Chem. Senses 24: 155-160,
1999
© Oxford University Press
Sensory Perception is Related to the Rate of Change of Volatile Concentration In-nose during Eating of Model Gels
Samworth Flavour Laboratory, Division of Food Sciences, University of Nottingham, Sutton Bonington Campus, Loughborough LE12 5RD, UK 1 Corporate R&D, Firmenich SA, Route des Jeunes 1, Geneva 8, Switzerland
Correspondence to be sent to: Inger Baek, Samworth Flavour Laboratory, Division of Food Sciences, University of Nottingham, Sutton Bonington Campus, Loughborough LE12 5RD, UK. e-mail: scxinb{at}szn1.agric.nottingham.ac.uk
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and methodsResultsReferences |
|---|
The relationship between perceived aroma and the volatile concentration measured in-nose was investigated during eating of a model food. Sensory ranking and time–intensity analysis (TI) were used to measure perceived aroma, while in-nose volatile concentration was monitored by atmospheric pressure ionization mass spectrometry, which produced time release data. A gelatine–sucrose gel with a range of gelatine concentrations (2-8% w/w) and flavoured with furfuryl acetate was used as the model food. Senosry scaling showed decreased flavour intensities and TI showed a decrease in the flavour perceived over time, as the gelatine concentration increased. Studies in model systems and in people demonstrated that the different rates of release observed for different gelatine concentrations were not due to binding of volatile to protein in the gel, nor to mucous membranes, but were due to different rates of gel breakdown in-mouth. There were no significant differences in the maximum in-nose volatile concentrations for the different gelatine concentrations, so the amount of volatile present did not correlate well with the sensory analysis. However, the rates of volatile release were different for the different gels and showed a good correlation with sensory data.
|
Introduction |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsReferences |
|---|
The link between the chemical signals (the stimuli) that cause flavour and the sensory perception (the response) they evoke has been the subject of much research. The concept of psychophysics was proposed 150 years ago to explain the observed relationship between the sweet response noted after stimulation of the taste buds with various sugars (for review see Hoppe, 1995
).
Stevens (1957)
developed and elaborated methods for determining stimulus magnitudes and response
sensations (often called direct scaling methods). The Power Law derived by Stevens has been
widely applied to correlate taste and odour sensations with the concentrations of flavour
chemicals in foods.
However, time–intensity analysis (TI) shows that there is a temporal dimension to
flavour perception.
Overbosch (1986)
proposed that the temporal dimension to aroma perception was due to adaptation. If the receptors
were subject to a constant level of stimulation by a volatile aroma compound, then the response
should decrease with time and, eventually, no response would be observed. The rate of
adaptation depends on the aroma molecule but, using data from the literature,
Overbosch (1986)
proposed a mathematical model that calculated the adaptive process and subtracted it from the
stimulus applied, to obtain the actual signal that triggered the receptor. He predicted that
significant adaptation could occur within the time taken to chew and swallow food. A refinement
of the model was published in 1989
(Overbosch and de Jong, 1989).
Overbosch (1986)
acknowledged that the situation that occurs with a dynamic stimulus was more complex but
showed an idealized plot of the relationship between stimulus and response. In his model, the
time to maximum intensity (Tmax) for both stimulus and response was
identical but subsequent work
(Linforth et al., 1998)
has suggested that another mechanism besides adaptation may occur in the early stages of eating.
Furthermore, the averaging methods of Overbosch may have obscured subtle differences in
Tmax between stimulus and response curves. It should also be recognized
that the
Overbosch models consider the aroma of single volatiles and take no account of potential
interactions between volatiles, which may change the relationship between stimulus and
response, as has been proposed in other models
(see for example Ennis, 1996).
The advent of methodology to follow volatile release as people eat foods [and at
concentrations that relate to the odour thresholds of many aroma compounds
(Linforth et al., 1996)]
has renewed interest in the relationship between aroma stimuli and perceived responses. There
is evidence that flavour compounds are delivered at different rates to the receptors in a wide
range of real foods, e.g. sugar confectionery
(Ingham et al., 1995),
cheese
(Delahunty and Piggott, 1995),
chocolate
(Roozen and Legger-Huysman, 1995)
and tomatoes
(Linforth et al., 1994).
Using a direct inlet system for atmospheric pressure ionization-mass spectrometry (API-MS)
(Linforth and Taylor, 1998),
the release of volatiles from foods can be measured in real time by sampling air from the nose or
mouth of people eating foods, with detection at the 10 to 100 ppbv level in the gas phase. The
data obtained can be considered analogous to sensory TI data and, for convenience, have been
termed time release (TR) curves.
The models of Overbosch
(Overbosch, 1986;
Overbosch and De Jong, 1989)
provide one explanation for linking volatile concentration with aroma perception and are
applicable to the TR and TI data now available with the real time assays. TI, however, is a
scientific tool for following perceptual changes in trained panellists and it is unlikely that the
typical consumer is consciously aware of the TI profile of foods when they judge flavour quality
and/or intensity. The question remains, therefore, whether these time-related volatile profile
changes during eating have any relevance to the consumer and their perception of food flavour
and whether the measurement of TR data is useful in developing food products for the general
public or just a scientific curiosity.
The purpose of this paper was to investigate how the release profiles of volatiles were related
to aroma intensity as perceived by sensory scaling methods. Some TI was carried out as a
comparison. Gelatine–sucrose gels were chosen as a model food as they are known to
deliver flavour at different rates when the gel formulation is changed
(Guinard and Marty, 1995;
Wilson and Brown, 1997).
To avoid any problems of interaction between volatiles, a single volatile (furfuryl acetate, FFA)
was used which had an easily recognizable, characteristic flavour so that sensory panellists
could readily recognize the compound. Another advantage of FFA was that it shows relatively
low persistence in-mouth and therefore minimizes problems of carry over from one sample to
another.
|
Materials and methods |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsReferences |
|---|
Gelatine gel system
Gelatine (250 bloom strength; DGF STOESS, Germany) gels were prepared containing 20.5% w/w sucrose and 22.5% w/w glucose syrup (Cerestar, UK). The gels were prepared by dissolving the gelatine at 60°C, prior to mixing with the sucrose–glucose mix (which had been boiled and allowed to cool). The gels were flavoured by adding FFA (Firmenich SA, Switzerland), dispersed in propylene glycol, to the cooled gelatine–sugar mixture before gelation occurred. The gelatine gels contained different volatile and gelatine concentrations, as shown in Table 1. The gels were cut into cubes (6.0 ± 0.1 g) and stored overnight at 4°C. The following day, the gels were equilibrated at room temperature before being presented to the panellists.
|
Sensory evaluation of gels: scaling
Sixteen trained panellists were asked to rate the relative maximum intensity of FFA perceived whilst eating gelatine gels. The samples were randomly coded with three-digit numbers and presented in random order. Panellists were asked to rate the samples relative to two reference samples: an unflavoured 5% w/w gelatine gel, representing average textural properties, and a flavoured 2% w/w gelatine gel, which caused the highest sensory perception. The two gels represented 0 and 10 on the intensity scale respectively. The panellists were trained to focus on the flavour and ignore differences in sweetness and texture. Each panellist assessed one set of gel samples and waited at least 2 min between samples. Odourless water and dry crackers were available to remove traces of gels between samples. The mean intensity values from the 16 panellists were calculated for each gel.
Sensory evaluation of gels: time intensity
Eleven trained panellists were asked to rate the perceived flavour intensity with time while eating the gels. TI was performed simultaneously with TR data collection and the TI signal was combined with the MS data using an analogue channel of the mass spectrometer (one data point collected every second). The resulting traces were processed to yield Tmax, Imax and gradient data. Each panellist was given one set of gels and had 4 min between samples during which they could use water or crackers to remove traces of gel from the mouth. The breath of each panellist was monitored by API-MS before the next gel was consumed and no significant carry over between samples was observed.
Instrumental analysis: model system
The release of FFA was measured from a cube of gel as it dissolved in 100 ml of distilled water (37°C) in a 125 ml flask with stirring. Samples of the headspace (HS) were drawn into the API-MS at 12 ml/min. Release profiles for FFA were monitored in duplicate for up to 45 min.
Instrumental analysis: TR during eating
Panellists were given samples of the gels to eat while the FFA content in-nose was monitored
by sampling the air flow from one nostril over a 2 min period (no specific eating instructions
were given). The breath was introduced (25–30 ml/min) into a modified API source
(Linforth and Taylor, 1998)
fitted on a Platform II Quadrupole mass spectrometer (Micromass,
Altrincham, UK), where the FFA was ionized by a 4 kV corona discharge. The mass
spectrometer was used in positive ion mode, with a cone voltage of 20 V, and set to monitor
m/z 80.8 (a fragment ion of FFA). The raw breath by breath traces were converted into
TR
curves by smoothing the peak height data and converting peak height into actual concentration in
air (nanolitres of volatile per litre of air; ppbv) after calibration of the API-MS interface with a
series of FFA standards. From the TR curves, the parameters maximum intensity
(Imax) and time to maximum intensity (Tmax) were
calculated
along with the rate of volatile release, which was defined as the gradient on the up-slope
t25 (time to 25% of Imax) and t75
(time to 75% of Imax).
|
Results |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsReferences |
|---|
Sensory evaluation of gelatine gels containing FFA
Initially, the gel samples containing FFA were evaluated sensorially to ensure that this
volatile showed release characteristics from the gelatine–sucrose system similar to the
volatiles used previously (benzaldehyde, D-limonene, ethyl butyrate:
Guinard and Marty, 1995;
commercial banana flavour:
Wilson and Brown, 1997).
Sixteen panellists each ate
a gel and recorded their relative overall flavour intensities using two reference gels. Figure 1
shows the mean flavour intensities (sixteen panellists) from five gel samples containing
2–8% w/w gelatine but with FFA at the same concentration (300 mg/kg). There was a
clear trend of decreasing sensory intensity with increasing gelatine concentration and the
sensory values for the 2% w/w and 8% w/w gels were significantly different (P<
0.001, ANOVA), which agrees with previous work
(Guinard and Marty, 1995;
Wilson and Brown, 1997).
|
Sensory evaluation: TI
TI was carried out using gels of different gelatine concentrations (2–8% w/w), all of
which contained the same amount of FFA (300 mg/kg). The TI data were averaged using a
method similar to that described by
Overbosch et al. (1986).
The TI data
were first normalized in the intensity direction to the numeric average of
Imax, followed by averaging in the time dimension (Figure 2). The TI
profiles showed a
significant difference between the different gels, with the Imax decreasing
(P < 0.001, ANOVA) and the Tmax increasing (P <
0.001, ANOVA) as gelatine concentration increased. Again, this confirmed the results obtained
with
banana flavour by
Wilson and Brown (1986)
and, despite the different volatiles,
the increase in gelatine concentration in both studies resulted in a decrease in sensory perception.
The increase in Tmax might be explained by an increase in melting point
(Wilson and Brown, 1997),
a different rate of dissolution and/or a different rate of breakdown
due to chewing efficiency
(Brown et al., 1995).
|
, 3.5%, 
, 5%, x, 6.5% and *, 8% w/w from gelatine gels. Each curve
is the mean result for eleven panellists each eating one sample of each gel.The differences in Imax, however, might be due to binding of the volatiles to gelatine, which reduced the amount of volatile available for release—a situation that has been reported for various volatiles and proteins (see e.g. Nawar, 1971
;
Solms et al., 1973;
Landy et al., 1995).
Alternatively, the decrease in
Imax with increasing gelatine concentration could be due to a slower rate of
release in
mouth, with the high gelatine gels releasing the same amount of volatile as a low gelatine gel, but
over a longer time period. Since the time in-mouth is limited, the slower release rate might
produce a lower Imax value for high gelatine gels. Instrumental analysis: model system
To investigate which of these mechanisms was valid, a model system was set up. Gels
containing the same concentration of FFA (300 mg/kg) but different gelatine concentrations (2, 5
and 8% w/w) were placed in water and stirred, and the FFA release was measured by monitoring
the HS until the gel had completely dissolved. The results (Figure 3)
demonstrated that the three
gels released FFA at different rates, with the softest gel (2% w/w gelatine) becoming fully
dissolved after 10 min, whereas it took 40 min for the harder gel (8% w/w gelatine) to dissolve
completely. However, irrespective of gelatine concentration, all gels reached the same
concentration of FFA in the HS when fully dissolved, showing there was no binding of FFA to
gelatine under these conditions and that slower release was the most likely explanation for the
difference in TI performance of the gels. This agrees with work by
Harrison and Hills (1996),
who found that increasing the gelatine and sucrose concentrations of a gel
system
resulted in reduced release rate for a water-soluble dye which they used as a marker for gel
dissolution.
|
, 8% w/w. Duplicate release curves are shown for each gel concentration.Instrumental analysis: TR
The model system work suggested that the rate of release was responsible for the differences
in TI. Binding of FFA to gelatine could not explain the TI differences but there was a possibility
that FFA could be binding to membranes in the mouth and nose so that only a portion of the FFA
release was actually transported to the olfactory receptors. To test this hypothesis, the in-nose
concentration of FFA during eating was monitored for a series of gels with different volatile
and gel concentrations using one panellist to minimize person-to-person variation. If no
significant binding of the volatile was occurring, plotting the maximum concentration of FFA
in-nose against the gel FFA concentration should produce a linear plot. Figure 4 shows the data
obtained from three replicates of three different gel concentrations with five FFA concentrations.
Despite the variable nature of the eating event, the lines are linear, suggesting that no significant
binding of FFA occurs between the release event in-mouth and perception in-nose. The above
experiment was repeated using three extra panellists. The relationships between the gel volatile
concentration and breath volatile concentration were still linear for each panellist (data not
shown), but there were substantial differences between panellists in the amount of FFA released
from the same gels, presumably because of the different chewing patterns of the panellists
(Brown et al., 1995).
|
Comparison between instrumental and sensory data
Gels containing 300 mg/kg FFA were prepared with gelatine concentrations of 2, 5 and 8% w/w, and simultaneous TI and TR data were collected from 11 panellists who ate one sample of each gel. The mean values for the two sets of data are plotted in Figure 5 and show that there is a clear decrease in sensory perception with increasing gelatine concentration; however, the changes in volatile concentration in-nose are less clear. Statistical analysis of the the three gel concentrations showed no significant difference in the TR Imax values but a significant difference (P < 0.001, ANOVA) for the TI values. These findings suggest that, in this system, perception is not directly related to the maximum volatile concentration in-nose as might be predicted by the basic Power Law. This supports the necessity to modify the Power Law to take account of temporal changes. However, whereas Overbosch based his model on data obtained from a constant volatile stimulus, these data were obtained from a food during eating where the stimulus is dynamic and changes rapidly over a period of just over 1 min (see TR data in Figure 5).
|
To test whether the Overbosch (1986
)
adaptation concept was applicable to
the data reported here, the ratios of the TI:TR Tmax values were calculated
from each of the individual curves that made up the mean values in Figure 5 and then plotted
against the TR Tmax values. If adaptation was occurring, then the receptors
should become progressively less receptive to an increase in TR so that the TI peak should
occur before the TR peak and the ratio of TI:TR should be <1. Figure 6
shows that the data are
scattered but the line of best fit (R = 0.421) reveals that the TI:TR ratio is in fact >1
until the TR Tmax reaches 0.6 min, after which TI:TR is <1. Although
these
results need to be treated with caution, due to the scatter and the fact that they represent only one
volatile, they suggest that there are two processes taking place, an initial lag phase where the
Tmax for perception occurs later than the Tmax of the
stimulus and then an adaptive phase where the Tmax for perception occurs
earlier than the Tmax of the stimulus. However, similar trends have been
noted with other volatiles
(Linforth et al., 1998)
and further modification to the models
proposed by
Overbosch (1986)
and
Overbosch and de Jong (1986)
may be necessary to describe the full range of events that occur over the time course of
eating.
|
Another parameter that can be extracted from the data is the rate of release, which expresses the FFA concentration in-nose as a function of time. The rate of release was defined as the gradient (1). The points t75 and t25 were used because, in most cases, this region of the TR curve was essentially linear whereas the situation was more complex above, and below, these values.
 | (1) |
To compare the relationships between the TR Imax, the TI Imax, the sensory scaling score and the gradient, the values were all expressed on a relative scale (the value for the 2% w/w gel was taken as 100%). A plot of the relative values against gel concentration (Figure 7) suggested that the gradient correlated better with the sensory values than TR Imax. To confirm this trend, the actual values were analysed statistically. The changes in the actual TR Imax values were not statistically significant. There was, however, a significant decrease in the instrumental gradient (P < 0.01) and in the two sensory parameters TI Imax value (P < 0.001) and sensory scaling score (P < 0.001) with increasing gelatine concentration. These experiments demonstrate that the temporal aspects of volatile release are related to aroma perception and that the ability to measure volatile concentration in-nose simultaneously with aroma perception provides new opportunities for examining the relationship between the volatile stimulus and the aroma response.
|
, sensory score, 
, TI Imax
and
|
Acknowledgments |
|---|
This work was sponsored by the UK Ministry of Agriculture Fisheries and Foods and the Biological and Biotechnological Science Research Council through a LINK scheme which included Firmenich, Micromass and Stable Micro Systems as industrial partners. Thanks are due to Brian Harvey and Isabelle Cayeux (Firmenich, Geneva) for many useful discussions.
|
References |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsReferences |
|---|
Brown, W.E., Dauchel, C. and Wakeling, I. (1995) Influences of chewing efficiency on texture and flavour perception of food.J.Text. Studies , 27, 433–450.
Delahunty, C.M. and Piggott, J.R. (1995) Current methods to evaluate contribution and interactions of components to flavour of solid foods using hard cheese as an example.Int. J. Food Sci. Tech. , 30 , 555 –570.
Ennis, D.M. (1996) A general molecular model for the effect of multi- component mixtures on biological systems .Food Chem. , 56 ,329 –335.
Guinard, J.X. and Marty, C. (1995) Time–intensity measurement of flavor release from a model gel system: effect of gelling agent type and concentration. J. Food Sci. ,60 , 727 –730.
Harrison, M. and Hills, B.P. (1996) A mathematical model to describe flavour release from gelatine gels. Int. J. Food Sci. Tech. ,31 , 167 –176.
Hoppe, K. (1995) Validity of the psychophysical Power Law. In Rothe, M. and Kruse, H.P (eds), Proceedings of the 4th Wartburg Symposium: Aroma, Perception, Formation, Evaluation. Eigenverlag Deutches Institut für Ernährungsforschung, Potsdam-Rehbrücke, pp. 131–144.
Ingham, K.E., Linforth, S.T. and Taylor, A.J. (1995) The effect of eating on the rate of aroma release from mint-flavoured sweets.Lebensm. Wissenschaft. Technol. , 28 , 105 –110.
Landy, P., Druaux, C. and Voilley, A. (1995) Retention of aroma compounds by proteins in aqueous solution.Food Chem. , 54 ,387 –392.
Linforth, R.S.T. and Taylor, A.J. (1998) Apparatus and method for the analysis of trace constituents in gases. European Patent, EP 0819 937 A2.
Linforth, R.S.T., Savary, I., Pattenden, B. and Taylor, A.J. (1994) Volatile compounds found in expired air during eating of fresh tomatoes and in the headspace of tomatoes. J. Sci. Food Agric. , 65 , 241 –247.
Linforth, R.S.T., Ingham, K.E. and Taylor, A.J. (1996) Time course profiling of volatile release from foods during the eating process. In Taylor, A.J. and Mottram, D.S. (eds), Flavour Science: Recent Develop- ments. Royal Society of Chemistry, Cambridge pp. 361–368.
Linforth, R.S.T., Baek, I. and Taylor A.J. (1998) Simultaneous instrumental and sensory analysis of volatile release from gelatine and pectin/gelatine gels. Food Chem., in press.
Nawar, W.W. (1971) Some variables affecting composition of headspace aroma. J. Agric. Food Chem., 19, 1057.
Overbosch, P. (1986)
A theoretical model for perceived intensity in human taste and smell as a function of
time. Chem. Senses, 11,315–329.
Overbosch, P., van den Enden, J.C. and Keur, B.M.
(1986)
An improved method for measuring perceived intensity/time relationships in human taste
and smell. Chem. Senses, 11, 331–338.
Overbosch, P. and de Jong, S. (1989) A theoretical model for perceived intensity in human taste and smell. II. Temporal integration and reaction times. Physiol. Behav., 45, 607 –613.[Medline]
Roozen, J.P. and Legger-Huysman, A. (1995) Sensory analysis and oral vapour gaschromatography of chocolate flakes. In Rothe, M. and Kruse, H.P. (eds), Proceedings of the 4th Wartburg Symposium: Aroma, Perception, Formation, Evaluation. Eigenverlag Deutches Institut für Ernährungsforschung, Potsdam-Rehbrücke, pp. 627–632.
Solms, J., Ismail, F.O. and Beyler, B. (1973) Interaction of volatiles with food components. J. Can. Inst. Food Sci. Technol., 6, 810.
Stevens, S.S. (1957) On the psychophysical law. Psychol. Rev., 64, 153–181.[Web of Science][Medline]
Wilson, C.E. and Brown, W.E. (1997) Influence of food matrix structure and oral break down during mastication on temporal perception of flavour. J. Sensory Studies, 12, 69–86.
Accepted December 9, 1998
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  K Marshall, D. G. Laing, A. Jinks, and I Hutchinson The Capacity of Humans to Identify Components in Complex Odor-taste Mixtures Chem Senses, July 1, 2006; 31(6): 539 - 545. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
D. J. Cook, T. A. Hollowood, R. S.T. Linforth, and A. J. Taylor Oral Shear Stress Predicts Flavour Perception in Viscous Solutions Chem Senses, January 1, 2003; 28(1): 11 - 23. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 R. Watson, C. J. Wright, T. McBurney, A. J. Taylor, and R. S. T. Linforth Influence of harvest date and light integral on the development of strawberry flavour compounds J. Exp. Bot., October 1, 2002; 53(377): 2121 - 2129. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||