it, I don’t like it’ tests, to complex modeling

procedures that uncover sensory reasons for prefer-

ence and relate these to the physical and chemical

properties of the food. Both qualitative and quantita-

tive techniques are described below for measurement

of sensory acceptability.

Hedonic Testing

0011Hedonic trials are performed in order to obtain quan-

tified information on, for example, how much people

like a product or how much a new prototype is pre-

ferred over the others. They are conducted with large

representative samples of the population and encom-

pass both preference and acceptability tests.

Preference Tests

0012The most commonly employed preference test is the

Paired Preference test, which is used when one wants

to examine the preference of one product directly

against a second, as in the case of product improve-

ments, or determine parity or preference over a com-

petitive brand. Participants in a paired preference test

receive two samples simultaneously, and the panelist

is asked to identify the sample that is preferred. The

major advantage of this test is that it is simple, and

participants can easily grasp the task. The Nonforced

Preference test is identical to that described above,

except that it allows a no-preference option. This can

yield valuable insights into the consumer’s state of

mind, as there may be participants who genuinely

do not have any preference; however, this option

does complicate the analysis, the power of the test is

decreased, and it can give some consumers an ‘easy

way out,’ as they do not have to, and do not, make a

choice. In the case of Preference ranking tests, con-

sumers are asked to rank several products in either

descending or ascending order of preference or liking.

The advantage of this test is, again, that it is intui-

tively simple for the consumer and can be conducted

quickly and with very little effort, but the data from

different sets of overlapping products cannot be com-

pared. Visual and tactile rankings are normally quite

straightforward, but multiflavor and taste compari-

sons can be fatiguing.

Acceptability Tests

0013Acceptability tests still investigate liking for products

but do not provide a direct comparison of liking over

5132 SENSORY EVALUATION/Food Acceptability and Sensory Evaluation

other products. The most common acceptability scale

is the nine-point hedonic scale, which has labeled

categories for indicating a level of like or dislike.

The advantages of this scale is that it is simple and

easy to use, with a reportedly high level of reliability

and stability independent of region and, to some

extent, panel size. The neutral ‘neither like or dislike’

category is a valid response for some consumers,

although some do show ‘end use avoidance,’ where

the participants avoid using the extremes of the scale.

This serves as a warning to those who may be tempted

to truncate the scale further to seven or even five

points. Hedonic measurements can also be achieved

using scales such as smiley faces, but the pictures may

be distracting to the participant, and the results often

prove unreliable. Acceptability ratings may also be

given on unstructured line scales, anchored by a like

or dislike at each end, and magnitude estimation

scales have also been proposed for the measurement

of food acceptability. Variations on acceptability

scaling include the just-right scale, which measures

the desirability of a specific attribute, e.g., a scale may

be anchored with ‘not sweet enough’ at the left, ‘just

right’ in the center, and ‘much too sweet’ at the right.

However, as consumers must have a consensus as to

the understanding of the attribute in question, the use

of this scale is often confined to a few simple attri-

butes that are widely understood. The analysis of this

scale can also pose some problems, and the assump-

tion is made that the population is homogenous when

in fact there may be two or more consumer segments

that prefer different levels of the attribute(s) in

question.

Preference Mapping: Internal

0014 Internal preference mapping is based solely on the

acceptability information gained from hedonic tests

(commonly the nine-point hedonic scale) and provides

a multidimensional representation of the samples and

the position of consumers in relation to the samples. It

provides an illustration of the consumer samples likes

and dislikes, and the intensity of these, and addition-

ally allows a visual segmentation of the samples likes

and dislikes to be made. An internal preference map

may be derived by multivariate statistical techniques

such as principal-components analysis and is rela-

tively straightforward to interpret. However, one po-

tential limitation of this method is that it assumes that

sensory perception is equal over all consumers and

that differences in the rating assigned to products are

due to varying preferences and not differences in per-

ception. Figure 1 illustrates an internal preference

map for eight cheeses; the further a consumer is from

the center of the plot, the more discriminatory their

evaluation was. From the plot, we can determine that

cheese 3 was the least liked, and cheese 8 was most

liked.

0015Whilst an invaluable part of determining food sens-

ory acceptability and preference, hedonic tests and

their representation by internal preference mapping

give no indication of why a product is liked and thus

are sometimes combined with other sensory tests,

such as a descriptive sensory analysis in a procedure

known as External Preference Mapping, or analytical

information, such as that obtained from gas chroma-

tography–mass spectrometry experiments.

Preference Mapping: External

0016The technique of external preference mapping

(PREFMAP) requires an objective characterization

of products’ sensory attributes, achieved by descrip-

tive sensory analysis (conducted using qualified and

trained panelists), which is then related to hedonic

acceptability ratings for the product obtained from a

representative sample of consumers. Descriptive sens-

ory analysis has been recommended as the most ap-

propriate sensory tool to provide the sensory attribute

measures, because there is no prior knowledge con-

cerning the most important sensory characteristics.

Statistical modeling of the data, using techniques

such as response surface analysis or partial least-

squares regression, allows the complex nature of the

relationships between consumer preference data and

descriptive data to be established.

0017There are a number of models/approaches that can

be used for external preference mapping. The most

simple of these is the vectorial model, which operates

on the assumption that ‘more of attribute X is better’.

−0.25

Principal component 1 (24%)

−0.25 −30

−30

0.25

Principal component 2 (9%)

fig0001Figure 1 Internal preference map illustrating consumer prefer-

ence for cheeses (consumers are represented by dots, and

products are represented by numbers).

SENSORY EVALUATION/Food Acceptability and Sensory Evaluation 5133

In practice, this model has obvious limitations as liking

for a product attribute will increase until the point of

maximum pleasantness (the ‘bliss point’), beyond

which an increasing intensity may begin to impact

negatively on acceptability. The ideal point model,

which is more complex, identifies the optimum com-

bination of product attributes (‘X and Y are ideal’) for

maximum liking and assumes that some products have

too much of an attribute, and some products have too

little. This technique is useful for market researchers

who are involved in new product development and

trying to identify gaps in the market place (i.e., the

map can indicate consumer groups whose preferences

are not currently met by any of the products on the

market place). Less commonly used and more com-

plex models used for external preference mapping

include the elliptical ideal point model, and the

elliptical ideal point model with rotation.

0018 Consumers are inherently complex, and thus at-

tempting to describe consumer behavior using a limi-

ted number of models is difficult, with a large number

of consumers who are not explained by the resultant

maps. To overcome this problem, the AUTOFIT

model was developed, which selected the most ac-

ceptable model for each consumer, with some con-

sumers fitted by the vectorial model, some by the

ideal point model, and so on. However, this method

results in 30–50% of consumers still not being ex-

plained in the analysis and has obvious limitations.

The question of discarding consumers for the reason

that they do not ‘fit’ the analysis has also been de-

bated and should be considered carefully.

0019 Preference mapping can also be combined with

segmentation techniques, such as cluster analysis.

These analyses are conducted prior to creating the

external preference map, and the scores assigned by

each of the consumer segments input into the analy-

sis. This allows consumers with similar preference

behavior to be identified. An example of an external

preference map for Cheddar cheese, using clustering,

is illustrated in Figure 2. In this trial, few consumers

liked the light cheese, which was perceived as ‘bitter’

and ‘rubbery.’ Some consumers (cluster 4) liked

premium and vegetarian cheese, whose preference

appeared to be driven by a ‘buttery’ flavor and

‘smooth’ texture. The map also illustrates consumer

needs not met by the products currently on the

market. One group of consumers (cluster 5) would

prefer sweeter and more caramel-like cheese.

0020 In general, these statistical modeling techniques

have brought a wealth of information about product

liking; however, some practical considerations are

worthy of mention here. At present, at least six prod-

ucts must be used in the external analysis, and all

consumers must assess all products, as missing values

cannot be accounted for. The sample range must also

be considered, and their properties in the context of

each other evaluated. The reliability of the preference

map result is dependent on the quality of the external

(trained panel) data, and often it is necessary to

weight all attributes equally (standardise), as some

attributes that are highly discriminating (e.g., color)

can skew the analysis. Whilst most commonly used to

understand sensory preference, preference mapping

can also be used to understand physical or chemical

reasons for consumer preference. Finally, when

choosing a modeling method, it is worthwhile to

remember a commonly used rule of thumb: ‘80% of

the modeling benefit comes from the first 20% of

model complexity.’

0021Overall, preference mapping is increasingly being

used within the food industry as a tool to reduce the

risk of failure in new product development, increase

liking, and secure market share for existing products.

Focus Groups

0022Focus groups are an increasingly common method for

eliciting qualitative information about foods’ sensory

properties and how these may relate to food accept-

ability. The group should be a carefully planned

session designed to obtain the maximum amount of

relevant information on the product/attribute of

interest. A moderator skilled in consumer research

and the nuances of group dynamics should coordinate

the effort to ensure that the result is a balanced and

informative representation of all the participants’

views. The information from focus groups is often

Sweet

Premium

Buttery

Vegetarian

Smooth

Supermarket

Shiny

Color

Light

Bitter

Rubbery

Astringent

−6

X = 43%, 10% Y =65%, 16%

Cluster 3

Cluster 1

Cluster 5

Caramel

Moldy

Grainy

Med mature

Farmhouse

Vintage

Salty

Pungent

Acid

Crumbly

Cluster 4

Cluster 2

fig0002Figure 2 External preference map illustrating consumer

clusters’ preference for cheeses and their associated sensory

attributes.

5134 SENSORY EVALUATION/Food Acceptability and Sensory Evaluation

used as a starting point in sensory research and can be

used as the basis for developing sensory question-

naires/tests in order to elicit quantitative acceptability

information about products.

Repertory Grid Method

0023 Individual or one-to-one interviews are an alternative

method for eliciting qualitative information about

products’ sensory attributes and acceptability in situ-

ations where group dynamics, are perceived to be a

problem. The repertory grid method (RGM), which

has its origins in the field of psychology, is now com-

monly used to collect qualitative sensory data on

product attributes and acceptability. The interviewer

should be trained and familiar with the method in

order to ensure that the maximum amount of product

information is gained. Products are presented to the

interviewee in triads (groups of three), and the inter-

viewee is subsequently asked to describe the similar-

ities and differences between the products in terms of

their sensory properties. When no more attributes can

be generated, the next triad of products is evaluated.

The proposed advantage of RGM is that the continu-

ally changing frame of reference to which the partici-

pant is exposed allows a greater insight into the

details and intricacies of the product attributes than

simply asking the participant to ‘describe the sensory

properties of this product.’ The qualitative informa-

tion gained from this procedure is often used as the

basis for quantitative research, for example in free-

choice profiling studies (discussed below) or for fur-

ther sensory questionnaires/tests.

Free-choice Profiling

0024 Free-choice profiling (FCP) is a variation of descrip-

tive sensory analysis and can also be combined with

acceptability information to ultimately determine

foods sensory acceptability. FCP allows consumers

to generate their own terms to describe product per-

ceptions, as in the case of descriptive sensory analysis

and PREFMAP; the information can then be related

to hedonic measurements, thus allowing an under-

standing of the sensory attributes driving accept-

ability.

0025 Generally, for the attribute elicitation stage, a

method such as RGM is used to maximize the amount

of information gained. Once the vocabulary of attri-

butes to use has been decided upon, participants score

products for the attributes on line scales. In addition,

consumers rate the acceptability of the product. The

attribute data are then analyzed by a procedure

known as generalized procrustes analysis (GPA).

Through stages of translation, scaling, and rotation/

reflection, the GPA adjusts for several sources of

variation, such as assessors using different levels and

ranges of the measurement scales. The resulting plots

demonstrate consumer perceptions of products,

which are then further analyzed in conjunction with

the hedonic data in order to determine reasons for

consumer preference. Figure 3 illustrates children’s

preference for the sensory attributes of vegetables.

Children used FCP and generated individual vocabu-

laries to describe vegetables and rated these attributes

on line scales. In addition, they were also asked to

rate the acceptability of the vegetables. Sensory attri-

bute data were then analyzed by GPA, and the subse-

quently identified consensus attributes were then

correlated with the first two preference dimensions

of an internal preference map (produced using the

acceptability data). Vegetables perceived as ‘big’ and

‘hard,’ such as cauliflower were least liked whilst

those perceived as ‘soft’ and ‘small’ were most liked.

0026The advantage of using FCP to determine accept-

ability is that the descriptive sensory measurements

(conducted using highly qualified panellists who can

be expensive to maintain) are not required. Partici-

pants need only to be able to use a scale and be

consumers of the product under evaluation. It also

provides a more ‘consumer orientated’ opinion on the

products. However, sometimes, the handling of indi-

vidual ballots for each participant in an FCP study

can prove time-consuming, and the interpretation of

the results by the sensory analyst can also be challen-

ging. Although, ultimately, the attribute information

is not as discriminatory as that obtained from a

trained panel, FCP measurements when related to

hedonic information in this way provide valuable

acceptability information about product sensory

attributes.

Future Role of Sensory Evaluation in

Determining Food Acceptability

0027The area of sensory evaluation and its role in deter-

mining food acceptability have increased rapidly and

will continue to do so for many years to come. The

use of validated statistical models will increasingly

allow the results of sensory tests (such as FCP, de-

scriptive analysis, time–intensity flavor evaluations)

to be related to hedonic measurements of accept-

ability, uncovering not just the ‘what is liked’ but

the ‘why.’ Reverse control modeling, which allows a

feedback loop to be established between chemical/

physical, sensory, and acceptability data will be used

increasingly to understand product acceptability. The

integration of the sensory analyst into both the tech-

nical and marketing departments of organizations

will extend further the boundaries of the sensory

science discipline. Already, we have research that

SENSORY EVALUATION/Food Acceptability and Sensory Evaluation 5135

illustrates how generated product concepts ‘fit’ the

sensory characteristics of products, and how these

may be matched better. Such techniques will be used

increasingly in order to minimize risk in an increas-

ingly competitive and global food industry.

See also: Sensory Evaluation: Sensory Characteristics of

Human Foods; Appearance; Texture; Aroma; Taste