Chem. Senses 27: 307-317,
2002
© Oxford University Press 2002
Selection of Odorants for Memory Tests on the Basis of Familiarity, Perceived Complexity, Pleasantness, Similarity and Identification
INRA, Unité Mixte de Recherches sur les Arômes, 17 Rue Sully, BP 86510, 21065 Dijon Cedex, France 1 Jan Van Scorelstraat 55, 3583 CK Utrecht, The Netherlands
Correspondence to be sent to: Sylvie Issanchou, Unité Mixte de Recherches sur les Arômes, INRA, 17 Rue Sully, BP 86510, 21065 Dijon Cedex, France. e-mail: issan{at}arome.dijon.inra.fr
 |
Abstract |
|---|
 Top Abstract IntroductionStep 1: selection of...Step 2: division of...Step 3: validation of...ConclusionReferences |
|---|
In a procedure for the selection of two equivalent sets of familiar and two equivalent sets of unfamiliar odours for use in odour memory studies, 24 naïve subjects were first asked to rate the familiarity, perceived complexity and pleasantness of 54 a priori unfamiliar odours and 57 a priori familiar odours and to identify the latter. After selection of the 40 most familiar and the 40 least familiar odours, the subjects sorted each of these two sets into groups of similar odours. Their results were analysed by multidimensional scaling and cluster analysis and each set was divided into two recognition sets that had the same degree of similarity between target and distractor odours and that had similar values of familiarity, pleasantness, perceived complexity (familiar and unfamiliar sets) and identifiability (familiar sets). Finally, recognition tasks were performed in order to check the equivalence in memory performance of both the two familiar and the two unfamiliar recognition sets.
|
Introduction |
|---|
|
TopAbstractIntroductionStep 1: selection of...Step 2: division of...Step 3: validation of...ConclusionReferences |
|---|
Recognition tasks are commonly used for studying odour memory. In a first acquisition stage subjects smell target odours that they then have to recognize among distractor odours in a second recognition stage. The present paper describes an odour selection procedure for recognition experiments. This procedure was used for selecting two equivalent sets of familiar odours and two equivalent sets of unfamiliar odours in order to study the impact of different learning and/or retrieval procedures on recognition performance using within-subject designs in later experiments (Sulmont et al., 1998
; Sulmont and Senouci,
1999).
Several odour characteristics may influence recognition performance. An
earlier study (Rabin and Cain,
1984) showed that odours that were identified more accurately
during the acquisition stage were also more accurately recognized during the
recognition stage. That study also found a correlation between odour
familiarity and recognition performance. No relationship was found between
pleasantness and recognition performance
(Lawless and Cain, 1975), but
many studies have observed a strong relationship between pleasantness and
familiarity (Jellinek and Köster,
1979,
1983;
Issanchou et al.,
1987; Porcherot,
1995; Ayabe-Kanamura et
al., 1998). To our knowledge, the impact of odour complexity
on recognition performance has never been studied, but the fact has been
pointed out that the perceived complexity of a stimulus partly determines the
arousal potential of this stimulus
(Berlyne, 1960). Thus, the
perceived complexity of an odour might influence the strength of the mnesic
trace of this odour by attracting the attention of the subject towards it.
Finally, the similarity of odours has been shown to have a strong influence on
recognition performance (Engen and Ross,
1973; Lawless and Cain,
1975; Jones et al.,
1978). These last authors found a correlation between judgements
of similarity and the confusions in a recognition task. The more the
distractor odours were similar to the target odours, the more they were
confused during the recognition stage.
These different odour characteristics (idendifiability, familiarity, pleasantness, perceived complexity and similarity) were taken into account in the selection of two equivalent recognition sets of familiar odours and two equivalent recognition sets of unfamiliar odours. The selection procedure consisted of three steps. The aim of the first step was to select 40 odours rated as familiar and 40 odours rated as unfamiliar by subjects with no previous experience in sensory analysis or odour research. Odour pleasantness, odour perceived complexity and familiar odour identifiability were also measured during this step. The aim of the second step was to divide each set of 40 odours into two recognition sets that had the same degree of similarity between the target and distractor odours and that had similar values of familiarity, pleasantness, perceived complexity (familiar and unfamiliar sets) and identifiability (familiar sets). The aim of the third step was to check whether each of the two familiar sets on one hand and each of the two unfamiliar sets on the other gave rise to a same range of recognition performance.
|
Step 1: selection of 40 familiar odours and 40 unfamiliar odours |
|---|
|
TopAbstractIntroductionStep 1: selection of...Step 2: division of...Step 3: validation of...ConclusionReferences |
|---|
Materials and methods
Usually, odour familiarity is measured by a scale ranging from `not
familiar' to `very familiar' (Rabin and
Cain, 1984; Issanchou et
al., 1987) and odour perceived complexity is measured by a
scale ranging from `few odour notes' to `many odour notes' (Jellinek and
Köster, 1979,
1983;
Issanchou et al.,
1987). However, the fact that familiarity and, in particular,
perceived complexity could include several subdimensions was pointed out in an
earlier study (Porcherot,
1995). That author proposed several scales for measuring each of
these dimensions and their questionnaire was adopted in the present experiment
and its validity will be discussed.
Since recall of odour names is very difficult
(Lawless and Engen, 1977), the
identifiability of familiar odours was evaluated by a multiple-choice
procedure.
Subjects
Twenty-four subjects with no previous experience in sensory analysis and no
self-reported problems in their sense of smell were recruited. They were
balanced for gender and age: young subjects (four females and four males of
mean age 23 years and range 20-25 years), middle-aged subjects (four females
and five males of mean age 39 years and range 35-44 years) and elderly
subjects (four females and three males of mean age 68 years and range 64-73
years). The subjects were paid for their participation.
Odorants
Two approximately equally sized sets of odorants were selected: a set of 54
odorants supposed to have an unfamiliar odour, i.e. an odour not present in
the subjects' daily environment and a set of 57 odorants supposed to have a
familiar odour, i.e. an odour often smelled in the subjects' everyday life.
This latter set consisted of essential oils, food and non-food flavours, food
and non-food products and monomolecular chemicals
(Table 1), whereas the
unfamiliar set contained single molecules and mixtures of a few monomolecular
chemicals (Table 2).
|
|
Each odorant was diluted in order to obtain a solution similar in odour intensity to a solution of butan-1-ol at 0.20 ml/l. This concentration was chosen because it has a weak intensity and, thus, limits olfactory adaptation during tests. The odorants were diluted in mineral oil with the exception of alcohol vinegar, bleach, Viandox® and caramel, which were diluted in distilled water. The odorant concentrations are presented in Tables 1 and 2.
The odorants were prepared 1 week before the first session. The odorous solutions were poured into 60 ml brown glass flasks (12.5 ml per flask). Each flask contained a 110 x 55 cm piece of absorbent tissue (absorbent sheet type P110, OSI, France) in order to increase the exchange area between the solution and the flask air. The flasks were stored at 5°C except flasks of Viandox® and surimi flavour, which were stored at -10°C in order to avoid bacterial contamination. One hour before sessions they were placed at 20.5 ± 0.5°C. A three-digit random number coded each flask, which was different in each session.
The subjects were instructed to open the jar and smell the odorant by breathing normally, without sniffing. The subjects were allowed to smell the odorants as many times as they needed while completing the tests. A break of 30 s was imposed on the subjects after each odorant in order to prevent olfactory adaptation. If a subject perceived no odour when smelling an odorant, he or she did not complete the test for this odorant.
Measurement of familiarity, perceived complexity and
pleasantness
The subjects were asked to rate the familiarity, perceived complexity and
pleasantness of each odour during three sessions of 
80 min duration.
After smelling an odorant, the subjects answered questions on eight items:
four on familiarity, three on perceived complexity and one on pleasantness.
They gave their answers on 12 cm linear scales labelled at each end of the
scale. The familiarity items [familiar (this odour is unfamiliar/very
familiar), oftenmet (do you smell this odour rarely/very often), known (this
odour is composed of unknown product(s)/well-known product(s)) and memories
(this odour recalls you few memories/many memories)] and complexity items
[complex (this odour is simple/complex), describe (this odour seems to you
easy to describe/difficult to describe) and notes (this odour is composed of
few odour notes/many odour notes)] were chosen from the questionnaire
mentioned above (Porcherot,
1995). The familiarity item `this stimulus is unknown/very known'
used in the earlier study (Porcherot,
1995) was changed to `this odour is composed of unknown
product(s)/well-known products(s)'. Indeed, some unfamiliar odours could be
perceived as unusual mixtures of familiar odours (for example, a mixture of
garlic and strawberry). No item definition was provided to the subjects, who
were encouraged to answer according to their own interpretation. Odour
pleasantness was measured by the question `Do you like this odour?' with
answers `not at all/a lot' (liking).
The presentation order of the items was the same for all odorants and all subjects. In order to limit the number of pages of the questionnaire, two items appeared on each page, but items measuring the same concept were never presented consecutively. The presentation order of the 111 odorants followed a Williams Latin square design. Thirty-one one odorants were evaluated during the first session and 40 during the second and third sessions.
Identification of odours
The subjects were asked to identify the odour of the 57 odorants supposed
to be familiar during a session of 90 min.
After smelling an odorant, the subjects had to find the name of its odour amongst a list of 68 descriptors sorted in alphabetical order. This list contained 60 expected labels, i.e. labels that were supposed to be the names of the familiar odours (Table 1) and eight distractor labels. With regard to the chemicals the expected labels were the names usually associated with their odour (for example, mushroom for oct-1-en-3-ol), with regard to the essential oils and flavours the expected labels were the names given by the manufacturer and with regard to the natural products the expected labels were the names of the product (bleach, olive oil, etc.). Three labels were added to the initial list of 57 expected labels because the odour of three odorants could be identified by two names. According to our experience essential oil of buchu could be perceived either as `odour of cat's urine' or as `odour of blackcurrant', orange flavour could be perceived either as `odour of orange' or as `odour of grapefruit' and bitter almond flavour could be perceived either as `odour of bitter almond' or as `odour of paste' (a particular adhesive colle blanche in French). The distractor labels were coconut, exotic fruit, lilac, musty, onion, polish, soap and tomato. The presentation order of the odorants followed a Williams Latin square design. No identification was asked for the 54 odorants supposed to have an unfamiliar odour as no expected label was available for their odours.
Experimental conditions
The tests were conducted in a sensory room equipped according to a known
standard (AFNOR, 1987). A red
light was used during odorant evaluation in order to mask potential colour
differences. The room temperature was 20.5 ± 0.5°C. The subjects
answered questionnaires on the FIZZ data acquisition system (FIZZ software,
Biosystèmes, Couternon, France). The three sessions for measuring
familiarity, perceived complexity and pleasantness and the identification
session took place on separate days.
Data analysis
All statistical analyses were conducted using SAS/STAT®
(SAS, 1989).
Responses on the 12 cm linear scales for measurement of familiarity, perceived complexity and pleasantness were converted into scores varying from 0 (left of the scale) to 100 (right of the scale). Individual scores were averaged per item and per odorant.
When a subject associated the odour of an odorant with its expected label, the answer was scored as correct. The frequency of correct identification was determined for each odorant.
Results and discussion
Measurement of familiarity, perceived complexity and
pleasantness
Selection of relevant indices of familiarity and perceived
complexity. Linear regressions performed between the familiarity items
(familiar, known, oftenmet and memories) showed a strong correlation between
those items (R2 > 0.85 and P < 0.001). One
could argue that, due to the fixed order of item presentation, a response to
one item may have systematically influenced the response to a later one.
Namely, if one claimed to experience an odour quite often (item oftenmet), one
might be inclined to claim that many memories were associated with it (item
memories). However, a high correlation between familiarity items by using
different random orders of items was also found in another study
(Porcherot, 1995). A global
index of familiarity (familiarity) was calculated by averaging the scores of
the items oftenmet, memories, known and familiar.
A strong negative relationship was found between the item describe and the index familiarity, indicating that an odour perceived as easy to describe was also perceived as familiar (R2 = 0.91 and P < 0.001). The subjects may have understood the question `this odour seems to you easy to describe/difficult to describe' in the sense of `do you know the name of this odour?' In other words, this question would measure a familiarity dimension (the subjects' feeling of knowing the odour's name) rather than a psychological complexity dimension (the subjects' feeling of being puzzled when smelling the odour). These results led us to remove the item describe in further odour selection.
No relationship was found between the items complex and notes, indicating
that these items measure different dimensions of perceived complexity
(R2 = 0.03 and P > 0.05). According to an
earlier study (Berlyne, 1960)
complexity increases with the number of elements and the dissimilarity between
perceived elements and inversely varies `with the degree to which several
elements are responded to as a unit', i.e. with a subject's ability to arrange
the perceived elements in a meaningful unit. Our results suggest that the item
notes reflects the number of perceived notes in an odour, while the item
complex reflects the degree to which a subject is able to interpret an odour
meaningfully. On the basis of these results, the items complex and notes were
taken into account separately in further odour selection.
In summary, the familiarity of an odour appears to be unidimensional, while
the perceived complexity of an odour seems to include at least two dimensions:
the number of perceived odour notes and a subject's ability to interpret an
odour meaningfully. Since no item definition was provided to the subjects, one
could claim that the words `simple' and `complex' (item complex) and the
notion of `odour notes' (item notes) were too ambiguous to warrant a uniform
interpretation by the subjects. It was recently proposed that the perceived
complexity of an odour should be measured by the question `this odour is
ordinary/surprising' (Lévy,
1998). That author found a strong correlation between this
question and the question `this odour is simple/elaborated', which is close to
our complex question. However, more work is needed in order to specify further
whether the item complex or the question proposed above
(Lévy, 1998) leads to
the most reliable results and to determine whether naïve subjects are
able to rate the number of perceived odour notes.
Relationships between familiarity, perceived complexity and
pleasantness. A strong correlation was found between the indices
familiarity and liking indicating that odours rated as familiar were liked,
while odours rated as unfamiliar were disliked
(Figure 1). In fact, the link
between familiarity and pleasantness judgement of non-odorous stimuli
(Zajonc, 1968) as well as
odorous stimuli (Jellinek and Köster,
1979,
1983;
Issanchou et al.,
1987; Porcherot,
1995; Ayabe-Kanamura et
al., 1998) is well documented. More interestingly, the
familiarity index is more strongly correlated with the complex index than with
the notes index (Figure 1). The
more an odour was rated as familiar the more it was rated as simple, while the
number of perceived odour notes remained relatively independent from odour
familiarity. An earlier study found a similar result
(Porcherot, 1995). As
mentioned in the previous paragraph, the item notes seems to reflect the
number of perceived notes in an odour and the item complex seems to reflect a
subject's ability to meaningfully interpret an odour. According to other work
a subject's ability to arrange perceived stimuli in a meaningful unit depends
heavily on the subject's knowledge about relationships between these stimuli
(Berlyne, 1960). Hence, it is
not surprising to find a stronger relationship between the indices familiarity
and complex than between the indices familiarity and notes.
|
With regard to the relationship between familiarity and pleasantness, some odorants appeared to be quite dissociated from the others (the encircled odorants in Figure 1). The odours of bleach, trimethylamine, isovaleric acid and blue cheese flavour were disliked and at the same time were rated as familiar. These odorants were not selected in the final sets because their odours might have been perceived as very different from the other familiar odours and, thus, might have been easier to memorize and/or recognize during future recognition tests. Such outliers were not found with regard to the relationship between familiarity and perceived complexity indices.
Identification of odours
The frequency of correct identification varied from 0.92 for bleach to 0.0
for chocolate flavour (mean = 0.43 ± 0.06). Since nobody associated the
odour of chocolate flavour with the label chocolate, but 21 and 17% of the
subjects respectively associated this odour with the labels caramel and
vanilla, this odorant was eliminated. Indeed, the odours of caramel (liquid
caramel) and of vanilla (vanillin) were already included in the familiar
set.
Specific hyposmia
Two odorants, geraniol and 3,4,5,6,6-pentamethylhept-3-en-2-one, were
eliminated because approximately one-third of the subjects seemed to have a
specific hyposmia for these odorants, namely 24 and 29% of the subjects did
not perceive geraniol and 3,4,5,6,6-pentamethyl hept-3-en-2-one respectively.
Geraniol has been previously reported to exhibit specific hyposmia in humans
(Amoore, 1977), but so far no
data on specific hyposmia for 3,4,5,6,6-pentamethyl-hept-3-en-2-one have been
found in the literature. The percentage of non-perception was below 20% for
each of the other odorants.
Selection of 40 familiar odours and 40 unfamiliar odours
After elimination of the odorants mentioned above, the 40 odorants with the
highest familiarity scores and the 40 odorants with the lowest familiarity
scores were selected as preliminary sets of familiar and unfamiliar odours
respectively. Surprisingly, the familiar set contained some odorants that had
been chosen by the experimenter as supposedly unfamiliar odours, namely
buccoverte forte base, piperonal, 
-ionone, petylyn, galbanum,
4-methyl-acetophenone, verdox and calone. A panel of 10 subjects trained in
sensory profiling was asked to describe their odours in order to check their
identifiability. Five of the odorants were associated with one or two
consensual labels by the panel: rose or lily of the valley for buccoverte
forte base, vanilla or almond for piperonal, violet for -ionone [this
label agrees with an earlier description (Fenaroli, 1971)], orange blossom or
marshmallow for petylyn and bitter almond or toilet cleaner for
4-methylacetophenone. No consensual label was found for galbanum, verdox and
calone. Consequently, they were eliminated from the familiar set and replaced
by the three odorants with the highest familiarity scores not yet selected.
The unfamiliar set contained some odorants that had been chosen by the
experimenter as supposedly familiar odours, namely musk fragrance, methional,
styrene, ethanol, naphthalene, octanoic acid, essential oil of pepper and
essential oil of incense. Nevertheless, some subjects were able to identify
their odours by their correct label. Consequently, these odorants were
eliminated from the unfamiliar set and replaced by the eight odorants with the
lowest familiarity scores not yet selected.
According to a Student's t-test, the familiarity scores of the
odours of the familiar set (mean = 61.1 and = 8.5) were significantly
higher than the familiarity scores of the odours of the unfamiliar set (mean =
35.3 and = 5.2) (t = 16.4 and P < 0.001).
Nevertheless, the difference between the minimal familiarity score of the
familiar set (46.7) and the maximal familiarity score of the unfamiliar set
(43.9) was very small. The familiarity scores were distributed along a
continuum rather than divided into two well-defined classes.
|
Step 2: division of each set of 40 odorants into two recognition sets |
|---|
|
TopAbstractIntroductionStep 1: selection of...Step 2: division of...Step 3: validation of...ConclusionReferences |
|---|
Materials and methods
The aim of this step was to divide each set of 40 odorants into two
recognition sets that had the same degree of similarity between the target and
distractor odours and that had similar values of familiarity, pleasantness,
perceived complexity (familiar and unfamiliar sets) and identifiability
(familiar sets). Odour similarities were measured by a sorting task for each
set (Lawless, 1989;
MacRae et al.,
1990).
Odorants
During the previous step the concentrations of the odorous solutions were
chosen by the experimenter in order to obtain an approximately equal odour
intensity for all odorants. However, a pre-test was carried out to procure
iso-intense odorous solutions in order to reduce the influence of odour
intensity on the sorting criteria and on memorization in later recognition
tests. Ten subjects (nine females and one male with range 2-55 years) who were
all experienced in sensory profiling were recruited. A range of five
increasing concentrations of butan-1-ol (0.05 ml/l, 0.10 ml/l, 0.20 ml/l, 0.40
ml/l and 0.80 ml/l) was used as a reference. The subjects rated the odour
intensity of each odorous solution on a 13-point scale, points 3, 5, 7, 9 and
11 of which respectively corresponded to the butan-1-ol solutions at 0.05,
0.10, 0.20, 0.40 and 0.80 ml/l. Odorous solutions rated as less intense than
the butan-1-ol solution at 0.20 ml/l and odorous solutions rated as more
intense than the butan-1-ol solution at 0.40 ml/l were re-prepared at higher
or lower concentrations respectively. Three successive trials (rating and
concentration adjustment) were necessary in order to reach these criteria for
all solutions. The final concentrations are given in Tables
1 and
2. In addition to this
pre-test, the odorous solutions' purity was checked by gas chromatography
olfactometry (sniffing odours at the sniffing port of a gas
chromatograph).
The odorants were diluted, stored and presented using the same procedure as that used during step 1.
Sorting task
The sorting task was performed by the 24 subjects recruited for step 1. Two
sessions, of 75 min duration took place on separate days, 1 month after
step 1. Half of the subjects received the familiar set during the first
session and the unfamiliar set during the second session and the other half
proceeded in the reverse order. The subjects sorted the 40 odorants of a set
into groups of samples having a similar odour during each session. They were
allowed to form as many groups as they wanted, but at least two groups. To
help them to perform this task, they could take notes about their sorting
criteria. When the subjects had finished they were encouraged to review their
groups and to check whether they were satisfied with the grouping they had
made. The experimental conditions were the same as those used during step
1.
Data analysis
A similarity matrix was prepared for each set of 40 odorants by summing the
number of times each pair of odorants was sorted into the same group over all
subjects. The greater the number in a cell of the matrix was, the greater the
presumed subjective odour similarity of the odorants intersecting at that
location. The matrices were analysed by the non-metric multidimensional
scaling (MDS) procedure of the SAS. The four-dimensional representation (fit
of spatial representation = 0.91 and stress = 0.12) and the six-dimensional
representation (fit of spatial representation = 0.82 and stress = 0.12) were
selected for the familiar and unfamiliar sets respectively. A new matrix was
prepared for each set with the 40 odorants in rows and their coordinates on
the selected MDS dimensions in columns. These new matrices were submitted to a
hierarchical clustering (the CLUSTER procedure of the SAS). Six and five
clusters were chosen for the familiar and unfamiliar sets respectively.
Results and discussion
After step 2, four odorants of each set had to be removed for different reasons and, as a result, the final recognition sets only contained nine target and nine distractor odorants instead of the 10 target and 10 distractor odorants initially planned. Terpinyl acetate, linalyl acetate, nonyl acetate and 4-methoxy-1-methylbenzene were removed from the unfamiliar set, because gas chromatography olfactometry revealed that these odorants were impure. Essential oil of buchu was removed from the familiar set because 62% of the subjects did not perceive its odour during the sorting tasks (this was probably due to an error of dilution). As was mentioned at the end of step 1, the difference between the minimal familiarity scores of the familiar set and the maximal familiarity scores of the unfamiliar odour set was very small. In order to increase this difference between the sets, odorants of the familiar set which had obtained the lowest score of familiarity, i.e. (2E,6Z)nona-2,6-dienal, (Z)hex-1-en-3-ol and ethylfenchol, were removed.
The clusters obtained from the familiar and the unfamiliar sets are presented in Figure 2a and b respectively. The odorants of each cluster were divided over four subsets for each set, with two subsets of nine target odorants and two subsets of nine distractor odorants. For example, the four odorants of the second familiar cluster were divided over the four subsets at the rate of one odorant per subset. Table 3 presents the final recognition sets, i.e. the familiar recognition sets Fa and Fb and the unfamiliar sets Uc and Ud. Each recognition set comprised one subset of nine target odorants and one subset of nine distractor odorants. Table 4 shows the means of the familiarity, liking, complex and notes scores for each recognition set and the means of the frequency of correct identifications for each familiar recognition set. A Student's t-test performed by type of odour yielded no significant difference between either the two familiar recognition sets or the two unfamiliar recognition sets (P > 0.05).
|
|
|
|
|
Step 3: validation of the recognition sets |
|---|
|
TopAbstractIntroductionStep 1: selection of...Step 2: division of...Step 3: validation of...ConclusionReferences |
|---|
Materials and methods
Recognition tasks were performed in order to check whether sets Fa and Fb on the one hand and sets Uc and Ud on the other led to a same range of performance level.
Subjects
Twenty subjects who had not participated in the previous tasks were
recruited. They were divided into two groups balanced for gender and age:
group familiar (four females and six males with range 24-56 years) and group
unfamiliar (four females and six males with range 25-51 years).
Recognition task
The subjects of group familiar performed two recognition tasks, one with
set Fa and one with set Fb. The subjects of group unfamiliar performed two
recognition tasks, one with set Uc and one with set Ud. The presentation order
of the sets within each group was counterbalanced over subjects. A break of 1
week was taken between each task.
Each recognition task consisted of two sessions of 30 min duration, 7
days apart. The subjects received the nine target odorants of a set during the
first session (acquisition stage). The subjects were asked to answer to the
questions familiar, complex, notes and liking (orientation task) for each
odorant. The subjects were told to memorize the odours. The subjects then
received 18 odorants in a randomized order during the second session
(recognition stage). Half of the stimuli were the same as in the learning
phase (target odorants) and half were new (distractor odorants). After
smelling an odorant, the subjects had to say whether they had smelt its odour
during the first session or not (yes/no).
As the recognition sets were to be used later for making a between-subject comparison on recognition performance, both the presentation order of the nine target odorants during the acquisition stage and the presentation order of the 18 target plus distractor odorants during the recognition stage were the same for all subjects.
The odorants were diluted, stored and presented using the same procedure as that used during step 1. The experimental conditions were the same as those used during step 1.
Data analysis
Recognition performances were determined according to signal detection
theory (Banks, 1970), i.e. by
computing the frequency of hits (target odours correctly recognized), the
frequency of false alarms (distractor odours incorrectly recognized) and
d' scores per subject and per recognition set. The index
d' represents the index of detectability and it is obtained by
the formula d' = Zhits —
Zfalse_alarms, where Zhits and
Zfalse_alarm are the standard frequencies of hits and
false alarms under the normal curve (Engen, 1971).
The difference in the d' scores, frequency of hits and frequency of false alarms between the two recognition sets was assessed by a related-sample Student's t-test for each group (familiar or unfamiliar).
Results and discussion
According to the results of a related-sample Student's t-test, no difference was observed between the familiar recognition sets or between the unfamiliar recognition sets, except for the frequency of false alarms of the familiar sets (Table 5). The frequency of false alarms for the set Fb was significantly higher than the frequency of false alarms for the set Fa.
|
In order to determine the reason for this, the frequency of false alarms was calculated for each familiar distractor odorant. The three odorants that were associated with the highest frequency of false alarms (50%) all belonged to set Fb: essential oil of eucalyptus, coffee flavour and apple flavour. The results of the identification task performed during step 1 showed that the odour of essential oil of eucalyptus (distractor odorant Fb) was identified 11 times as odour of mouthwash, which was a target odour of set Fb and that the odour of apple flavour (distractor odorant Fb) was identified four times as odour of strawberry, which was a target odour of set Fb. In order to reduce the frequency of false alarms of set Fb, apple flavour was exchanged with peach flavour (distractor odorant Fa) and essential oil of eucalyptus was exchanged with lavender fragrance (distractor odorant Fa). This last exchange should have increased the frequency of false alarms of set Fa. Indeed, according to the identification results, the odour of eucalyptus was identified seven times as odour of pine, a target odour of set Fa. It was hoped that these exchanges would harmonize the frequencies of false alarms between sets Fa and Fb. Unfortunately, no further check of this was possible under the circumstances.
|
Conclusion |
|---|
|
TopAbstractIntroductionStep 1: selection of...Step 2: division of...Step 3: validation of...ConclusionReferences |
|---|
Several odour characteristics (familiarity, pleasantness, complexity, identifiability and similarity) were taken into account in order to select recognition sets that had the same level of difficulty. The results of the third step pointed out another factor that might be taken into account in this procedure, namely the matrix of identification errors. Notwithstanding the limitations in the applicability of the resulting odour recognition sets outside the French culture, the selection procedure itself could be used anywhere. However, more work is needed in order to determine which factors are the best predictors of the range of recognition performances.
|
Acknowledgments |
|---|
We thank Ton Teerling (International Flavors & Fragrances, The Netherlands) for the supply of unfamiliar odorants and Annie Hay and Christophe Martin for their help during the experiments.
|
References |
|---|
|
TopAbstractIntroductionStep 1: selection of...Step 2: division of...Step 3: validation of...ConclusionReferences |
|---|
AFNOR (ed.) (1987) Directives Générales pour L'implantation de Locaux Destinés à L'analyse Sensorielle. Association Française de Normalisation, Paris.
Amoore, J.E. (1977) Specific anosmia and
the concept of primary odors. Chem. Senses,2
, 267-281.
Ayabe-Kanamura, S., Schicker, I., Laska, M., Hudson, R., Distel,
H., Kobayakawa, T. and Saito, S. (1998)
Differences in perception of everyday odors: a Japanese—German
cross-cultural study. Chem. Senses,23
, 31-38.
Banks, W. (1970) Signal detection theory and human memory. Psychol. Bull.,2 , 81-99.
Berlyne, D.E. (ed.) (1960) Conflict, Arousal and Curiosity. McGraw-Hill, New York.
Engen, T. and Ross, B.M. (1973) Long-term memory of odors with and without verbal descriptions.J. Exp. Psychol. , 100,221 -227.[Web of Science][Medline]
Issanchou, S., Köster, E.P. and Teerling, A. (1987) Caractéristiques de l'arôme et préférence. Sci. Aliments,7 , 53-58.
Jellinek, J.S. and Köster, E.P. (1979) Perceived fragrance complexity and its relation to familiarity and pleasantness. J. Soc. Cosm. Chem.,30 , 253-262.
Jellinek, J.S. and Köster, E.P. (1983) Perceived fragrance complexity and its relationship to familiarity and pleasantness II. J. Soc. Cosm. Chem.,34 , 83-97.
Jones, F.N., Roberts, K. and Holman, E.W. (1978) Similarity judgements and recognition memory for some common spices. Percept. Psychophys.,24 , 2-6.[Web of Science][Medline]
Lawless, H.T. (1989) Exploration of
fragrance categories and ambiguous odors using multidimensional scaling and
cluster analysis. Chem. Senses, 14,349
-360.
Lawless, H.T. and Cain, W.S. (1975)
Recognition memory for odors. Chem. Senses,1
, 331-337.
Lawless, H. and Engen, T. (1977) Associations to odors: interference, mnemonics, and verbal labeling.J. Exp. Psychol. Human Learn. Mem. , 3,52 -59.
Lévy, C. (1998) La dynamique des préférences: application aux boissons. Thesis, University of Burgundy, Dijon, France.
MacRae, A., Howgate, P. and Geelhoed, C.
(1990) Assessing the similarity of odours by sorting and
triadic comparison. Chem. Senses,15
, 691-699.
Porcherot, C. (1995) Contribution à la mesure de la familiarité et de la complexité d'arômes alimentaires: pertinence de ces concepts pour expliquer les préférences des consommateurs. Thesis, University of Burgundy, Dijon, France.
Rabin, M.D. and Cain, W.S. (1984) Odor recognition: familiarity, identifiability and encoding consistency. J. Exp. Psychol. Learn. Mem. Cognit.,10 , 316-325.[Web of Science][Medline]
SAS (ed.) (1989) SAS/STAT® User's Guide. SAS Institute, Cary, NC.
Sulmont, C. and Senouci, K. (1999) Effect of an incidental learning versus an intentional learning on performance at a recognition test. In European Symposium on Olfaction and Cognition, 10-12 June, Lyon, France.
Sulmont, C., Issanchou, S. and Köster, E.P. (1998) Effect of nature of learning and nature of stimuli on performance at a recognition test. Chem. Senses,24 , 116.
Weast, R.C. and Tune, G.L. (ed.) (1971)Fenaroli's Handbook of Flavor Ingredients . The Chemical Rubber Co., Cleveland, OH.
Zajonc, R. (1968) Attitudinal effects of mere exposure. J. Person. Soc. Psychol.,9 , 1-27.[Web of Science][Medline]
Accepted December 12, 2001
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  S. Delplanque, D. Grandjean, C. Chrea, L. Aymard, I. Cayeux, B. Le Calve, M. I. Velazco, K. R. Scherer, and D. Sander Emotional Processing of Odors: Evidence for a Nonlinear Relation between Pleasantness and Familiarity Evaluations Chem Senses, June 1, 2008; 33(5): 469 - 479. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
V. Chen and B. P. Halpern Retronasal but Not Oral-Cavity-Only Identification of "Purely Olfactory" Odorants Chem Senses, February 1, 2008; 33(2): 107 - 118. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. Plailly, B. Tillmann, and J.-P. Royet The Feeling of Familiarity of Music and Odors: The Same Neural Signature? Cereb Cortex, November 1, 2007; 17(11): 2650 - 2658. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
C. Sulmont-Rosse, S. Issanchou, and E.P. Koster Odor Naming Methodology: Correct Identification with Multiple-choice versus Repeatable Identification in a Free Task Chem Senses, January 1, 2005; 30(1): 23 - 27. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||