Chem. Senses 27: 747-755,
2002
© Oxford University Press 2002
Demystifying Wine Expertise: Olfactory Threshold, Perceptual Skill and Semantic Memory in Expert and Novice Wine Judges
1 Centre for Viticulture & Oenology, Lincoln University, Canterbury, New Zealand 2 Department of Psychology, University of Otago, Dunedin, New Zealand
Correspondence to be sent to: Wendy V. Parr, Animal & Food Sciences Division, Lincoln University, PO Box 84, Lincoln University, Canterbury, New Zealand. e-mail: parrw1{at}lincoln.ac.nz
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and methodsResults and discussionReferences |
|---|
We investigated recognition and identification of wine-relevant odours as a function of domain-specific expertise. Eleven wine experts and 11 wine novices participated in tasks measuring olfactory threshold, odour recognition, odour identification, and consistency of odour naming. Twenty-four wine-relevant odorants were sampled orthonasally by each participant in the semantic (identification; consistency of naming) and episodic (recognition) memory tasks. Results showed superior olfactory recognition by expert wine judges, despite their olfactory sensitivity and bias measures being similar to those of novices. Contrary to predictions based on reports of an association between odour memory and semantic processing, wine experts did not perform better than novices on the verbal memory tasks. Further, ability to recognize odours and ability to name odours were not positively correlated, although the novices' data showed a trend in this direction. The results imply that the source of superior odour recognition in wine experts was not enhanced semantic memory and linguistic capabilities for wine-relevant odours. One interpretation of the data is that wine experts were less susceptible than wine novices to verbal overshadowing. When forced to identify the odorants, experts' superior perceptual skills protected them from verbal interference, whereas novices' generated verbal representations of the odours were emphasized at the expense of the odorant itself. This has implications for training in wine-evaluation skills.
|
Introduction |
|---|
|
TopAbstractIntroductionMaterials and methodsResults and discussionReferences |
|---|
What underlies a wine lover's ability to identify a favourite vintage with their nose? Olfaction is clearly an important process when evaluating complex mixtures such as wine where much flavour is aroma (Thorngate, 1997
). The
orthonasally experienced, volatile component of a wine is typically perceived
prior to tasting and can play a key role in affecting the judgement of quality
that the wine receives. The nose provides considerable information about the
type, such as cultivar or physical aspects of terroir, the wine's age,
condition and overall quality. The judge may not be consciously aware of the
contributions of the nose because there is no easy parallel to the shutting of
eyes to exclude visual input. In some evaluative contexts, however, such as
winemaking, olfaction alone may be employed as a quality control tool. For
example, the nose may be relied upon for early detection of wine faults such
as excess acetic acid or hydrogen sulphide. The recognition and identification
of relevant odours is therefore important to fault detection and exemplifies
the role of cognitive processes in wine evaluation.
Despite their obvious importance, there has been little systematic
investigation of the cognitive components of olfaction in relation to wine
expertise (Parr, 2000). The
last decade has seen increased interest however
(Morrot and Brochet, 1999),
notably concerning associations between colour and odour
(Morrot et al.,
2001). The present study investigated olfactory sensitivity, odour
recognition, odour identification and consistency of odour labelling in expert
and novice wine judges. The questions we asked were: Are wine experts more
accurate than novices at recognizing and identifying wine-relevant smells? Is
the greater ability of wine experts to recognize and identify odours a result
of their enhanced sensitivity or their semantic knowledge?
Fundamental psychological research on olfaction and cognition suggests that
olfaction is a particularly important sense to understand in relation to wine
evaluation, not least because humans are considered to have relatively
impoverished language for describing odours
(Engen, 1982). Odours
frequently evoke idiosyncratic, autobiographical memories which are often
learned in childhood (Chu and Downes,
2000), are often associated with emotion
(Herz, 1997;
Epple and Herz, 1999), and can
be resistant to unlearning or relearning
(Lawless and Engen, 1977),
including in expert perfumers (Ishii
et al., 1997).
The major theoretical basis for the present study concerns the relation
between odour memory and language. Relatively poor performance by human adults
when recognizing and labelling everyday odours has frequently been reported
(Cain and Potts, 1996) and
typically interpreted as having its source in poor semantic memory
(Rabin and Cain, 1984;
Lehrner et al.,
1999). Lehrner et al., for example, investigated memory
for everyday odours across several age ranges. For adults, they reported
positive correlations between odour identification and odour recognition
(r = +0.69), and between naming consistency and odour recognition
(r = +0.54). The notion that linguistic limitations underlie poor
odour recognition and identification has been adopted by many wine
professionals and incorporated into their learning and teaching programmes. It
is common practice when learning to assess wines that students are encouraged
to engage in a matching process whereby perceived smells and tastes are
matched to a linguistic tool such as the Wine Aroma Wheel
(Noble et al.,
1984).
There is no direct evidence, however, that wine judges' semantic (verbal)
memories are the source of their ability to recognize wine components such as
a fault. To the contrary, emphasis on the linguistic component of olfactory
cognition may come at a price, especially when emphasized early in a wine
professional's development. In keeping with this notion, oenologist Emile
Peynaud is reported to have said that fluency is often a screen for inaccuracy
(Brochet, 1999). Olfactory
perceptual ability (smelling) and language (naming an odour) may be
associated, not in a facilitative way, but in an inhibitory way
(Melcher and Schooler, 1996;
Lorig, 1999). More
specifically, verbal and perceptual processes related to olfaction may
interfere with one another, the degree of interference being mediated by
expertise in the particular domain such as wine. Melcher and Schooler
investigated memory for wine tastes as a function of expertise. They concluded
that verbalization, rather than enhancing learning of complex stimuli, may
have an insidious, disruptive effect that they term `verbal overshadowing'.
Further, their data demonstrated the disruption to be a function of expertise
of participants.
In the present study, we investigated olfactory-guided judgements in expert
and novice wine judges within a detection theory framework. Evaluations such
as detecting an off-note in wine are analogous to other diagnostic problems
that are intrinsically probabilistic. A fundamental characteristic of such
tasks is that many variables contribute to the `evidence' for a decision.
Detection theory permits the ability to detect and recognize a smell to be
measured independently from motivational factors that can influence the
judging and deciding aspects of the task
(MacMillan and Creelman,
1991). The present study aimed to simulate odour-discrimination
tasks that occur within the typical wine-evaluation situation. Odour
identification was employed as a measure of explicit, semantic memory, while
odour recognition was used as a measure of explicit, episodic memory
(Elsner, 2001). Semantic memory
is assumed based on a person's general knowledge and experience with an
odorant (Savic and Berglund,
2000). In contrast, episodic memory is not necessarily based on a
verbal representation but has its basis in perceptual and possibly imaging
processes (Lehrner, 1993;
Herz, 2000).
In order to provide a basis for assessing accuracy in judgements of the
bouquet of a wine, the stimuli employed in the present study comprised
compounds typically found in wine (Lenoir,
1995; Bende and Nordin,
1997). The compounds were selected on the following basis. They
were compounds that had perceived odour notes with well-established veridical
names in prior published literature [e.g. the Atlas of Odor Character
Profiles (ASTM, 1985)]
and/or were included in the tool-kit of chemical compounds known as Le nez
du vin (Lenoir, 1995)
that is available for learning about wine aroma. The odorants spanned the
categories of wine faults such as excess acetic acid, primary characters
(those pertaining to the grape such as floral and fruity notes), secondary
characters (those pertaining to fermentation and winemaking procedures), and
maturation characters such as mushroom or leather.
Superior performance of expert participants was expected on the tasks
involving odour naming, odour recognition, and consistent use of an odour name
(Lehrner et al.,
1999). Such expertise was not expected in a group of novice
participants. We expected, however, that experts and non-experts would not
differ on ability to detect 1-butanol
(Bende and Nordin, 1997;
Morrot, 1999).
|
Materials and methods |
|---|
|
TopAbstractIntroductionMaterials and methodsResults and discussionReferences |
|---|
Subjects
Twenty-two adults, 11 experts and 11 novices, classified on the basis of
their experience with wines, participated in the study. The groups were
matched for age, gender, dietary and smoking status. Exact matching proved
difficult. There were five female novices, six male novices, four female
experts, and seven male experts. Age range was 25-55 years for novices and
25-58 years for experts. There was no significant difference between the
groups in terms of age. Each group contained one participant who was an
occasional smoker. The remaining participants were non-smokers. Experts were
defined in accordance with several previous studies
(Melcher and Schooler, 1996;
Bende and Nordin, 1997). A
person was defined as an expert if they fitted at least one of the following
categories:
- established winemakers;
- wine-science researchers and teaching staff who were regularly involved in
wine-making and/or wine evaluation;
- wine professionals (e.g. Master of Wine, wine judges, wine writers, wine
retailers).
- graduate students in Viticulture and Oenology who had relevant professional
experience (e.g. had participated in more than one vintage; had run
wine-tasting classes);
- persons with an extensive (> 10 years) history of wine involvement (e.g.
family history, extensive wine cellar, regular involvement in formal wine
tastings).
Novices in the present study were defined as those persons who drank wine
regularly but had participated in little formal wine evaluation or winemaking
at the time of the study. The novice group included wine and food students who
were outside the criteria for inclusion in the expert group. Relative to most
studies in the sensory literature where novices and experts have been compared
(Chollet and Valentin, 2000),
the present group of novices could be defined as `intermediates' rather than
as novices (Melcher and Schooler,
1996). For example, the study by Chollet and Valentin compared
senior wine students as experts, and senior students in other faculties as
novices, when they investigated individual differences in people evaluating
Burgundian red wines. The aim of comparing experts to intermediates rather
than to complete novices was to provide a relatively stringent test of the
issue of whether the assumed greater semantic knowledge of experts influenced
their olfactory discrimination.
Materials
The stimuli employed in the olfactory-detection threshold task were
prepared as described by Lehrner et al.
(Lehrner et al.,
1999). Beginning with a 4% solution of 1-butanol in distilled
water, serial dilution progressed in 10 steps of successive thirds (dilution
factor 3). The 11 concentrations, ranging from 4% (dilution step 0) to
0.00007% (weakest concentration; dilution step 10), were stored in glass
bottles with tightly fitting, plastic screw lids. Each contained 
7 ml of
fluid. Four identical bottles, each containing distilled water, were also
prepared.
The 28 stimuli used as odorants were chemical compounds. Due to the
difficulty in assessing what it means to be `right' when describing the
bouquet of a wine, the odorants used were compounds typically found in wine
(Lenoir, 1995;
Bende and Nordin, 1997), rather
than actual wines. They were selected to provide perceived odour notes from
the categories of wine faults, primary characters, secondary characters, and
aged characters (see Table 1
for a complete list). The concentration and dilution medium selected for each
odorant were based on data provided in the published literature. Prior to the
experiment proper, the odorants were rated by another 11 adults (drawn from
the same subject pool as the novices) on a 100 mm Visual Analogue Scale
(Savic and Berglund, 2000)
with respect to quality and intensity of the particular odour note. Compounds
that received mean ratings on the scale >80 mm or <20 mm were not
included. Odorants were contained in 10 ml, amber glass bottles with
polypropylene screw lids. Odorants were kept in a refrigerator when not in use
and taken out to warm up to ambient room temperature (20°C) before an
experimental session began.
|
Procedure
Participants were tested individually. Nineteen of the 22 participants were
tested in a purpose-built, sensory-evaluation laboratory that was designed
according to the guidelines of the American Society for Testing and Materials
(ASTM, 1986). Ambient
temperature of the room was maintained at 20 ± 2°C. The remaining
three participants were tested at the wineries at which they were owners or
employees. For two of the three people who participated off-campus, the
conditions at their winery were similar to those within the Sensory Laboratory
at Lincoln University. The third participant who was tested off-campus took
part in an outdoor setting that was relatively free of interference in terms
of noise, visual stimulation, ambient odour, and adverse weather conditions
such as wind. Each participant was given a code number, seated comfortably,
and general instructions were given. Novices and experts alternated in terms
of participation order. Alternating category order was employed to
counterbalance any effects from changes in the headspace of chemical stimuli
over time. Participants were advised that the study involved naming and
remembering wine-relevant smells. The participant was then seated within a
booth that included a plain white table on which the stimuli were handled.
Odour-detection threshold, odour identification and odour recognition were
performed in that order, in a single session that lasted 60 min. Specific
instructions preceded each individual task. To estimate odour detection
threshold, solutions of n-butyl alcohol in distilled water were used
in a two-alternative, forced-choice procedure
(Bende and Nordin, 1997;
Lehrner et al., 1999)
involving an ascending staircase method of limits. Starting at the lowest
concentration, an odorant bottle was presented to the participant in the
booth, accompanied by an identical bottle that contained distilled water only.
Participants were encouraged to sniff each bottle bi-rhinally. An inter-trial
interval of 30 s occurred between successive trials. The distilled-water-only
bottle was presented as the left or the right sample equally often. When a
correct choice was made, the same concentration of odorant was presented to
the participant until four consecutive correct responses were given. This
concentration was taken as an estimate of the participant's detection
threshold. A different bottle of distilled water was presented alongside each
of the four consecutive presentations of the same concentration of
odorant.
Cognitive tasks
Twenty-four odorants were selected randomly for each participant from the
28 odorants that comprised the larger sample set. Of these, 12 were selected
randomly to comprise `old' stimuli, with the remaining 12 designated as `new'
stimuli in the subsequent recognition test. Participants smelled in succession
12 odorants from the 24-item stimulus set selected for them and were asked to
remember the smells. During testing, odorants other than the one being sniffed
were kept tightly sealed and an extraction fan minimized diffusion of odours
into the testing room. A stimulus presentation rate of 45 s was employed,
during which time a participant sniffed the compound ad libitum and
attempted to name the odour as specifically as possible. Participants were
reminded of the wine context. An inter-trial interval of 30 s followed.
Following presentation of the 12 stimuli designated as `old', a retention
interval of 10 min occurred. During the interval, the participant was invited
to chat about their wine-relevant experience. Twenty-four odorants were then
presented in random order. They comprised the 12 previously presented odorants
(old) and 12 new. Participants judged whether each odorant was old or new,
gave a confidence rating for the recognition judgement, named the odorant as
specifically as possible, and finally, gave a confidence rating to reflect
their certainty that the name provided was the veridical name. The confidence
rating scale comprised a horizontal line scale, numbered 1-5, with the words
`extremely confident' positioned below the 5, and `not at all confident' below
the 1.
Quantitative analyses
The score obtained for each participant's odour-detection threshold comprised the dilution number corresponding to the 1-butanol concentration correctly chosen over distilled water in four consecutive trials. A high number represents a low threshold.
Olfactory recognition
Based on the theory of signal detection (TSD), hit rates, false-alarm
rates, and measures of discriminability and bias were calculated
(MacMillan and Creelman,
1991). A `hit' was defined as a `yes' response to an old
(previously presented) odorant, and a false alarm (FA) was defined as a `yes'
to a new odorant. The signal detection approach treats the odour recognition
task in the same way as a memory recognition task, where new and old items
vary along a psychological dimension of memory strength or familiarity. The
groups of new and old items are represented by normal probability
distributions on the familiarity dimension. A yes/no response in the
recognition task is based on the assumption that the judge establishes a
criterion, C, on the psychological dimension. If the familiarity of
an odour is greater than the criterion, the judge responds `yes', and if it is
weaker than C, the judge responds `no'. Discriminability is the
distance between the means of the probability distributions for new and old
odours. The measure of discriminability calculated was the recognition index
d' and the measure of bias was the criterion measure,
C. Discriminability, d', is calculated from hit rates
and false-alarm rates (see below). As a measure of the ability to recognize
odours, it has two main advantages. First, it is not confounded with response
bias (a tendency to say `yes'), which can be measured separately in terms of
the location of the criterion, C, on the familiarity dimension. Note
that with higher values of C there is a tendency to say `yes', and
that C can vary independently of the distance, d',
between the probability distributions for new and old odours. Second,
d' varies on an equal-interval scale and is not bounded in the
same way as is the traditional measure of accuracy, percent correct.
 | (1) |
 | (2) |
). This involved adding 0.5 to each frequency of hits
and false alarms and dividing by N + 1, where N is the
number of old or new stimuli.
In the context of TSD, confidence judgements in the recognition task can be
interpreted as the person making graded responses that reflect their degree of
experience with each odorant. In detection theory, this is analogous to
employing multiple criteria within a single task or situation so that the
levels of confidence correspond to movements in the bias parameter
(C) (Lawless and Heymann,
1998). Memory operating characteristic (MOC) curves for the
probability of calling a previously presented odorant `old' versus the
probability of calling a previously presented odorant `new' were constructed
for the groups (Cain and Potts,
1996) by working out a mean hit and false-alarm rate for each
confidence interval. The smooth curves in
Figure 2 were fitted to the
data points using nonlinear regression. Their equation is based on the
distributions of stimulus effect assumed by TSD, with parameters for
detectability, d', and the ratio of the variance of the two
distributions.
|
Semantic memory
The name(s) given for each odorant was scored for correctness
(Cain, 1979;
Lehrner et al.,
1999), and for consistency of usage. For correctness, veridical
labels were scored 2 (e.g. `pear' for pear), a near miss (e.g. `cloves' for
cinnamon) was given a score of 1, and a far miss (e.g. `citrus' for a buttery
note) was scored zero, giving a maximum of 48 points for correct
identification for each participant. Consistency of naming was scored as 1
when an odorant was named with the same label at initial presentation and at
recognition testing. A score of zero was given when a different label was used
across the two situations (e.g. `spicy' at presentation and `marzipan' at
testing). Proportions correct were derived for each participant from their
correctness of naming score and from their consistency of naming score.
The relationships between olfactory performances were assessed using Pearson's correlations. They were performed on the individual data observed for each task.
|
Results and discussion |
|---|
|
TopAbstractIntroductionMaterials and methodsResults and discussionReferences |
|---|
Sensitivity to 1-butanol
Estimates of the participants' detection thresholds for n-butyl
alcohol concentrations ranged between 0.00007% (dilution step 10) and 0.016%
(dilution step 5) in both groups. Consistent with previous results
(Bende and Nordin, 1997),
detection thresholds did not differ between groups, t(20) = 0.31,
P = 0.76 (experts: mean = 8.18, SD = 1.40; novices: mean = 8.36, SD =
1.36). The means are in keeping with the reported range for olfactory
threshold of 1-butanol as 2-5 p.p.m.
(Moskowitz et al.,
1974) as dilution step 8 represents a concentration of 0.0002%.
Correlation coefficients were calculated between threshold scores and the
cognitive tasks. There were significant inverse correlations between estimates
of experts' thresholds and odour recognition (r = -0.82, P
< 0.05), and between estimates of novices' thresholds and odour
identification (r = -0.60, P < 0.05). That is, for
experts, higher thresholds for n-butyl alcohol detection (lower
dilution steps) were positively associated with accuracy of odour recognition,
supporting the notion that superior sensitivity in experts was not the source
of any enhanced olfactory memory performance. For novices, the higher the
estimated threshold for n-butyl alcohol detection, the greater their
accuracy for identification. These associations are not easy to interpret.
However, Lehrner et al. (Lehrner
et al., 1999) also reported significant negative
associations between olfactory threshold and both odour memory and odour
identification in their adult sample.
Olfactory recognition and semantic memory
There was a significant difference in odour recognition, as measured by the discriminability index d', as a function of wine-relevant expertise, t(20) = 2.13, P < 0.05. Experts showed superior recognition of olfactory stimuli (mean = 2.26, SD = 0.49) when compared with novices (mean = 1.63, SD = 0.85). Figure 1 and Table 2 show detection thresholds and recognition results. Not only was episodic memory enhanced in experts as shown by their enhanced recognition ability, but experts also demonstrated less within-group variability as reflected in the standard deviation measures. There was no difference between groups in the bias measure, t(20) = 0.184, P > 0.05. Therefore the difference between the groups reflected a true difference in recognition ability and not a difference in tendency to report having experienced the odorant before.
|
|
A two-way ANOVA on the hit rate versus false-alarm rate data for experts
versus novices produced a significant interaction between these two variables,
F(1,20) = 6.10, P < 0.05.
Table 2 shows the `mirror
effect' where experts' hit rates were overall higher and their false-alarm
rates were overall lower compared to those of novices. This result is similar
to that found in both human and non-human memory studies where increasing the
difficulty of a discrimination task results in an increased false-alarm rate
(Wixted, 1992). It is also
consistent with the conclusion that novices found the task overall more
difficult than experts. The symmetrical change in hit and false-alarm rates is
consistent with no difference in response bias (at least as defined by the
measure, C), F(1,20) = 0.034. The greater difficulty in
discrimination for novices was reflected in a lower d' than for
experts, F(1,20) = 4.55, P < 0.05.
Memory operating characteristic (MOC) curves for the recognition data as a function of expertise are shown in Figure 2. The values of the d' parameter for the fitted functions were 3.35 and 2.55, respectively, for the experts and novices. The values of r were 1.0 in both cases, showing that the variances of the distributions for old and new odorants were equal and symmetrical.
Table 2 reports proportions
correct for the semantic memory tasks, namely identification and consistency
of labelling an odorant, as a function of expertise. Each group's mean
identification performance was similar to, or slightly better than, that
reported in the literature concerning humans' identification of everyday
odorants (Cain and Potts,
1996). However, contrary to experimental hypotheses, there was no
evidence of superior olfactory identification by experts, t(20) =
1.31, P > 0.05 (expert mean = 0.51, SD = 0.12; novice mean = 0.45,
SD = 0.10). Nor was there an effect of expertise on consistency of labelling
the wine-relevant odorants, t(20) = 0.30, P > 0.05
(expert mean = 0.55, SD = 0.14; novice mean = 0.53, SD = 0.22). To investigate
the relationship between measures of semantic memory (identification and
naming consistency) and episodic memory (recognition), Pearson's correlations
were performed. The results are reported in
Table 3.
|
The correlation coefficients show no significant relation between odour
recognition and odour identification or between odour recognition and
consistency of naming for either experts or novices. However, there is a trend
in these directions for novices that may have reached significance were a
larger sample size employed in the study. There is no such trend for odour
recognition and identification in the expert data. There is also a trend
toward a positive association between odour recognition and consistency of
naming for experts. This suggests that, for experts, consistent use of a name
is more important than its `objective' or veridical name in advantaging odour
recognition. Lehrner et al. reported a similar result
(Lehrner et al.,
1999).
Olfactory performance by odorant
Descriptive statistics were gathered on the odorants used in the experiment. Each odorant could be employed a maximum of 22 times (i.e. once per subject) as the 24 odorants used for any particular subject were selected at random from the larger stimulus set of 28 odorants. Presentation frequencies over the duration of the study ranged from 9 (cloves) to 19 (vinegar; pine), with a median of 15.5. Table 4 shows mean recognition accuracy and mean identification score as a function of odorant. Proportions correct for recognition ranged from 1.0 for anise and nutty/sweet to 0.63 for green/vegetal. Proportions correct for identification ranged from 0.63 for caramel to 0.23 for cloves. When odorants were sorted to provide an order of recognizability and an order of identifiability, there was no particularly salient outlier in either list. No relation was found between an odorant's recognizability and its identifiability, Pearson's r = -0.08. For example, vanilla was recognized with 95% accuracy but identified less frequently than the mean identifiability score.
|
The major finding of the present study is the demonstration of superior explicit recognition by wine experts for wine-relevant odorants. This superiority did not have its source in bias (criterion), sensitivity (detection threshold), or semantic memory as measured by odour identification and naming consistency. This implies that the locus of superior recognition of wine-relevant odours in the present study appears to be perceptual, or sensory-based memory (e.g. olfactory imaging).
There are several differences between the present study and most published
research concerning odour recognition and identification
(Lehrner et al.,
1999). First, the present study involved a contextualized
situation where domain-specific olfactory expertise was investigated, rather
than olfactory expertise in general. Second, the present study's sample size
was relatively small. The aim of employing a small N was to
demonstrate any effects that were sufficiently robust to be detectable with
small groups of participants. This resulted in the performance of participants
on odour recognition and identification tasks being dissociated by the
variable of wine expertise.
Finally, the present study involved experts and novices who presumably differed more with respect to experientially gained knowledge than semantic knowledge (e.g. wine theory). The present novices were intermediates in relation to wine education and it is conceivable that perceptual skill opportunities separated the novices from experts more than linguistic skills or semantic knowledge. The specific style of wine-evaluation experience encountered by many in the novice group largely involved analytical techniques with a strong linguistic base. That is, students are encouraged to deconstruct a wine into its particular characters (e.g. odours, tastes and mouth-feel components), identifying and verbally labelling each individual character that has been detected. This approach contrasts with more synthetic wine-evaluation approaches where a wine may be considered as a whole or Gestalt. It is conceivable that evaluation of a wine as a whole, rather than deconstructing it, places greater emphasis on perceptual skill than on linguistic skill.
One possible interpretation of the present data involves Melcher and
Schooler's concept of verbal overshadowing
(Melcher and Schooler, 1996).
Verbal overshadowing is a form of memory illusion that is assumed to occur
when people are forced to name complex stimuli (i.e. those that are difficult
to capture in words, such as tastes and smells), particularly when the
relevant perceptual and linguistic skills are not equally developed. An
example of verbal overshadowing is where a verbal representation of a stimulus
(e.g. the word `aniseed') is remembered at the expense of the actual stimulus
itself (e.g. the odour of aniseed). When a person has both perceptual and
verbal expertise in a domain (as could be expected from the experts in the
current study), their susceptibility to verbal overshadowing is assumed
reduced because experts can shift between reliance on verbal or perceptual
expertise without consequence. Such a notion could account for the present
data from the expert group. To date, verbal overshadowing has been examined in
situations where it has been argued that perceptual expertise exceeded verbal
expertise (Melcher and Schooler,
1996; Schooler and Engster-Schooler, 1990). The present study, to
our knowledge, is the first to demonstrate attenuated episodic (recognition)
memory for odorants in a situation where verbal expertise was similar across
groups. The notion that information processing of odours can be interfered
with by concurrent use of language has been argued on the basis of
electro-physiological as well as behavioural data. Lorig hypothesizes that
language perception and odour information processing share a similar neural
substrate so that when we are called upon to simultaneously process odour and
language information, interference occurs
(Lorig, 1999).
Although language serves memory well under many situations, language may
also be an insidious source of memory disruption in situations for which it is
not well suited, such as when remembering smells. The type and degree of
disruption appear dependent on an individual's domain-specific expertise. It
is conceivable that in some areas of expertise, semantic memory plays a large
role in the early stages of skill development, but that qualitatively
different processes are involved subsequently as expertise advances. There is
a precedent in the literature for arguing for such a qualitative change in
processing with increasing expertise. For example, in the specific area of
cognitive research involved with problem solving, work with expert systems
(i.e. artificial intelligence simulations of human performance) has been used
to argue for qualitatively different processes underlying medical diagnoses by
experts and novices. It has been argued that inexperienced physicians follow
rules (e.g. `if three out of five symptoms are present, diagnose as X'),
whereas senior medical specialists may operate at a more global or `intuitive'
level (Reisberg, 1997).
In conclusion, an accumulating body of evidence suggests that verbal codes
are not essential, or even necessarily activated
(Herz, 2000), for successful
odour-guided cognition. In keeping with this notion, our data suggest that
perceptual skill, at least in relation to olfaction, is critical to wine
expertise. Further, emphasizing verbal skill (e.g. forced naming of a
perceived odour and/or matching to a linguistic tool) may interfere with
olfactory performance in some situations (e.g. in the absence of
well-developed perceptual expertise). From an applied perspective, some things
may be better left unsaid (Schooler and
Engstler-Schooler, 1990): that is, current training methods in
wine evaluation would be unwise to emphasize linguistic skills in the absence
of well-developed, relevant perceptual skill.
|
Acknowledgments |
|---|
The work was funded by a Lincoln University Doctoral Scholarship, a Lincoln University Fund for Excellence award, and a Claude McCarthy Fellowship awarded by the New Zealand Vice-Chancellors' Committee (2001) to Wendy Parr. We would like to thank Bedoukian Research Inc. for generously donating many of the odorants, and Dr Rob Sherlock, Graeme Steans and Janette Busch for assistance with development of materials.
|
References |
|---|
|
TopAbstractIntroductionMaterials and methodsResults and discussionReferences |
|---|
ASTM (1985) Atlas of Odor Character Profiles. ASTM Publications, Philadelphia.
ASTM (1986) Physical Requirement Guidelines for Sensory Evaluation Laboratories. ASTM STP 913. ASTM Publications, Philadelphia.
Bende, M. and Nordin, S. (1997) Perceptual learning in olfaction: professional wine tasters versus controls. Physiol. Behav., 62,1065 -1070.[Medline]
Brochet, F. (1999) The taste of wine in consciousness. Journal International des Sciences de la Vigne et du Vin: Special Issue: Wine Tasting,19 -22.
Cain, W.S. (1979) To know with the nose:
keys to odor identification. Science,203
, 467-470.
Cain, W.S. and Potts, B.C. (1996)
Switch and bait: Probing the discriminative basis of odor identification
via recognition memory. Chem. Senses,21
, 35-44.
Chollet, S. and Valentin, D. (2000) Le degré d'expertise a-t-il une influence sur la perception olfactive? Quelques éléments de réponse dans le domaine du vin. L'Anne psychologique, 100,11 -36.
Chu, S. and Downes, J. (2000)
Odour-evoked autobiographical memories: psychological investigations of
Proustian phenomena. Chem. Senses,25
, 111-116.
Elsner, R.J.F. (2001) Odor memory and aging. Aging Neuropsychol. Cogn.,8 , 270-284.
Engen, T. (1982) The Perception of Odors. Academic Press, New York.
Epple, G. and Herz, R.S. (1999) Ambient odors associated to failure influence cognitive performance in children. Dev. Psychobiol., 35,103 -107.[Web of Science][Medline]
Herz, R.S. (1997) Emotion experienced during encoding enhances odor retrieval cue effectiveness. Am. J. Psychol., 110,489 -506.[Web of Science][Medline]
Herz, R.S. (2000) Verbal coding in olfactory versus nonolfactory cognition. Mem. Cognit.,28 , 957-964.[Web of Science][Medline]
Herz, R.S. and Engen, T. (1996) Odor memory: review and analysis. Psychon. Bull. Rev.,3 , 300-313.[Web of Science]
Ishii, R., Kemp, S., Gilbert, A. and O'Mahony, M. (1997) Variation in sensory conceptual structure: an investigation involving the sorting of odor stimuli. J. Sens. Stud., 12,195 -214.
Lawless, H.T. and Engen, T. (1977) Associations to odors: interference, mnemonics and verbal labeling.J. Exp. Psychol. Hum. Learn. Mem. , 3,52 -59.
Lawless, H.T. and Heymann, H. (1998)Sensory Evaluation of Food: Principles and Practices . Chapman & Hall, New York.
Lehrner, J. (1993) Gender differences in
long-term odor recognition memory: verbal versus sensory influences and
consistency of label use. Chem. Senses,18
, 17-26.
Lehrner, J., Gluck, J. and Laska, M.
(1999) Odor identification, consistency of label use,
olfactory threshold and their relationships to odor memory over the human
lifespan. Chem. Senses, 24,337
-346.
Lenoir, J (1995) Le nez du vin. Éditions Jean Lenoir, France.
Lorig, T.S. (1999) On the similarity of odor and language perception. Neurosci. Biobehav. Rev.,23 , 391-398.[Web of Science][Medline]
Macmillan, N. and Creelman, C.D. (1991)Detection Theory: a User's Guide . Cambridge University Press, Cambridge.
Melcher, J. and Schooler, J. (1996) The misremembrance of wines past: verbal and perceptual expertise differentially mediate verbal over-shadowing of taste memory. J. Mem. Lang., 35,231 -245.
Morrot, G. (1999) Can we improve taster performance? Journal International des Sciences de la Vigne et du Vin: Special Issue: Wine Tasting,29 -35.
Morrot, G. and Brochet, F. (1999) What our nose can tell us. Journal International des Sciences de la Vigne et du Vin: Special Issue: Wine Tasting,15 -17.
Morrot, G., Brochet, F. and Dubourdieu, D. (2001) The color of odors. Brain Lang.,79 , 309-320.[Web of Science][Medline]
Moskowitz, H.R., Dravnieks, A., Cain, W. and Turk, A.
(1974) Standardized procedure for expressing odor intensity.
Chem. Senses Flavor, 1,235
-237.
Noble, A., Arnold, R., Masuda, B., Pecore, S., Schmidt, J.
and Stern, P. (1984) Progress towards a standardized
system of wine aroma terminology. Am. J. Enol.
Viticulture, 35,107
-109.
Parr, W.V. (2000) Sensory evaluation of wine: a psychological perspective. Proceedings of Lincoln University Annual Grape and Wine School, Canterbury, NZ. Lincoln University, pp. 91-99.
Rabin, M.D. and Cain, W.S. (1984) Odor recognition: familiarity, identifiability, and encoding consistency. J. Exp. Psychol. Learn. Mem. Cogn.,10 , 316-325.[Web of Science][Medline]
Rapp, A. and Mandery, A. (1985) Wine aroma. Experientia, 42,873 -884.
Reisberg, D. (1997) Cognition: Exploring the Science of the Mind. W.W. Norton, New York.
Savic, I. and Berglund, H. (2000)
Right-nostril dominance in discrimination of unfamiliar, but not familiar,
odours. Chem. Senses, 25,517
-523.
Schooler, J.W. and Engstler-Schooler, T.Y. (1990) Verbal overshadowing of visual memories: some things are better left unsaid. Cognit. Psychol.,22 , 36-71.[Web of Science][Medline]
Snodgrass, J. and Corwin, J. (1988) Pragmatics of measuring recognition memory: applications to dementia and amnesia. J. Exp. Psychol. Gen.,117 , 34-50.[Web of Science][Medline]
Thorngate, J.H. (1997) The physiology of
human sensory response to wine. Am. J. Enol. Viticulture,48
, 271-279.
Wixted, J.T. (1992) Subjective memorability and the mirror effect. J. Exp. Psychol. Learn. Mem. Cogn.,18 , 681-690.[Web of Science][Medline]
Accepted July 30, 2002
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  L. Perrin, R. Symoneaux, I. Maitre, C. Asselin, F. Jourjon, and J. Pages Comparison of Conventional Profiling by a Trained Tasting Panel and Free Profiling by Wine Professionals Am. J. Enol. Vitic., December 1, 2007; 58(4): 508 - 517. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||