Chem. Senses 28: 11-23,
2003
© Oxford University Press 2003
Oral Shear Stress Predicts Flavour Perception in Viscous Solutions
Samworth Flavour Laboratory, Division of Food Sciences, The University of Nottingham, Sutton Bonington Campus, Loughborough, Leicestershire LE12 5RD, UK
Correspondence to be sent to: David Cook, Division of Food Sciences, The University of Nottingham, Sutton Bonington Campus, Loughborough, Leicestershire LE12 5RD, UK. e-mail: david.cook{at}nottingham.ac.uk
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and methodsResultsDiscussionConclusionsReferences |
|---|
The perception of sweetness and flavour were studied in viscous solutions containing 50 g/l sucrose, 100 p.p.m. iso-amyl acetate and varying concentrations of three hydrocolloid thickeners (guar gum,
-carrageenan and hydroxypropylmethyl cellulose). Zero-shear viscosity
of the samples ranged from 1 to 5000 mPas. Perception of both sweetness and
aroma was suppressed at thickener concentrations above c*
(coil overlap concentration, the point at which there is an abrupt increase in
solution viscosity as thickener concentration is increased). Sensory data for
the three hydrocolloids was only loosely correlated with their concentration
relative to c* (c/c* ratio),
particularly above c*. However, when perceptual data were
plotted against the Kokini oral shear stress (
), calculated from
rheological measurements, data for the three hydrocolloids aligned to form a
master-curve, enabling the prediction of flavour intensity in such systems.
The fact that oral shear stress can be used to model sweetness and aroma
perception supports the hypothesis that somatosensory tactile stimuli can
interact with taste and aroma signals to modulate their perception.
Key words: flavour intensity, hydrocolloid thickener, oral shear stress, sensory perception, viscosity
|
Introduction |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionConclusionsReferences |
|---|
Increasing the viscosity of liquid foods with hydrocolloid thickeners is known to change the sensory properties of such systems. Both taste and aroma perception can be suppressed by increasing concentrations of hydrocolloid, although effects are specific to the taste-modality, flavouring and hydrocolloid concerned (Pangborn et al., 1973
; Pangborn and
Szczesniak, 1974). With sweetness perception, however, the effect
appears to be more generic. Investigators have reported a decrease in
perceived sweetness intensity in solutions thickened with a range of
hydrocolloids (Vaisey et al.,
1969; Moskowitz and Arabie,
1970; Christensen,
1980; Paulus and Haas,
1980; Izutsu et al.,
1981; Baines and Morris,
1987; Malkki et al.,
1993). This effect has been observed for both sugars and
high-intensity sweeteners of wide-ranging chemical nature and different modes
of taste chemoreception (Cook et
al., 2002).
Baines and Morris (Baines and Morris,
1987) investigated both sweetness and strawberry flavour
perception in solutions thickened with guar gum. They concluded that the key
determinant of flavour suppression in such systems was the thickener
concentration relative to its coil-overlap concentration
(c*). This is the point at which there is an abrupt
increase in solution viscosity as thickener concentration is increased. At a
molecular level, this phenomenon is interpreted as the point at which the
hydrocolloid chains begin to overlap in solution, reducing freedom of
molecular movement and resulting in a sharp increase in viscosity. At
concentrations below c*, Baines and Morris found that
sweetness and strawberry aroma perception were relatively unchanged, whilst
both were progressively suppressed at thickener concentrations above
c*. A theory was developed to model flavour perception in
a range of hydrocolloid matrices in terms of thickener concentration relative
to c* (Baines and
Morris, 1988). Baines and Morris proposed that perceptual changes
might be linked to inefficient mixing in solutions above
c*, inhibiting the transport of small taste and aroma
molecules to their respective receptors. Hollowood et al.
(Hollowood et al.,
2002) used real-time analysis of in-nose volatile release by
atmospheric pressure ionization mass spectrometry (API-MS)
(Taylor et al., 2000)
to show that retronasal benzaldehyde release was not substantially changed by
hydroxypropylmethyl cellulose (HPMC) at up to 2.1 times the
c* concentration. Perception of almond flavour was
nonetheless shown to decline sharply above c*, in
accordance with the results of Baines and Morris. Similar findings were
reported for ethyl butyrate and strawberry flavour perception at up to
3.5c* HPMC concentration. These results were explained in
terms of a taste—aroma interaction
(Noble, 1996;
Davidson et al.,
1999), with the perceived drop in sweetness of the system driving
the decline in perceived aroma, even though the concentration of aroma
compound reaching the olfactory receptors remained broadly constant. The key
question then concerns the mechanism behind the drop in perceived sweetness in
such systems.
When foods are consumed, perceived flavour is a result of the simultaneous
stimulation of three principal sensory systems
(Cerf-Ducastel and Murphy,
2001): taste, olfaction and the trigeminal system. The latter
comprises chemical, thermal and tactile stimulation of the somatosensory
system. One such tactile stimulus, oral viscosity, is perceived by
mechano-receptors in the mouth and in particular on the tip of the tongue
(Guinard and Mazzucchelli,
1996). Whilst taste—aroma interactions have been well
studied and are widely accepted, potential interactions between somatosensory
information and taste or olfaction have received less attention. The current
study investigated the relationship between oral viscosity and the perception
of sweetness and aroma in hydrocolloid systems.
Many workers have studied the oral perception of viscosity, yet few have
looked at the way that this interacts with food flavour. Two groups
(Cussler et al.,
1979; Kokini et al.,
1982) investigated the effects of increasing viscosity on taste
perception. However, in each case, the focus was on predicting how changes in
viscosity would affect diffusion rates in the hydrocolloid matrix, thus
reducing the flux of taste molecules to the tongue's surface. Cussler et
al. (Cussler et al.,
1979) reported some success in fitting sensory data by this
method; however, their diffusion coefficients were predicted theoretically
from high-shear viscosity measurements and not determined experimentally. They
also found that consideration of diffusion factors alone failed to predict the
perception of several taste compounds in thickened solution, in which cases
they assumed mass transfer to be non rate-limiting. However, there was no
clear rationale behind which compounds/tastes could be modelled in this
fashion and which could not.
Hydrocolloid solutions are non-Newtonian in nature, which means that they have different apparent viscosities (the ratio of shear stress to shear rate) dependent on the shear stress applied. Because of this, the apparent viscosity of such solutions is normally quoted together with the measured shear rate (the velocity gradient set up in a solution under applied stress). Hydrocolloids are typically shear-thinning in nature, meaning that as the applied shear stress is increased, the apparent viscosity decreases. The zero-shear viscosity, normally extrapolated from experimental data, is the viscosity as the shear rate tends to zero, and is therefore the highest apparent viscosity for shear-thinning fluids.
The sensory thickness, or oral viscosity, of shear-thinning hydrocolloids
thus depends on the shear stress applied to the fluid in-mouth and the
resultant shear rate. Wood (Wood,
1968) correlated the perceived texture of hydrocolloids with their
rheological flow properties and concluded that the stimulus associated with
the oral evaluation of viscosity was a shear stress developed in mouth at a
constant shear rate of 
50 s-1. Shama and Sherman
(Shama and Sherman, 1973)
looked at the perception of viscosity in a range of fluid and semi-solid
foods. They concluded that a much wider range of shear rates were operative
in-mouth (10-1000 s-1), dependent on the flow characteristics of
the food. The stimulus associated with the evaluation of fluid foods appeared
to be the shear rate developed at a constant shear stress of 10 Pa; for
highly viscous foods, the stimulus was reported to be the shear stress
developed at a constant shear rate of 10 s-1. Between these
extremes, there was a curved space defining the limits of oral viscosity
evaluation within the shear stress—shear rate plane. Richardson et
al. (Richardson et al.,
1989) continued attempts to correlate sensory perception of
thickness with rheological measurements. They reported that small deformation
measurements of dynamic viscosity (
*) under oscillatory shear
at 50 rad/s correlated directly with the perceived thickness of both
true solutions and weak gels. However, there was no clear underlying mechanism
for this correlation, and the authors concluded that the relevant sensory
stimulus was probably a related `large-deformation' property, such as the peak
shear stress developed in moving the sample from rest.
Jozef Kokini and co-workers used fluid dynamics to calculate theoretically
the shear-stress on the tip of the tongue resulting from the manipulation of
fluid samples in-mouth. This approach was developed in a series of
publications (Kokini et al.,
1977; Kokini and Cussler,
1983; Kokini,
1985). The calculation assumes a `parallel-plate' model
(Figure 1), with a cylindrical
plug of fluid between the tongue and roof of the mouth. The two plates are
slowly squeezed towards one another by a normal force W;
simultaneously, the two plates move steadily relative to one another at
velocity V. Calculation of the resultant shear stress was solved
mathematically (deMartine and Cussler,
1975), leading to equation (1)
(Figure 1). Kokini made what he
decided were reasonable assumptions for the forces, dimensions and velocities
involved (Figure 1), and held
these to be relatively constant for all samples. Under these conditions, the
principal variables in the equation predicting oral shear stress are
m and n, the power law constants for the fluid concerned.
Elejalde and Kokini (Elejalde and Kokini,
1992) applied this technique for predicting oral shear stresses to
the evaluation of a range of hydrocolloid solutions and real fluid foods. They
were able to show that sensory perception of thickness can be predicted by the
Kokini oral shear stress, based on simple rheological measurements. In
addition, they concluded that the shear stress in mouth is in fact the sensory
mechanism used for the oral evaluation of viscosity, since the exponent of the
psychophysical law relating sensory viscosity to shear stress was close to 1.
They were able to use this approach to model sets of data published by other
research groups.
|
Until now, no one had taken the further step of demonstrating a link between oral shear stress and the overall perception of flavour, as proposed in this paper.
|
Materials and methods |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionConclusionsReferences |
|---|
Stimuli
Solutions were flavoured with 50 g/l sucrose and 100 p.p.m. iso-amyl
acetate, and contained one of eight concentrations of each thickener used in
the study (HPMC, guar gum or -carrageenan; 24 samples in total).
Thickener concentrations were calculated relative to their
c* values, and were 0 (i.e. water),
c*/4, 3c*/4, c*,
5c*/4, 7c*/4,
10c*/4 and 14c*/4. Measurement of
c* for these samples has been described previously
(Cook et al., 2002).
Resultant values were 5.7 g/l (HPMC), 1.9 g/l (guar gum) and 4.8 g/l
(-carrageenan). The experimental design was weighted around
c* to focus on perceptual changes in this region, but
included a wide range of viscous stimuli, from 1 mPas (water) to 5 Pas
zero-shear viscosity (0) for the most viscous samples.
Hydrocolloids were dispersed in water using an overhead paddle stirrer at
200-600 r.p.m. Guar gum (purified grade; SKW Biosystems, La
Ferté-sous-fociaire, France) was dispersed in water at room
temperature, HPMC (Methocel, The Dow Chemical Company, Langhorne, PA) at
95°C and -carrageenan (Red Carnation Gums Ltd, Basildon, Essex,
UK) at 60°C. All three hydrocolloid dispersions were subsequently cooled
to 5°C and stirred for a further 6 h to ensure adequate hydration of
polymer chains. Samples for sensory analysis were prepared by weighing
appropriate quantities of hydrocolloid, water and sucrose into sample bottles
and mixing thoroughly on a roller bed (SRT2; Stuart Scientific, Redhill, UK).
A flavour concentrate of iso-amyl acetate (230 µl) in food-grade absolute
ethanol with yellow food colouring (total volume 10 ml) was prepared and added
to each solution at a rate of 0.5 ml per 100 ml to yield a final concentration
of 100 p.p.m. iso-amyl acetate. Flavoured solutions were then rolled for a
further 2 h, to ensure adequate mixing, prior to presentation for sensory
analysis. The yellow colouring acted as a check on adequate mixing of the
solutions and also masked some of the visual differences between samples (e.g.
degree of clarity).
Sample rheology
The flow characteristics of each solution were measured at 25°C on a
Bohlin CS-10 applied stress rheometer (Bohlin Instruments, Lund, Sweden).
Double-gap geometry was used for the less-viscous solutions, and cone and
plate geometry for samples with zero-shear viscosities of >600 mPas. The
resultant shear rate range was between 1 and 100 s-1. Measurements
were performed in duplicate on the same batch of solutions prepared for
sensory analysis. Flow curves were fitted to the Cross equation
(Cross, 1965) using CS-10
software, to enable calculation of the apparent viscosity at any shear rate in
the range and extrapolation to the zero-shear viscosity (y-axis
intercept). The power law region of each flow curve was fitted to equation (2)
in order to estimate the power law parameters of the samples.
 | (2) |
is the apparent viscosity, 
is the shear rate, m
is the consistency index and n is the power law index. API-MS analysis of volatile release from samples
Real-time in-nose release of iso-amyl acetate during consumption of the
samples was measured using the MS-Nose
(Micromass, Manchester, UK). Five
panellists consumed two replicates of each hydrocolloid solution, according to
a fixed protocol. They were asked to breathe in, sip 8 ml of solution from a
spoon, close their mouth and swallow the sample, then exhale and continue to
breathe normally whilst resting their nose on the MS-Nose nasal sampling
tube. Air from the nose was sampled directly to the source of an API-MS at 30
ml/min (Taylor et al.,
2000) and the release of iso-amyl acetate followed by monitoring
m/z 131 (the mass to charge ratio for the molecular ion). The in-nose
concentration of iso-amyl acetate was calculated by comparison with the signal
for a calibrant of known concentration
(Taylor et al.,
2000).
Subjects
An established and trained panel of 15 assessors (5 male, 10 female) aged
between 30 and 55 years, was used in the study. The panel had been recruited
and selected on the basis of their sensory acuity, in particular their ability
to distinguish between concentrations of the same stimulus. They received
specific training in the use of magnitude estimation against a fixed modulus
and had >2 years of experience in conducting such tasks [see, for example
(Hollowood et al.,
2002)].
Sensory analysis
Magnitude estimation against a fixed modulus was used to appraise the sweetness and flavour intensities of the solutions. The aqueous sample was used as the reference, and was assigned a score of 100 for both sweetness and iso-amyl acetate (`banana'/`pear-drop') flavour. Panelists were instructed to assign a score to each sample in turn, to represent the perceived intensity of a particular sensory attribute relative to the reference (= 100). To minimize the potential for halo-dumping effects, the panel were advised that they would have the opportunity to score the same samples for both sweetness and flavour in separate sessions, but to focus solely on the sensory attribute being scored in that particular session. Solutions were presented in groups of four for appraisal against a fresh reference sample. Presentation was ordered by a randomized complete block design and each panelist scored the reference as a sample twice in every panel session. The 24 samples were scored in duplicate for both sweetness and flavour (three panel sessions for each attribute).
Samples were presented in coded plastic cups and evaluations conducted in isolated booths under controlled lighting conditions. Solutions were tasted from a spoon (8 ml). Plain crackers and water were supplied to assist in cleansing the palate between tastings.
Sensory data analysis
The mean magnitude estimation scores from 10 sets of sensory data were selected for modelling. The criterion for inclusion was that the panelists' mean score for the reference when presented as a sample was 100 ± 25%. Consistency of scores when presented with replicates of the same sample was also taken into account when further selecting the 10 best sets of sensory data for modelling.
Mean sensory data was plotted in a conventional psychophysics log—log
format against c/c* and measured rheological properties of
the solutions [consistency index, oral shear stress () and the apparent
viscosity at shear rates of 0, 50 and 100 s-1]. These plots were
fitted with two linear regressions, corresponding to below
c* and above c* series.
Design Expert 6.0.2 software (Statease, Minneapolis, MN) was used to model the mean sensory data and to perform an analysis of variance (ANOVA).
Model validation experiment
Models for sensory perception, developed to fit the initial data set, were tested in a subsequent experiment using xanthan (1, 4 and 7.5 g/l) and methyl cellulose (4 and 8 g/l) thickeners, as well as different concentrations of HPMC (12c*/4) and guar gum (18c*/4). These were chosen to present a range of viscous stimuli across the breadth of the original model.
Xanthan gum (food grade; Red Carnation Gums Ltd) was dispersed at room temperature and methyl cellulose (The Dow Chemical Company) at 60°C. All other details of solution preparation were as described above.
Sensory and rheological testing was conducted as described previously. To calculate apparent viscosities at a range of shear rates for xanthan samples, the power law (equation 2) was used to fit flow curves, as these could not be adequately fitted with the Cross equation.
|
Results |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionConclusionsReferences |
|---|
Sample rheology
Rheological data for the samples are detailed in
Table 1 (mean of two replicate
analyses). Values for the Kokini oral shear stress () of each sample were
calculated from the power law parameters fitted to each flow curve, and using
the equation and assumptions shown in
Figure 1.
|
API-MS analysis of volatile release from samples
API-MS analysis showed that the in-nose release of iso-amyl acetate from
samples was relatively independent of thickener concentration
(Figure 2). The pooled area
data for the first three exhalations after swallowing reflect the total amount
of volatile release from the sample (>99%), since iso-amyl acetate is not
persistent on the breath (Linforth and
Taylor, 2000). The first exhalation peak concentration is
indicative of the maximum stimulus reaching the olfactory receptors.
|
Thus, neither the maximum concentration nor the total amount of iso-amyl acetate reaching the olfactory receptors was substantially affected by thickener type or concentration.
Sensory analysis
The perception of both sweetness and iso-amyl acetate flavour was suppressed by increasing concentrations of each of the three hydrocolloid thickeners (Table 2).
|
Mean sensory data for flavour and sweetness perception were plotted on a log—log scale against c/c* (Figure 3). There were small reductions in mean sensory scores at thickener concentrations below c*, with sweetness and flavour suppression increasing sharply above the c* concentration. Whilst the data could be fitted with a polynomial function, it was felt that results were best portrayed by two straight lines, intersecting in the region of c*, since it is difficult to propose a mechanism that would result in a curve on a double logarithmic plot. The plots were fitted with two linear regressions, corresponding to the below c* and above c* series, and an R2 value was calculated for the fit to the Boolean function thus formed. This process was repeated for plots of the sensory data against a range of measured rheological properties of the solutions, in place of the c/c* ratio. Table 3 shows the resultant fit between these rheological parameters and the sensory data. The Kokini oral shear stress was the factor which showed the best fit for sweetness and flavour perception, closely followed by the apparent viscosity at a shear rate of 50 s-1 (Table 3 and Figure 4).
|
|
|
A two-factor ANOVA of the above c* sweetness data (with
log c/c* and thickener type as factors) showed significant
effects of both factors and a significant interaction term (P <
0.05; Table 4A). This indicates
that the thickeners behaved differently with respect to sweetness suppression
as the c/c* ratio increased. The interaction graph
generated by Design Expert (Figure
5) shows clearly that -carrageenan did not suppress
sweetness as much as guar or HPMC, when comparing samples of the same
c/c* ratio.
|
|
The corresponding results for iso-amyl acetate flavour also showed a
significant interaction between thickener type and c/c*
ratio (P < 0.05; Table
4B). HPMC had a greater suppressive effect on flavour perception
than did guar or -carrageenan in viscous solutions of the same
c/c* ratio (Figure
5B).
When using the oral shear stress as a factor in place of c/c*, results for flavour perception showed only a significant effect of the oral shear stress (P < 0.0001), and no significant effect of thickener type or interaction terms (Table 4D and Figure 5D). For sweetness, there was no longer a significant interaction term, but thickener type was still significant (P < 0.001) in addition to the oral shear stress (P < 0.0001); Table 4C and Figure 5C). Guar gum had a more suppressive effect on sweetness, relative to its oral shear stress, than carrageenan or HPMC.
A two-factor ANOVA with interactions was computed from the raw sensory data
for both sweetness and flavour (with thickener type and
c/c* ratio as categorical factors). Fisher's least
significant difference (LSD) for comparing mean values was calculated to be
14.6 for sweetness and 15.7 for flavour. With reference to
Table 1, mean sensory scores
differing by more than the LSD values quoted were found to be significantly
different from one another (P < 0.05). Thus, relative to the
reference (in water), each hydrocolloid resulted in a significant reduction in
perceived sweetness at the 3c*/4 concentration. For banana
flavour, there were different behaviours with hydrocolloid type. HPMC and guar
significantly reduced flavour perception relative to the reference at
7c*/4 and above. -Carrageenan had a significant
effect at 5c*/4 and above.
Model validation experiment
Mean sensory scores for perceived sweetness and flavour of seven new samples (including xanthan and methyl cellulose thickened solutions) fitted closely with the original models based on oral shear stress (Table 5 and Figure 6). Incorporation of the new data resulted in only minor changes to the original models; indeed, the model for sweetness could be improved by including the new points (R2 = 0.88).
|
|
|
Discussion |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionConclusionsReferences |
|---|
By comparing the sensory and in-nose aroma release experiments, the conclusion is that the drop in flavour perception which occurred above c* (Figure 3) was not attributable to a drop in odour stimulus (in-nose iso-amyl acetate concentration) at the olfactory receptors. This is in agreement with findings elsewhere (Hollowood et al., 2002
). Sensory perception began to drop off steeply at thickener
concentrations above c*
(Figure 3). However, there was
no corresponding reduction in aroma release in-nose, in terms of either
concentration or total amount integrated over time. It is important to
consider total area and first exhalation concentration data jointly
(Figure 2), since both stimulus
intensity and temporal aspects of release play a role in determining perceived
odour magnitude (Overbosch,
1989; Baek et al.,
1999).
The first exhalation peak concentration of iso-amyl acetate in-nose was
more or less constant, averaged across all panellists, up to the
10c*/4 concentration of thickener. A reduction in mean
in-nose concentration was observed at the 14c*/4 thickener
concentration, although this was not found to be statistically significant
owing to the degree of variation inherent in nose-space release measurements
(error bars are omitted from Figure
2 for clarity). This results from variations in swallowing
pressure, and from individual differences in physiology and mouth movements
(Burdach and Doty, 1987;
Buettner and Schieberle, 2000).
Thus, some panellists actually showed an increase in release from the most
viscous solutions while others gave the opposite result; on average, there was
no clear trend with thickener concentration. In contrast, trends in sensory
perception were clear-cut and consistently fell in intensity with increased
viscosity.
Sensory data plotted against thickener c/c*
(Figure 3) showed a similar
form to previously published results (Baines and Morris,
1987,
1988;
Hollowood et al.,
2002). However, above c*, perceptual data for
the three different hydrocolloids was correlated only loosely with the
c/c* ratio. Results of the ANOVA for both sweetness and
flavour data above c*
(Table 4A,B) showed significant
interactions between thickener type and c/c* ratio,
confirming that the behaviours of individual hydrocolloids differed when
plotted against c/c*.
The rheological factor which gave the best overall correlation with the
sensory perception of sweetness and flavour was the Kokini oral shear stress
(Table 3). Other rheological
parameters, notably the apparent viscosity at a shear rate of 50
s-1, showed good correlation with sensory data, in agreement with
previous findings (Wood,
1968). This is not too surprising, since the oral shear stress is
calculated from and thus related to some of the other factors investigated.
Its advantage, relative to using estimates of oral viscosity based on one
shear rate, is that it accounts for the shear-thinning nature of hydrocolloids
over a range of shear rates, providing a better model for the range of shear
rates encountered in the oral evaluation of viscosity
(Shama and Sherman, 1973;
Christensen and Casper, 1987).
An added attraction of using the oral shear stress to model sensory data is
that it has been shown to be the stimulus involved in the oral evaluation of
viscosity (Elejalde and Kokini,
1992) and thus has real meaning in a somatosensory sense.
ANOVA of the above c* data
(Table 4D) showed that banana
flavour perception could be predicted solely from the oral shear stress, with
no significant effect of thickener type. Thus, relating flavour perception to
the oral shear stress removed the variation between hydrocolloids that was
evident in the c/c* model
(Table 4B). The situation was
not as clear cut with sweetness perception. Although looking at sweetness in
terms of the oral shear stress removed the interaction term from the model
(Table 4C), there was still a
significant effect of thickener type (guar gum suppressing sweetness slightly
more than HPMC or -carrageenan in samples of similar oral shear
stress). One factor that could contribute to this discrepancy is the question
of individual taste of the hydrocolloids
(Izutsu and Wani, 1985).
Whilst the hydrocolloids used in the investigation were of high purity and
selected on the basis of their relative tastelessness, an appraisal of the
taste of hydrocolloid solutions in isolation was not included in our research.
Although subtle differences in taste due to hydrocolloid type cannot be ruled
out as a source of minor variation in the taste data, we would not have
expected the thickeners all to behave so similarly if there were substantial
taste components present. Such taste effects would be expected to be both
concentration and hydrocolloid dependent, yet although the concentrations of
individual thickeners ranged from, for example, 3.4 g/l of guar to 10 g/l of
HPMC in the 7 c*/4 samples, the sweetness suppression was
of the same order of magnitude for each
(Table 2). It seems reasonable
to conclude that the taste component of hydrocolloids might account for subtle
differences in behaviour between thickeners, but is not the principal factor
involved in sweetness reduction.
The perceptual models based on the oral shear stress, which had been
developed in the main experiment, were sufficiently robust to predict the
sweetness and flavour suppressing behaviour of the additional hydrocolloids
used in the model-validation experiment
(Table 5). The final models
(Figure 6) thus adequately
described the behaviours of five different hydrocolloids. The prediction of
flavour perception from xanthan solutions is of particular note, since it
demonstrates that the concepts involved can be applied to thickeners other
than random-coil polysaccharides. Xanthan adopts a rod-like conformation in
solution, and the ability of these rods to align under shear is thought to
explain the particularly shear-thinning nature of the gum
(Cuvelier and Launay, 1986).
This behaviour is taken into account when predicting oral shear stress, so for
a specified zero-shear viscosity, the oral shear stress would be much lower
for xanthan than for HPMC or guar gum. This may well explain the reportedly
good flavour release properties of xanthan
(Baines and Morris, 1988).
Significantly, we still observe an inflection in perception when plotting
sensory data against the oral shear stress, at around = 3.7 Pa (the
point on the oral shear-stress axis corresponding to c*).
This inflection might arise from two linear effects of differing origin, which
have first a gradual and then a more drastic effect on flavour perception.
Such effects might be physical (e.g. reduction in tastant mass transfer;
blocking of receptor sites by hydrocolloid), neurological (complex integration
of signals at multi-modal neurons) or psychophysical (a result of higher
cortical processing) in nature.
The overall pattern of our results seems to be readily reproducible
(Baines and Morris, 1987;
Hollowood et al.,
2002). It may be that what we are seeing is the interplay between
the senses of taste, aroma and texture as viscosity is manipulated in a model
system. Since iso-amyl acetate flavour perception decreased in the presence of
a consistent in-nose stimulus, it is feasible that a perceptual mechanism, or
interaction of the senses, is operative. Evidence of taste—aroma
interactions is now well documented
(Noble, 1996) in situations
where there is `congruency' between the taste and aroma concerned
(Schifferstein and Verlegh,
1996). This is clearly the case for iso-amyl acetate and sweet
taste. The observed suppression of iso-amyl acetate flavour in the presence of
viscous stimuli may be explained by this taste—aroma interaction, with
the drop in perceived sweetness driving the perception of a congruent aroma
(Hollowood et al.,
2002). Taking this concept further, and bearing in mind that the
shear stress developed in-mouth is the stimulus for the oral perception of
viscosity (Elejalde and Kokini,
1992), our data supports the possibility of a sensory input for
viscosity that is a part of the flavour experience of foods. Somatosensory
information might then interact at a neural or psychological level with the
perception of taste and aroma to produce texture—taste— aroma
interactions.
Alternatively, our models based on oral shear stress may work simply
because they take into account the individual shear-thinning characters of the
hydrocolloids (exemplified by m and n). These parameters
might adequately predict an underlying physical mechanism for the reduction in
sweetness. For example, if shear-thinning behaviour controlled the flux of
tastant molecules across a boundary layer of fluid to the tongue's surface,
then it might predict taste perception for mass transfer reasons, rather than
by altering oral shear stress. The drop in aroma perception would then be
attributed to a reduced taste—aroma interaction between sweetness and a
sweet-associated volatile (e.g. iso-amyl acetate). To explore this
possibility, we used the technique of Cussler et al.
(Cussler et al.,
1979) to model our results for sweetness against theoretically
derived diffusion coefficients for the respective solutions
(Clough et al., 1962)
(data not shown). Correlation was very poor (R2 < 0.2),
suggesting that diffusion/mass-transfer considerations alone were insufficient
to explain the observed changes in perception.
To look at the nature of texture—taste—aroma interactions more
closely, we re-examined data from Hollowood et al.
(Hollowood et al.,
2002). This experiment modelled sweetness and flavour perception
using a large three-factor design of solutions thickened with HPMC, sweetened
with sucrose and flavoured with benzaldehyde. Looking first at non-thickened
samples containing 100 p.p.m. benzaldehyde, we plotted the
sucrose—benzaldehyde interaction in aqueous solution
(Figure 7). Over the 30-50 g/l
range of sucrose, the increase in benzaldehyde flavour was practically linear
with increasing sweetness perception. We then identified thickened solutions
containing 50 g/l sucrose which the model predicted to be iso-sweet to aqueous
solutions of 30-50 g/l sucrose and plotted the resultant flavour perception on
the same axes (Figure 7). It
can be seen that iso-sweet solutions did not lead to the same benzaldehyde
flavour score when one was presented in water and the other was thickened and
its sweetness partially suppressed by HPMC. Initially, as HPMC was added to
the system (increasing from right to left in
Figure 7), the flavour score
for the HPMC solution remained higher than would have been estimated on the
basis of a pure sweetness—aroma interaction. Subsequently, at 10
g/l HPMC, the hydrocolloid system came to have a lower flavour intensity than
would have been estimated from the results for the aqueous system. Clearly,
these discrepancies suggest that taste—aroma interactions alone are not
driving the reduction in flavour perception. A more complex integration of the
senses might explain this interplay of texture, taste and flavour.
|
There is considerable neurological evidence to support functional
convergence of sensory information relating to taste, olfaction and
somatosensory stimuli. Cerf-Ducastel et al.
(Cerf-Ducastel et al.,
2001) provided a recent overview of the evidence from neurological
studies in both rodents and primates. They concluded that there is clearly a
functional convergence of the gustatory, lingual somatosensory and olfactory
inputs, which together constitute the neural substrate for a multimodal
representation of flavour information. Sensory pathways are known to overlap
widely from the periphery, so that gustatory nerves, including the chorda
tympani, may respond to both taste and tactile stimulation, originating
simultaneously during food intake. Related fields of research are concurrently
developing theories founded in a multi-sensory approach to perception. Driver
and Spence (2000) described
cross-modal integration of sensory information and considered this to be the
rule, rather than the exception, in real-world perception.
So, if flavour is a multimodal percept, arising from signals sharing common neural pathways, integrated and interpreted by the brain through a logical process aimed at optimal decision making, it should come as little surprise that changes in one stimulus can impact on the perception of another. The results presented here can thus be viewed as complementary to contemporary views on the multimodal nature of flavour perception.
|
Conclusions |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionConclusionsReferences |
|---|
Aroma and flavour perception in hydrocolloid thickened solutions can be predicted by the Kokini oral shear stress. This supports the hypothesis that a sensory signal for viscosity, corresponding to the shear stress generated inmouth, can modulate the perception of taste and aroma. Equally, the success of the models developed here may result from physical effects related to the power law properties of fluids, which are incorporated in the calculation of oral shear stress. Either way, the success of Elejalde et al. (Elejalde et al., 1992) in applying the Kokini model to the perception of thickness in real foods suggests that the principles involved might be applicable when studying flavour perception in more complex systems.
|
Acknowledgments |
|---|
The authors wish to thank Firmenich SA, Geneva, Switzerland and BBSRC for their support of this work.
|
References |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionConclusionsReferences |
|---|
Baek, I., Linforth, R.S.T., Blake, A. and Taylor, A.J. (1999) Sensory perception is related to the rate of change of volatile concentration in-nose during eating of model gels.Chem. Senses , 24,155 -160.
Baines, Z.V. and Morris, E.R. (1987) Flavour/taste perception in thickened systems: the effect of guar gum above and below c*. Food Hydrocoll.,3 , 197-205.
Baines, Z.V. and Morris, E.R. (1988) Effects of polysaccharide thickeners on organoleptic attributes. In Phillips, G.O., Williams, P.A. and Wedlock, D.J. (eds), Gums and Stabilisers for the Food Industry 4. IRL Press, Oxford, pp.192 -201.
Buettner, A. and Schieberle, P. (2000) Exhaled odorant measurement (EXOM)—a new approach to quantify the degree of in-mouth release of food aroma compounds. Lebensm.-Wiss. Technol.-Food Sci. Technol., 33,553 -559.
Burdach, K.J. and Doty, R.L. (1987) The effects of mouth movements, swallowing and spitting on retronasal odor perception. Physiol. Behav., 41,353 -356.[CrossRef][Medline]
Cerf-Ducastel, B. and Murphy, C. (2001)
fMRI activation in response to odorants orally delivered in aqueous
solutions. Chem. Senses, 26,625
-637.
Cerf-Ducastel, B., Van de Moortele, P-F., Macleod, P., Le Bihan,
D. and Faurion, A. (2001) Interaction of gustatory
and lingual somatosensory perceptions at the cortical level in the human: a
functional magnetic resonance imaging study. Chem. Senses,26
, 371-383.
Christensen, C.M. (1980) Effects of solution viscosity on perceived saltiness and sweetness. Percept. Psychophys., 28,347 -353.[Web of Science][Medline]
Christensen, C.M. and Casper, L.M. (1987) Oral and non-oral perception of solution viscosity. J. Food. Sci., 52,445 -447.[CrossRef]
Clough, S.B., Read, H.E., Metzner, A.B. and Behn, V.C. (1962) Diffusion in slurries and in non-Newtonian fluids. A.I.Ch.E. J., 8,346 -350.[CrossRef]
Cook, D.J., Hollowood, T.A., Linforth, R.S.T. and Taylor, A.J. (2002) Perception of taste intensity in solutions of random-coil polysaccharides above and below c*. Food Qual. Prefer., 13,473 -480.[CrossRef]
Cross, M.M. (1965) Rheology of non-Newtonian fluids: a new flow equation for pseudoplastic systems.J. Colloid Sci. , 20,417 .[CrossRef]
Cussler, E.L., Kokini, J.L., Weinheimer, R.L. and Moskowitz, H.R. (1979) Food texture in the mouth.Food Technol. , 33,89 -92.
Cuvelier, G. and Launay, B. (1986) Concentration regimes in xanthan gum solutions deduced from flow and viscoelastic properties. Carbohydr. Polym.,6 , 321-333.
Davidson, J.M., Hollowood, T.A., Linforth, R.S.T. and Taylor, A.J. (1999) The effect of sucrose on the perceived flavour intensity of chewing gum. J. Agric. Food Chem., 47,4336 -4340.[CrossRef][Web of Science][Medline]
deMartine, M.L. and Cussler, E.L. (1975) Predicting subjective spreadability, viscosity and stickiness.J. Pharm. Sci. , 64,976 -982.[CrossRef][Web of Science][Medline]
Driver, J. and Spence, C. (2000) Multisensory perception: beyond modularity and convergence.Curr. Biol. , 10,R731 -R735.[CrossRef][Web of Science][Medline]
Elejalde, C.C. and Kokini, J.L. (1992) The psychophysics of pouring, spreading and in-mouth viscosity.J. Text. Studies , 23,315 -336.
Guinard, J.X. and Mazzucchelli, R. (1996) The sensory perception of texture and mouthfeel.Trends Food Sci. Technol. , 7,213 -219.[CrossRef]
Hollowood, T.A., Linforth, R.S.T. and Taylor, A.J.
(2002) The effect of viscosity on the perception of
flavour. Chem. Senses, 27,583
-591.
Izutsu, T. and Wani, K. (1985) Food texture and taste—a review. J. Text. Studies,16 , 1-28.
Izutsu, T., Taneya, S.I., Kikuchi, E. and Sone, T. (1981) Effect of viscosity on perceived sweetness intensity of sweetened sodium carboxymethyl-cellulose solutions. J. Text. Studies, 12,259 -273.
Kokini, J.L. (1985) Fluid and semi-solid food texture and texture taste interactions. Food Technol., 39,86 -94.
Kokini, J.L. and Cussler, E.L. (1983) Predicting the texture of liquid and melting semi-solid foods.J. Food. Sci. , 48,1221 -1225.[CrossRef]
Kokini, J.L., Kadane, J.B. and Cussler, E.L. (1977) Liquid texture perceived in the mouth. J. Text. Studies, 8,195 -218.
Kokini, J.L., Bistany, K., Poole, M. and Stier, E. (1982) Use of mass-transfer theory to predict viscosity—sweetness interactions of fructose and sucrose solutions containing tomato solids. J. Text. Studies,13 , 187-200.
Linforth, R.S.T. and Taylor, A.J. (2000) Persistence of volatile compounds in the breath after their consumption in aqueous solutions. J. Agric. Food Chem.,48 , 5419-5423.[CrossRef][Web of Science][Medline]
Mälkki, Y., Heiniö, R.L. and Autio, K. (1993) Influence of oat gum, guar gum and carboxymethyl cellulose on the perception of sweetness and flavor. Food Hydrocoll., 6,525 -532.
Moskowitz, H. R. and Arabie, P. (1970) Taste intensity as a function of stimulus concentration and solvent viscosity. J. Text. Studies, 1,502 -510.
Noble, A.C. (1996) Taste—aroma interactions. Trends Food Sci. Technol.,7 , 439-444.[CrossRef]
Overbosch, P. (1989) A theoretical model for perceived intensity in human taste and smell as a function of time.Physiol. Behav. , 45,607 -613.[CrossRef][Medline]
Pangborn, R.M. and Szczesniak, A.S. (1974) Effect of hydrocolloids and viscosity on aromatic flavour compounds. J. Text. Studies,4 , 467-482.
Pangborn, R.M., Trabue, I. M. and Szczesniak, A.S. (1973) Effect of hydrocolloids on oral viscosity and basic taste intensities. J. Text. Studies,4 , 224-241.
Paulus, K. and Haas, E.M. (1980) The
influence of solvent viscosity on the threshold values of primary tastes.Chem. Senses
, 5,23
-32.
Richardson, R.K., Morris, E.R., Ross-Murphy, S.B., Taylor, L.J. and Dea, I.C.M. (1989) Characterization of the perceived texture of thickened systems by dynamic viscosity measurements. Food Hydrocoll., 3,175 -191.
Schifferstein, H.N.J. and Verlegh, P.W.J. (1996) The role of congruency and pleasantness in odor-induced taste enhancement. Acta Psychol.,94 , 87-105.[CrossRef][Medline]
Shama, F. and Sherman, P. (1973) Identification of stimuli controlling the sensory evaluation of viscosity. J. Text. Studies, 4,111 -118.
Taylor, A.J., Linforth, R.S.T., Harvey, B.A. and Blake, A. (2000) Atmospheric pressure chemical ionisation mass spectrometry for in vivo analysis of volatile flavour release.Food Chem. , 71,327 -338.[CrossRef]
Vaisey, M., Brunon, R. and Cooper, J. (1969) Some sensory effects of hydrocolloid sols on sweetness. J. Food Sci., 34,397 -400.[CrossRef]
Wood, F.W. (1968) Psychophysical studies on the consistency of liquid foods. In SCI Monograph No. 27. Rheology and Texture of Foodstuffs. Society of Chemical Industry, London, p. 40.
Accepted September 24, 2002
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  J. Lim and B. G. Green Tactile Interaction with Taste Localization: Influence of Gustatory Quality and Intensity Chem Senses, February 1, 2008; 33(2): 137 - 143. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||