Sei sulla pagina 1di 9

Trends in Food Science & Technology 38 (2014) 104e112

Review

Temporal
Dominance of Likewise, it was pointed out by Piggott (1994) and
Dijksterhuis (1996) that elaborated processes involved in
food consumption, including mastication, salivation,
Sensations: A review tongue movement, swallowing and so on modified the
perception of both the intensity and the quality of aroma,
taste, flavor and texture from one moment to another.
Rossella Di Monaco*, Therefore, to truly understand how products are perceived
by consumers during the course of consumption and
Chengcheng Su, Paolo Masi and consequently, how their hedonic reaction is affected, tem-
Silvana Cavella poral sensory methodologies should be adopted (Lawless
& Heymann, 2010, chap. 8; Piggott, 2000).
Department of Agricultural and Food Science, Different temporal methodologies have been devel-
University of Naples, Via Universit
a 100, 80055 oped over the last few decades, some of them have
Portici, NA, Italy (Tel.: D39 081 2539456; fax: D39 largely been applied in literature, such as Time Intensity
081 7754942; e-mail: rossella.dimonaco@unina.it) and more recently Temporal Dominance of Sensations,
whereas others have been less used, such as Progressive
Profile (Jack, Piggott, & Paterson, 1994); Sequential
Food and beverages undergo a series of physical and chemical Profiling (Methven et al. 2010) and Dual Attribute
reactions during mastication, salivation and so on, moreover Time Intensity (Duizer, Bloom, & Findlay, 1996;
the perception of aroma, taste, flavor and texture changes Duizer, Bloom, & Findlay, 1997). This review focuses
too. Thus, conventional static sensory methods, which require on Temporal Dominance of Sensations (TDS), a rela-
judges to average their dynamic sensations and give only a sin- tively new dynamic sensory methodology. TDS will be
gle point evaluation, are bound to miss some significant prod- briefly compared to other static and dynamic descriptive
uct information. To overcome this drawback, different sensory techniques.
dynamic sensory methods have been developed and refined.
This review focuses on Temporal Dominance of Sensations
(TDS), a relatively new dynamic sensory methodology.
Temporal Dominance of Sensations
Temporal Dominance of Sensations (TDS) was devel-
oped at the Centre Europeen des Sciences du Go^ ut in
Introduction
the LIRIS lab in 1999 and was first presented at the Pang-
Conventional descriptive sensory techniques, such as Sen-
born Symposium by Pineau, Cordelle, and Schlich (2003).
sory Profiling, require judges to time-average or integrate
TDS studies the sequence of dominant sensations of a prod-
their changing perceived sensations to make only single
uct during a certain time period (Pineau et al., 2009). More
point evaluations (Cliff & Heymann, 1993), thus
precisely, it consists of identifying and sometimes rating the
providing an overall impression of attribute maximum in-
intensity of sensations perceived as dominant until the
tensity or averaged intensity instead of the time course of
perception ends. Subjects have to select a new dominant
a sensation. Although in many cases, this average mea-
attribute whenever they perceive a change in dominant sen-
surement might suffice to tackle problems, some vital in-
sations. Dominant has been defined as the sensation that
formation is bound to be missed because the process of
captures ones attention, the most striking perception, or the
tasting a food or drinking a beverage is not static. The
new sensation that pops up at a given time but not neces-
savoring of a food does not begin when it arrives in our
sarily the most intense sensation (Labbe, Schlich, Pineau,
mouth; it often lasts after food removal; it is influenced
Gilbert, & Martin, 2009; Pineau et al., 2009), or as a
by the pattern of taste stimulation and it takes time for
stand-alone definition (Le Reverend, Hidrio, Fernandes,
us to perceive its maximum intensity (Halpern, 1991).
& Aubry, 2008; Meillon, Urbano, & Schlich, 2009). TDS
judges do not require lengthy training and moreover several
* Corresponding author. attributes can be evaluated simultaneously by TDS.
0924-2244/$ - see front matter 2014 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.tifs.2014.04.007
R. Di Monaco et al. / Trends in Food Science & Technology 38 (2014) 104e112 105

Test procedure 2012), which enables meaningful product descriptions to


Different tests ranging from the triangular test, taste be obtained from TDS data. Goupil de Bouille, Pineau,
recognition, odor identification to intensity ranking are Meyners, Martin, and Schlich (2010) analyzing 15 TDS
used to select subjects. The optimum number of judges studies, found that judges behave differently with respect
have not been studied or verified to date. Good practice to how many and to which attributes they use, when the
could be to use rather more judges than that of conventional attributes list contains more than six. The same re-
descriptive methods (Pineau et al. 2012). Qualified judges searchers also found that judges use on average about 4
then undergo several training sessions, during which they attributes to describe a single sample and that attribute or-
are usually first introduced to the notion of the temporality der did not affect the attribute chosen. Pineau and co-
of sensations, then to the definition of dominance of sensa- workers proposed to keep the number of attributes to
tions. Since, as Meyners (2010) stated, dominance and in- around 8 to 10 based on the results of 21 TDS studies
tensity are two different but complementary concepts, in (Pineau et al., 2012).
TDS it is useful to use judges unaware of classical descrip- Tasting protocol must be agreed on and standardized by
tive methods. the judges, particularly when the tasting mode, such as in-
It is worth mentioning that judges need to be highly mouth manipulation, could bring about obvious different
motivated and dedicated to TDS measurement since they sensory perceptions (Su, 2013).
are required to continuously make a choice among several Once the judges have been trained in products sensory
attributes to determine the sequence of dominant sensations evaluation, no more than three sessions are required for
and in most cases their intensities as well (Ng et al., 2012; them to become familiar with TDS procedure (Meillon
Pineau et al., 2012). However, intensity ranking was found et al., 2009).
not to be really necessary (Pineau et al., 2012) and evidence Judges start the evaluation while putting samples into
shows that dominance rates alone can provide important their mouths. Once the chronometer starts, they identify
temporal information (Dinnella, Masi, Naes, & the sensations perceived as dominant and sometimes rate
Monteleone, 2013; Paulsen, Ns, Ueland, Rukke, & their intensities if required while performing the tasting
Hersleth, 2013). Meillon et al. (2009) suggested that judges protocol. The evaluation ends when judges can no longer
should not be over-trained, in order to avoid individual perceive sensations, and stop the chronometer (Meillon
typed-response which is a tendency for subjects to quote et al., 2009). Otherwise, data acquisition stops automati-
descriptors in the same order for all products. In TDS meth- cally after a certain period depending on the products and
odology the training should be more oriented to sample testing protocols. If TDS data acquisition is realized with
description in order to improve the list of dominant attri- the use of FIZZ software, all the selected attributes are
butes (Pineau et al. 2012). Thus, if the judges are not shown on the same computer screen with their correspond-
familiar with sample category, evaluation processes, and ing buttons or unstructured scales ranging from weak to
above all, the definition of the attributes, considerable strong when attribute intensities need to be assessed
time will still need to be invested in training. Consensus alike. Biosystemes was the first historical partner to
must be reached among judges on the definition of each develop TDS application but nowadays, Fizz is not the
attribute. only software that runs TDS evaluations, in fact most of
A selected list of attributes describing examined prod- the sensory software provides a TDS module (EyeQuestion,
ucts is generated during training. For attribute selection, Compusense, TimeSens.).
different methods have been adopted by different re- It must be pointed out that a chosen attribute is regis-
searchers. The most common way of building an attribute tered as dominant until another attribute is chosen. Dur-
list is to firstly provide judges samples with different ing a testing, judges are free to choose the same
perceived in-mouth properties to define the product scale, attribute for several times as long as they think it is
asking them to taste the samples and note down all the dominant, but they have to take into account that only
perceived sensations; secondly collect and compare the one attribute can be selected at each time. Conversely,
answers of assessors under the guidance of a panel leader they might not use all the attributes provided (Pineau
in group discussions, during which hedonic, quantitative et al., 2009). Based on 21 TDS studies from the TDS
and irrelevant descriptors are eliminated, and synonyms base, Pineau et al. (2012) found that judges tend to
are combined; finally, only the most frequently cited attri- choose a relatively constant number of attributes per
butes are selected and kept for further TDS analysis evaluation (average 4) regardless of the total number
(Albert, Salvador, Schlich, & Fiszman, 2012; Lenfant, of attributes on the list. To limit the preferential use of
Loret, Pineau, Hartmann, & Martin, 2009; Meillon the first attributes listed, descriptors must be presented
et al., 2009, 2010; Paulsen et al., 2013). Additionally, randomly across the panel but remain the same for a
attribute lists can be developed on the basis of initial sen- given judge during the entire evaluation, thus facilitating
sory profiling (Bruzzone, Ares, & Gimenez, 2013; Labbe him/her in the learning of the terms and scoring inten-
et al. 2009; Ng et al. 2012; Teillet, Schlich, Urbano, sities during an evaluation (Meillon et al., 2009;
Cordelle, & Guichard, 2010; Sokolowsky & Fischer, Pineau et al., 2012).
106 R. Di Monaco et al. / Trends in Food Science & Technology 38 (2014) 104e112

Analysis and interpretation of results


Usually the data collected during TDS are: name of the
dominant attribute; the time when an attribute is selected as
dominant; the duration of the dominance can be calculated
a posteriori and the intensity scored for each dominant attri-
bute is recorded only for protocols with intensity scales
(Labbe et al., 2009).
TDS curves show the dominance rates of attributes
against time for each sample (Bruzzone et al., 2013;
Meillon et al., 2010). Whether the TDS curves are normal-
ized or not, the Y-axis always stands for the dominance rate,
which is the percentage of selections of an attribute as
dominant at a particular time point (Ng et al., 2012). It is
calculated by dividing the number of citations of an attri-
bute (all replications) by the number of runs
(judge  replication). These dominance rates can be seen
as a reflection of consensus among judges and therefore a
measurement of panel performance (Pineau et al., 2009).
More precisely, since only one attribute can be selected
as dominant at any one time by any one judge, the sum
of the dominance rates of all the attributes at each point
of time for that judge is 1. The higher the dominance rate
of the attribute, the better the agreement among judges
(Albert et al., 2012). It is important to bear in mind that
TDS curves rely solely on the selection of an attribute as
dominant or not, no data concerning attribute intensity is
shown.
Fig. 1a gives an example of a TDS curve based on a
study on polenta sticks. To assist the interpretation of
TDS results, two lines representing the chance level and
Fig. 1. a) TDS curves of all the evaluated attributes for fried polenta
significance level are drawn on TDS curves (Fig. 1b (Di stick; b) TDS curves of dominant attributes for fried polenta stick.
Monaco et al., in preparation)).
Chance level is defined as follows (1):
when they are above significance level (Labbe et al.,
P0 1=p 1 2009; Lenfant et al., 2009; de Loubens et al., 2011;
Pineau et al., 2009; Teillet et al., 2010).
where p is the number of attributes, standing for the domi- In some cases, for example, when the investigation is
nance rate that an attribute can obtain by chance (Pineau focusing on what happens just before swallowing, TDS
et al., 2009). Significance level, calculated from the confi- curves are preferably standardized; this leads to a better
dence interval of a binomial proportion based on a normal display of perception evolution and greater consensus
approximation, indicates the minimum value that must be across the whole panel (Lenfant et al., 2009; Pineau,
reached for the dominance rate to be considered as signif- Lenfant, Hartmann, & Martin, 2008). The duration of
icantly higher than the chance level (Albert et al., 2012; mastication -from the moment when the sample is placed
Deleris et al., 2011; Labbe et al., 2009; Meillon et al., into judges mouths, to the moment when the sample is
2010; Pineau et al., 2009). It is calculated as follows (2): ready to be swallowed e could be measured before each
formal session by using a chronometer. However, in other
Ps P0 1:645P0 1  P0 =n 2
1=2
situations, such as when in-mouth duration time is impor-
tant, non-standardized data are preferable (Meyners,
where Ps is the lowest significant proportion value 2010). Therefore, the choice between standardization and
(a 0.05) at any point of time for a TDS curve, and n is non-standardization depends on the objective of the study.
the number of runs (judges  replicates). Although applicable to liquid products (Ng et al., 2012),
Different studies have different panel level definition standardization of TDS curves is difficult to apply to
criteria. For instance, Albert et al. (2012) considers TDS studies on lingering (Pineau et al., 2008). Furthermore, af-
curves consistent at panel level when they rise from be- ter standardization, the X-axis of the TDS curve does not
tween chance and significance levels to above the latter, represent true consumption time but the mastication period
while others maintain that TDS curves are consistent from first scoring to swallowing (Lenfant et al., 2009), or,
R. Di Monaco et al. / Trends in Food Science & Technology 38 (2014) 104e112 107

from when judges click the start button to the moment they time and obtained an additional curve called the average
click the stop button or when recording stops automatically TDS curve. Used in combination with a representation of
(Ng et al., 2012). In addition to TDS curves, TDS differ- TDS curves for attributes for all products, the average
ence curves are commonly plotted for sample comparisons TDS curve gave information on the general sequence of
(Albert et al., 2012; Bruzzone et al., 2013; Lenfant et al., dominant sensations during the consumption of a certain
2009; Meillon et al., 2009; Pineau et al., 2009). The curves product category. Later on, Deleris and fellow workers
are drawn by subtracting the dominance rates of two sam- (2011) extracted mean duration D (the cumulative duration
ples for each attribute at each point of time. The difference for which a given attribute was selected) and mean time T
in dominance rate is only plotted when it is considered (the first time point at which a given attribute was selected)
significantly different from 0 according to a classical test for all judges in order to understand the influence of swal-
of comparison of binomial proportions (Pineau et al., lowing on aroma perception of alcoholic beverages. Thus, it
2009). Fig. 2 illustrates the comparison between two coffee can be seen that various parameters can be extracted from
based frozen creams using TDS difference curves (Su, raw TDS data and recombined to fulfill different research
2013). targets.
The total dominance of an attribute is also used for TDS In contrast to the simple process of data extraction, the
data interpretation. During an evaluation, this parameter is statistical analysis of TDS data is troublesome. In most
defined as the area between significant level and TDS research, TDS data analysis is restricted to the description
curves. It takes into account both the dominance rate and of TDS curves, in such cases comparisons between
duration of an attribute. Despite its common application different samples can be made by the qualitative evaluation
in TI (Cliff & Heymann, 1993), it was not applied to of the curves (Lawless & Heymann, 2010, chap. 8). There
TDS analysis until Bruzzone et al. (2013) conducted a were no valid statistical inference tests until 2010, when
research on yoghurt texture perception, in which a sample Meyners and Pineau, having found two important limita-
map based on attribute dominance was generated to better tions of the test for binomial proportions proposed by
study the influence of different yoghurt formulations on Pineau et al. (2009) proposed an approach based on
texture characteristics. Although on one hand this proce- randomization tests. Firstly, it relies on a sample approxi-
dure is a quick and effective way to interpret representa- mation involving the normal distribution that cannot be
tions of both samples and attributes, on the other hand it applied to the TDS data based on small samples and small
could be considered as just another tool to convey the elicitation probabilities; secondly but no less important, the
same results of a classical sensory profiling, since any tem- test assumes independent samples, even though in the TDS
poral aspect of the data is removed if the areas under the experiments usually each subject evaluates all the samples.
curves are used. Again, the use of PCA with only five vari- In the new method, TDS data were unfolded into matrices
ables, as reported by Bruzzone et al. (2013), is worthy of so as to calculate the Euclidean distance, thus allowing the
consideration. comparison of global and paired samples (Meyners &
In a study on cereals, Lenfant et al. (2009) averaged the Pineau, 2010), in addition to the assessment of panel agree-
dominance rates of each attribute for all products at a given ment (Meyners, 2011). Even though, according to the au-
thors, this approach, when applied to real data, gave good
results, the main limitations are the long computing time
and the dependency on the randomization used. The authors
also proposed a graphical method to show the many p-
values derived from it.
To assess panel and assessor performance, Pineau and
fellow workers proposed the use of aggregated frequency
values of attribute dominance in a given number of time pe-
riods rather than at all time points collected (Pineau,
Neville, & Lepage, 2011). Although it involves the applica-
tion of a simple ANOVA model to TDS data, making it
easy to implement and fast to run, this approach has not
been used in any other published research to date, due to
a lack of validity (Dinnella et al., 2013).
To enable the evaluation of product differences,
Dinnella et al. (2013) introduced the investigation of
the residuals of ANOVA models by using histograms,
qeq and scatter plots, which validated the proposed
testing procedure. The procedures of computation and
Fig. 2. TDS difference curves of two different coffee based frozen calculation of frequency values are as follows: first, the
creams. TDS curve of each sample must be inspected visually
108 R. Di Monaco et al. / Trends in Food Science & Technology 38 (2014) 104e112

and compared in order to split the entire evaluation Application of TDS and comparison with other
period into smaller equaled intervals based on the results methods
as proposed by Pineau et al., (2011); second, the domi- TDS has already been used for various products with
nant attributes for each sample must be summarized in different texture properties, ranging from liquid to semi-
a table of dominant attributeesample pairs; thus, the fre- solid and solid food items (Table 1).
quency value for each dominant attributeesample pair In spite of its incapability of discriminating the tastes of
from each judge in each repetition in each interval are water (Teillet et al. 2010), TDS proved its efficiency in dis-
computed. Then, the summarized frequency value for tinguishing other product categories, such as wines
each dominant attribute in all corresponding samples differing in bitterness and other attributes (Sokolowsky &
for each judge in each interval based on the whole gener- Fischer, 2012), yoghurt with different texture properties re-
ated set of aggregated frequency values must be calcu- sulting from different formulations (Bruzzone et al., 2013)
lated. A judge whose summarized frequency values and so on. Additionally, TDS enabled the evaluation of a
equal zero, indicating that he never uses this dominant sequence of sensations generated by the oral trajectory dur-
attribute to describe any corresponding samples, is ex- ing mastication of breakfast cereals (Lenfant et al., 2009),
empted from further analysis. In this way, judges whose fish sticks (Albert et al., 2012) and biscuits with different
descriptive vocabularies are not in consensus with the fiber and fat contents (Laguna, Varela, Salvador, &
rest of the panel are excluded. Then, the frequency values Fiszman, 2013). In a recent study of Varela, Pintor, and
generated in all replicates by each remaining judge for Fiszman (2014) TDS highlighted the effect of hydrocol-
each dominant attributeesample pair in each time inter- loids in changing the dynamic perception of ice cream
val must be independently analyzed by using one-way texture attributes.
ANOVA with sample as the factor. The sample difference Studies on gels (Labbe et al., 2009), dealcoholized wines
distinguishing capacity of each judge is then assessed by (Meillon et al., 2009) and blackberry squashes (Ng et al.,
the generated p values. Finally, the pre-processed data 2012) revealed that TDS can even provide information on
must be submitted to a regular mixed model ANOVA sensations perceived after food consumption, such as a
with subject, sample and interaction effects, with the refreshing sensation, which suggests its promising applica-
latter two effects considered as random. To check the val- tion in the investigation of food experience or emotion (Ng
idity of the ANOVA model, the distributions of the resid- et al., 2012).
uals are further analyzed by histograms, qeq plots and TDS and TI methods, both of which are dynamic tests,
plots of estimated values vs. the residuals. have been compared in some studies. In 2008, Le Reverend
At present, the procedure proposed by Dinnella et al. et al. characterized the specific sensory properties of six hot
(2013) has not been applied in other published papers. beverage products using both TDS and TI. By using three
Probably the reason being that although apparently simple different statistical approaches, they found that TI and
and useful at first glance, in practice there are many limita- TDS produced similar information in terms of differences
tions and unresolved issues. between products, attributes, and evolution over time
The first important contradiction of this data analysis despite TDS being quicker in obtaining information. How-
is the condensation of temporal dominance data into ever, TI was found to be more suitable for the determina-
just 3 or 4 intervals. This is conceptually wrong, at this tion of the kinetics of one specific attribute whereas TDS
point it would be more appropriate to use another method could be used to illustrate product perception as a function
such as a quick flash profiling, or a repeated CATA on of time. Besides, extra information concerning the
these few attributes - giving all applicable attributes, sequence of sensations and multidimensionality, where
not just the one that is dominant, and thus more informa- interaction between attributes is required, can be provided
tion. Another important limitation is the choice of the in- by TDS.
tervals e how does this affect results and interpretation? The same conclusions were drawn by Vazquez-Ara ujo,
It is suggested that this is done by close inspections of Parker, and Woods (2013). They compared TI, TDS and
the curves e how to avoid selection bias and self- drinking profile in order to evaluate the flavor of beers.
fulfilling prophecies? Besides, by considering, for each The drinking profile method (DP) allows the intensity of
attribute, only some samples and/or some subjects for several attributes during different sips to be recorded,
the analysis is contradictory to the fact that subjects with each sip being spaced several seconds apart. Jack
have to select one dominant attribute at a time and within et al. (1994) first used this procedure, called progressive
a fixed list. The proposal to exclude judges from the profiling. Both TI and TDS highlighted significant temporal
panel based on their evaluation is absolutely in contradic- differences among the samples, whereas DP gave the same
tion with the suggestions of Meillon et al. (2009) early information found from descriptive analysis, without
stated on panel training. Moreover, this proposal is not revealing the flavor evolution after five sips.
able to cope with the whole dataset, but only to some spe- TDS and TI were also compared by Pineau et al. (2009)
cific areas for which ANOVA assumptions are not by studying five different liquid dairy products. Similar
broken. conclusions were drawn, TDS and TI globally provided
R. Di Monaco et al. / Trends in Food Science & Technology 38 (2014) 104e112 109

Table 1. Published TDS studies.a

Food category Number of judges Other methodb Author Year


Liquid Hot beverages 12 TI Le Reverend et al. 2008
Liquid dairy products 16 TI Pineau et al. 2009
Dealcoholized red wines 16 SP Meillon et al. 2009
Red wines 8 e Meillon et al. 2010
Water 16 SP Teillet et al. 2010
Flavored vodka 10 e Deleris et al. 2011
White wines 18 QDA, TI Sokolowsky & Fischer 2012
Blackcurrant squashes 11 QDA Ng. et al. 2012
Espresso coffee 16 HS, NS Barron et al. 2012
Coffee 13 e Dinnella et al. 2013
Beer 12 TI, DP Vazquez-Ara ujo et al. 2013
Semi-solid Gels 12 SP Labbe et al. 2009
Gelled candies 12 SP Saint-Eve et al. 2011
Cheese 16 e de Loubens et al. 2011
Yoghurt 10 QDA Bruzzone et al. 2013
Ice cream 14 e Varela et al. 2014
Solid Wheat flakes 25 e Lenfant et al. 2009
Fish sticks 9 KASP Albert et al. 2012
Biscuits 13 e Laguna et al. 2013
Nuts 40 e Hutchings et al. 2014
Food combination EVOO with vegetables 13 SP Dinnella et al. 2012
Salmon-sauce 9 QDA Paulsen et al. 2013
Notes.
a
A broad search for papers related to the TDS, from the 2003 until the 2013, using major databases (i.e., Web of Knowledge, SCOPUS) was
conducted.
b
TI e Time Intensity; SP e Sensory Profiling; QDA e Quantitative Descriptive Analysis; HS e Headspace; NS e Nose-space; DP e Drinking
Profile; KASP e Key Attribute Sensory Profiling.

compliant and coherent results, but in general they are not previous studies on products with different texture proper-
designed for the same needs. TI focuses on the evolution of ties (white wines (Sokolowsky & Fischer, 2012), fish sticks
the intensity of one attribute at a time, thus being more suit- (Albert et al., 2012)), especially on semi-solid products,
able when the intensity of a specific attribute needs to be including gels (Labbe et al., 2009), gelled candies (Saint-
followed carefully over time. When several attributes Eve et al., 2011) and yoghurt (Bruzzone et al., 2013).
have to be compared over time, TDS seems better suited Nevertheless, the qualitative and quantitative informa-
because the process of picking dominant attribute is some- tion provided by static and dynamic techniques are not al-
what of a comparison process. Furthermore, TDS is a less ways the same. By studying 9 different gels using sensory
time-consuming and multi-attribute temporal method profiling and TDS, Labbe et al. (2009) found that the dy-
which highlights the interactions among attributes namic perception differs from the immediate perception
(Deleris et al., 2011). of products with long lasting multi-sensory perception.
Apart from this, Sokolowsky and Fischer research on Similarly, by analyzing the impact of two Italian extra-
white wines (2012) shows that TI tends to exaggerate the virgin olive oils on the perceived sensory profiles of pureed
bitter intensity due to the focus on one taste only and beans and tomatoes, Dinnella, Masi, Zoboli, and
halo-dumping effects, which is prevented in TDS analysis. Monteleone (2012) found that the importance of a sensation
In fact, the evaluation of the bitterness of wine during clas- during food consumption is not necessarily the same as that
sical descriptive analysis (DA) did not reflect the maximum indicated by intensity ratings from static sensory profiles. In
intensity of the same taste as evaluated by TI. Results from other words, sensations with a moderate intensity or even
DA were higher compared to those from TDS, meaning that less can also be dominant in the dynamic profile; unripe
the globally perceived intensity of bitterness appears to be fruit flavor is an example of an unexpected attribute in an
more affected by the periods of dominance. extra-virgin Tuscan oil, which was perceived as dominant
Despite the different psychological mechanisms even when perceived at low intensity level.
involved in evaluating global (focused on one single rating) Compared to static sensory techniques, TDS has obvious
or temporal (choice of one attribute from a list over time) advantages thanks to its temporal component. In the study
sensations, good agreement between conventional descrip- of partially dealcoholized red wines (Meillon et al.,
tive profiling data and TDS curves was demonstrated in 2009), TDS generates double the number of attributes
110 R. Di Monaco et al. / Trends in Food Science & Technology 38 (2014) 104e112

that discriminate tested samples and enables the recognition alone are enough to obtain important information.
of temporal differences between wines. Similarly, studies Furthermore, several unpublished studies indicate a
on flavored gels (Labbe et al., 2009) and blackberry delay in the peaks of dominance curves when intensity
squashes (Ng et al., 2012) confirm the capacity of TDS to is included. Taking into consideration both this and the
estimate qualitative changes of dominant attributes during difficulty of assessing dominance and intensity simulta-
and after food consumption, which cannot be obtained neously, until now buttons have been used instead of in-
with conventional sensory profile. Moreover, TDS helps tensity scales to omit intensity evaluation in some
to better understand the dynamics of phenomena involved research studies.
in texture-flavor interactions (Saint-Eve et al., 2011) and (4) As indicated by some studies, judges are able to use
food component interactions (Dinnella et al., 2012), and different attribute types (taste, texture and aroma) in a
is therefore more suitable for the interpretation of consumer single evaluation. Nevertheless, the exact influence of
responses and complex sensation perceptions. Last, but not listing different sensory properties in the same list re-
least, TDS method is more rapid and the fact that attributes mains unknown.
are not evaluated independently could be in some cases an (5) Since TDS data can involve different attributes, in
advantage (Saint-Eve et al., 2011; Teillet et al. 2010). different orders and with different times of elicitations,
Despite all the merits that TDS possesses, it only enables the assessment of the repeatability at individual level is
the evaluation of dominant sensations (Ng et al., 2012; problematical. Although computing frequency values
Saint-Eve et al., 2011) and cannot therefore replace con- proves to be a useful approach of performance evalua-
ventional sensory profiling in the context of product devel- tion, especially when several sensations are perceived
opment today, as conventional sensory profiling allows as dominant during the same time interval, further in-
evaluation of a larger number of attributes. However, vestigations are still required to improve the
TDS can complement static descriptive analysis data since assessment.
the latter only yields single point information about a (6) Though it is suggested that using TDS together with in-
perception, which actually changes constantly during food nose measurement of aroma concentration could make a
consumption due to complex interactions among food com- difference in establishing the relationship between aroma
ponents, mastication style, initial food structure, etc. perception and aroma release, it remains hard to relate sen-
(Dinnella et al., 2012; Ng et al., 2012; Sokolowsky & sory and instrumental data due to the variety and
Fischer, 2012). complexity of the phenomena involved. In-depth research
As pointed out by Meyners (2010), TDS is conceptually needs to be carried out to validate this connection.
different from any method focused on perception intensity
such as TI or sensory profiling, therefore it cannot be
considered as a replacement, but simply as a method useful Conclusion
to analyze and interpret another sensory dimension. Despite the fact that Temporal Dominance of Sensations
To summarize, TDS and conventional sensory profiling has never been applied to studies of food appearance or
are designed to satisfy different needs and are generally orthonasal-odor properties, Temporal Dominance of Sensa-
recommended to be used together as complementary tech- tions has proven to be less time-consuming than time-
niques to obtain a more rounded product profile. intensity and conventional sensory profiling. In addition,
it provides supplementary information on the sequence of
Future developments sensations and qualitative changes perceived during the
As a newly developed temporal methodology, TDS still food consumption process that are not measurable with
suffers some gaps in knowledge and further research needs TI or conventional sensory analysis.
to be conducted in order to clarify the following aspects: The efficiency of TDS in distinguishing weak-sensory-
difference samples and describing the evolution of different
(1) How the dominance is defined and interpreted by the sensory perceptions (texture, taste and aroma) during and
judges may have an impact on TDS results, and, after food consumption have been evidenced by studies
accordingly to Meyners (2010), further research is on complex products that arouse dynamic sensory.
needed to understand how the dominance is conceptu- Furthermore, TDS is believed to be more appropriate to
alized by both judges and consumers. explain consumer responses than static descriptive analysis
(2) The optimum number of judges and replications for due to its temporal element. In addition, the dynamic infor-
TDS analysis have not been studied or verified to mation from TDS may contribute to the understanding of
date. Common practice uses slightly more judges than liking drivers (Varela et al. 2014). Besides, it is noteworthy
that of conventional methods, and three replicates are that even though liking was a temporal phenomena, only
almost always conducted in order to have no fewer few researchers attempted its measurement through a tem-
than 30 evaluations per product. poral procedure (Sudre, Pineau, Loret, & Martin, 2012).
(3) Although some researchers include intensity ranking in It is not difficult to say that TDS is a very useful sensory
TDS analysis, it has been shown that dominance rates methodology when taking into account all its advantages
R. Di Monaco et al. / Trends in Food Science & Technology 38 (2014) 104e112 111

and the ever increasing need to acquire temporal sensory aroma release and perception during the consumption of alcoholic
information on food products. beverages. Chemistry Senses, 36, 701e713.
Di Monaco, R., Volpe, S., Su, C. C., Masi, P., & Cavella, S. Effects of
However, it is necessary to investigate it together with storage conditions and different cooking procedures on dominant
more kinds of product to explore the application scope of attribute of polenta stick (in preparation).
TDS. And attention still needs to be paid when TDS is Dijksterhuis. (1996). Time-intensity methodology: review and
used for the following reasons: preview. In Proceedings, COST96 Meeting: Interaction of food
matrix with small ligands influencing flavour and texture, Dijion,
(1) It is not easy for judges to keep in mind all the attri- France, Nov. 20, 1995.
Dinnella, C., Masi, C., Naes, T., & Monteleone, E. (2013). A new
butes simultaneously during an evaluation and a non- approach in TDS data analysis: a case study on sweetened coffee.
negligible part of the panel can no longer handle the Food Quality and Preference, 30(1), 33e46.
whole list when the number of attributes exceeds 10. Dinnella, C., Masi, C., Zoboli, G., & Monteleone, E. (2012). Sensory
As a result, it is recommended to include a maximum functionality of extra-virgin olive oil in vegetable foods assessed by
of 10 attributes in a list to reduce the impact of different temporal dominance of sensations and descriptive analysis. Food
Quality and Preference, 26, 141e150.
judge behavior towards the list whereas such a limit Duizer, L. M., Bloom, K., & Findlay, C. J. (1996). Dual-attribute time-
does not exist in regular descriptive analysis. In a intensity measurement of sweetness and peppermint perception of
word, the selection or discarding of attributes could chewing-gum. Journal of Food Science, 61(3), 636e638.
be a real difficulty when using TDS. Duizer, L. M., Bloom, K., & Findlay, C. J. (1997). Dual-attribute time-
(2) Since only a maximum of 10 attributes can be intensity sensory evaluation: a new method for temporal
measurement of sensory perceptions. Food Quality and
analyzed in one evaluation, so TDS has to be com- Preference, 8(4), 261e269.
bined with other sensory methods in order to get a Goupil de Bouille, A., Pineau, N., Meyners, M., Martin, N., &
comprehensive and complete picture of the products Schlich, P. (2010). How do panelists use the list of attributes during
being analyzed. a temporal dominance of sensations experiment?. In Proceedings
(3) Since the primary information collected in TDS anal- of the 11th European Symposium on Statistical Methods for the
Food Industry (AgroStat); Benevento, Italy, February 23e26, 2010
ysis is the choice of a certain attribute as dominant, (pp. 179e186) Afragola (Na): Academy School.
and only this information will be used later it is difficult Halpern, B. P. (1991). More than meets the tongue: temporal
to define the attribute sequences that judges should characteristic of taste intensity and quality. In H. T. Lawless, &
reproduce. B. P. Klein (Eds.), Sensory science theory and applications in foods
(4) Although no training in intensity scoring is required, (pp. 95). New York: Dekker.
Hutchings, S. C., Foster, K. D., Grigor, J. M. V., Bronlund, J. E., &
judges should receive a sound training on the recogni- Morgenstern, M. P. (2014). Temporal dominance of sensations: a
tion and definition of different sensory qualities. This comparison between younger and older subjects for the
is because even though a consensus on the attributes perception of food texture. Food Quality and Preference, 31,
used to differentiate products can be reached, judges 106e115.
may use them differently, which will eventually bring Jack, F. R., Piggott, J. R., & Paterson, A. (1994). Analysis of textural
changes in hard cheese during mastication by progressive
differences to the results. profiling. Journal of Food Science, 59, 539e543.
(5) The attribute order across the entire panel should be Labbe, D., Schlich, P., Pineau, N., Gilbert, F., & Martin, N. (2009).
balanced so as to reduce the potential attribute position Temporal dominance of sensations and sensory profiling: a
bias, i.e., attributes at the top of the list tend to be comparative study. Food Quality and Preference, 20, 216e221.
selected earlier than those at the bottom. Laguna, L., Varela, P., Salvador, A., & Fiszman, S. (2013). A new
sensory tool to analyse the oral trajectory of biscuits with different
fat and fibre contents. Food Research International, 51, 544e553.
Lawless, H. T., & Heymann, H. (2010). Sensory evaluation of food:
Principles and practices (2nd ed.). New York: Springer.
References
Le Reverend, F. M., Hidrio, C., Fernandes, A., & Aubry, V. (2008).
Albert, A., Salvador, A., Schlich, P., & Fiszman, S. (2012). Comparison Comparison between temporal dominance of sensations and time
between temporal dominance of sensations (TDS) and key- intensity results. Food Quality and Preference, 19, 174e178.
attribute sensory profiling for evaluating solid food with Lenfant, F., Loret, C., Pineau, N., Hartmann, C., & Martin, N. (2009).
contrasting textural layers: fish sticks. Food Quality and Perception of food oral breakdown: the concept of sensory
Preference, 24, 111e118. trajectory. Appetite, 53, 659e667.
Barron, D., Pineau, N., Matthey-Doret, W., Ali, S., Sudre, J., de Loubens, C., Panouille, M., Saint-Eve, A., Deleris, I., Trelea, I. C., &
Germain, J. C., et al. (2012). Impact of crema on the aroma release Souchon, I. (2011). Mechanistic model of in vitro salt release from
and the in-mouth sensory perception of espresso coffee. Food & model dairy gels based on standardized breakdown test simulating
Function, 3(9), 893e986. mastication. Journal of Food Engineering, 105, 161e168.
Bruzzone, F., Ares, G., & Gimenez, A. (2013). Temporal aspects of Meillon, S., Urbano, C., & Schlich, P. (2009). Contribution of the
yoghurt texture perception. International Dairy Journal, 29, temporal dominance of sensations (TDS) method to the sensory
124e134. description of subtle differences in partially dealcoholized red
Cliff, M., & Heymann, H. (1993). Development and use of time- wines. Food Quality and Preference, 20, 490e499.
intensity methodology for sensory evaluation e a review. Food Meillon, S., Viala, D., Medel, M., Urbano, C., Guillot, G., &
Research International, 26(5), 375e385. Schlich, P. (2010). Impact of partial alcohol reduction in Syrah
Del
eris, I., Saint-Eve, A., Guo, Y., Lieben, P., Cypriani, M. L., wine on perceived complexity and temporality of sensations and
Jacquet, N., et al. (2011). Impact of swallowing on the dynamics of link with preference. Food Quality and Preference, 21, 732e740.
112 R. Di Monaco et al. / Trends in Food Science & Technology 38 (2014) 104e112

Methven, L., Rahelu, K., Economou, N., Kinneavy, L., Ladbrooke- is a good attribute list? Food Quality and Preference, 26,
Davis, L., Kennedy, O. B., et al. (2010). The effect of consumption 159e165.
volume on profile and liking of oral nutritional supplements of Pineau, N., Lenfant, F., Hartmann, C., & Martin, N. (2008). Temporal
varied sweetness: sequential profiling and boredom tests. Food dominance of sensation: some statistical improvements on TDS
Quality and Preference, 21, 948e955. curves. In 10eme journ e es europe ennes Agro-industrie et m
e thodes
Meyners, M. (2010). On the design, analysis and interpretation of statistiques. 22e25 January 2008, Louvain La Neuve, Belgium.
temporal dominance of sensations data. In Proceedings of the 11th Pineau, N., Neville, T., & Lepage, M. (2011). Panel performance tool
European Symposium on Statistical Methods for the Food Industry for temporal dominance of sensations studies. In Abstract book of
(AgroStat); Benevento, Italy, February 23e26, 2010 (pp. 45e53). 9th Pangborn Sensory Science Symposium, Toronto, Canada.
Afragola (Na): Academy School. Pineau, N., Schlich, P., Cordelle, S., Mathonniere, C., Issanchou, S.,
Meyners, M. (2011). Panel and panelist agreement for product Imbert, A., et al. (2009). Temporal dominance of Sensations:
comparisons in studies of temporal dominance of sensations. Food construction of the TDS curves and comparison with time-
Quality and Preference, 22, 365e370. intensity. Food Quality and Preference, 20, 450e455.
Meyners, M., & Pineau, N. (2010). Statistical inference for temporal Saint-Eve, A., Deleris, I., Panouille, M., Dakowski, F., Cordelle, S.,
dominance of sensations data using randomization tests. Food Schlich, P., et al. (2011). How texture influences aroma and taste
Quality and Preference, 21, 805e814. perception over time in candies. Chemosensory Perception, 4,
Ng, M., Lawlor, J. B., Chandra, S., Chaya, C., Hewson, L., & Hort, J. 32e41.
(2012). Using quantitative descriptive analysis and temporal Sokolowsky, M., & Fischer, U. (2012). Evaluation of bitterness in white
dominance of sensations analysis as complementary methods for wine applying descriptive analysis, time-intensity analysis, and
profiling commercial blackcurrant squashes. Food Quality and temporal dominance of sensations analysis. Analytica Chimica
Preference, 25, 121e134. Acta, 732, 46e52.
Paulsen, M. T., Ns, T., Ueland, ., Rukke, E. O., & Hersleth, M. Su, C. (2013). Application of temporal dominance of sensations (TDS)
(2013). Preference mapping of salmon-sauce combinations: the to the study of a coffee based frozen cream (Master thesis in Food
influence of temporal properties. Food Quality and Preference, 27, Innovation and Product Design). University of Naples, Federico II.
120e127. Sudre, J., Pineau, N., Loret, C., & Martin, N. (2012). Comparison of
Piggott, J. R. (1994). Understanding flavour quality: difficult or methods to monitor liking of food during consumption. Food
impossible? Food Quality and Preference, 5, 167e171. Quality and Preference, 24, 179e189.
Piggott, J. R. (2000). Dynamism in flavour science and sensory Teillet, E., Schlich, P., Urbano, C., Cordelle, S., & Guichard, E. (2010).
methodology. Food Research International, 33, 191e197. Sensory methodologies and the taste of water. Food Quality and
Pineau, N., Cordelle, S., & Schlich, P. (2003). Temporal dominance of Preference, 21, 967e976.
sensations: a new technique to record several sensory attributes Varela, P., Pintor, A., & Fiszman, S. (2014). How hydrocolloids affect
simultaneously over time. In Abstract book of 5th Pangborn the temporal oral perception of ice cream. Food Hydrocolloids,
Sensory Science Symposium, Boston, MA, USA, July 20e24, 36, 220e228.
2003, P121. Vazquez-Ara ujo, L., Parker, D., & Woods, E. (2013). Comparison of
Pineau, N., Goupil de Bouille, A., Lepage, M., Lenfant, F., Schlich, P., temporal-sensory methods for beer flavor evaluation. Journal of
Martin, N., et al. (2012). Temporal dominance of sensations: what Sensory Studies, 28, 387e395.

Potrebbero piacerti anche