Perceptual Blending in Odor Mixtures Depends on the Nature of Odorants and Human Olfactory Expertise
Advance Access publication August
Perceptual Blending in Odor Mixtures Depends on the Nature of Odorants and Human Olfactory Expertise
S. Barkat 1
E. Le Berre 0
G. Coureaud 0
G. Sicard 0
T. Thomas-Danguin 0
0 Centre des Sciences du Gout et de l'Alimentation, UMR6265 CNRS, UMR1324 INRA, Universite de Bourgogne, AgroSup Dijon , 21065 Dijon Cedex , France
1 Neurosciences Sensorielles Comportement Cognition, UMR5020, CNRS , Universite Lyon 1, 69007 Lyon , France
Our olfactory system is confronted with complex mixtures of odorants, often recognized as single entities due to odor blending (e.g., coffee). In contrast, we are also able to discriminate odors from complex mixtures (e.g., off-odors). Therefore, the olfactory system is able to engage either configural or elemental processes when confronted with mixtures. However, the rules that govern the involvement of these processes during odor perception remain poorly understood. In our first experiment, we examined whether simple odorant mixtures (binary/ternary) could elicit configural perception. Twenty untrained subjects were asked to evaluate the odor typicality of mixtures and their constituents. The results revealed a significant increase in odor typicality in some but not all mixtures as compared with the single components, which suggest that perceptual odor blending can occur only in specific mixtures (configural processing). In our second experiment, we tested the hypothesis that general olfactory expertise can improve elemental perception of mixtures. Thirty-two trained subjects evaluated the odor typicality of the stimuli presented during the first experiment, and their responses were compared with those obtained from the untrained panelists. The results support the idea that general training with odors increases the elemental perception of binary and ternary blending mixtures.
configural processing; expertise; odor mixtures; perceptual blending; quality; training; typicality
The processing of complex stimuli by the olfactory system is
a central issue in the understanding of odor perception in
natural conditions because the odors we perceive come
mostly from complex mixtures of odorants. The perception
of single odorants and mixtures is a product of both
interactions at the level of olfactory receptors and interactions
during neural processing of olfactory information. In the
case of a mixture of odorants, competition may occur at
the olfactory receptors level as well as inhibitory interactions
at the neural level. Therefore, the perception of an odorant
mixture is not a simple sum of the percepts of the unmixed
components (Laing and Jinks 2001).
Studies in animal models have investigated odor mixture
processing and have mainly focused on odor discrimination
(Derby et al. 1996; Deisig et al. 2006; Kay and Stopfer 2006;
Coureaud et al. 2008). These studies have demonstrated that
a binary mixture can be perceived in at least 2 ways. First,
each component of the mixture remains separate and
identifiable. Such processing has been qualified as dissociative,
analytical, or elemental (Derby et al. 1996). Second, the
mixture is perceived as an entity, conveying a unique quality not
present in its single components. This phenomenon has been
called associative, synthetic, or configural processing (Derby
et al. 1996). Kay et al. (2005) suggested that configural
processing might be weak or robust, depending on whether the
odor of the whole mixture partially smells similar to the odor
of the mixture’s constituents or does not smell at all like the
constituents. These data support the idea that a mixture of
odorants can elicit a novel odor percept through configural
processing (i.e., perceptual odor blending). However, in
humans, there is little scientific evidence for the perception of
a specific quality carried by a mixture. According to Olsson
(1994), a binary mixture percept does not form a quality that
is dissimilar from the odorant quality of its chemical
components, but rather one that falls in between the 2.
Nevertheless, it has been suggested that perceptual blending may
happen in specific mixtures, especially those containing more
than 4 components in which the odorants may lose their
individuality and produce new odor sensations (Laing
1991, 1994; Jinks and Laing 2001). Recently, Le Berre,
Thomas-Danguin, et al. (2008) showed that a binary and
a ternary mixture could be perceived as more typical of
a pineapple odor than their components. Moreover, recent
results obtained in newborn rabbits with the same binary
mixture as the one eliciting a pineapple odor in humans, have
strongly suggested that this mixture is processed as a
configuration (Coureaud et al. 2008, 2009, 2011). Taken together,
these results are consistent with the idea that a mixture of
odorants could induce an odor note different from the
one carried by its components.
In spite of recent neurophysiological data showing that
binary odorant mixtures can stimulate cortical neurons not
stimulated by their individual components (Silbering and
Galizia 2007; Grossman et al. 2008; Howard et al. 2009; Deisig
et al. 2010), the rules that govern the involvement of either
elemental or configural processes during odor perception
remain poorly investigated. One possible explanation of this
lack of investigation, especially in humans, could be the
difficulty quantifying odor quality (Wise et al. 2000). Indeed,
methods that address odor quality in mixtures of odorants
need to be carefully selected, especially when perceptual
interactions affect the mixture odor quality, which is likely the case
when perceptual odor blending occurs. If components
contribute to an odor blend, the main character descriptor of the odor
can be used to measure the impact of the components on the
perception of the blend (Bult et al. 2002). Therefore, to be able
to describe odors with different degrees of blending, one can
choose a detailed aroma-profiling task involving both single
component descriptors and a main character descriptor.
However, such a procedure engages panelists in an analytical
perceptual processing strategy, which would decrease putative
synthetic processing and consequently the blending effect
(Le Berre, Thomas-Danguin, et al. 2008). Moreover, an
aroma-profiling task requires an odor reference for each
descriptor that needs to be presented at the beginning of the
testing session. Such an exposition can also modulate the latter
perception and evaluation of blending mixtures (Le Berre,
Thomas-Danguin, et al. 2008). Thus, the choice of sensory
method is a critical step in investigating blending processes
in odor mixtures.
The creation of new odors is the common goal of
perfumers and flavorists. These olfactory experts memorize
and identify quantities of various odorant materials before
creating new fragrances and aromas. One could ask if such
experience and training with odors could affect their
perception of odor mixtures. However, professionals who deal with
odors daily (perfumers, flavorists, and oenologists) are not
only exposed to a wide variety of odors but are also
systematically confronted with descriptions and verbalizations of
their olfactory perceptions. These abilities require both a
perceptual and semantic knowledge of odors (Lawless 1984).
Several studies have underlined that training and experience
with odors do not improve olfactory discrimination,
identification (Chambers and Smith 1993; Roberts and Vickers
1994; Livermore and Laing 1996), or detection thresholds
(Parr et al. 2002), but other experiments have shown positive
effects of training on olfactory performance (Clapperton and
Piggott 1979; Rabin 1988). Bende and Nordin (1997)
reported that expert oenologists did not achieve better
performance than did untrained subjects during a detection
task, but they had greater abilities to discriminate and
identify specific odors. These findings suggest that experience and
training could have an impact on mixture processing and
perception. In a complementary way, Le Berre,
ThomasDanguin, et al. (2008) and Le Berre et al. (2010) reported that
a pre-exposure of na¨ıve subjects to out-of-mixture
components could further influence blending mixture perception;
however, it remains unknown whether general experience
and extensive sensory training can modify the ways experts
perceive mixtures in which perceptual blending occurs.
The aim of the present study was 2-fold. We first wanted to
confirm, with an untrained (na¨ıve) panel, that perceptual
blending occurs in ‘‘chemically simple’’ odorant mixtures
(2 or 3 odorants) and leads to an increase in mixture-induced
odor character, using 2 different sensory tasks: a ranking
task and a rating task. The second objective was to test
the hypothesis that extensive training, namely olfactory
expertise, alters the perception of mixtures in which
perceptual blending occurs. To do so, we followed an experimental
procedure that relied on odor typicality evaluations. This
procedure prevents panelists from using an analytical
perceptual processing strategy (Le Berre, Thomas-Danguin,
et al. 2008; Le Berre et al. 2010). This task could be
considered as a similarity rating between an actually sniffed odor
and an internal representation. Thus, typicality should
reflect the quality of the main character of the odor in the
case of blending mixtures. We performed 2 separate
experiments. In the first one, untrained panelists were asked to rate
the odor quality of 2 binary and 1 ternary mixture formulated
to elicit a pineapple odor. Through typicality ranking and
rating tasks, the panelists were asked to evaluate the pineapple
typicality of the mixtures, their components, and other
pineapple odor references. The pineapple typicality of the mixtures
was compared with that of the components to evidence the
blending process. In the second experiment, a group of trained
subjects (students in oenology) performed the same tasks with
the same stimuli. A comparison of their results with those
obtained by the na¨ıve panel was performed to evaluate the impact
of expertise on odor mixture perception.
Materials and methods
Both experiments relied on the same protocol, with the
exception of different subjects.
In the first experiment, 20 na¨ıve students (10 women and 10
men, M = 23 years, standard deviation [SD] = 3 years) were
involved. They were considered to be na¨ıve subjects because
they did not have any special expertise in olfaction or sensory
In the second experiment, the subjects included 32
oenology students (15 women and 17 men, M = 26 years, SD = 6
years) from the Oenology Faculty of Bordeaux (France).
These subjects were considered to be expert subjects because
they were trained in wine tasting and description of wines.
Ten odorous stimuli were tested (Table 1). A binary mixture
designated F1 (ethyl isobutyrate + ethyl maltol), a ternary
mixture designated F2 (ethyl isobutyrate + ethyl maltol +
allyl-a-ionone), and another binary mixture designated F3
(ethyl caproate+ furaneol) were formulated by flavorists
to produce a pineapple odor. The 5 components were also
evaluated singly, as were 2 references that might produce
a pineapple odor: a single odorant (allyl caproate) and
an essential oil of pineapple designated HE (provided by
Euracli). The 6 pure odorants were purchased from
Ten strips of filter paper (1 · 16 cm, Granger-Veyron) were
prepared 24 h before the sensory session. A total of 500 lL of
each odorant solution (Table 1) was poured onto one end of
each strip, and the strips were stored separately at the bottom
of a closable 70 mL Pyrex test tube at ambient temperature.
The tests were conducted in a quiet well-ventilated room
under daylight. Both groups of subjects were assigned 2 distinct
tasks. First, the subjects had to smell each tube to sort the
tubes from the most pineapple-like odor to the least
pineapple-like odor, according to their resemblance to a pineapple
odor (internal reference). The encoded tubes were presented
to the subjects in a random order. The subjects were allowed
to smell the stimuli as many times as they wanted until they
were satisfied with their ranking. Subjects were instructed to
close the tube after each evaluation to prevent odor
dissemination in the room. Second, when the ranking was
completed, the tubes were again shuffled and presented to the
subjects. They were asked to smell each tube and rate, for
each stimulus, first, the typicality of the pineapple odor
on a dedicated labeled 9-point scale (from 1, ‘‘not typical
at all’’ to 9, ‘‘extremely typical’’). Second, they rated the
edibility of the sample on another dedicated labeled 9-point
scale (from 1, ‘‘not edible at all’’ to 9, ‘‘extremely edible’’).
Only the typicality scores are shown and discussed in this
All statistical analyses were conducted using SAS 9.1 release
(SAS Institute Inc.).
For ranking data, the Kendall’s coefficient of concordance
(W) was calculated to evaluate the concordance between
subjects on the samples’ ranking. For ranking and rating
data, a two-way analysis of variance (ANOVA) was
performed (‘‘Subject,’’ ‘‘Odorant’’) with subjects as random
factor using the SAS GLM procedure. Preplanned contrasts (no
adjustment of alpha for multiple comparisons) between the
typicality of the mixtures and their components were
performed using least squares means comparisons. To compare
the responses of the trained and untrained subjects
(‘‘Group’’ factor), a three-way ANOVA (Subject, Group,
Odorant) with interactions was performed with Subject
(nested in group) as random factor (SAS GLM procedure).
Pearson correlation coefficients were calculated using the
CORR procedure to compare typicality ratings and
rankings. For all data analyses, the effects were considered to
be significant when P < 0.05.
Dilutions in ethylic alcohol (90 ) (%)
Composition Type Component
Component Reference Reference Mixture
EI + EM + AI
30% EI + 70% EM
20.5% EI + 50% EM + 29.5% AI
50% CE + 50% FU
The aim of this experiment was to confirm that odor
blending could occur in an odorant mixture and lead to an increase
in the odor typicality of the mixture-specific odor. Thus,
a group of na¨ıve panelists performed rating and ranking
tasks to evaluate the pineapple typicality of 10 samples.
In regard to the ranking task (Figure 1), Kendall’s
coefficient of concordance calculated on the scores was highly
significant (W = 0.38, P < 0.0001), which indicated that the
subjects agreed on the ranking of the samples. Allyl caproate
(CA, Figure 1) was perceived as the most pineapple-like odor
(M = 9.01, SD = 1.65). A two-way ANOVA (Subject,
Odorant) on the typicality ranking scores indicated a significant
effect of the Odorant factor (F9,171 = 11.6, P < 0.0001).
Preplanned contrasts indicated that the F1 binary mixture was
considered to be significantly more pineapple-like than its
components EM (P < 0.041) and EI (P < 0.0005). Similar
results were obtained for the F2 ternary mixture (EM
P < 0.050, AI P < 0.018, and EI P < 0.0006). On the contrary,
even if the F3 binary mixture was ranked as more typical
than FU (P < 0.0054), its other component (CE) was ranked
as more pineapple-like than the mixture (P < 0.004).
Regarding the results of the rating task (Figure 2), allyl
caproate (CA) was rated the most typical of the pineapple
odor (M = 7.5, SD = 2.0). A two-way ANOVA (Subject,
Odorant) on the typicality scores indicated a significant
effect of the Odorant factor (F9,171 = 11.3, P < 0.0001).
Preplanned contrasts showed that the F3 binary mixture was
perceived as more typical of the pineapple odor than its
FU component (P < 0.0002) but significantly less typical
than its CE component (P < 0.042). In contrast, the F1
and F2 mixtures were rated as significantly more typical
of the pineapple odor than their components. Specifically,
F1 was rated as more typical than EM (P < 0.014) and EI
(P < 0.008). Similarly, F2 was rated as more typical than
AI (P < 0.024), EM (P < 0.011), and EI (P < 0.006).
Correlation analysis of the typicality ratings and rankings
indicated that the more typical an odorant, the higher its rank
(r(198) = 0.31, P < 0.0001). In other words, the results
obtained in the rating analysis were in accordance with those
obtained using the ranking methodology.
The aim of the second experiment was to test the hypothesis that
training and sensory olfactory expertise could influence the
perception of odor blending mixtures. A panel of trained subjects
evaluated the 10 stimuli presented in the previous experiment
(with untrained subjects) by following exactly the same
methodology. Such a strategy ensured that the only difference
between the 2 experiments was the training level of the panelists.
Kendall’s coefficient of concordance calculated from the
ranking scores was highly significant (W = 0.42, P <
0.0001) and indicated a global agreement between the
subjects. As with the na¨ıve subjects, the trained subjects perceived
allyl caproate (CA, Figure 3) as eliciting the most typical
pineapple-like odor (M = 8.6, SD = 1.6). A two-way ANOVA
(Subject, Odorant) on the typicality ranking scores indicated
a significant effect of the Odorant factor (F9,279 = 23.2, P <
0.0001). Preplanned contrasts revealed that both F1 and F2
mixtures were perceived to be as pineapple-like as the EI
component (P > 0.3). The F1 binary mixture obtained a higher
typicality rank than EM (P < 0.0006), whereas the F2 ternary
mixture obtained a higher typicality rank than EM (P <
0.003) and AI (P < 0.0001). The F3 binary mixture obtained
a higher mean rank than FU (P < 0.003) but a lower mean
rank than its CE component (P < 0.0001).
Figure 1 The means of the ranking scores of the stimuli sorted by the
naı¨ve subjects (n = 20) in the second experiment. CA (allyl caproate), CE
(ethyl caproate), F1 (EI + EM), F2 (EI + EM + AI), F3 (CE + FU), EM (ethyl
maltol), AI (allyl-a-ionone), HE (pineapple essential oil), EI (ethyl isobutyrate),
and FU (furaneol). Asterisks indicate significant differences between 2
stimuli: (*) = P < 0.1; * = P < 0.05. Error bars represent 95% confidence
interval on mean.
Figure 2 The means of the typicality ratings of the samples compared with
those of a pineapple odor on a 9-point scale. The data were obtained with
the ‘‘naı¨ve’’ subjects (n = 20) in the second experiment. CA (allyl caproate),
CE (ethyl caproate), F2 (EI + EM + AI), F1 (EI + EM), F3 (CE + FU), AI
(allyla-ionone), EM (ethyl maltol), EI (ethyl isobutyrate), HE (pineapple essential
oil), and FU (furaneol). Asterisks indicate significant differences between 2
stimuli: * = P < 0.05. Error bars represent 95% confidence interval on mean.
Figure 3 The means of the ranking scores of the stimuli sorted by the
expert subjects (n = 32) in the third experiment. CA (allyl caproate), CE (ethyl
caproate), HE (pineapple essential oil), F1 (EI + EM), F2 (EI + EM + AI), EI
(ethyl isobutyrate), F3 (CE + FU), EM (ethyl maltol), AI (allyl-a-ionone), and FU
(furaneol). Asterisks indicate significant differences between 2 stimuli: NS =
Non-Significant; ** = P < 0.01. Error bars represent 95% confidence
interval on mean.
Regarding the results of the rating task, ethyl caproate (CE,
Figure 4) and allyl caproate (CA) were rated as the most
typical of the pineapple odor (M = 7.3, SD = 1.7 and M = 7.2, SD
= 2.1, respectively). A two-way ANOVA (Subject, Odorant)
on the typicality rating scores indicated a significant effect of
the Odorant factor (F9,277 = 24.9, P < 0.0001). Preplanned
contrasts showed that the F3 binary mixture was perceived
to be more typical of the pineapple odor than its FU
component (P < 0.0001) but significantly less typical than its CE
component (P < 0.0001). The F1 and F2 mixtures were rated
to be as typical of the pineapple odor as their EI component
(P > 0.2). The F1 binary mixture was rated as more typical
than EM (P < 0.0001), and the F2 ternary mixture was rated
as more typical than AI (P < 0.0001) and EM (P < 0.0001).
Moreover, a correlation calculated between the typicality
ratings and rankings indicated that the more typical an odorant,
the higher its rank (r(316) = 0.75, P < 0.0001). Again, the
results obtained with the rating analysis were in accordance with
those obtained with the ranking methodology.
To test the hypothesis that sensory expertise could
influence the perception of odor blending mixtures, we compared
the evaluations performed by trained and untrained subjects.
We performed a three-way ANOVA (Subject, Group, and
Odorant) for each task (ranking and rating); the factor
Group represented the 2 different groups of subjects (na¨ıve
vs. trained). As expected, the results indicated a significant
effect of Odorants on both tasks (ranking: F9,450 = 29.1,
P < 0.0001; rating: F9,448 = 29.5, P < 0.0001). However, there
was no significant effect of the Group factor on either task
(ranking: F1,50 = 1.6, P > 0.2; rating: F1,50 = 0.5, P > 0.5).
However, a significant interaction Group · Odorant was
observed for both tasks (ranking: F9,450 =3.2, P < 0.002; rating:
F9,448 = 3.7, P < 0.0003). This result indicated that the
Figure 4 The means of the typicality ratings of the samples compared with
those of a pineapple odor on a 9-point scale performed by the expert
subjects (n = 32) in the third experiment. CE (ethyl caproate), CA (allyl
caproate), F1 (EI + EM), HE (pineapple essential oil), EI (ethyl isobutyrate), F2
(EI + EM + AI), F3 (CE + FU), EM (ethyl maltol), AI (allyl-a-ionone), and FU
(furaneol). Asterisks indicate significant differences between 2 stimuli: NS =
Non-Significant; ** = P < 0.01. Error bars represent 95% confidence
interval on mean.
2 groups of subjects did not evaluate the pineapple odor
typicality of some stimuli in the same way. Interestingly,
differences between the 2 groups of subjects were observed for
some mixtures and some components. The untrained
subjects found the F1 and F2 mixtures to be more typical of
the pineapple odor than all their components, whereas the
trained subjects did not. An important difference was that
the trained subjects rated ethyl isobutyrate (EI) as more
typical of the pineapple odor compared with the na¨ıve subjects
(rating: M = 5.0 vs. 3.5, P < 0.011; ranking: M = 5.6 vs. 3.7,
P < 0.007). Consequently, for the trained subjects, the
differences between the F1 mixture and the EI component did not
reach the level of significance.
Our results confirmed, through distinct psychophysical
procedures (rating and ranking), that, in human na¨ıve subjects,
certain mixtures of odorants could be judged as more typical
of pineapple odor than each of their single components. Our
findings also revealed differences in the perception of such
mixtures between na¨ıve subjects and experts who had received
general sensory analytical training (oenology students).
The quantification of odor quality in humans, especially
the comparison of the odor quality of a mixture with its
components, suffers from several difficulties (Olsson and Cain
2000; Wise et al. 2000). It has been reported that in
experimental investigations of mixture aroma quality, a single
attribute describing the main character of the aroma cannot
sufficiently reflect the contributions of all the components
to the aroma (Bult et al. 2002). In particular, it has been
argued that the use of a single attribute might obscure
perceptual interactions between odors. Indeed, the
perceptual processing strategies engaged by subjects during odor
mixture sensory testing could affect their perceptions and
responses (Le Berre, Thomas-Danguin, et al. 2008). Panelists
who were provided with specific descriptors that directed
them in rating specific feature intensities were able to
recognize the unique contribution of each manipulated
component of a complex aroma mixture (Bult et al. 2002; Le
Berre et al. 2010). In contrast, it has been shown that humans
have great difficulty in deciding whether an odor is present,
or not, in mixtures containing up to 3 or 4 odors. The limited
capacity of such an identification process is as few as 3 or 4
components, regardless of the chemical complexity of the
mixture (Laing and Francis 1989; Laing and Glemarec
1992; Livermore and Laing 1996).
In the present experiment, our objective was not to engage
the panelists in any direct analytical strategy but rather to
engage them in a synthetic strategy that might reflect a more
natural way of perceiving everyday odors, especially food
odors. We therefore postulate that the measurement of the
odor typicality of a mixture compared with the typicality
of its components relies on a holistic perception of odors
and may thus reveal putative perceptual blending processes
in odorant mixture perception. Rosch (1973) showed that
color categories are structured along a typicality gradient
from prototypes in that some exemplars were better and more
representative than others. Prototypes, as representations, are
stable within a subject’s memory and are shared across a
subject’s memory as pieces of knowledge. Chrea et al. (2005)
confirmed this theory with odors and demonstrated that
odor categories were universally organized around some
prototypes. Thus, we suggest that the typicality of an odor
reflects the degree of qualitative similarity between the actual
odor perception of a stimulus and the internal memorized
representation of this odor. In a typicality rating task, subjects
evaluate the perceptual distance between their actual
perception of a stimulus and their memorized representation. It is
likely that a higher odor typicality of a mixture, as compared
with its components, reflects a better match between the
mixture percept and the memorized odor representation. Namely,
the mixing of components in definite proportions leads to the
fusion of individual odors to create a combination with more
specific odor quality characteristics. This fusion can be seen as
a perceptual blending of individual odors in the mixture, this
has also been proposed for the cross-modal interaction
between spoken speech and the moving mouth (McGurk and
MacDonald 1976) or during multisensory integration of the
chemical senses in flavor perception (Veldhuizen et al. 2010).
In our data of untrained subjects, 2 of 3 mixtures were
found to be more typical of pineapple odor than their
individual components. Mixing the 2 odorants EI and
EM, both slightly typical of pineapple odor, caused an
increase in pineapple typicality of the mixture (F2) as
compared with both individual components. Similar results
were obtained with the F3 ternary mixture. These findings,
which were replicated in the present study using 2 different
sensory tasks (rating and ranking), suggest that odor
blending occurs but only in specific mixtures of odorants. One can
argue that the F2 mixture could be more typical of pineapple
than its unmixed components because it has 2 key odor notes
of pineapple rather than that it becomes more similar to
some main character of pineapple. Nevertheless, results
obtained in newborn rabbits with this F2 mixture support also
the idea that it is processed as a partial configuration
(Coureaud et al. 2008, 2009, 2011). Moreover, when considering
the results obtained for the third mixture (F3), the pineapple
odor typicality was found to be lower than the typicality of
one of its components (ethyl caproate; fruity note) and
higher than the other (furaneol; caramel note). It is likely that the
perception induced by this F3 binary mixture is in line with
the rule proposed by Olsson (1994) that a binary mixture
percept forms a quality positioned between the odorant qualities
of its chemical components. This F3 mixture could thus carry
the fruity and caramel notes, also carried, respectively, by EI
and EM, but with no perceptual fusion; this suggestion could
explain why pineapple typicality was not enhanced in this
mixture. Taken together, these results are consistent with
the idea that some odorant combinations are probably more
inclined to elicit perceptual interactions, thus conferring an
odor quality modification to the mixture. This theory is in
agreement with previous findings that showed that blending
was optimal for a specific ratio of odorants (Le Berre, Ishii,
et al. 2008; Coureaud et al. 2011).
Several authors have suggested that the olfactory system
could use both configural and elemental processes, according
to the complexity of the mixture. Thus, it has been proposed
that binary odorant mixtures could not produce configural
effects in humans because a mixture of 2 components causes
little loss of components’ qualities and no emergent ones
(Cain and Drexler 1974; Laing and Willcox 1983; Derby
et al. 1996). However, for more complex mixtures (more than
3 components), evidence of configural effects could be
produced by the subjects’ poor ability to accurately discriminate
and identify more than 3 components in a mixture (Laing
and Francis 1989; Laing and Livermore 1992). In our case,
perceptual blending occurred in binary and ternary mixtures,
which may support a weak configural processing (Kay et al.
2005) of such chemically simple mixtures. Indeed, in our
pineapple-like mixtures, the components were perceived as
slightly typical of the pineapple odor. Therefore, the
perception of a mixture as significantly more typical than its
components could account only for an incomplete perceptual
blending, which leads to an increase in the pineapple
typicality. Here, incomplete perceptual blending occurred in 2 of
the studied mixtures (F1 and F2). A similar conclusion
has been drawn from studies with newborn rabbits exposed
to the F2 mixture (Coureaud et al. 2008, 2009). In these
studies, when the pups had learned the odor of the mixture, they
responded to it and to the odor of the constituents. However,
after they had learned one constituent’s odor, they responded
to this odor but not to the mixture’s odor (at least for a
certain ratio of components; Coureaud et al. 2011). This result
suggests that even though the mixture was perceived to be
different from its components, information about the
individual components remained perceptible in the mixture. This
conclusion seems to be in accordance with a weak configural
processing of the mixture, evidenced through an incomplete
perceptual odor blending. This specificity of olfactory
perception could correlate with neurobiological observations
(Malnic et al. 1999; Duchamp-Viret et al. 2003).
Reports on the role of expertise in the perception of an
odor mixture mainly focus on the capacity of discrimination
and identification of odors in or out of mixtures. However,
these results did not provide clear evidence of the impact of
such specific training and exposure to odor discrimination
and identification. Whereas several studies have
demonstrated no effect of training and experience (Chambers
and Smith 1993; Roberts and Vickers 1994; Livermore
and Laing 1996), others have shown positive effects of
training on olfactory performance (Clapperton and Piggott 1979;
Rabin 1988). It has also been recognized that expertise,
especially olfactory expertise such as that used by perfumers or
oenologists, is based on 2 confounded cognitive abilities:
perceptual and semantic learning (Holley 2002; Chollet et al.
2005). In fact, experts are exposed to odors daily and are
confronted with the task of describing and verbalizing their
olfactory perceptions. Our results shed light on the impact of
such expertise on the perception of mixtures in which
perceptual blending occurs. They suggest that trained analytical
subjects (‘‘experts’’) do not perceive these blending mixtures
(the F1 and F2 mixtures) as significantly more typical than
their components, in contrast to na¨ıve subjects. In particular,
the expert subjects perceived one of the mixtures’ shared
odorants (ethyl isobutyrate) as more typical of the pineapple
odor than did the na¨ıve subjects, leading to nonsignificant
differences between both mixtures and this odorant. This
result supports the idea that olfactory expertise can modify the
configural perception of a mixture and lead the olfactory
system to turn toward a more elemental perception. This finding
is in agreement with a recent report from Le Berre et al.
(2010) and with data from animal studies in which olfactory
enrichment improved the recognition and discrimination of
individual components in mixtures (Mandairon et al. 2006).
In all typicality rating tasks, even with expert subjects, the
F1 and F2 mixtures were rated as moderately typical of the
pineapple odor (the typicality mean scores were largely
between 5 and 6). However, one of the components (ethyl
isobutyrate) was rated as poorly typical of the pineapple odor by
the na¨ıve participants, whereas the expert subjects found this
odorant to be more typical of the pineapple odor. This finding
could also explain why the expert subjects did not show a
difference in typicality ratings between the F1 and F2 mixtures
and this component. One can argue that experts usually
undergo specific training sessions in which they use odor
references elicited by single chemical compounds. This training
could explain why the consistency between ranking and rating
was much higher for the expert subjects than the na¨ıve subjects
(see correlation coefficients). Therefore, it could be assumed
that the expert subjects were more accurate but also more
likely to have a more sharply defined internal reference for
the pineapple odor. Indeed, the expert subjects both rated
and ranked the pineapple essential oil much higher than
did the na¨ıve subjects. Moreover, ethyl isobutyrate is often
used as an example of a fruity odor during the training
sessions undergone by the expert subjects. Thus, the members
of the expert panel might have been quite familiar with the
odor of ethyl isobutyrate and were thus more inclined to find
a perceptual similarity between this component and the
mixtures. Indeed, it is highly conceivable that odor typicality is
linked to familiarity with the odor, even if there are counter
examples that suggest that familiarity is not the only
determinant of typicality (Chrea et al. 2005). In addition, Lawless
et al. (1991) showed that odor category boundaries are often
fuzzy and can vary depending on the context. In our study,
such a context could be induced by the presentation of
multiple odor quality exemplars. Indeed, as has been
demonstrated for taste stimuli (O’Mahony 1991), the presentation
of multiple quality exemplars serves to sharpen the fuzzy
edges of taste quality classes and causes observers to be more
decisively inclusive/exclusive of potential category members.
To conclude, our experiments emphasized that perceptual
odor blending could occur in specific mixtures composed of
2 or 3 odorants; the results account for configural or weak
configural processing of odor mixtures. Moreover, our data have
shown that olfactory expertise, such as the one developed by
perfumers, flavorists, and oenologists could affect mixture
processing and, in some cases, prevent perceptual odor blending.
Our results suggest that compared with the olfactory systems
of na¨ıve subjects, the specific training and exposure to odors
experienced by expert subjects leads the olfactory system to
engage more readily an elemental processing of odor mixtures.
S.B. received personal support from the Edmond Roudnitska ’s
Foundation for this work.
We are particularly thankful to Prof. Gilles De Revel (Oenology
Faculty of Bordeaux, France) for allowing us to test his panel of
experts, Cle´mence Lhermey for conducting the expert subjects
phase, and Prof. Tyler S. Lorig for his helpful advice.
Bende M , Nordin S. 1997 . Perceptual learning in olfaction: professional wine tasters versus controls . Physiol Behav . 62 ( 5 ): 1065 - 1070 .
Bult JH , Schifferstein HN , Roozen JP , Boronat ED , Voragen AG , Kroeze JH . 2002 . Sensory evaluation of character impact components in an apple model mixture . Chem Senses . 27 : 485 - 494 .
Cain WS , Drexler M. 1974 . Scope and evaluation of odor counteraction and masking . Ann N Y Acad Sci . 237 : 427 - 439 .
Laing DG , Jinks AL . 2001 . Psychophysical analysis of complex odor mixtures . Chimia . 55 : 413 - 420 .
Chambers E , Smith EA. 1993 . Effects of testing experience on performance of trained sensory panelists . J Sens Stud . 8 : 155 - 166 .
Chollet S , Valentin D , Abdi H. 2005 . Do trained assessors generalize their knowledge to new stimuli . Food Qual Prefer . 16 : 13 - 23 .
Chrea C , Valentin D , Sulmont-Ross e´ C, Nguyen DH , Abdi H. 2005 . Semantic, typicality and odor representation: a cross-cultural study . Chem Senses . 30 : 37 - 49 .
Laing DG , Livermore BA . 1992 . Perceptual analysis of complex chemical signals by humans . In: Doty RL, Muller-Schwartz D , editors. Chemical signals in vertebrates VI. New York: Plenum Press . p. 587 - 593 .
Laing DG , Willcox ME . 1983 . Perception of components in binary odor mixtures . Chem Senses . 7 : 249 - 264 .
Lawless H. 1984 . Flavor description of white wine by 'expert' and non expert wine consumers . J Food Sci . 49 : 120 .
Clapperton JF , Piggott JR . 1979 . Flavour characterization by trained and untrained assessors . J Inst Brew. 85 : 275 - 277 .
Lawless H , Glatter S , Hohn C. 1991 . Context-dependent changes in the perception of odor quality . Chem Senses . 16 : 349 - 360 .
Coureaud G , Gibaud D , Le Berre E , Schaal B , Thomas-Danguin T. 2011 . Proportion of odorants impacts the configural versus elemental perception of a binary blending mixture in newborn rabbits . Chem Senses . 36 : 693 - 700 .
Le Berre E , Ishii A , B e´no N, Chabanet C , Etie´ vant P , Thomas-Danguin T. 2008 . Just noticeable differences in components concentrations modify the odor quality of a blending mixtures . Chem Senses . 33 : 389 - 395 .
Coureaud G , Hamdani Y , Schaal B , Thomas-Danguin T. 2009 . Elemental and configural processing of odour mixtures in the newborn rabbit . J Exp Biol . 212 : 2525 - 2531 .
Le Berre E , Jarmuzek E , Be´ no N, Etie´ vant P , Prescott J , Thomas-Danguin T. 2010 . Learning influences the perception of odor mixtures . Chemosens Percept . 3 : 156 - 166 .
Coureaud G , Thomas-Danguin T , Le Berre E , Schaal B. 2008 . Perception of odor blending mixtures in the newborn rabbit . Physiol Behav . 95 : 194 - 199 .
Le Berre E , Thomas-Danguin T , Beno N , Coureaud G , Etievant P , Prescott J. 2008 . Perceptual processing strategy and exposure influence the perception of odor mixtures . Chem Senses . 33 : 193 - 199 .
Deisig N , Giurfa M , Lachnit H , Sandoz JC . 2006 . Neural representation of olfactory mixtures in the honeybee antennal lobe . Eur J Neurosci . 24 : 1161 - 1174 .
Deisig N , Giurfa M , Sandoz JC . 2010 . Antennal lobe processing increases separability of odor mixture representations in the honeybee . J Neurophysiol . 103 : 2185 - 2194 .
Livermore A , Laing DG . 1996 . Influence of training and experience on the perception of multicomponent odor mixtures . J Exp Psychol Hum Percept Perform . 22 : 267 - 277 .
Malnic B , Hirono J , Sato T , Bock LB . 1999 . Combinatorial receptor codes for odors . Cell . 96 : 713 - 723 .
Derby CD , Hutson M , Livermore BA , Lynn WH . 1996 . Generalization among related complex odorant mixtures and their components: analysis of olfactory perception in the spiny lobster . Physiol Behav . 60 : 87 - 95 .
Mandairon N , Stack C , Kiselycznyk C , Linster C. 2006 . Olfactory enrichment improves the recognition of individual components in mixtures . Physiol Behav . 89 : 379 - 384 .
Duchamp-Viret P , Duchamp A , Chaput MA . 2003 . Single olfactory sensory neurons simultaneously integrate the components of an odor mixture . Eur J Neurosci . 18 : 2690 - 2696 .
Grossman KJ , Mallik AK , Ross J , Kay LM , Issa NP . 2008 . Glomerular activation patterns and the perception of odor mixtures . Eur J Neurosci . 27 : 2676 - 2685 .
Holley A. 2002 . Cognitive aspects of olfaction in perfumer practice . In: Rouby C, Schaal B , Dubois D , Gervais R , Holley A, editors. Olfaction, taste and cognition . Cambridge : Cambridge University Press . p. 16 - 26 .
Howard JD , Plailly J , Grueschow M , Haynes JD , Gottfried JA . 2009 . Odor quality coding and categorization in human posterior piriform cortex . Nat Neurosci . 12 : 932 - 939 .
Jinks A , Laing DG . 2001 . The analysis of odor mixtures by humans: evidence for a configurational process . Physiol Behav . 72 : 51 - 63 .
Kay LM , Crk T , Thorngate J. 2005 . A redefinition of odor mixture quality . Behav Neurosci . 119 : 726 - 733 .
Kay LM , Stopfer M. 2006 . Information processing in the olfactory systems of insects and vertebrates . Semin Cell Dev Biol . 17 : 433 - 442 .
Laing DG . 1991 . Perception of complex smells . In: Dulbecco R, editor. Encyclopedia of human biology . Vol. 6 . 1st edn . New York : Academic Press. p. 759 - 767 .
Laing DG . 1994 . Perceptual odor interactions and objective mixture analysis . Food Qual Prefer . 5 : 75 - 80 .
McGurk H , MacDonald J. 1976 . Hearing lips and seeing voices . Nature . 264 : 746 - 748 .
Olsson MJ . 1994 . An interaction model for odor quality and intensity . Percept Psychophys . 55 : 363 - 372 .
Olsson MJ , Cain WS . 2000 . Psychometrics of odor quality discrimination: method for threshold determination . Chem Senses . 25 : 493 - 499 .
O'Mahony M. 1991 . Descriptive analysis and concept alignment . In: Lawless HT , Klein BP, editors. Sensory science theory and applications in foods. New York: Marcel Dekker . p. 223 - 268 .
Parr WV , Heatherbell D , White KG . 2002 . Demystifying wine expertise: olfactory threshold, perceptual skill and semantic memory in expert and novice wine judges . Chem Senses . 27 : 747 - 755 .
Rabin MD . 1988 . Experience facilitates olfactory quality discrimination . Percept Psychophys . 44 : 532 - 540 .
Roberts AK , Vickers ZM . 1994 . A comparison of trained and untrained judges' evaluation of sensory attributes intensities and liking of Cheddar cheeses . J Sens Stud . 9 : 1 - 20 .
Rosch E. 1973 . On the internal structure of perceptual and semantic categories . In: Moore TE, editor. Cognitive development and the acquisition of language . New York : Academic Press. p. 110 - 144 .
Silbering AF , Galizia CG . 2007 . Processing of odor mixtures in the Drosophila antennal lobe reveals both global inhibition and glomerulus-specific interactions . J Neurosci . 27 : 11966 - 11977 .
Laing DG , Francis GW . 1989 . The capacity of humans to identify odors in mixtures . Physiol Behav . 46 : 809 - 814 .
Veldhuizen MG , Shepard TG , Wang MF , Marks LE . 2010 . Coactivation of gustatory and olfactory signals in flavor perception . Chem Senses . 35 : 121 - 133 .
Laing DG , Glemarec A. 1992 . Selective attention and the perceptual analysis of odor mixtures . Physiol Behav . 52 : 1047 - 1053 .
Wise PM , Olsson MJ , Cain WS . 2000 . Quantification of odor quality . Chem Senses . 25 : 429 - 443 .